Antidepressants for low back pain and spine‐related leg pain

Michael C Ferraro; Donna M Urquhart; Giovanni E Ferreira; Michael A Wewege; Christina Abdel Shaheed; Adrian C Traeger; Jan L Hoving; Eric J Visser; James H McAuley; Aidan G Cashin

doi:10.1002/14651858.CD001703.pub4

. 2025 Mar 10;2025(3):CD001703. doi: 10.1002/14651858.CD001703.pub4

Antidepressants for low back pain and spine‐related leg pain

Michael C Ferraro ^1,^2,^✉, Donna M Urquhart ³, Giovanni E Ferreira ^4,⁵, Michael A Wewege ^1,⁶, Christina Abdel Shaheed ⁷, Adrian C Traeger ^4,⁵, Jan L Hoving ⁸, Eric J Visser ⁹, James H McAuley ^1,², Aidan G Cashin ^1,²

Editor: Cochrane Central Editorial Service

PMCID: PMC11890917 PMID: 40058767

Abstract

Background

Antidepressants are commonly used to treat low back pain and spine‐related leg pain. However, their benefits and harms are uncertain.

This is an update of a 2008 Cochrane review of antidepressants for non‐specific low back pain.

Objectives

To assess the benefits and harms of antidepressants for non‐specific low back pain and spine‐related leg pain.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, Embase, ClinicalTrials.gov, World Health Organization International Clinical Trials Registry Platform, and EU Clinical Trials Register from inception to 14 November 2024.

Selection criteria

We included randomised controlled trials that compared antidepressants with placebo, usual care, or no treatment/waiting list. Participants were 18 years of age or older with non‐specific low back pain or spine‐related leg pain of any duration. We excluded participants with low back pain due to spinal fracture, inflammatory disease, aortic dissection, malignancy, or infection. Primary outcomes were pain intensity and disability, measured at short‐term follow‐up (> 4 to 14 weeks post‐randomisation), and total adverse events. Secondary outcomes were serious adverse events, withdrawals due to adverse events, depressive symptoms, and health‐related quality of life.

Data collection and analysis

Two review authors independently screened records to determine study inclusion, extracted data, and evaluated risk of bias using RoB 1 tool. Where possible, we conducted meta‐analyses. We used GRADE to assess the certainty of evidence.

Main results

We included 26 randomised controlled trials. Eighteen studies included 2535 participants with non‐specific low back pain, seven studies included 329 participants with spine‐related leg pain, and one study included 68 participants with either condition. Most participants had pain lasting more than three months, with a mean duration between 18 months and 20 years. Mean ages ranged from 27 to 59 years. Studies evaluated serotonin and norepinephrine reuptake inhibitors (SNRIs; eight studies), selective serotonin reuptake inhibitors (SSRIs; two studies), tricyclic antidepressants (TCAs; 14 studies), tetracyclic antidepressants (TeCAs; two studies), or 'other antidepressants' (two studies). All studies were placebo‐controlled. Outcomes were measured at short‐term follow‐up in 73% of studies. All included studies had at least one domain judged at high risk of bias, with 69% at high risk of attrition bias.

Non‐specific low back pain (benefits)

Moderate‐certainty evidence demonstrated that SNRIs probably have a small effect on pain intensity (mean difference (MD) (0 to 100 scale) ‐5.25, 95% confidence interval (CI) ‐7.17 to ‐3.34; I² = 0; 4 studies, 1415 participants) and a trivial effect on disability (MD (0 to 24 scale) ‐0.91, 95% CI ‐1.30 to ‐0.51; I² = 0; 4 studies, 1348 participants) at short‐term follow‐up.

Low‐certainty evidence showed that SSRIs may have little to no effect on pain intensity (MD 1.20, 95% CI ‐4.90 to 7.30; I² = 0; 3 studies, 199 participants) and disability (MD ‐2.20 (0 to 100 scale), 95% CI ‐8.11 to 3.71; 1 study, 92 participants) at short‐term follow‐up.

Moderate‐certainty evidence demonstrated that TCAs probably have little to no effect on pain intensity (MD ‐2.00, 95% CI ‐7.25 to 3.24; I² = 31%; 4 studies, 417 participants), but probably have a small effect on disability (MD (0 to 24 scale) ‐1.76, 95% CI ‐2.70 to ‐0.82; I² = 0; 3 studies, 330 participants) at short‐term follow‐up.

The effects of TeCAs (MD ‐4.50, 95% CI ‐17.59 to 8.59; 1 study, 52 participants) and other antidepressants (MD ‐5.40, 95% CI ‐23.08 to 12.28; 1 study, 39 participants) on pain intensity at short‐term follow‐up are unclear (very low‐certainty evidence). No studies assessed the effects of TeCAs or other antidepressants on disability.

Spine‐related leg pain (benefits)

The effects of SNRIs on pain intensity (MD ‐46.10, 95% CI ‐89.29 to ‐2.91; 1 study, 11 participants) and disability (MD (0 to 100 scale) ‐4.40, 95% CI ‐20.25 to 11.45; 1 study, 11 participants) at short‐term follow‐up are very uncertain (very low‐certainty evidence).

Low‐certainty evidence showed TCAs may have a large effect on pain intensity at short‐term follow‐up (MD ‐23.00, 95% CI ‐32.12 to ‐13.88; 1 study, 60 participants), and a moderate effect on disability (MD (0 to 100 scale) ‐13.00, 95% CI ‐19.42 to ‐6.58; 1 study, 60 participants).

There were no studies that assessed the effects of SSRIs, TeCAs, or other antidepressants in people with spine‐related leg pain.

Non‐specific low back pain and spine‐related leg pain (harms)

Moderate‐certainty evidence demonstrated that SNRIs probably increase the risk of adverse events (risk ratio (RR) 1.17, 95% CI 1.07 to 1.27; I² = 0%; 5 studies, 1510 participants), but it is unclear whether they increase the risk of serious adverse events (Peto odds ratio (OR) 1.75, 95% CI 0.79 to 3.89; 5 studies, 1510 participants; very low‐certainty evidence).

It is unclear whether TCAs increase the risk of adverse events (RR 1.76, 95% CI 0.79 to 3.90; 7 studies, 474 participants; low‐certainty evidence) or serious adverse events (Peto OR 6.64, 95% CI 0.41 to 106.72; I² = 0%; 1 study, 142 participants; very low‐certainty evidence).

It is unclear whether SSRIs (RR 1.83, 95% CI 0.14 to 24.19; I² = 95%; 2 studies, 107 participants; very low‐certainty evidence) or TeCAs increase the risk of adverse events (RR 0.93, 95% CI 0.79 to 1.09; 1 study, 52 participants; very low‐certainty evidence). No studies assessed the risk of serious adverse events for these classes.

No studies measured total adverse events for other antidepressants. It is unclear whether other antidepressants increase the risk of serious adverse events (Peto OR 0.90, 95% CI 0.16 to 4.96; 1 study, 42 participants; very low‐certainty evidence).

Authors' conclusions

We found that in people with non‐specific low back pain, SNRIs probably have small effects on pain intensity, trivial effects on disability, and are probably associated with adverse effects. TCAs probably do not reduce low back pain intensity, but may have a small effect on disability. The effects of antidepressants on spine‐related leg pain are uncertain, though SNRIs and TCAs might be prioritised over other classes for future investigations. Evidence for the safety of SSRIs, TCAs, TeCAs, and other antidepressants in non‐specific low back pain and spine‐related leg pain remains unclear.

Plain language summary

Do antidepressants help people with non‐specific low‐back pain and spine‐related leg pain?

Key messages

• Compared with a placebo (inactive or 'dummy' pill), serotonin and norepinephrine reuptake inhibitors (a class of antidepressant) probably provide small reductions in pain intensity and trivial improvements in function in people with low back pain. Some people will probably experience unwanted effects when taking these medicines.

• Compared with a dummy pill, tricyclic antidepressants (a class of antidepressant) probably provide small improvements in function in people with low back pain, but probably have little to no effect on pain intensity.

• We are uncertain about the effects of any antidepressant for the treatment of spine‐related leg pain.

What are low back pain and spine‐related leg pain?

Low back pain is a leading cause of disability around the world. Most cases of low back pain are called 'non‐specific' because they are not caused by clear damage to the spine. Many people with low back pain also experience pain that radiates into the leg.

How do antidepressants treatlow back pain and spine‐related leg pain?

Antidepressants are a group of medicines that were originally developed to treat depression. The most common classes are serotonin and norepinephrine reuptake inhibitors, selective serotonin reuptake inhibitors, and tricyclic antidepressants. Antidepressants are thought to relieve pain by blocking pain signals in the nervous system. Some people who take antidepressants might experience unwanted effects, such as dry mouth and nausea.

What did we want to find out?

We wanted to find out if antidepressants are more effective than a dummy pill, standard care, or no treatment, at relieving pain and increasing function in people with low back pain or spine‐related leg pain. We also looked at whether antidepressants were associated with unwanted effects.

What did we do?

We searched for studies that compared antidepressants with a dummy pill, standard care, or no treatment, in people with low back pain or spine‐related leg pain. We compared and summarised the results of the studies and rated our confidence in the evidence, based on factors such as study methods and sizes.

What did we find? This is an update of a Cochrane review first published in 2008. We found 26 studies involving 2932 people: 18 studies included 2535 people with low back pain; seven studies included 329 people with spine‐related leg pain, and one study included 68 people with either condition. The average age of participants ranged from 27 to 59 years. Participants had 'chronic' symptoms (lasting more than 3 months) in 62% of studies. The study periods lasted from 1 day to 6 months. Studies investigated serotonin and norepinephrine reuptake inhibitors, selective serotonin reuptake inhibitors, tricyclic antidepressants, tetracyclic antidepressants, or other types of antidepressants (grouped as 'other antidepressants'), all compared with a dummy pill. The studies were conducted around the world, but most were in high‐income countries, including the USA, United Kingdom, France, Japan, and Australia. Almost half the included studies were funded by pharmaceutical companies.

Main results

We found that serotonin and norepinephrine reuptake inhibitors probably provide small reductions in pain intensity and trivial improvements in function in people with low back pain.

We also found that tricyclic antidepressants probably provide small improvements in function, but are unlikely to reduce pain intensity, in people with low back pain.

We do not know if selective serotonin reuptake inhibitors, tetracyclic antidepressants, or other antidepressants reduce pain intensity and improve function in people with low back pain.

For people with spine‐related leg pain, we do not know if any antidepressant reduces pain and increases function.

Serotonin and norepinephrine reuptake inhibitors probably increase the risk of experiencing unwanted effects. It is unclear whether the other antidepressant classes increase the risk of unwanted effects and serious unwanted events.

What are the limitations of the evidence?

Our confidence in the evidence for serotonin and norepinephrine reuptake inhibitors and tricyclic antidepressants is only moderate because not all studies provided data about everything we were interested in.

We have little or no confidence in the rest of the evidence in this review because not all studies provided data about everything we were interested in, the studies were very small, and there were not enough studies to be certain about the results of our outcomes of interest.

The results of further research could differ from the results of this review.

How current is this evidence?

This review updates our previous review. The evidence is current to November 2024.

Summary of findings

Summary of findings 1. Summary of findings table ‐ Serotonin and norepinephrine reuptake inhibitors compared to placebo for adults with non‐specific low back pain and spine‐related leg pain.

Serotonin and norepinephrine reuptake inhibitors compared to placebo for adults with non‐specific low back pain and spine‐related leg pain
Patient or population: adults with non‐specific low back pain and spine‐related leg pain Setting: outpatient (medical institutions and university medical centres) Intervention: serotonin and norepinephrine reuptake inhibitors  Comparison: placebo
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Outcomes	Risk with placebo	Risk with serotonin and norepinephrine reuptake inhibitors	Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Pain intensity (non‐specific low back pain) Scale from: 0 to 100 follow‐up: range 4 weeks to 16 weeks	The mean pain intensity was ‐17.5	MD 5.25 lower (7.17 lower to 3.34 lower)	‐	1415 (4 RCTs)	⊕⊕⊕⊝ Moderate^a	Serotonin and norepinephrine reuptake inhibitors probably have a small effect on pain intensity in people with non‐specific low back pain.
Pain intensity (spine‐related leg pain) Scale from: 0 to 100 follow‐up: range 4 weeks to 16 weeks	The mean pain intensity was 59.7	MD 46.1 lower (89.29 lower to 2.91 lower)	‐	11 (1 RCT)	⊕⊝⊝⊝ Very low^b,^c	The evidence is very uncertain about the effect of serotonin and norepinephrine reuptake inhibitors on pain intensity in people with spine‐related leg pain.
Disability (non‐specific low back pain) assessed with: Roland Morris Disability Questionnaire Scale from: 0 to 24 follow‐up: range 4 weeks to 16 weeks	The mean disability was ‐1.9	MD 0.9 lower (1.3 lower to 0.5 lower)	‐	1348 (4 RCTs)	⊕⊕⊕⊝ Moderate^a	Serotonin and norepinephrine reuptake inhibitors probably have a trivial effect on disability in people with non‐specific low back pain.
Disability (spine‐related leg pain) assessed with: Oswestry Disability Index Scale from: 0 to 100 follow‐up: range 4 weeks to 16 weeks	The mean disability was 22.5	MD 4.4 lower (20.25 lower to 11.45 higher)	‐	11 (1 RCT)	⊕⊝⊝⊝ Very low^b,^d	The evidence is very uncertain about the effect of serotonin norepinephrine reuptake inhibitors on disability in people with spine‐related leg pain.
Total adverse events (non‐specific low back pain and spine‐related leg pain)	558 per 1000	652 per 1000 (597 to 708)	RR 1.17 (1.07 to 1.27)	1510 (5 RCTs)	⊕⊕⊕⊝ Moderate^a	Serotonin and norepinephrine reuptake inhibitors probably increase the risk of adverse events in people with non‐specific low back pain and spine‐related leg pain.
Serious adverse events (non‐specific low back pain and spine‐related leg pain)	12 per 1000	21 per 1000 (9 to 45)	OR 1.75 (0.79 to 3.89)	1510 (5 RCTs)	⊕⊝⊝⊝ Very low^a,^e,^f	The evidence is very uncertain about the risk of serious adverse events with serotonin and norepinephrine reuptake inhibitor use in people with non‐specific low back pain and spine‐related leg pain.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  CI: confidence interval; MD: mean difference; OR: odds ratio; RR: risk ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_447751292574340661.

Open in a new tab

^a Downgraded once for risk of bias: high risk of attrition bias and unclear risk of selection bias. ^b Downgraded twice for risk of bias: high risk of attrition and reporting bias, unclear risk of selection bias. ^c Downgraded once for imprecision: 95% CI compatible with a large effect and a trivial effect. ^d Downgraded once for imprecision: 95% CI compatible with a large effect and no effect. ^e Downgraded once for inconsistency: incomplete overlap of 95% CIs. ^f Downgraded twice for imprecision: 95% CI compatible with a large risk and decreased risk.

Summary of findings 2. Summary of findings table ‐ Selective serotonin reuptake inhibitors compared to placebo for adults with non‐specific low back pain and spine‐related leg pain.

Selective serotonin reuptake inhibitors compared to placebo for adults with non‐specific low back pain and spine‐related leg pain
Patient or population: adults with non‐specific low back pain and spine‐related leg pain Setting: outpatient (primary care, orthopaedic, and rheumatology clinics) Intervention: selective serotonin reuptake inhibitors Comparison: placebo
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Outcomes	Risk with placebo	Risk with selective serotonin reuptake inhibitors	Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Pain intensity (non‐specific low back pain) Scale from: 0 to 100 follow‐up: range 4 weeks to 16 weeks	The mean pain intensity was 43.2	MD 1.2 higher (4.9 lower to 7.3 higher)	‐	199 (3 RCTs)	⊕⊕⊝⊝ Low^a,^b	Selective serotonin reuptake inhibitors may result in little to no difference in pain intensity in people with non‐specific low back pain.
Pain intensity (spine‐related leg pain) ‐ not measured	‐	‐	‐	‐	‐
Disability (non‐specific low back pain) assessed with: Oswestry Disability Index Scale from: 0 to 100 follow‐up: range 4 weeks to 16 weeks	The mean disability was 52.4	MD 2.2 lower (8.11 lower to 3.71 higher)	‐	92 (1 RCT)	⊕⊕⊝⊝ Low^c,^d	Selective serotonin reuptake inhibitors may result in little to no difference in disability in people with non‐specific low back pain.
Disability (spine‐related leg pain) ‐ not measured	‐	‐	‐	‐	‐
Total adverse events (non‐specific low back pain and spine‐related leg pain)	630 per 1000	1000 per 1000 (88 to 1000)	RR 1.83 (0.14 to 24.19)	107 (2 RCTs)	⊕⊝⊝⊝ Very low^e,^f,^g	The evidence is very uncertain about the risk of adverse events with selective serotonin reuptake inhibitor use in people with non‐specific low back pain and spine‐related leg pain.
Serious adverse events (non‐specific low back pain and spine‐related leg pain) ‐ not measured	‐	‐	‐	‐	‐
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  CI: confidence interval; MD: mean difference; RR: risk ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_447753863375202215.

Open in a new tab

^a Downgraded once for risk of bias: high risk of performance, detection, attrition, reporting, and other bias. ^b Downgraded once for indirectness: around a quarter of participants met DSM‐III criteria for major depression. ^c Downgraded once for risk of bias: high risk of other bias and unclear risk of performance, detection, attrition, and reporting bias. ^d Downgraded once for indirectness: 60% of participants met DSM‐III criteria for major depression. ^e Downgraded twice for risk of bias: high risk of performance, detection, attrition, and reporting bias.  ^f Downgraded once for inconsistency: incomplete overlap of 95% CIs. ^g Downgraded once for imprecision: 95% CI compatible with a trivial risk and a large risk.

Summary of findings 3. Summary of findings table ‐ Tricyclic antidepressants compared to placebo for adults with non‐specific low back pain and spine‐related leg pain.

Tricyclic antidepressants compared to placebo for adults with non‐specific low back pain and spine‐related leg pain
Patient or population: adults with non‐specific low back pain and spine‐related leg pain Setting: outpatient (primary care clinics), inpatient (hospital rheumatology and neurology departments) and community  Intervention: tricyclic antidepressants  Comparison: placebo
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Outcomes	Risk with placebo	Risk with tricyclic antidepressants	Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Pain intensity (non‐specific low back pain) Scale from: 0 to 100 follow‐up: range 4 weeks to 16 weeks	The mean pain intensity was 38.2	MD 2 lower (7.25 lower to 3.24 higher)	‐	417 (4 RCTs)	⊕⊕⊕⊝ Moderate^a	Tricyclic antidepressants probably result in little to no difference in pain intensity in people with non‐specific low back pain.
Pain intensity (spine‐related leg pain) Scale from: 0 to 100 follow‐up: range 4 weeks to 16 weeks	The mean pain intensity was 33	MD 23 lower (32.12 lower to 13.88 lower)	‐	60 (1 RCT)	⊕⊕⊝⊝ Low^b	Tricyclic antidepressants may have a large effect on pain intensity in people with spine‐related leg pain.
Disability (non‐specific low back pain) assessed with: Roland Morris Disability Questionnaire Scale from: 0 to 24 follow‐up: range 4 weeks to 16 weeks	The mean disability was 7.4	MD 1.76 lower (2.7 lower to 0.82 lower)	‐	330 (3 RCTs)	⊕⊕⊕⊝ Moderate^a	Tricyclic antidepressants probably have a small effect on disability in people with non‐specific low back pain.
Disability (spine‐related leg pain) assessed with: Oswestry Disability Index Scale from: 0 to 100 follow‐up: range 4 weeks to 16 weeks	The mean disability was 35	MD 13 lower (19.4 lower to 6.6 lower)	‐	60 (1 RCT)	⊕⊕⊝⊝ Low^b	Tricyclic antidepressants may have a moderate effect on disability in people with spine‐related leg pain.
Total adverse events (non‐specific low back pain and spine‐related leg pain)	351 per 1000	618 per 1000 (277 to 1000)	RR 1.76 (0.79 to 3.90)	474 (7 RCTs)	⊕⊕⊝⊝ Low^c,^d	The evidence is uncertain about the risk of adverse events with tricyclic antidepressant use in people with non‐specific low back pain and spine‐related leg pain.
Serious adverse events (non‐specific low back pain and spine‐related leg pain)	0 per 1000	0 per 1000 (0 to 0)	OR 6.64 (0.41 to 106.72)	142 (1 RCT)	⊕⊝⊝⊝ Very low^e,^f	The evidence is very uncertain about the risk of serious adverse events with tricyclic antidepressant use in people with non‐specific low back pain and spine‐related leg pain, due to the small number of studies and few events. We could not estimate absolute risk as there were no events in the control group.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  CI: confidence interval; MD: mean difference; OR: odds ratio; RR: risk ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_447774841358666296.

Open in a new tab

^a Downgraded once for risk of bias: high risk of attrition, reporting, and other bias. ^b Downgraded twice for risk of bias: high risk of reporting bias and unclear risk of selection, performance, detection, attrition, and other bias. ^c Downgraded once for risk of bias: high risk of attrition bias and unclear risk of selection bias. ^d Downgraded once for imprecision: 95% CI compatible with an increased risk and a decreased risk.  ^e Downgraded twice for risk of bias: high risk of attrition and other bias. ^f Downgraded twice for imprecision: 95% CI compatible with an increased risk and a decreased risk.

Summary of findings 4. Summary of findings table ‐ Tetracyclic antidepressants compared to placebo for adults with non‐specific low back pain and spine‐related leg pain.

Tetracyclic antidepressants compared to placebo for adults with non‐specific low back pain and spine‐related leg pain
Patient or population: adults with non‐specific low back pain and spine‐related leg pain Setting: outpatient (primary care) Intervention: tetracyclic antidepressants Comparison: placebo
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Outcomes	Risk with placebo	Risk with tetracyclic antidepressants	Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Pain intensity (non‐specific low back pain) Scale from: 0 to 100 follow‐up: range 4 weeks to 16 weeks	The mean pain intensity was 38.5	MD 4.5 lower (17.59 lower to 8.59 higher)	‐	52 (1 RCT)	⊕⊝⊝⊝ Very low^a,^b	The evidence is very uncertain about the effect of tetracyclic antidepressants on pain intensity in people with non‐specific low back pain.
Pain intensity (spine‐related leg pain) ‐ not measured	‐	‐	‐	‐	‐
Disability (non‐specific low back pain) ‐ not measured	‐	‐	‐	‐	‐
Disability (spine‐related leg pain) ‐ not measured	‐	‐	‐	‐	‐
Total adverse events (non‐specific low back pain and spine‐related leg pain)	969 per 1000	901 per 1000 (765 to 1000)	RR 0.93 (0.79 to 1.09)	52 (1 RCT)	⊕⊝⊝⊝ Very low^a,^c	The evidence is very uncertain about the risk of adverse events with tetracyclic antidepressant use in people with non‐specific low back pain and spine‐related leg pain.
Serious adverse events (non‐specific low back pain and spine‐related leg pain) ‐ not measured	‐	‐	‐	‐	‐
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  CI: confidence interval; MD: mean difference; RR: risk ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_447775718545968868.

Open in a new tab

^a Downgraded twice for risk of bias: high risk of performance, detection, attrition, and reporting bias. ^b Downgraded once for imprecision: 95% CI compatible with a moderate effect and no effect. ^c Downgraded once for imprecision: 95% CI compatible with an increased risk and a decreased risk.

Summary of findings 5. Summary of findings table ‐ Other antidepressants compared to placebo for adults with non‐specific low back pain and spine‐related leg pain.

Other antidepressants compared to placebo for adults with non‐specific low back pain and spine‐related leg pain
Patient or population: adults with non‐specific low back pain and spine‐related leg pain Setting: outpatient (pain clinic and veteran medical centre) Intervention: other antidepressants Comparison: placebo
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Outcomes	Risk with placebo	Risk with other antidepressants	Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Pain intensity (non‐specific low back pain) Scale from: 0 to 100 follow‐up: range 4 weeks to 16 weeks	The mean pain intensity was 53.4	MD 5.4 lower (23.08 lower to 12.28 higher)	‐	39 (1 RCT)	⊕⊝⊝⊝ Very low^a,^b	The evidence is very uncertain about the effect of other antidepressants on pain intensity in people with non‐specific low back pain.
Pain intensity (spine‐related leg pain) ‐ not measured	‐	‐	‐	‐	‐
Disability (non‐specific low back pain) ‐ not measured	‐	‐	‐	‐	‐
Disability (spine‐related leg pain) ‐ not measured	‐	‐	‐	‐	‐
Total adverse events (non‐specific low back pain and spine‐related leg pain) ‐ not measured	‐	‐	‐	‐	‐
Serious adverse events (non‐specific low back pain and spine‐related leg pain)	150 per 1000	137 per 1000 (27 to 467)	OR 0.90 (0.16 to 4.96)	42 (1 RCT)	⊕⊝⊝⊝ Very low^c,^d	The evidence is very uncertain about the risk of serious adverse events with other antidepressant use in people with non‐specific low back pain and spine‐related leg pain.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  CI: confidence interval; MD: mean difference; OR: odds ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_447750180547171168.

Open in a new tab

^a Downgraded once for risk of bias: high risk of attrition bias and unclear risk of selection bias. ^b Downgraded twice for imprecision: 95% CI compatible with a large effect and no effect. ^c Downgraded once for risk of bias: high risk of attrition bias and unclear risk of selection and reporting bias. ^d Downgraded twice for imprecision: 95% CI compatible with an increased risk and decreased risk.

Background

Description of the condition

Low back pain is a major global health problem. It is usually defined by the location of its symptoms, located between the costal margin and the inferior gluteal folds (Knezevic 2021). Low back pain is common – its one‐month prevalence has been estimated at 23.2% (standard deviation (SD) 2.9%) – and peaks at 85 years of age (Ferreira 2023; Vos 2020). For the past 30 years, low back pain has been the greatest contributor to years lived with disability globally; in 2020, there were 500 million cases worldwide, projected to increase to more than 800 million by 2050 (Ferreira 2023). Low back pain places a considerable burden on individuals, society, and the economy – it is responsible for the highest total expenditure on health care in the United Kingdom (28,000 million pounds sterling (GBP) per year) (Hong 2013), and the USA (134,500 million US dollars in 2016) (Dieleman 2016).

Low back pain has a heterogenous presentation and for around 90% of patients, a pathological cause of nociception cannot be reliably determined (Koes 2006). Low back pain is commonly accompanied by pain in one or both legs, affecting between 48% and 74% of primary care patients, depending on the definition used (Harrisson 2020). Pain radiating from the spine into the leg has varied sources, characteristics, and distributions, and so may be classed as somatic‐referred pain, radicular pain with radiculopathy, or radicular pain without radiculopathy (Bogduk 2009; IASP 2012). In 2023, a working group from the Neuropathic Pain Special Interest Group of the International Association for the Study of Pain proposed 'spine‐related leg pain' as an umbrella term to encompass these three case definitions (Schmid 2023). Outcomes associated with spine‐related leg pain are poorer than those associated with low back pain alone, with patients reporting increased pain and disability, and reduced quality of life (Konstantinou 2013).

Description of the intervention

Antidepressants are a broad group of medicines used primarily for the treatment of clinical depression. These medicines are classed according to their presumed mechanism of action, with the main groups as follows:

serotonin and norepinephrine reuptake inhibitors (SNRIs) (e.g. duloxetine, milnacipran, venlafaxine);
selective serotonin reuptake inhibitors (SSRIs) (e.g. escitalopram, fluoxetine, sertraline);
tricyclic antidepressants (TCAs) (e.g. amitriptyline, desipramine, nortriptyline);
tetracyclic antidepressants (TeCAs) (e.g. maprotiline, mianserin, mirtazapine);
'other antidepressants' (e.g. agomelatine, bupropion, trazodone).

Early observations of the analgesic properties of antidepressants have led to their 'off‐label' use to treat pain (Paoli 1960). Internationally, prescriptions of antidepressants are common for people with low back pain: approximately a quarter of Americans with chronic low back pain are prescribed an antidepressant within the first three months of diagnosis (Ivanova 2011), and 12.3% of people with low back pain in Portugal report the use of an antidepressant to manage their low back pain (Gouveia 2017).

How the intervention might work

The pharmacological targets of antidepressants are diverse, and accordingly, their effects may be exerted through the modulation of neurochemical, neuroplastic, and cognitive‐neuropsychological pathways (Harmer 2017). The precise mechanisms that underpin the analgesic effects of antidepressants are poorly understood, and likely vary over classes (McCleane 2008). One theory is that their inhibition of monoamine reuptake in the central nervous system leads to increased activity of descending serotonin and noradrenaline anti‐nociceptive pathways (Mika 2013; Urits 2019). The SNRI duloxetine and TCA amitriptyline have been demonstrated to relieve neuropathic pain via two independent noradrenergic mechanisms: one is rapid and mediated by descending controls from the brain to the spinal cord; the other is delayed, peripheral, and involves anti‐neuroimmune actions (Kremer 2018).

While people with chronic pain often experience depressive symptoms concomitantly (Orhurhu 2019; Rayner 2016), the analgesic effects of antidepressants may be independent of their effects on depression – analgesia is often achieved at lower doses, and after a shorter duration of therapy, compared with antidepressive effects (Hirschfeld 2005; Micó 2006). Due to their wide‐ranging effects, antidepressants may be prescribed for low back pain with or without leg pain to provide pain relief, reduce comorbid symptoms of depression, or improve sleep (Koes 2018).

Why it is important to do this review

Antidepressants are an established pharmacological intervention to treat the symptoms of low back pain and spine‐related leg pain. The use of antidepressants is rapidly increasing, with an increase in total prescriptions of 3.9 million (6.8%) in the United Kingdom between 2014 and 2015 (Health and Social Care Information Centre 2016). There is evidence that for back pain specifically, the use of antidepressants is rising, demonstrated by increases in prescriptions for this indication between 1999 and 2010 in the USA (Mafi 2013), and between 2004 and 2014 in Australia (Mathieson 2018).

The use of antidepressants to treat the symptoms of chronic low back pain is endorsed by six of eight international guidelines (Oliveira 2018); however, their specific recommendations are conflicting. In the USA, for example, the American College of Physicians endorses duloxetine, an SNRI, as second‐line therapy for chronic low back pain (Qaseem 2017), whereas in the United Kingdom, the National Institute for Health and Care Excellence (NICE) guideline for low back pain advises against the use of any antidepressant (NICE 2016). SNRIs or TCAs are recommended by five of 11 international clinical practice guidelines for the treatment of spine‐related leg pain (Price 2024).

There is concern that recommendations for the use of antidepressants for low back pain and spine‐related leg pain are supported by a limited evidence base. The previous version of this review found no clear evidence to support the use of any antidepressant (Urquhart 2008). However, we found no studies in people with spine‐related leg pain for the 2008 version, and the data were insufficient to robustly evaluate benefits and harms. Furthermore, several eligible trials have been published since 2008. A 2023 Cochrane network meta‐analysis found moderate‐certainty evidence for the efficacy of duloxetine in chronic pain, but did not include separate analyses for non‐specific low back pain or spine‐related leg pain (Birkinshaw 2023). Given the uncertainty of the benefits and harms of these medicines for low back pain and spine‐related leg pain, a rigorous, up‐to‐date review is needed.

Objectives

To assess the benefits and harms of antidepressants for non‐specific low back pain and spine‐related leg pain.

Methods

Criteria for considering studies for this review

Types of studies

We included parallel and cross‐over randomised controlled trials (RCTs). Inclusion was restricted to randomised trials because they are the best design to minimise bias when evaluating the effectiveness of an intervention. We considered studies reported as full‐text articles and those published as abstracts only. We also included data from trial registry reports, as the exclusion of such data may overestimate the efficacy of an intervention (Bagg 2020). There were no language restrictions.

Types of participants

We included studies in adults, aged 18 years and above, with non‐specific low back pain or spine‐related leg pain, of any duration. Non‐specific low back pain was defined as pain between the costal margin and the inferior gluteal fold, with or without associated leg pain (Dionne 2008). Spine‐related leg pain was defined as somatic‐referred pain, radicular pain with radiculopathy, or radicular pain without radiculopathy, or other descriptions suggestive of those definitions. Included participants may or may not have reported depressive symptoms.

We excluded participants with low back pain that was due to specific medical conditions, such as spinal fracture, inflammatory disease, aortic dissection, malignancy, or infection.

Types of interventions

We included studies comparing a systemically administered dose of an antidepressant medicine with a placebo, an active placebo (i.e. a medicine with no effect on pain that mimics the active drug's side effects), continuation of usual care, no treatment, or a waiting list. Trials comparing an antidepressant plus another medicine with a placebo plus the other medicine were also included.

Participants in the antidepressant and comparator arms may have received co‐interventions such as oral medicines, physical therapies, and psychological therapies. We included studies where participants received such co‐interventions if they were applied equally across treatment arms.

We grouped comparisons of antidepressants versus reference treatments into the following categories:

antidepressant versus placebo;
antidepressant versus usual care;
antidepressant versus no treatment/waiting list.

We classed antidepressants using codes from the Anatomical Therapeutic Chemical (ATC) classification system controlled by the World Health Organization Collaborating Centre for Drug Statistics Methodology (WHO 2022).

Types of outcome measures

Primary outcomes

The selection of outcomes was informed by the core outcome domains for clinical trials in low back pain (Chiarotto 2018). The primary outcomes were pain intensity, disability, and total adverse events:

Participant‐reported pain intensity, measured using a visual analogue scale (VAS), a numerical rating scale (NRS), a Likert scale, a rating scale within a composite measure of pain (e.g. McGill Pain Questionnaire), or an ordinal scale.
Disability, defined as low back‐specific disability, measured on a subjective rating scale (e.g. Oswestry Disability Index (ODI) and Roland Morris Disability Questionnaire (RMDQ)), a rating scale within a composite measure of disability, or an ordinal scale.
Total adverse events, measured by the number of participants who experienced an adverse event. We defined an adverse event as "any untoward medical occurrence associated with the use of a drug in humans, whether or not considered drug related" (FDA 2020).

Secondary outcomes

Serious adverse events, measured by the number of participants who experienced a serious adverse event. We defined a serious adverse event as "death, threat to life, in‐patient hospitalization or prolongation of existing hospitalization, a persistent or significant incapacity or substantial disruption of the ability to conduct normal life functions" (FDA 2020).
Withdrawals due to adverse events, measured by the number of participants who discontinued treatment due to adverse events.
Depressive symptoms, measured using a formal rating scale of depressive symptoms (e.g. Beck Depression Inventory and Montgomery‐Åsberg Depression Rating Scale).
Health‐related quality of life, measured using a validated quality of life scale (e.g. EuroQol‐5D, 12‐item Short‐Form Health‐Related Qualify of Life Survey).

Search methods for identification of studies

Electronic searches

For this updated review, we included the previous review as a source of studies, and searched the following databases from inception to 14 November 2024, with no language restrictions:

Cochrane Central Register of Controlled Trials (CENTRAL; 2024, Issue 11);
MEDLINE (Ovid; 1946 to 14 November 2024);
Embase (Ovid; 1947 to 14 November 2024).

In this update, we amended our original search strategies using terms developed by the Cochrane Musculoskeletal group. We did not search the PsycINFO database for the current version.

We also searched ClinicalTrials.gov, the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP), and the EU Clinical Trials Register (www.clinicaltrialsregister.eu).

If we needed information to confirm the eligibility of a study, we contacted study authors, and placed studies in 'awaiting classification' if we were unable to obtain the information. Where studies were ongoing, we kept records and reported them as such.

The complete search strategy is found in Appendix 1.

Searching other resources

We checked reference lists of all primary studies and review articles for additional references. We contacted experts in the field for unpublished and ongoing studies. Where necessary, we contacted study authors for additional information. We screened the titles of all included studies and citations on the Retraction Watch database to ensure none were retracted (Center for Scientific Integrity 2018).

Data collection and analysis

Selection of studies

Two review authors (MF, and GF, MW, CAS, AT, JH, or AC) independently screened titles and abstracts of the potentially‐relevant studies identified in the search. We coded them as 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. Two review authors (MF, and GF, MW, CAS, AT, JH, or AC) independently screened the full texts of retrieved studies for inclusion, and recorded reasons for exclusion of the ineligible studies. We resolved any disagreement through discussion or, if required, we consulted a third review author (JM). We identified and excluded duplicates, and collated multiple reports of the same study so that each study, rather than each report, was the unit of interest in the review. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram (Page 2020) and the Characteristics of excluded studies table.

Data extraction and management

We used a data collection form for study characteristics and outcome data, which we piloted on one study in the review. Working independently, two review authors (MF, and GF, MW, AT, or AC) extracted study characteristics from the included studies. A third review author (JM) spot‐checked study characteristics for accuracy against the trial report. We extracted the following study characteristics.

Methods: study design, total duration of study, number of study centres, location and country of origin, study setting, and date of study.
Participants: number, mean age, age range, gender/sex, condition, duration of symptoms, severity of pain, diagnostic criteria, study inclusion and exclusion criteria, and characteristics that stratify health opportunities outlined in PROGRESS (O'Neill 2014): place of residence, race/ethnicity, culture/language, occupation, religion, education, socioeconomic status, social capital.
Interventions: intervention(s) (including type, duration, dosage and run‐in period), comparison, rescue analgesia, concomitant medications, and excluded medications.
Outcomes: primary and secondary outcomes of interest to this review, time points reported, and baseline data.
Notes: funding for trial, and notable declarations of interest of trial authors.

Two review authors (MF, and GF, MW, AT, or AC) independently extracted outcome data from the included studies. We extracted the number of events and number of participants per treatment group for dichotomous outcomes, and means and standard deviations and number of participants per treatment group for continuous outcomes. We noted in the results section if data were not reported in a usable way and when data were transformed or estimated from a graph. We resolved disagreements by consensus or by involving a third review author (JM). One review author (MF) transferred data into the Review Manager (RevMan) file (RevMan 2024).

Where numeric data were not reported, we used WebPlotDigitizer, in duplicate, to extract data from graphs or figures.

We selected data to extract based on the following criteria:

where studies reported more than one relevant measure of pain intensity, we extracted data in the following order of preference: 100‐mm VAS, 11‐point NRS, rating scale from a composite measure of pain, ordinal scale with more than six levels, ordinal scale with fewer than six levels;
where studies reported more than one relevant measure of disability, we extracted data in the following order of preference: Oswestry Disability Index (ODI), Roland‐Morris Disability Questionnaire (RMDQ), rating scale for back‐specific disability from a composite measure of disability, ordinal scale with more than six levels, ordinal scale with fewer than six levels;
where studies reported more than one relevant measure of depressive symptoms, we prioritised data obtained using the scale employed in the greatest number of included studies;
where studies reported more than one relevant measure of health‐related quality of life, we prioritised data obtained using the scale employed in the greatest number of included studies;
where both final values and change‐from‐baseline values were reported for the same outcome, we extracted final values;
where both unadjusted and adjusted values for the same outcome were reported, we extracted adjusted values;
for all outcomes, we prioritised data analysed as intention‐to‐treat (ITT) rather than per‐protocol;
where multiple time points were recorded, we extracted data for the time point closest to four weeks, 16 weeks, six months, and 12 months.

We prioritised information from records according to the following hierarchy: i) primary publication; ii) conference abstract; iii) trial registration.

Assessment of risk of bias in included studies

Two review authors (MF, and GF, MW, AT, or AC) independently assessed the risk of bias for each study using Cochrane's risk of bias (RoB) 1 tool, using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). We resolved any disagreements by discussion or by involving another author (JM). We assessed the following risk of bias domains:

random sequence generation (selection bias);
allocation concealment (selection bias);
blinding of participants and personnel (performance bias);
blinding of outcome assessment (detection bias), for self‐reported outcomes;
incomplete outcome data (attrition bias);
selective outcome reporting (reporting bias);
other bias, such as unequal application of co‐interventions across treatment arms, early stopping, and unplanned interim analyses.

We classified each potential source of bias as high, low, or unclear risk, and provided information from the study report, together with a justification for our judgment in the risk of bias table. We summarised the risk of bias judgements across different studies for each of the domains listed.

Measures of treatment effect

We analysed dichotomous data as risk ratios or Peto odds ratios if the number of events was small (approximately less than 10% incidence), and used 95% confidence intervals (CIs). We analysed continuous data as mean difference (MD) and 95% CIs. We entered data presented as a scale with a consistent direction of effect across studies. We converted aggregate outcome data for pain to a common 0 (no pain) to 100 (worst pain) scale to improve clinical interpretability of results (Dworkin 2008; Ostelo 2008). When different scales were used to measure the same conceptual outcome (e.g. disability, depressive symptoms, HRQoL), we calculated standardised mean differences with corresponding 95% CIs.

We used criteria from Chou 2017 to define small, moderate, and large between‐group effects for pain and disability. For pain intensity, we defined effects of less than 5 points as trivial, 5 to 10 points on a 0‐ to 100‐point VAS as small, 10 to 20 points on a 0‐ to 100‐point VAS as moderate, and more than 20 points on a 0‐ to 100‐point VAS as large. For disability measured using the Oswestry Disability Index, we defined effects of less than 5 points as trivial, 5 to 10 points as small, more than 10 to 20 points as moderate, and more than 20 points as large. For disability measured using the Roland Morris Disability Questionnaire, we defined less than 1 point as trivial, 1 to 2 points as small, more than 2 to 5 points as moderate, and more than 5 points as large. Where SMDs were used to combine different scales, we defined between‐group effects of 0.2 to 0.5 as small, more than 0.5 to 0.8 as moderate, and more than 0.8 as large.

Unit of analysis issues

The unit of analysis was the participant for all included trials. Where multiple trial arms were reported in a single trial, we included only the eligible arms. If two comparisons (e.g. drug A versus placebo and drug B versus placebo) were combined in the same meta‐analysis, we split the control group to avoid double‐counting. For cross‐over trials, we planned to only include data from the first phase because low back pain and spine‐related leg pain have a variable course (Kongsted 2015), carry‐over effects of antidepressants are possible, and wash‐out of antidepressants may induce a withdrawal effect in the second period (Henssler 2024).

Dealing with missing data

We contacted investigators or study sponsors to verify key study characteristics and obtain missing numerical outcome data where possible (e.g. when a study was identified as an abstract only or when data were not available for all participants). Where this was not possible, and the missing data were thought to introduce serious bias, we planned to explore the impact of including such studies in the overall assessment of results by a sensitivity analysis. We clearly described any assumptions and imputations to handle missing data.

For dichotomous outcomes (e.g. number of withdrawals due to adverse events), we calculated the withdrawal rate using the number of participants randomised in the group as the denominator.

For continuous outcomes (e.g. mean change in pain score), we calculated the MD based on the number of participants analysed at that time point. If the number of participants analysed was not presented for each time point, we planned to use the number of randomised participants in each group at baseline.

We computed missing standard deviations from other statistics, such as standard errors, CIs, or P values, according to the methods recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2023). If we were unable to calculate standard deviations, we imputed them (e.g. from other studies in the meta‐analysis).

Assessment of heterogeneity

We assessed the clinical and methodological diversity of the included studies in terms of participants, interventions, outcomes, and study characteristics, to determine whether meta‐analyses were appropriate. We attempted to deal with clinical heterogeneity by performing separate analyses for each antidepressant class and condition.

To estimate statistical heterogeneity, we calculated the Chi² statistic, the between‐study variance (Tau²), and the proportion of that variance not due to sampling error (I²). For analyses with 10 or more studies, we planned to calculate 95% prediction intervals to indicate the potential effect of the intervention in a new population similar to that included in the meta‐analysis (Riley 2011). We used these measures, along with visual inspection of the forest plots, to form judgements about heterogeneity.

As recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2023), we interpreted an I² value of: 0% to 40% as heterogeneity that might 'not be important'; 30% to 60% as representing 'moderate' heterogeneity; 50% to 90% as representing 'substantial' heterogeneity; and 75% to 100% as representing 'considerable' heterogeneity. As noted in the Cochrane Handbook (Deeks 2023), we considered that the importance of I² depends on: (i) the magnitude and direction of effects and (ii) the strength of evidence for heterogeneity. We interpreted the Chi² test to indicate evidence of statistical heterogeneity where P values were 0.10 or lower.

If we identified substantial heterogeneity in the main analyses, we planned to report it and investigate possible causes by following the recommendations in section 10.10 of the Cochrane Handbook (Deeks 2023).

Assessment of reporting biases

We planned to create and examine funnel plots to explore possible small study biases for analyses that included 10 or more studies. In interpreting funnel plots, we planned to examine the different possible reasons for funnel plot asymmetry, as outlined in section 13 of the Cochrane Handbook for Systematic Reviews of Interventions, and relate this to the results of the review (Page 2023). For analyses including 10 or more studies, we planned to undertake formal statistical tests to investigate funnel plot asymmetry, and follow the recommendations in section 13.3 of the Cochrane Handbook (Page 2023).

To assess outcome reporting bias, we checked trial protocols against published reports. For studies published after 1 July 2005, we screened the Clinical Trial Register at the International Clinical Trials Registry Platform of the World Health Organization for the a priori trial protocol. Where selective reporting of outcomes was present, we reported this under the 'selective outcome reporting (reporting bias)' domain in the risk of bias assessment.

Data synthesis

We performed meta‐analyses using random‐effects models. We only performed meta‐analyses where meaningful; that is, where we considered that the treatments, participants, and underlying clinical question were similar enough for pooling to make sense.

For the following antidepressant medicine classes, we performed separate analyses for non‐specific low back pain and spine‐related leg pain:

serotonin and norepinephrine re‐uptake inhibitors (SNRIs);
selective serotonin re‐uptake inhibitors (SSRIs);
tricyclic antidepressants (TCAs);
tetracyclic antidepressants (TeCAs);
other antidepressants.

For outcomes assessing benefit (pain intensity, disability, depressive symptoms, health‐related quality of life), we performed analyses for these follow‐up time points:

immediate‐term: outcomes reported at up to four weeks post‐randomisation;
short‐term: outcomes reported at more than four to 16 weeks post‐randomisation (primary time point);
mid‐term: outcomes reported at between four and 12 months post‐randomisation;
long‐term: outcomes reported at more than 12 months post‐randomisation.

Where multiple time points were used within each period, we took the latest date.

For outcomes assessing harms (total adverse events, serious adverse events, withdrawals due to adverse events), we performed analyses using data measured at the end of the treatment period.

The primary analyses included all trials of each class of antidepressant medicine compared with placebo at short‐term follow‐up, regardless of the studies' risk of bias.

We analysed data using RevMan, and where additional functions were not supported in RevMan, we planned to use R software.

Subgroup analysis and investigation of heterogeneity

For the primary outcomes, we conducted the following subgroup analyses for the investigation of important heterogeneity for the primary time point where data were available.

Medicine dose: standard dose range (SDR), less than SDR, and above SDR according to the Prescriber's Digital Reference (https://pdr.net).
Symptom duration (acute (0 to 6 weeks)/subacute (6 to 12 weeks)/chronic (> 12 weeks)).

We used the formal test for subgroup interactions in RevMan but were cautious in interpreting these, as advised in section 10 of the Cochrane Handbook (Deeks 2023). We compared the magnitude of the effects between the subgroups by means of assessing the overlap of the CIs of the summary estimate.

Sensitivity analysis

Where sufficient data were available, we conducted the following sensitivity analyses to investigate the robustness of effects on the primary outcomes.

Risk of selection bias: we explored the impact of risk of bias by repeating the analyses and excluding studies rated at high or unclear risk of selection bias.
Risk of detection bias (for self‐reported outcomes): we explored the impact of risk of bias by repeating the analyses and excluding studies rated at high or unclear risk of detection bias.
Missing data: we planned to explore the impact of imputing measures of variance by repeating the analyses and excluding studies where we imputed measures of variance.
Unpublished data: we explored the impact of including data from trial registry reports with no primary publication by repeating the analyses with the inclusion of studies from trial registries.
Type of comparator: we explored the impact of combining active and inert placebos by repeating the analyses and including only studies using an inert placebo as a comparator.

Summary of findings and assessment of the certainty of the evidence

Two authors (MF and AC) independently assessed the certainty of the evidence. We used the five GRADE considerations (risk of bias, inconsistency, imprecision, indirectness, and publication bias) to assess the certainty of a body of evidence as it related to the studies which contributed data to the analyses for the prespecified outcomes. GRADE uses the following criteria to describe the confidence in the evidence:

high: we are very confident that the true effect lies close to that of the estimate of the effect;
moderate: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of effect, but there is a possibility that it is substantially different;
low: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect;
very low: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

We decreased the GRADE rating by one (‐ 1), two (‐ 2), or three (‐ 3) levels, up to a maximum of three downgrades (indicating very low certainty), based on the level of concern the evidence raised.

Our judgements were guided by methods described in chapter 14 of the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2023), and interpreted in line with recommendations from Santesso 2020. We used GRADEpro software to prepare the summary of findings (SoF) tables (GRADEpro GDT). We justified all decisions to downgrade the certainty of the evidence for each outcome presented in the SoF tables using footnotes, with comments to aid the reader's understanding of the review where necessary.

We created SoF tables using the following outcomes, for each class of antidepressants, and each condition:

pain intensity;
disability;
total adverse events;
serious adverse events.

The main comparisons in the SoF tables were antidepressants versus placebo at short‐term follow‐up for outcomes assessing benefits (pain, disability), and at the end of the treatment period for outcomes assessing harms (total adverse events, serious adverse events).

Results

Description of studies

Results of the search

The results of the search are illustrated in Figure and the full search strategy is provided in Appendix 1. Searches for the previous version of this review were conducted in November 2008 and updated on 14 November 2024 for the current version. The current search identified 7217 records across three databases (CENTRAL, MEDLINE, and Embase). A further 118 records were identified in trial registries (ClinicalTrials.gov; WHO ICTRP; www.clinicaltrialsregister.eu). After removal of duplicates, we screened the titles and abstracts of 6770 records. Of these, we assessed the full‐texts of 83 articles for eligibility, and excluded 35 articles (see Excluded studies). A further two studies were terminated (NCT03364075; NCT00227292), one study awaits classification (see Studies awaiting classification), and one study is ongoing (see Ongoing studies). We linked 18 articles to already included studies. Thus, we have included a total of 26 studies in this review: 10 identified in the previous searches and included in the 2008 version of the review (Urquhart 2008), and 16 identified in the searches for this update (see Included studies). All included studies were available as published trial reports, except for Afilal 2020 (conference abstract and medical specialisation dissertation), Johnson 2011 (conference abstract and clinical trial registry) and NCT01225068 (trial registry record).

Included studies

Details of key study characteristics are available in Characteristics of included studies.

Study design

All 26 included studies were RCTs. Twenty studies used a parallel design (Afilal 2020; Alcoff 1982; Atkinson 1998; Atkinson 1999; Atkinson 2007; Dickens 2000; Goodkin 1990; Gould 2020; Jenkins 1976; Konno 2016; Kurniawati 2020; Marks 2014; NCT01225068; Pirbudak 2003; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b; Treves 1991; Urquhart 2018; Vanelderen 2015), and six studies used a cross‐over design (Johnson 2011; Katz 2005; Khoromi 2007; Pheasant 1983; Schliessbach 2018; Schukro 2016). A single study had six intervention arms (Atkinson 2007), three studies had three intervention arms (Khoromi 2007; Schliessbach 2018; Skljarevski 2009a), and three studies had two intervention arms (Atkinson 1999; Treves 1991; Vanelderen 2015). Study follow‐up periods ranged from two hours (Schliessbach 2018) to six months (Pirbudak 2003; Urquhart 2018).

Country of origin and number of centres

Three studies were multinational (Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b): Skljarevski 2009a was conducted in high‐income (USA) and upper‐middle income (Argentina) countries; Skljarevski 2010a was conducted in high‐income (France, Germany, Netherlands) and upper‐middle income countries (Brazil, Mexico); and Skljarevski 2010b was conducted in high‐income (Germany, Netherlands, Poland, Spain, USA) and upper‐middle income countries (Russia). The remaining studies were conducted in a single country: 21 in high‐income countries, including Australia (Urquhart 2018), Austria (Schukro 2016), Belgium (Vanelderen 2015), France (Treves 1991), Japan (Konno 2016), Switzerland (Schliessbach 2018), Turkey (Pirbudak 2003), the USA (Alcoff 1982; Atkinson 1998; Atkinson 1999; Atkinson 2007; Goodkin 1990; Gould 2020; Johnson 2011; Katz 2005; Khoromi 2007; Marks 2014; NCT01225068; Pheasant 1983), the United Kingdom (Dickens 2000; Jenkins 1976); and two in lower‐middle income countries, Morocco (Afilal 2020) and Indonesia (Kurniawati 2020).

Five studies were multicentre (Alcoff 1982; Goodkin 1990; Konno 2016; Skljarevski 2010a; Skljarevski 2010b), with the number of centres ranging from two (Alcoff 1982; Goodkin 1990) to 58 (Konno 2016). Twelve studies were single‐centre (Afilal 2020; Dickens 2000; Gould 2020; Jenkins 1976; Katz 2005; Khoromi 2007; Kurniawati 2020; Marks 2014; Pheasant 1983; Schliessbach 2018; Schukro 2016; Vanelderen 2015). The remaining studies did not clearly report the number of study centres. Details on study settings are provided in Characteristics of included studies.

Study funding and author declarations of interest

Eleven studies received some form of industry funding (Dickens 2000; Goodkin 1990; Jenkins 1976; Johnson 2011; Katz 2005; Khoromi 2007; Konno 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b; Treves 1991). Nine studies received funding from national or state grants (Alcoff 1982; Atkinson 1998; Atkinson 1999; Atkinson 2007; Goodkin 1990; Gould 2020; Khoromi 2007; Schliessbach 2018; Urquhart 2018), three studies received internal or institutional funding (Goodkin 1990; Kurniawati 2020; Vanelderen 2015), and four studies did not report a funding source (Afilal 2020; NCT01225068; Pheasant 1983; Pirbudak 2003).

Four studies had one or more authors who declared relationships with industry (Marks 2014; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b), involving employment, stock options, and consultancy. Six studies declared no conflicts of interest (Afilal 2020; Kurniawati 2020; Schliessbach 2018; Schukro 2016; Urquhart 2018; Vanelderen 2015), and the remaining studies provided no information relating to author conflicts of interest.

Participants

A total of 2932 participants were randomised across the included studies, with study sample sizes ranging from 11 participants (Marks 2014) to 458 participants (Konno 2016). Eighteen studies included 2535 participants with clinically‐diagnosed low back pain (Alcoff 1982; Atkinson 1998; Atkinson 1999; Atkinson 2007; Dickens 2000; Goodkin 1990; Gould 2020; Jenkins 1976; Johnson 2011; Katz 2005; Konno 2016; Kurniawati 2020; Pheasant 1983; Schliessbach 2018; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b; Urquhart 2018). Seven studies included 329 participants with spine‐related leg pain (Afilal 2020; Khoromi 2007; Marks 2014; NCT01225068; Pirbudak 2003; Schukro 2016; Vanelderen 2015), of which four required radiographic confirmation of symptoms (Afilal 2020; Marks 2014; Pirbudak 2003; Vanelderen 2015); one required radiographic or clinical confirmation (Khoromi 2007); one required clinical confirmation (Schukro 2016); and one did not specify any confirmatory criteria (NCT01225068). A single study included 68 participants with either low back pain or spine‐related leg pain (Treves 1991).

Nineteen studies required a minimum symptom duration of three months (Afilal 2020; Atkinson 1998; Atkinson 1999; Atkinson 2007; Dickens 2000; Gould 2020; Johnson 2011; Katz 2005; Khoromi 2007; Konno 2016; Kurniawati 2020; Marks 2014; NCT01225068; Schliessbach 2018; Schukro 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b; Urquhart 2018), with the mean reported symptom duration of enroled participants ranging from 18 months (Schukro 2016) to 20 years (Goodkin 1990). One study required symptoms of less than three months' duration (Pirbudak 2003), two studies included participants with both acute and chronic symptoms (Alcoff 1982; Goodkin 1990), and four studies did not specify any required duration of symptoms (Jenkins 1976; Pheasant 1983; Treves 1991; Vanelderen 2015). Across the studies that reported baseline pain intensity, scores ranged from the equivalent of around 4 out of 10 in Urquhart 2018, to 6 out of 10 in Goodkin 1990 for low back pain, and 6 out of 10 in Marks 2014 to 8 out of 10 in Pirbudak 2003 for spine‐related leg pain. Sixteen studies excluded participants with clinical depression or significant depressive symptoms (Atkinson 1998; Atkinson 1999; Atkinson 2007; Dickens 2000; Gould 2020; Khoromi 2007; Konno 2016; Kurniawati 2020; NCT01225068; Schliessbach 2018; Schukro 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b; Urquhart 2018; Vanelderen 2015), and one study enroled participants with significant depressive symptoms (Dickens 2000).

All studies included both female and male participants, except for Atkinson 1998 and Johnson 2011, which included males only. In studies that included female participants, the average proportion of females was 51% and ranged from 5% in Jenkins 1976 to 95% in Afilal 2020. Fourteen studies set an upper age limit to restrict participation: 48 years (Pirbudak 2003); 49 years (Kurniawati 2020); 65 years (Atkinson 1998; Atkinson 1999; Atkinson 2007; Khoromi 2007); 70 years (Gould 2020; NCT01225068); 75 years (Urquhart 2018); and 80 years (Afilal 2020; Konno 2016; Schliessbach 2018; Schukro 2016; Vanelderen 2015). The mean age of included participants ranged from around 27 in Jenkins 1976 to 59 years in Konno 2016. Ten studies reported the race/ethnicity of included participants (Alcoff 1982; Atkinson 1998; Atkinson 1999; Atkinson 2007; Goodkin 1990; Gould 2020; Konno 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b). Across all studies, enroled participants were mostly white, except for Konno 2016, which included only Asian participants.

Interventions and comparators

Included studies evaluated five antidepressant classes. Eight studies tested the serotonin and norepinephrine reuptake inhibitors (SNRIs) duloxetine (Johnson 2011; Konno 2016; Schukro 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b) and milnacipran (Marks 2014; NCT01225068); two studies tested the selective serotonin reuptake inhibitor (SSRI) paroxetine (Atkinson 1999; Dickens 2000); fourteen studies tested the tricyclic antidepressants (TCAs) amitriptyline (Kurniawati 2020; Pheasant 1983; Pirbudak 2003; Urquhart 2018; Vanelderen 2015), clomipramine (Afilal 2020; Treves 1991), desipramine (Atkinson 2007; Gould 2020), imipramine (Alcoff 1982; Jenkins 1976; Schliessbach 2018), and nortriptyline (Atkinson 1998; Khoromi 2007); two studies tested the tetracyclic antidepressants (TeCAs) maprotiline (Atkinson 1999) and imipramine (Schliessbach 2018); and two studies tested other antidepressants: trazodone (Goodkin 1990) and bupropion (Katz 2005).

The route of antidepressant administration was oral for all included studies except Afilal 2020 and Treves 1991, which used intravenous administration. Antidepressants were administered as a single dose in one study (Schliessbach 2018), as 10‐ to 14‐day courses in four studies (Afilal 2020; Kurniawati 2020; Treves 1991; Vanelderen 2015), as four‐ to six‐week courses in five studies (Goodkin 1990; Jenkins 1976; Johnson 2011; Pheasant 1983; Schukro 2016), as seven‐ to 10‐week courses in seven studies (Alcoff 1982; Atkinson 1998; Atkinson 1999; Dickens 2000; Katz 2005; Khoromi 2007; NCT01225068), as 12‐ to 14‐week courses in seven studies (Atkinson 2007; Gould 2020; Konno 2016; Marks 2014; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b), and as six‐month courses in two studies (Pirbudak 2003; Urquhart 2018). Seventeen studies used a run‐in phase where the antidepressant dose was gradually increased before commencing the target dose (Afilal 2020; Alcoff 1982; Atkinson 1998; Atkinson 1999; Atkinson 2007; Goodkin 1990; Gould 2020; Johnson 2011; Katz 2005; Khoromi 2007; Konno 2016; Marks 2014; NCT01225068; Schukro 2016; Skljarevski 2009a; Skljarevski 2010a; Treves 1991).

All studies used placebo as a comparator, with six studies using an active placebo (Atkinson 1999; Atkinson 2007; Gould 2020; Pheasant 1983; Schliessbach 2018; Urquhart 2018). Three studies used additional analgesia as part of the study treatment in both the intervention and placebo groups (Afilal 2020; Kurniawati 2020; Pirbudak 2003). Eleven studies permitted continuation of simple analgesic medicines during the study treatment period (Atkinson 1998; Atkinson 1999; Atkinson 2007; Dickens 2000; Goodkin 1990; Gould 2020; Johnson 2011; Skljarevski 2009a; Skljarevski 2010a; Urquhart 2018; Vanelderen 2015). Rescue medicines were permitted in nine studies, which consisted of restricted, episodic use of simple or opioid analgesics (Jenkins 1976; Khoromi 2007; Konno 2016; Pheasant 1983; Schukro 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b; Vanelderen 2015).

Outcomes

We attempted to contact the authors of nine studies to request outcome data: Atkinson 2007, Johnson 2011, and Katz 2005 responded but reported that the required data were not available; Schliessbach 2018 and Schukro 2016 did not reply to our emails. We were unsuccessful in attempts to reach the authors of Alcoff 1982, Khoromi 2007, Pheasant 1983, and Treves 1991 because we could not find current contact details.

Included studies reported data for three of our four time periods of interest: 12 studies at immediate‐term follow‐up (≤ 4 weeks post‐randomisation) (Afilal 2020; Jenkins 1976; Johnson 2011; Katz 2005; Kurniawati 2020; Pirbudak 2003; Schliessbach 2018; Schukro 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b; Vanelderen 2015); 19 studies at short‐term follow‐up (> 4 to 16 weeks post‐randomisation) (Atkinson 1998; Atkinson 1999; Atkinson 2007; Dickens 2000; Goodkin 1990; Gould 2020; Johnson 2011; Katz 2005; Khoromi 2007; Konno 2016; Marks 2014; NCT01225068; Pheasant 1983; Pirbudak 2003; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b; Urquhart 2018; Vanelderen 2015); and two studies at mid‐term follow‐up (Pirbudak 2003; Urquhart 2018) (> 4 to 12 months post‐randomisation). No included studies measured outcomes at long‐term follow‐up (> 12 months post‐randomisation).

Primary outcomes

Pain intensity

Twenty‐five studies assessed low back or leg pain intensity using a range of outcome measures: five studies used a 0 to 10 NRS (Katz 2005; Khoromi 2007; Schliessbach 2018; Skljarevski 2009a; Vanelderen 2015); four studies used a 10‐cm VAS (Afilal 2020; Jenkins 1976; Pirbudak 2003; Schukro 2016); five studies used a 100‐mm VAS (Dickens 2000; Goodkin 1990; Marks 2014; NCT01225068; Urquhart 2018); one study used an unspecified VAS (Kurniawati 2020); four studies used the 0 to 10 Brief Pain Inventory (Johnson 2011; Konno 2016; Skljarevski 2010a; Skljarevski 2010b); four studies used the Descriptor Differential Scale (Atkinson 1998; Atkinson 1999; Atkinson 2007; Gould 2020); one study used the Short Back Pain Questionnaire (Alcoff 1982); and one study used a verbal rating scale (Treves 1991).

Disability/function

Thirteen studies measured low back‐specific disability or function: four studies used the ODI (Dickens 2000; Khoromi 2007; Marks 2014; Pirbudak 2003); eight studies used the RMDQ (Afilal 2020; Atkinson 2007; Gould 2020; Konno 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b; Urquhart 2018); and one study used a self‐administered 'activity questionnaire' (Pheasant 1983).

Total adverse events

While most studies mentioned adverse events, only 17 studies reported the number of participants who experienced an adverse event for each treatment arm (Afilal 2020; Atkinson 1998; Atkinson 1999; Atkinson 2007; Gould 2020; Johnson 2011; Khoromi 2007; Konno 2016; Kurniawati 2020; Marks 2014; NCT01225068; Schukro 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b; Urquhart 2018; Vanelderen 2015).

Secondary outcomes

Serious adverse events

Serious adverse events were reported in nine studies (Goodkin 1990; Gould 2020; Johnson 2011; Konno 2016; Marks 2014; NCT01225068; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b).

Withdrawals due to adverse events

Withdrawals due to adverse events were reported for each treatment arm in 18 studies (Afilal 2020; Alcoff 1982; Atkinson 1998; Atkinson 1999; Atkinson 2007; Goodkin 1990; Gould 2020; Jenkins 1976; Katz 2005; Khoromi 2007; Konno 2016; Marks 2014; Schukro 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b; Urquhart 2018; Vanelderen 2015).

Depressive symptoms

Thirteen studies measured depressive symptoms: 11 studies used the Beck Depression Inventory (Atkinson 1998; Atkinson 1999; Atkinson 2007; Goodkin 1990; Katz 2005; Khoromi 2007; Marks 2014; Schliessbach 2018; Skljarevski 2009a; Skljarevski 2010a; Urquhart 2018); one study used the Hospital Anxiety and Depression Scale (Afilal 2020); and one study used the Montgomery‐Åsberg Depression Rating Scale (Dickens 2000).

Health‐related quality of life

Nine studies measured health‐related quality of life: one study used the European Quality of Life Questionnaire–5 Dimensions (EQ‐5D) (Urquhart 2018); three studies used the 36‐item Short Form Survey (SF‐36) (Khoromi 2007; Marks 2014; Schukro 2016); four studies used both the EQ‐5D and SF‐36 (Konno 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b); and one study used the Quality of Wellbeing scale (Atkinson 1998).

Excluded studies

We excluded 35 studies for the following reasons: 20 due to an ineligible participant population; 11 due to an ineligible study design; one because it did not include an outcome of interest; two due to an ineligible intervention; and one due to an ineligible comparator. Details are provided in Characteristics of excluded studies.

Risk of bias in included studies

We provide a summary of our judgements of the included studies' risk of bias in Figure.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

We judged 12 studies as having a low risk of selection bias, with adequate methods reported for the generation of the random sequence, and allocation concealment (Atkinson 1998; Atkinson 1999; Atkinson 2007; Dickens 2000; Gould 2020; Katz 2005; Konno 2016; Schliessbach 2018; Schukro 2016; Skljarevski 2009a; Urquhart 2018; Vanelderen 2015).

Thirteen studies were at unclear risk of selection bias because it was unclear how randomisation was accomplished (Alcoff 1982; Goodkin 1990; Jenkins 1976; Johnson 2011; Khoromi 2007; Kurniawati 2020; Marks 2014; NCT01225068; Pheasant 1983; Pirbudak 2003; Skljarevski 2010a; Skljarevski 2010b; Treves 1991). One study used appropriate methods to generate the random sequence, but it was unclear how allocation was concealed (Afilal 2020).

Blinding

We judged 15 studies as having a low risk of bias due to blinding (performance bias and detection bias) because they clearly reported participants as blinded; used study medicines that were indistinguishable from placebo; and either reported that tests that assessed the adequacy of blinding were successful or reported no other information suggesting that blinding was broken (Afilal 2020; Alcoff 1982; Atkinson 1998; Atkinson 2007; Goodkin 1990; Gould 2020; Katz 2005; Konno 2016; Marks 2014; Pheasant 1983; Schliessbach 2018; Schukro 2016; Skljarevski 2009a; Urquhart 2018; Vanelderen 2015).

Two studies were at high risk of bias for blinding, with both reporting that tests of blinding adequacy indicated unsuccessful blinding (Atkinson 1999; Khoromi 2007).

Nine studies were at unclear risk of bias for blinding: one due to a lack of detail regarding participant blinding (Dickens 2000); seven due to the indistinguishability of trial treatments (Jenkins 1976; Johnson 2011; Kurniawati 2020; Pirbudak 2003; Skljarevski 2010a; Skljarevski 2010b; Treves 1991); and one did not report information sufficient to assess whether blinding could have been broken (NCT01225068).

There were no assessor‐reported outcomes for evaluation of detection bias.

Incomplete outcome data

We judged three studies to have a low risk of attrition bias (Afilal 2020; Johnson 2011; Treves 1991), each reporting complete outcome data.

Nineteen studies were at high risk of attrition bias, primarily because they had substantial amounts of missing outcome data (Jenkins 1976; Khoromi 2007; Marks 2014; Pheasant 1983; Schliessbach 2018; Schukro 2016); imbalances in the amount of missing outcome data between intervention groups (Atkinson 1999; Goodkin 1990; NCT01225068; Urquhart 2018); between‐group differences in reasons for missingness (Alcoff 1982); and used inappropriate methods to account for missing data ('last observation carried forward') (Atkinson 1998; Atkinson 2007; Gould 2020; Katz 2005; Konno 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b).

Four studies were at unclear risk of attrition bias: one study did not report treatment withdrawals or loss to follow‐up by study arm (Dickens 2000); two studies did not report the number of participants who provided outcome data (Kurniawati 2020; Pirbudak 2003); and one study did not report the methods used to account for missing outcome data (Vanelderen 2015).

Selective reporting

We judged seven studies to have a low risk of reporting bias (Gould 2020; Konno 2016; NCT01225068; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b; Urquhart 2018): the study protocols were available and all prespecified outcomes of interest to the review were reported in the prespecified way. While Urquhart 2018 did not report data for one of two primary outcomes, these data were available from the authors.

Thirteen studies were at high risk of reporting bias: two because they were retrospectively registered (Afilal 2020; Schliessbach 2018); five because they reported outcomes incompletely so they could not be entered in a meta‐analysis (Alcoff 1982; Atkinson 1999; Jenkins 1976; Johnson 2011; Treves 1991); four because they did not report outcome data for prespecified endpoints (Atkinson 2007; Marks 2014; Schukro 2016; Vanelderen 2015); and two because they did not report the number of participants who provided outcome data (Kurniawati 2020; Pirbudak 2003).

Six studies were at unclear risk of reporting bias because they had no study protocol or registration, thus providing insufficient information with which to assess selective reporting (Atkinson 1998; Dickens 2000; Goodkin 1990; Katz 2005; Khoromi 2007; Pheasant 1983).

Other potential sources of bias

We judged two studies to have a low risk of other biases because they did not report unequal application of co‐interventions across treatment arms, early stopping, or unplanned interim analyses (Goodkin 1990; Skljarevski 2009a).

Three studies were at high risk of other biases: one because it reported significant between‐group differences in analgesic medicine use (Dickens 2000); and two because they were terminated before reaching their prespecified sample size (Gould 2020; Schliessbach 2018).

We judged 21 studies to be at unclear risk of other biases: while there was no early stopping or unplanned interim analyses, they reported insufficient information to assess the application of co‐interventions (Afilal 2020; Alcoff 1982; Atkinson 1998; Atkinson 1999; Atkinson 2007; Jenkins 1976; Johnson 2011; Katz 2005; Khoromi 2007; Konno 2016; Kurniawati 2020; Marks 2014; NCT01225068; Pheasant 1983; Pirbudak 2003; Schukro 2016; Skljarevski 2010a; Skljarevski 2010b; Treves 1991; Urquhart 2018; Vanelderen 2015).

Effects of interventions

See: Summary of findings 1 Summary of findings table ‐ Serotonin and norepinephrine reuptake inhibitors compared to placebo for adults with non‐specific low back pain and spine‐related leg pain; Summary of findings 2 Summary of findings table ‐ Selective serotonin reuptake inhibitors compared to placebo for adults with non‐specific low back pain and spine‐related leg pain; Summary of findings 3 Summary of findings table ‐ Tricyclic antidepressants compared to placebo for adults with non‐specific low back pain and spine‐related leg pain; Summary of findings 4 Summary of findings table ‐ Tetracyclic antidepressants compared to placebo for adults with non‐specific low back pain and spine‐related leg pain; Summary of findings 5 Summary of findings table ‐ Other antidepressants compared to placebo for adults with non‐specific low back pain and spine‐related leg pain

We present outcomes assessing benefit separately for non‐specific low back pain and spine‐related leg pain, by intervention and time point. We present harm outcomes for participants with both low back pain and spine‐related leg pain, by intervention only. No included studies compared antidepressants with usual care, no treatment, or a waiting‐list control.

Absolute and relative effects for key comparisons and outcomes are summarised with their GRADE ratings in Table; Table; Table; Table; Table.

Non‐specific low back pain

Serotonin and norepinephrine reuptake inhibitors (SNRIs) versus placebo

Primary outcomes

Pain intensity

Immediate‐term follow‐up (≤ 4 weeks)

Five studies assessed the effect of SNRIs on low back pain intensity compared with placebo in the immediate term (Johnson 2011; Konno 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b). Outcome data for Johnson 2011 were not reported for this time point and could not be obtained from study authors. We used WebPlotDigitizer to extract point estimates and standard errors for Konno 2016, and point estimates for Skljarevski 2010b. We calculated standard deviations for Skljarevski 2010a and Skljarevski 2010b from P values using the RevMan calculator. Our meta‐analysis demonstrated a small between‐group difference in pain intensity in favour of SNRIs, though the lower bound of the confidence interval was compatible with a trivial effect (mean difference (MD) ‐6.12, 95% confidence interval (CI) ‐8.42 to ‐3.82; 4 studies, 1424 participants; Analysis 1.1). Heterogeneity was minimal (Tau² = 2.40; Chi² = 7.08, degrees of freedom (df) = 5 (P = 0.22); I² = 29%). We judged the certainty of evidence as moderate, downgraded once for risk of bias (high risk of attrition bias and unclear risk of selection bias).

1.1 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 1: Non‐specific low back pain: pain intensity (0 to 100) at immediate‐term follow‐up (≤ 4 weeks)

Short‐term follow‐up (> 4 to 16 weeks; main comparison)

Five studies assessed the effect of SNRIs on low back pain intensity compared with placebo in the short term (Johnson 2011; Konno 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b). We were unable to include outcome data from Johnson 2011 in our analysis because they were not presented by cross‐over treatment phase and could not be obtained from study authors. Skljarevski 2009a compared three doses with placebo, so we divided the placebo group by the number of comparisons to avoid double‐counting. Our meta‐analysis demonstrated a small between‐group difference in pain intensity in favour of SNRIs, though the lower bound of the confidence interval was compatible with a trivial effect (MD ‐5.25, 95% CI ‐7.17 to ‐3.34; I² = 0; 4 studies, 1415 participants; Analysis 1.2; Figure). We judged the certainty of evidence as moderate, downgraded once for risk of bias (high risk of attrition bias and unclear risk of selection bias).

1.2 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 2: Non‐specific low back pain: pain intensity (0 to 100) at short‐term follow‐up (> 4 to 16 weeks)

All antidepressants versus placebo for non‐specific low back intensity (0 to 100) at short‐term follow‐up (> 4 to 16 weeks).

Our sensitivity analysis showed the estimate was robust to the removal of studies with unclear risk of selection and detection bias (Skljarevski 2010a; Skljarevski 2010b) (MD ‐4.53, 95% CI ‐6.88 to ‐2.17; I² = 0; 2 studies, 794 participants).

While minimal heterogeneity was observed in the main analysis, data were available to explore dose as a possible effect modifier for our planned subgroup analysis. This analysis did not show visible or statistical differences in effects between below (< 40 mg/day), within (40 mg to 60 mg/day), or above (< 60 mg/day) the standard duloxetine dosing range (test for subgroup differences: Chi² = 2.84, df = 2 (P = 0.24); I² = 29.7%). The observational nature of this analysis and small number of studies in each subgroup means the possibility of important differences cannot be excluded. Data were insufficient to perform the planned subgroup analysis for symptom duration.

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Disability

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks; main comparison)

Four studies assessed the effect of SNRIs on low back‐specific disability compared with placebo at short‐term in participants with low back pain (Konno 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b), measured using the RMDQ (0 to 24; higher scores indicate greater low back‐specific disability). Skljarevski 2009a compared three doses with placebo, so we divided the placebo group by the number of comparisons to avoid double‐counting. We calculated standard deviations for Skljarevski 2010a from P values using the RevMan calculator. Our meta‐analysis demonstrated a trivial between‐group difference in disability in favour of SNRIs (MD ‐0.91, 95% CI ‐1.30 to ‐0.51; I² = 0; 4 studies, 1348 participants; Analysis 1.4; Figure). We judged the certainty of evidence as moderate, downgraded once for risk of bias (high risk of attrition bias and unclear risk of selection bias).

1.4 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 4: Non‐specific low back pain: disability (RMDQ, 0 to 24) at short‐term follow‐up (> 4 to 16 weeks): SNRI

SNRIs versus placebo for non‐specific low back pain disability (RMDQ, 0 to 24) at short‐term follow‐up (> 4 to 16 weeks).

Our sensitivity analysis showed the estimate was robust to the removal of studies with unclear risk of selection and detection bias (Skljarevski 2010a; Skljarevski 2010b) (MD ‐0.92, 95% CI ‐1.39 to ‐0.45; I² = 0; 2 studies, 766 participants).

While minimal heterogeneity was observed in the main analysis, data were available to explore dose as a possible effect modifier via subgroup analysis. This analysis did not show visible or statistical differences in effects between below (< 40 mg/day), within (40 mg to 60 mg/day), or above (< 60 mg/day) the standard duloxetine dosing range (test for subgroup differences: Chi² = 1.05, df = 2 (P = 0.59); I² = 0%). The observational nature of this analysis and small number of studies in each subgroup means the possibility of important differences cannot be excluded. Data were insufficient to perform the planned subgroup analysis for symptom duration.

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Secondary outcomes

Depressive symptoms

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks)

Two studies assessed the effect of SNRIs on depressive symptoms compared with placebo in the short term (Skljarevski 2009a; Skljarevski 2010a), measured using the Beck Depression Inventory. Our meta‐analysis showed no evidence of a difference in depressive symptoms between SNRIs and placebo (MD ‐0.09, 95% CI ‐1.20 to 1.02; 2 studies, 613 participants; Analysis 1.8). Heterogeneity was minimal and mostly due to variation in true effects (Tau² = 0.70; Chi² = 6.64, df = 3 (P = 0.08); I² = 55%). We judged the certainty of evidence as moderate, downgraded once for risk of bias (high risk of attrition bias and unclear risk of selection, performance, and detection bias).

1.8 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 8: Non‐specific low back pain: depressive symptoms (BDI, 0 to 63) at short‐term follow‐up (> 4 to 16 weeks): SNRI

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Health‐related quality of life

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks)

Four studies assessed the effect of SNRIs on health‐related quality of life (HRQoL) compared with placebo in the short term (Konno 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b), measured on the EQ‐5D (0 to 1; higher scores indicate greater HRQoL). Our meta‐analysis showed a between‐group difference in HRQoL in favour of SNRIs (MD 0.03, 95% CI 0.01 to 0.06; 4 studies, 1401 participants; Analysis 1.13). Heterogeneity was minimal and mostly due to variation in true effects (Tau² = 0.00; Chi² = 6.74, df = 5 (P = 0.24); I² = 26%). We judged the certainty of evidence as moderate, downgraded once for risk of bias (high risk of attrition bias and unclear risk of selection, performance, and detection bias).

1.13 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 13: Non‐specific low back pain: HRQoL (EQ‐5D, 0 to 1) at short‐term follow‐up (> 4 to 16 weeks): SNRI

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Selective serotonin reuptake inhibitors (SSRIs) versus placebo

Primary outcomes

Pain intensity

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks; main comparison)

Three studies assessed the effect of SSRIs on low back pain intensity compared with placebo in the short term (Atkinson 1999; Atkinson 2007; Dickens 2000). Our meta‐analysis showed no evidence of a difference in pain intensity between SSRIs and placebo, with confidence intervals spanning a trivial effect and no effect at all (MD 1.20, 95% CI ‐4.90 to 7.30; I² = 0; 3 studies, 199 participants; Analysis 1.2; Figure). We judged the certainty of evidence as low, downgraded once for risk of bias (high risk of performance bias, detection bias, attrition bias, reporting bias, and other bias), and once for indirectness (60% of participants in Dickens 2000 fulfilled the Diagnostic and Statistical Manual of Mental Disorders 3rd Edition (DSM‐III) criteria for major depression).

Our sensitivity analyses showed the estimate was minimally impacted by the removal of studies at high or unclear risk of detection bias (MD 1.50, 95% CI ‐9.00 to 12.00; 1 study, 53 participants) (Atkinson 1999; Dickens 2000), and studies using active placebos (MD 0.00, 95% CI ‐9.83 to 9.83; 1 study, 92 participants) (Atkinson 1999; Atkinson 2007).

There were insufficient data to explore the effect of dose as a possible effect modifier for our planned subgroup analysis because Atkinson 2007 randomised participants to target desipramine serum concentrations rather than fixed doses. Data were also insufficient to perform the planned subgroup analysis for symptom duration.

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Disability

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks; main comparison)

Two studies assessed the effect of SSRIs on low back disability compared with placebo in the short term in participants with low back pain (Atkinson 2007; Dickens 2000). We were unable to include outcome data from Atkinson 2007 in our analysis because results were not provided in the trial report and could not be obtained from study authors. Dickens 2000 measured disability using the Oswestry Disability Index. Our analysis showed no evidence of a difference in disability between SSRIs and placebo, with confidence intervals spanning a small effect and no effect at all (MD ‐2.20 (0 to 100 scale), 95% CI ‐8.11 to 3.71; 1 study, 92 participants; Analysis 1.5). We judged the certainty of evidence as low, downgraded once for risk of bias (high risk of other bias and unclear risk of performance, detection, attrition, and reporting bias), and once for indirectness (60% of participants in Dickens 2000 fulfilled the DSM‐III criteria for major depression). Data were insufficient to perform the planned subgroup and sensitivity analyses.

1.5 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 5: Non‐specific low back pain: disability (ODI, 0 to 100) at short‐term follow‐up (> 4 to 16 weeks): SSRI

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Secondary outcomes

Depressive symptoms

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks)

One study assessed the effect of SSRIs on depressive symptoms compared with placebo in the short term (Dickens 2000), measured using the Montgomery‐Åsberg Depression Rating Scale (0 to 60; higher scores indicate greater depressive symptoms). Our analysis showed no evidence of a difference in depressive symptoms between SSRIs and placebo (MD ‐0.10, 95% CI ‐3.64 to 3.44; 1 study, 92 participants; Analysis 1.9). We judged the certainty of evidence as low, downgraded once for risk of bias (high risk of other bias and unclear risk of performance, detection, attrition, and reporting bias), and once for indirectness (60% of participants in Dickens 2000 fulfilled the DSM‐III criteria for major depression).

1.9 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 9: Non‐specific low back pain: depressive symptoms (MADRS, 0 to 60) at short‐term follow‐up (> 4 to 16 weeks): SSRI

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Health‐related quality of life

No studies measured or reported this outcome.

Tricyclic antidepressants (TCAs) versus placebo

Primary outcomes

Pain intensity

Immediate‐term follow‐up (≤ 4 weeks)

Four studies assessed the effect of TCAs on low back pain intensity compared with placebo in the immediate term (Jenkins 1976; Kurniawati 2020; Schliessbach 2018; Treves 1991). We were unable to analyse outcome data from two studies: Treves 1991 did not report a point estimate and associated measure of variance, and Schliessbach 2018 did not report outcome data by cross‐over phase; these could not be obtained from study authors. We used WebPlotDigitizer to extract point estimates for Jenkins 1976, and we imputed a conservative standard deviation of 3 (approximately one standard deviation higher than those of other included TCA studies). Our meta‐analysis showed no evidence of a difference in pain intensity between TCAs and placebo, with confidence intervals spanning both a moderate effect and no effect at all (MD 1.20, 95% CI ‐12.36 to 14.76; I² = 0%; 2 studies, 78 participants; Analysis 1.1). We judged the certainty of evidence as low, downgraded once for risk of bias (high risk of attrition and other bias, unclear selection, performance, and detection bias), and once for imprecision.

Short‐term follow‐up (> 4 to 16 weeks; main comparison)

Five studies assessed the effect of TCAs on low back pain intensity compared with placebo in the short term (Alcoff 1982; Atkinson 1998; Atkinson 2007; Gould 2020; Urquhart 2018). We were unable to include outcome data from Alcoff 1982 in our analysis because point estimates and associated measures of variance were not reported and could not be obtained from study authors. Data from Atkinson 2007 were provided for participants randomised to the low concentration desipramine group only. Gould 2020 was a 2x2 factorial trial so we analysed the four treatment groups as two separate comparisons to assess the independent effect of amitriptyline: amitriptyline plus cognitive behavioural therapy (CBT) versus placebo plus cognitive behavioural therapy (CBT); and amitriptyline versus placebo. Our meta‐analysis showed no evidence of a difference in pain intensity between TCAs and placebo, with confidence intervals spanning a small effect and no effect at all (MD ‐2.00, 95% CI ‐7.25 to 3.24; 4 studies, 417 participants; Analysis 1.2; Figure). Heterogeneity was minimal (Tau² = 11.03; Chi² = 5.81, df = 4 (P = 0.21); I² = 31%). We judged the certainty of evidence as moderate, downgraded once for risk of bias (high risk of attrition, reporting, and other bias).

Our sensitivity analysis showed the estimate was not robust to the removal of studies using active placebo (Atkinson 2007; Gould 2020; Urquhart 2018), although data from a single study only remained in the analysis (MD ‐8.40, 95% CI ‐16.69 to ‐0.11; 1 study, 78 participants).

There were insufficient data to explore the effect of dose as a possible effect modifier for our planned subgroup analysis because Atkinson 2007 and Gould 2020 (three of five comparisons from this analysis) randomised participants to target desipramine serum concentrations rather than fixed doses. Data were also insufficient to perform the planned subgroup analysis for symptom duration.

Mid‐term follow‐up (> 4 to 12 months)

One study assessed the effect of TCAs on low back pain intensity compared with placebo at mid‐term follow‐up (Urquhart 2018). Our analysis showed a small between‐group difference in pain intensity in favour of TCAs (MD ‐8.20, 95% CI ‐16.28 to ‐0.12; 1 study, 146 participants; Analysis 1.3). Confidence intervals were compatible with both a moderate effect and a trivial effect. We judged the certainty of evidence as low, downgraded once for risk of bias (high risk of attrition bias), and once for imprecision.

1.3 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 3: Non‐specific low back pain: pain intensity (0 to 100) at mid‐term follow‐up (> 16 to 52 weeks)

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Disability

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks; main comparison)

Five studies assessed the effect of TCAs on low back‐specific disability compared with placebo in the short term in participants with low back pain (Alcoff 1982; Atkinson 2007; Gould 2020; Pheasant 1983; Urquhart 2018). We were unable to include outcome data from two studies in our analysis: Alcoff 1982 measured disability on an ordinal scale but did not report a point estimate or associated measure of variance, and Pheasant 1983 did not report outcome data by cross‐over phase; these data could not be obtained from study authors. The remaining studies measured disability using the RMDQ. Measures of variance were not reported for Atkinson 2007 so we calculated standard deviations from the reported P value using the RevMan calculator. Our meta‐analysis demonstrated a small between‐group difference in disability in favour of TCAs (MD ‐1.76, 95% CI ‐2.70 to ‐0.82; I² = 0; 3 studies, 330 participants; Analysis 1.6; Figure). We judged the certainty of evidence as moderate, downgraded once for risk of bias (high risk of attrition, reporting, and other bias).

1.6 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 6: Non‐specific low back pain: disability (RMDQ, 0 to 24) at short‐term follow‐up (> 4 to 16 weeks): TCA

TCAs versus placebo for non‐specific low back pain disability (RMDQ, 0 to 24) at short‐term follow‐up (> 4 to 16 weeks).

Because Atkinson 2007 and Gould 2020 (two of three comparisons from this analysis) randomised participants to target desipramine concentrations, there were incomplete data to explore the effect of dose as a possible effect modifier via subgroup analysis. Data were insufficient to perform the planned subgroup analysis for symptom duration.

Mid‐term follow‐up (> 4 to 12 months)

One study assessed the effect of TCAs on low back‐specific disability compared with placebo at mid‐term follow‐up in participants with low back pain (Urquhart 2018), measured using the RMDQ. Our analysis showed no evidence of a difference in disability between TCAs and placebo, with confidence intervals spanning both a small effect and no effect (MD ‐1.20, 95% CI ‐2.73 to 0.33; 1 study, 146 participants; Analysis 1.7). We judged the certainty of evidence as low, downgraded once for risk of bias (high risk of attrition bias), and once for imprecision.

1.7 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 7: Non‐specific low back pain: disability (RMDQ, 0 to 24) at mid‐term follow‐up (> 16 to 52 weeks): TCA

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Secondary outcomes

Depressive symptoms

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks)

Three studies assessed the effect of TCAs on depressive symptoms compared with placebo in the short term (Alcoff 1982; Atkinson 1998; Urquhart 2018), measured using the Beck Depression Inventory. We were unable to include outcome data from Alcoff 1982 in our analysis because a point estimate and associated measure of variance were not reported, and these could not be obtained from study authors. Our meta‐analysis showed a trivial between‐group difference in depressive symptoms in favour of TCAs (MD ‐1.59, 95% CI ‐2.95 to ‐0.23; I² = 0%; 2 studies, 194 participants; Analysis 1.10). We judged the certainty of evidence as moderate, downgraded once for risk of bias (high risk of attrition bias).

1.10 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 10: Non‐specific low back pain: depressive symptoms (BDI, 0 to 63) at short‐term follow‐up (> 4 to 16 weeks): TCA

Mid‐term follow‐up (> 4 to 12 months)

One study assessed the effect of TCAs on depressive symptoms compared with placebo at mid‐term follow‐up (Urquhart 2018), measured using the Beck Depression Inventory. Our analysis showed no evidence of a difference in depressive symptoms between TCAs and placebo (MD ‐0.93, 95% CI ‐3.44 to 1.58; 1 study, 118 participants; Analysis 1.11). We judged the certainty of evidence as moderate, downgraded once for risk of bias (high risk of attrition bias).

1.11 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 11: Non‐specific low back pain: depressive symptoms (BDI, 0 to 63) at mid‐term follow‐up (> 16 to 52 weeks): TCA

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Health‐related quality of life

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks)

Two studies assessed the effect of TCAs on HRQoL compared with placebo in the short term (Atkinson 1998; Urquhart 2018). HRQoL was measured using the Quality of Well‐Being Scale in Atkinson 1998, and the EQ‐5D visual analogue scale in Urquhart 2018, so we used the standardised mean difference (SMD) as our effect measure. Our meta‐analysis showed no evidence of a difference in HRQoL between TCAs and placebo (SMD 0.07, 95% CI ‐0.43 to 0.58; 2 studies, 194 participants). Confidence intervals were compatible with a small effect and no effect at all. Heterogeneity was minimal and mostly due to variation in true effects (Tau² = 0.09; Chi² = 3.04, df = 1 (P = 0.08); I² = 67%; Analysis 1.14). We judged the certainty of evidence as low, downgraded once for risk of bias (high risk of attrition bias), and once for imprecision.

1.14 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 14: Non‐specific low back pain: HRQoL (SMD) at short‐term follow‐up (> 4 to 16 weeks): TCA

Mid‐term follow‐up (> 4 to 12 months)

One study assessed the effect of TCAs on HRQoL compared with placebo at mid‐term follow‐up (Urquhart 2018), measured using the EQ‐5D visual analogue scale (0 to 100; higher scores indicate greater HRQoL). Our analysis showed no evidence of a difference in HRQoL between TCAs and placebo (MD 3.80, 95% CI ‐1.77 to 9.37; 1 study, 118 participants; Analysis 1.15). Confidence intervals were compatible with both an effect and no effect at all. We judged the certainty of evidence as low, downgraded once for risk of bias (high risk of attrition bias), and once for imprecision.

1.15 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 15: Non‐specific low back pain: HRQoL (EQ‐5D, 0 to 100 VAS) at mid‐term follow‐up (> 16 to 52 weeks): TCA

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Tetracyclic antidepressants (TeCAs) versus placebo

Primary outcomes

Pain intensity

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks; main comparison)

One study assessed the effect of TeCAs on low back pain intensity compared with placebo in the short term (Atkinson 1999). Our analysis showed no evidence of a difference in pain intensity between TeCAs and placebo, with confidence intervals spanning both a moderate effect and no effect at all (MD ‐4.50, 95% CI ‐17.59 to 8.59; 1 study, 52 participants; Analysis 1.2; Figure). We judged the certainty of evidence as very low, downgraded twice for risk of bias (high risk of performance, detection, attrition, and reporting bias), and once for imprecision.

Data were insufficient to perform the planned subgroup analyses.

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Disability

No studies measured or reported this outcome.

Secondary outcomes

Depressive symptoms

No studies measured or reported this outcome.

Health‐related quality of life

No studies measured or reported this outcome.

Other antidepressants versus placebo

Primary outcomes

Pain intensity

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks; main comparison)

Two studies assessed the effect of other antidepressants on low back pain intensity compared with placebo in the short term (Goodkin 1990; Katz 2005). We were unable to include outcome data from Katz 2005 in our analysis because they were not presented by cross‐over treatment phase and could not be obtained from study authors. Our analysis showed no evidence of a difference in pain intensity between other antidepressants and placebo, with confidence intervals spanning both a large effect and no effect at all (MD 5.40, 95% CI ‐23.08 to 12.28; 1 study, 39 participants; Analysis 1.2; Figure). We judged the certainty of evidence as very low, downgraded once for risk of bias (high risk of attrition bias and unclear risk of selection bias) and twice for imprecision.

Data were insufficient to perform the planned subgroup analyses.

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Disability

No studies measured or reported this outcome.

Secondary outcomes

Depressive symptoms

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks)

One study assessed the effect of other antidepressants on depressive symptoms compared with placebo in the short term (Goodkin 1990), measured using the Beck Depression Inventory (0 to 63; higher scores indicate greater depressive symptoms). Our analysis showed no evidence of a difference in depressive symptoms between other antidepressants and placebo (MD 2.21, 95% CI ‐4.00 to 8.42; 1 study, 40 participants; Analysis 1.12). We judged the certainty of evidence as low, downgraded once for risk of bias (high risk of attrition bias and unclear selection and reporting bias), and once for imprecision.

1.12 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 12: Non‐specific low back pain: depressive symptoms (BDI, 0 to 63) at short‐term follow‐up (> 4 to 16 weeks): other antidepressants

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Health‐related quality of life

No studies measured or reported this outcome.

Spine‐related leg pain

Serotonin and norepinephrine reuptake inhibitors (SNRIs) versus placebo

Primary outcomes

Pain intensity

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks; main comparison)

Two studies assessed the effect of SNRIs on spine‐related leg pain intensity compared with placebo in the short term (Marks 2014; NCT01225068). Our main analysis showed a large between‐group difference in pain intensity in favour of SNRIs (MD ‐46.10, 95% CI ‐89.29 to ‐2.91; 1 study, 11 participants; Analysis 1.17; Figure). While the point estimate exceeded our threshold for a large effect, the confidence interval was consistent with both a large effect and a trivial effect. We judged the certainty of evidence as very low, downgraded twice for risk of bias (high risk of attrition and reporting bias and unclear selection bias), and once for imprecision.

1.17 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 17: Spine‐related leg pain: pain intensity (0 to 100) at short‐term follow‐up (> 4 to 16 weeks)

All antidepressants versus placebo for spine‐related leg pain (0 to 100) at short‐term follow‐up (> 4 to 16 weeks).

Our sensitivity analysis showed the estimate was impacted substantially by the addition of data from NCT01225068 (MD ‐16.13, 95% CI ‐67.17 to 34.91; 2 studies, 46 participants). The point estimate was reduced to a moderate effect, and while the upper bound of the confidence interval remained consistent with a large effect, the lower bound spanned no effect at all. Heterogeneity was high, due to both the variation of true effects and variation due to sampling error (Tau² = 1091.02; Chi² = 4.73, df = 1 (P = 0.03); I² = 79%).

Data were insufficient to perform the planned subgroup analyses.

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Disability

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks)

One study assessed the effect of SNRIs on low back‐specific disability compared with placebo in the short term in participants with spine‐related leg pain (Marks 2014), measured using the ODI (0 to 100; higher scores indicate greater low back‐specific disability). Our analysis showed no evidence of a difference in disability between SNRIs and placebo (MD ‐4.40, 95% CI ‐20.25 to 11.45; 1 study, 11 participants; Analysis 1.20; Figure). Confidence intervals were compatible with both a large effect and no effect at all. We judged the certainty of evidence as very low, downgraded twice for risk of bias (high risk of attrition and reporting bias and unclear selection bias), and once for imprecision.

1.20 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 20: Spine‐related leg pain: disability (ODI, 0 to 100) at short‐term follow‐up (> 4 to 16 weeks)

All antidepressants versus placebo for low back‐specific disability (ODI, 0 to 100) in patients with spine‐related leg pain at short‐term follow‐up (> 4 to 16 weeks).

Data were insufficient to perform the planned subgroup analyses.

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Secondary outcomes

Depressive symptoms

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks)

One study assessed the effect of SNRIs on depressive symptoms compared with placebo in the short term in participants with spine‐related leg pain (Marks 2014), measured using the Beck Depression Inventory. Our analysis showed no evidence of a difference in depressive symptoms between SNRIs and placebo (MD ‐1.70, 95% CI ‐13.40 to 10.08; 1 study, 11 participants; Analysis 1.23). Confidence intervals were compatible with an effect and no effect at all. We judged the certainty of evidence as very low, downgraded twice for risk of bias (high risk of attrition and reporting bias and unclear selection and other bias), and once for imprecision.

1.23 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 23: Spine‐related leg pain: depressive symptoms (BDI, 0 to 63) at short‐term follow‐up (> 4 to 16 weeks)

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Health‐related quality of life

Immediate‐term follow‐up (≤ 4 weeks)

No studies measured or reported this outcome.

Short‐term follow‐up (> 4 to 16 weeks)

One study assessed the effect of SNRIs on HRQoL compared with placebo in the short term in participants with spine‐related leg pain (Marks 2014), measured using the 36‐Item Short Form Health Survey (0 to 100; higher scores indicate greater health‐related quality of life). Our analysis showed no evidence of a difference in HRQoL between SNRIs and placebo (MD ‐2.70, 95% CI ‐12.42 to 7.02; 1 study, 11 participants; Analysis 1.24). Confidence intervals were compatible with an effect and no effect at all. We judged the certainty of evidence as very low, downgraded twice for risk of bias (high risk of attrition and reporting bias, and unclear selection and other bias), and once for imprecision.

1.24 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 24: Spine‐related leg pain: HRQoL (SF‐36, 0 to 100) at short‐term follow‐up (> 4 to 16 weeks)

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Selective serotonin reuptake inhibitors (SSRIs) versus placebo

No studies comparing SSRIs with placebo in participants with spine‐related leg pain were included.

Tricyclic antidepressants (TCAs) versus placebo

Primary outcomes

Pain intensity

Immediate‐term follow‐up (≤ 4 weeks)

Three studies assessed the effect of TCAs on spine‐related leg pain intensity compared with placebo in the immediate term (Afilal 2020; Pirbudak 2003; Vanelderen 2015). Vanelderen 2015 reported within‐arm change from baseline and a 95% confidence interval for the amitriptyline arm only, so we used the baseline placebo value as the control group value and imputed highest baseline standard error (0.4) as the outcome measure of variance for each arm. Standard deviations reported for Afilal 2020 were implausibly small, so we assumed these were standard errors reported incorrectly as standard deviations. We took a conservative approach and entered these as standard errors in the analysis. Our meta‐analysis showed a moderate between‐group difference in pain intensity in favour of TCAs (MD ‐15.18, 95% CI ‐30.97 to 0.61; 3 studies, 161 participants; Analysis 1.16). Confidence intervals spanned both a large effect and no effect. Heterogeneity was high, due to both the variation of true effects and variation due to sampling error (Tau² = 169.25; Chi² = 15.87, df = 2 (P = 0.0004); I² = 87%). The certainty of evidence was very low, downgraded once for risk of bias (high risk of reporting bias, unclear risk of selection bias), once for inconsistency, and once for imprecision.

1.16 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 16: Spine‐related leg pain: pain intensity (0 to 100) at immediate‐term follow‐up (≤ 4 weeks)

Short‐term follow‐up (> 4 to 16 weeks; main comparison)

Two studies assessed the effect of TCAs on spine‐related leg pain intensity compared with placebo in the short term (Khoromi 2007; Pirbudak 2003). We were unable to include outcome data from Khoromi 2007 in the analysis because they were not presented by cross‐over treatment phase and could not be obtained from study authors. Our analysis showed a large between‐group difference in pain intensity in favour of TCAs (MD ‐23.00, 95% CI ‐32.12 to ‐13.88; 1 study, 60 participants; Analysis 1.17; Figure). The point estimate exceeded our threshold for a large effect, and the lower bound of the confidence interval was consistent with a moderate effect. We judged the certainty of evidence as low, downgraded twice for risk of bias (high risk of reporting bias, and unclear risk of selection, performance, detection, attrition, and other bias).

Data were insufficient to perform the planned subgroup analyses.

Mid‐term follow‐up (> 4 to 12 months)

One study assessed the effect of TCAs on spine‐related leg pain intensity compared with placebo at mid‐term follow‐up (Pirbudak 2003). Our analysis showed a large between‐group difference in pain intensity in favour of TCAs (MD ‐27.00, 95% CI 36.12 to ‐17.88; 1 study, 60 participants; Analysis 1.18), with the lower bound of the confidence interval spanning a moderate effect. We judged the certainty of evidence as low, downgraded twice for risk of bias (high risk of reporting bias, and unclear risk of selection, performance, detection, attrition and other bias).

1.18 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 18: Spine‐related leg pain: pain intensity (0 to 100) at mid‐term follow‐up (> 16 to 52 weeks)

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Disability

Immediate‐term follow‐up (≤ 4 weeks)

Two studies assessed the effect of TCAs on low back‐specific disability compared with placebo in the immediate term in participants with spine‐related leg pain (Afilal 2020; Pirbudak 2003). Disability was measured using the RMDQ in Afilal 2020 and the ODI in Pirbudak 2003, so we used the SMD as our effect measure. Our meta‐analysis showed a small between‐group difference in disability in favour of TCAs (SMD ‐0.60, 95% CI ‐1.09 to ‐0.10; 2 studies, 122 participants; Analysis 1.19). Confidence intervals were compatible with a large effect and a trivial effect. Heterogeneity was minimal (Tau² = 0.06; Chi² = 1.85, df = 1 (P = 0.17); I² = 46%). We judged the certainty of evidence as low, downgraded twice for risk of bias (high risk of reporting bias and unclear risk of selection, performance, and detection bias).

1.19 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 19: Spine‐related leg pain: disability (SMD) at immediate‐term follow‐up (≤ 4 weeks)

Short‐term follow‐up (> 4 to 16 weeks; main comparison)

Two studies assessed the effect of TCAs on low back‐specific disability compared with placebo in the short term in participants with spine‐related leg pain (Khoromi 2007; Pirbudak 2003), measured using the ODI. We were unable to include outcome data from Khoromi 2007 in our analysis because they were not presented by cross‐over treatment phase and could not be obtained from study authors. Our analysis showed a moderate between‐group difference in disability in favour of TCAs (MD ‐13.00, 95% CI ‐19.42 to ‐6.58; 1 study, 60 participants; Analysis 1.20; Figure). Confidence intervals were compatible with both a moderate effect and a small effect. We judged the certainty of evidence as low, downgraded twice for risk of bias (high risk of reporting bias, and unclear selection, performance, attrition, and other bias).

Data were insufficient to perform the planned subgroup analyses.

Mid‐term follow‐up (> 4 to 12 months)

One study assessed the effect of TCAs on low back‐specific disability compared with placebo at mid‐term follow‐up in participants with spine‐related leg pain (Pirbudak 2003), measured using the ODI. Our analysis showed a large between‐group difference in pain intensity in favour of TCAs (MD ‐20.00, 95% CI ‐27.74 to ‐12.26; 1 study, 60 participants; Analysis 1.21). Confidence intervals were compatible with a large effect and a moderate effect. We judged the certainty of evidence as low, downgraded twice for risk of bias (high risk of reporting bias and unclear risk of selection, performance, detection, attrition, and other bias).

1.21 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 21: Spine‐related leg pain: disability (ODI, 0 to 100) at mid‐term follow‐up (> 16 to 52 weeks)

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Secondary outcomes

Depressive symptoms

Immediate‐term follow‐up (≤ 4 weeks)

One study assessed the effect of TCAs on depressive symptoms compared with placebo in the immediate term in participants with spine‐related leg pain (Afilal 2020), measured using the Hospital Anxiety and Depression Scale Depression subscale (0 to 21; higher scores indicate greater depressive symptoms). Our analysis showed no evidence of a difference in depressive symptoms between TCAs and placebo (MD 0.23, 95% CI ‐1.51 to 1.97; 1 study, 62 participants; Analysis 1.22). Confidence intervals were compatible with an effect and no effect at all. We judged the certainty of evidence as low, downgraded once for risk of bias (high risk of reporting bias and unclear risk of selection bias and other bias), and once for imprecision.

1.22 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 22: Spine‐related leg pain: depressive symptoms (HADS‐D, 0 to 21) at immediate‐term follow‐up (≤ 4 weeks)

Short‐term follow‐up (> 4 to 16 weeks; main comparison)

No studies measured or reported this outcome.

Mid‐term follow‐up (> 4 to 12 months)

No studies measured or reported this outcome.

Long‐term follow‐up (> 12 months)

No studies measured or reported this outcome.

Health‐related quality of life

No studies measured or reported this outcome.

Tetracyclic antidepressants (TeCAs) versus placebo

No studies comparing TeCAs with placebo in participants with spine‐related leg pain were included.

Other antidepressants versus placebo

No studies comparing other antidepressants with placebo in participants with spine‐related leg pain were included.

Non‐specific low back pain and spine‐related leg pain

Serotonin and norepinephrine reuptake inhibitors (SNRIs) versus placebo

Primary outcomes

Total adverse events

Eight studies assessed the number of participants experiencing adverse events with SNRI administration compared with placebo (Johnson 2011; Konno 2016; Marks 2014; NCT01225068; Schukro 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b). We were unable to include outcome data from Johnson 2011 and Schukro 2016 because they did not present outcome data by cross‐over treatment phase and these could not be obtained from study authors. Our meta‐analysis demonstrated an increased risk of adverse events with SNRIs (risk ratio (RR) 1.17, 95% CI 1.07 to 1.27; I² = 0%; 5 studies, 1510 participants; Analysis 1.25; Figure). We judged the certainty of evidence as moderate, downgraded once for risk of bias (high risk of attrition bias and unclear risk of selection bias).

1.25 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 25: Non‐specific low back pain and spine‐related leg pain: total adverse events

All antidepressants versus placebo for low back pain and spine‐related leg pain: total adverse events at short‐term follow‐up (> 4 to 16 weeks).

Our sensitivity analysis showed the estimate was robust to the removal of studies with unclear risk of selection bias (Marks 2014; Skljarevski 2010a; Skljarevski 2010b) (RR 1.18, 95% CI 1.06 to 1.31; I² = 0%; 2 studies, 862 participants), unclear detection bias (Skljarevski 2010a; Skljarevski 2010b) (RR 1.18, 95% CI 1.06 to 1.31; I² = 0%; 3 studies, 873 participants), and the addition of data from NCT01225068 (RR 1.17, 95% CI 1.08 to 1.27; I² = 0%; 6 studies, 1550 participants).

While minimal heterogeneity was observed in the main analysis, data were available to explore dose as a possible effect modifier via subgroup analysis. This analysis did not show visible or statistical differences in effects between below (< 40 mg/day), within (40 mg to 60 mg/day), or above (< 60 mg/day) the standard duloxetine dosing range (test for subgroup differences: Chi² = 0.13, df = 2 (P = 0.94); I² = 0%). The observational nature of this analysis and small number of studies in each subgroup means the possibility of important differences cannot be excluded. Data were insufficient to perform the planned subgroup analysis for symptom duration.

Secondary outcomes

Serious adverse events

Six studies assessed the number of participants experiencing serious adverse events with SNRI administration compared with placebo (Johnson 2011; Konno 2016; Marks 2014; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b). We were unable to include outcome data from Johnson 2011 because they did not present outcome data by cross‐over treatment phase and these could not be obtained from study authors. Our meta‐analysis showed no clear evidence of a difference in serious adverse events between SNRIs and placebo (Peto odds ratio (OR) 1.75, 95% CI 0.79 to 3.89; 5 studies, 1510 participants; Analysis 1.26). Confidence intervals were compatible with both large risk and a decreased risk, though heterogeneity was minimal (Chi² = 6.03, df = 5 (P = 0.30); I² = 17%). We judged the certainty of evidence as very low, downgraded once for risk of bias (high risk of attrition bias and unclear risk of selection, performance, detection, and other bias), once for inconsistency (incomplete overlap of confidence intervals), and twice for imprecision.

1.26 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 26: Non‐specific low back pain and spine‐related leg pain: serious adverse events

Withdrawals due to adverse events

Six studies assessed the number of participants who discontinued treatment due to adverse events with SNRI administration compared with placebo (Konno 2016; Marks 2014; Schukro 2016; Skljarevski 2009a; Skljarevski 2010a; Skljarevski 2010b). Our meta‐analysis demonstrated an increased risk of withdrawals due to adverse events with SNRIs (RR 2.36, 95% CI 1.66 to 3.36; I² = 0%; 6 studies, 1543 participants; Analysis 1.27). We judged the certainty of evidence as moderate, downgraded once for risk of bias (high risk of attrition and reporting bias, and unclear risk of selection, performance, detection, and other bias).

1.27 — Comparison 1: All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain, Outcome 27: Non‐specific low back pain and spine‐related leg pain: withdrawals due to adverse events

Selective serotonin reuptake inhibitors (SSRIs) versus placebo

Primary outcomes

Total adverse events

Two studies assessed the number of participants experiencing adverse events with SSRI administration compared with placebo (Atkinson 1999; Atkinson 2007). Our meta‐analysis showed no evidence of a difference in adverse events between SSRIs and placebo (RR 1.83, 95% CI 0.14 to 24.19; 2 studies, 107 participants; Analysis 1.25; Figure). Confidence intervals were compatible with both a high risk and decreased risk. Heterogeneity was high, due to variation of true effects (Tau² = 3.32; Chi² = 21.51, df = 1 (P < 0.001); I² = 95%). We judged the certainty of evidence as very low, downgraded twice for risk of bias (high risk of performance, detection, attrition, and reporting bias), once for inconsistency, and once for imprecision.

Our sensitivity analysis showed the estimate was impacted by the removal of one study at high risk of detection bias (Atkinson 1999) (RR 3.78, 95% CI 1.25 to 11.43; 1 study, 53 participants). Both the upper and lower bounds of the confidence interval were consistent with an increased risk of adverse events; however, the effect estimate remained imprecise.

Because Atkinson 2007 (one of two comparisons from this analysis) randomised participants to target fluoxetine concentrations, there were insufficient data to explore the effect of dose as a possible effect modifier via subgroup analysis. Data were also insufficient to perform the planned subgroup analysis for symptom duration.

Secondary outcomes

Serious adverse events

No studies measured or reported this outcome.

Withdrawals due to adverse events

Two studies assessed the number of participants who discontinued treatment due to adverse events with SSRI administration compared with placebo (Atkinson 1999; Atkinson 2007). Our meta‐analysis showed no clear evidence of a difference in withdrawals due to adverse events between SSRIs and placebo (RR 3.54, 95% CI 0.78 to 16.02; I² = 0%; 2 studies, 139 participants; Analysis 1.27). Confidence intervals were compatible with a large risk and a decreased risk. We judged the certainty of evidence as very low, downgraded twice for risk of bias (high risk of performance, detection, attrition, and reporting bias), and once for imprecision.

Tricyclic antidepressants (TCAs) versus placebo

Primary outcomes

Total adverse events

Nine studies assessed the number of participants experiencing adverse events with TCA administration compared with placebo (Afilal 2020; Atkinson 1998; Atkinson 2007; Gould 2020; Khoromi 2007; Kurniawati 2020; Treves 1991; Urquhart 2018; Vanelderen 2015). We were unable to include outcome data from two studies in our analysis: Khoromi 2007 did not present outcome data by cross‐over treatment phase and Treves 1991 did not report the number of participants experiencing adverse events per treatment group; these could not be obtained from study authors. Our meta‐analysis showed no evidence of a difference in adverse events between TCAs and placebo (RR 1.76, 95% CI 0.79 to 3.90; 7 studies, 474 participants; Analysis 1.25; Figure). Confidence intervals were compatible with both an increased risk and decreased risk. Heterogeneity was high, due to variation of true effects (Tau² = 0.99; Chi² = 105.26, df = 7 (P < 0.001); I² = 93%). We judged the certainty of evidence as low, downgraded once for risk of bias (high risk of attrition, reporting, and other bias), and once for imprecision.

Our sensitivity analysis showed the estimate was robust to the removal of two studies with unclear risk of selection bias (Afilal 2020; Kurniawati 2020) (RR 1.57, 95% CI 0.67 to 3.65; I² = 92%; 5 studies, 349 participants), detection bias (Kurniawati 2020) (RR 2.05, 95% CI 0.68 to 6.19; I²; 6 studies, 411 participants), and studies using active placebo (Atkinson 2007; Gould 2020; Urquhart 2018) (RR 2.34, 95% CI 0.46 to 11.80; I² = 96%; 4 studies, 222 participants), although substantial imprecision was introduced.

Because Atkinson 2007 and Gould 2020 (two of seven comparisons from this analysis) randomised participants to target desipramine concentrations, there were incomplete data to explore dose as a possible effect modifier via subgroup analysis. Data were also insufficient to perform the planned subgroup analysis for symptom duration.

Secondary outcomes

Serious adverse events

One study assessed the number of participants experiencing serious adverse events with TCA administration compared with placebo (Gould 2020). Our analysis showed no clear evidence of a difference in serious adverse events between TCAs and placebo (Peto OR 6.64, 95% CI 0.41 to 106.72; I² = 0%; 1 study, 142 participants; Analysis 1.26). Confidence intervals were compatible with both a large risk and decreased risk. We judged the certainty of evidence as very low, downgraded twice for risk of bias (high risk of attrition and other bias), and twice for imprecision.

Withdrawals due to adverse events

Eight studies assessed the number of participants who discontinued treatment due to adverse events with TCA administration compared with placebo (Afilal 2020; Alcoff 1982; Atkinson 1998; Atkinson 2007; Gould 2020; Khoromi 2007; Urquhart 2018; Vanelderen 2015). We were unable to include outcome data from Khoromi 2007 because they did not present data by cross‐over treatment phase and these could not be obtained from study authors. Our meta‐analysis showed no conclusive evidence of a difference in withdrawals due to adverse events between TCAs and placebo (RR 1.81, 95% CI 0.89 to 3.69; 8 studies, 665 participants; Analysis 1.27). Confidence intervals were compatible with a large risk and a decreased risk, though heterogeneity was minimal (Tau² = 0.19; Chi² = 8.51, df = 7 (P = 0.29); I² = 18%). We judged the certainty of evidence as very low, downgraded twice for risk of bias (high risk of attrition, reporting, and other bias, unclear risk of selection and other bias), and once for imprecision.

Tetracyclic antidepressants (TeCAs) versus placebo

Primary outcomes

Total adverse events

One study assessed the number of participants experiencing adverse events with TeCA administration compared with placebo (Atkinson 1999). Our analysis showed no evidence of a difference in adverse events between TeCAs and placebo (RR 0.93, 95% CI 0.79 to 1.09; 1 study, 52 participants; Analysis 1.25; Figure). Confidence intervals were compatible with both an increased risk and decreased risk. We judged the certainty of evidence as very low, downgraded twice for risk of bias (high risk of performance, detection, attrition, and reporting bias), and once for imprecision.

Data were insufficient to perform the planned subgroup analyses.

Secondary outcomes

Serious adverse events

No studies measured or reported this outcome.

Withdrawals due to adverse events

One study assessed the number of participants who discontinued treatment due to adverse events with TeCA administration compared with placebo (Atkinson 1999). Our analysis showed an increased risk of withdrawals due to adverse events with TeCA administration (RR 10.91, 95% CI 1.48 to 80.66; 1 study, 69 participants; Analysis 1.27). Confidence intervals were compatible with a large risk and a small risk. We judged the certainty of evidence as very low, downgraded twice for risk of bias (high risk of performance, detection, attrition, and reporting bias), and once for imprecision.

Other antidepressants versus placebo

Primary outcomes

Total adverse events

No studies measured or reported this outcome.

Secondary outcomes

Serious adverse events

One study assessed the number of participants experiencing serious adverse events with other antidepressants compared with placebo (Goodkin 1990). Our analysis showed no clear evidence of a difference in serious adverse events between other antidepressants and placebo (Peto OR 0.90, 95% CI 0.16 to 4.96; 1 study, 42 participants; Analysis 1.26). Confidence intervals were compatible with a large risk and a decreased risk. We judged the certainty of evidence as very low, downgraded once for risk of bias (high risk of attrition bias and unclear risk of selection and reporting bias), and twice for imprecision.

Withdrawals due to adverse events

Two studies assessed the number of participants who discontinued treatment due to adverse events with other antidepressant administration compared with placebo (Goodkin 1990; Katz 2005). For Katz 2005, we included data only from the first phase of the cross‐over study. Our meta‐analysis showed no clear evidence of a difference in withdrawals due to adverse events between other antidepressants and placebo (RR 3.45, 95% CI 0.59 to 20.10; 2 studies, 96 participants; Analysis 1.27). Confidence intervals were compatible with a large risk and a decreased risk. We judged the certainty of evidence as low, downgraded once for risk of bias (high risk of attrition bias and unclear risk of selection and reporting bias), and once for imprecision.

Discussion

Summary of main results

We included 26 studies that randomised 2932 participants with non‐specific low back pain or spine‐related leg pain. Nine studies, comprising 1914 participants with non‐specific back pain, and seven studies, comprising 329 participants with spine‐related leg pain, were new to this update. Included studies evaluated serotonin and norepinephrine reuptake inhibitors (SNRIs), selective serotonin reuptake inhibitors (SSRIs), tricyclic antidepressants (TCAs), tetracyclic antidepressants (TeCAs), and other antidepressants, all compared with an inert or active placebo, and mainly administered orally. Most studies assessed outcomes at short‐term follow‐up (> 4 to 16 weeks), our primary time point of interest.

For non‐specific low back pain at short‐term follow‐up, we found moderate‐certainty evidence from four studies that SNRIs probably have small effects on pain intensity, and trivial effects on disability (Table). There was moderate‐certainty evidence from four studies that TCAs probably have little to no effect on pain intensity, but probably have a small effect on disability (Table). We found inconclusive evidence for the effects of other antidepressants, SSRIs, and TeCAs on non‐specific low back pain, owing to small studies with imprecise estimates, and low‐ to very low‐certainty evidence (Table; Table; Table).

For spine‐related leg pain at short‐term follow‐up, there were no studies evaluating the effects of SSRIs, TeCAs, or other antidepressants. We found evidence of positive effects for SNRIs on pain intensity (Table), and TCAs on pain intensity and disability (Table), although the very small sample sizes and low‐ to very low‐certainty evidence suggests that the true effects are likely to be substantially different from our estimates.

We evaluated harm outcomes in participants with non‐specific low back pain and spine‐related leg pain collectively. We found moderate‐certainty evidence from five studies that SNRIs are probably associated with an increased risk of experiencing any adverse event (Table). The risk was 17% greater for SNRIs than the risk with placebo (95 more adverse events per 1000 participants), though it may have been as low as 7% or as high as 27% (39 to 151 more adverse events per 1000 participants). The risk of experiencing an adverse event following SSRI, TCA, and TeCA administration is unclear, due to imprecise estimates and very low‐ and low‐certainty evidence, respectively (Table; Table; Table). No studies assessed the number of total adverse events for other antidepressants. We evaluated the odds of experiencing a serious adverse event for SNRIs, TCAs, and other antidepressants, but imprecise effect estimates and low‐ to very low‐certainty evidence prevented us from drawing conclusions for this outcome.

Overall completeness and applicability of evidence

Our review considered studies evaluating any antidepressant in adults with non‐specific low back pain and spine‐related leg pain. Despite our broad study eligibility criteria, there were several factors that limited the completeness and applicability of the included evidence.

Population

Participant sex/gender was mostly balanced across studies. The mean age of participants in the included studies ranged from 27 to 59 years, and more than half the studies had an upper age limit to restrict participation. Study participants were therefore younger and more frequently male than the general low back pain population, given that low back pain's prevalence increases with age, and is higher in females than in males (Ferreira 2023). Exclusion criteria for comorbid health conditions were extensive, meaning that study participants may have had better overall health than those who present clinically.

Studies were predominantly conducted in high‐income countries, with just two small studies in low‐income countries, and two centres from multinational studies in middle‐income countries. Reporting of characteristics that stratify health opportunities (such as race or ethnicity, occupation, or socioeconomic status) was limited. The 10 studies that collected race or ethnicity data reported that most participants were white. The high concentration of white participants from high‐income countries could limit the generalisability of results, as both pain and treatments may be influenced by cultural, social, and economic factors (James 2023).

The criteria used to diagnose study participants varied. For studies of spine‐related leg pain, a lack of uniform inclusion criteria probably introduced a heterogeneous participant population who may or may not have met a neuropathic pain classification (Finnerup 2016). Almost three‐quarters of all studies excluded people experiencing pain for less than three months. Our findings are therefore most applicable to participants with persistent symptoms. Although we planned to evaluate whether antidepressant effects differed for acute and chronic symptoms, a lack of data precluded these analyses. Average baseline pain intensity ranged from 4 to 6 points out of 10 for participants with non‐specific low back pain and 6 to 8 points out of 10 for spine‐related leg pain, which may be representative of clinical populations. Participants with clinical depression or significant depressive symptoms were excluded from participation in around 60% of included studies, suggesting that our findings generally apply to people without severe comorbid depression.

Interventions

Studies that tested SNRI and TCA antidepressant classes comprised 31% and 46% of included evidence, respectively. There was limited evidence for SSRIs, TeCAs, and other antidepressants. Antidepressants were taken orally in 93% of studies, reflecting the most common route of administration. Dosing strategies varied, with around 50% of studies testing doses above or below standard ranges. Data were available to explore SNRI dose as a treatment effect modifier, with no differences between below, within, or above standard SNRI (duloxetine) dosing ranges observed for main low back pain outcomes, though we are unable to rule out true differences due to the low statistical power of this analysis. Most studies had a treatment duration of four weeks or greater, in line with antidepressant prescribing guidance for depression (NICE 2022). Co‐interventions were permitted by most studies, including continuation of simple analgesics, or episodic use of simple or opioid analgesics. However, reporting on their use was limited, making it difficult to establish whether treatment effects were unbiased (Abdel Shaheed 2021).

Comparators

We sought to include studies that compared an antidepressant to placebo, usual care, or no treatment, although only placebo‐controlled trials were identified. Because some antidepressants are associated with specific side effects (e.g. dry mouth, sedation), almost a quarter of studies used active placebos as comparators to maintain participant blinding. We explored whether use of active placebo modified treatment effects by excluding these studies from the main analyses. The effect of TCAs on low back pain was not robust to the removal of studies using active placebo. However, this analysis was limited by the very small number of participants.

Outcomes

While all but one study measured pain intensity, only 50% measured disability or function, and 35% measured health‐related quality of life. The absence of data for these outcomes is noteworthy as they comprise two of four core outcome domains for non‐specific low back pain (Chiarotto 2018). Only five studies provided data for all harm outcomes, and there was a general lack of detail regarding the methods used to capture adverse events. As a consequence, harms may have been under‐reported. Outcomes were measured within the period of antidepressant administration, meaning the duration of beneficial or adverse effects of these medicines remains unclear. Long‐term (> 12 months) follow‐up data were not available for any study.

Certainty of the evidence

We used GRADE to judge the certainty of the evidence. We found no high‐certainty evidence for any comparison, with most evidence rated as low or very low certainty.

We downgraded all evidence for risk of bias. Major sources of bias included inadequate or unclear methods to blind participants, missing outcome data, and selective reporting. We considered detection bias an important threat to the validity of our findings, as knowledge of treatment assignment is likely to influence the self‐reported outcomes of interest to this review. While analyses were robust to the exclusion of studies at unclear or high risk of detection bias, it remains possible that knowledge of assignment influenced treatment outcomes. It is likely that missing outcome data had a substantial impact on our results. We observed many instances where missingness in the outcome was probably related to its true value (e.g. loss to follow‐up due to lack of efficacy or adverse effects), and methods to account for missing data were inappropriate. Imputation using 'last observation carried forward' may have biased effect estimates (away from the null) for several large studies, especially in the likely presence of dropouts due to adverse events (Lachin 2016; Moore 2012). Finally, we judged three‐quarters of studies to be at unclear or high risk of reporting bias because they lacked a study protocol or registration that prespecified the reported outcomes, or failed to report data completely so that they could be entered into meta‐analyses.

We downgraded most analyses for imprecision. Limited numbers of participants and few events led to wide estimates that were compatible with both benefit and harm, or opposing clinical decisions. Estimates for harm outcomes were very imprecise, meaning the true risks of most antidepressant classes are unclear for people with non‐specific low back pain and spine‐related leg pain.

We downgraded just three studies for inconsistency as most analyses showed inconclusive evidence of heterogeneity. This probably resulted from our decision to separate analyses by antidepressant class and clinical condition.

The included studies generally met the population, intervention, comparison, and outcome specified for this review, so we did not find indirectness to be a problem. We considered three comparisons that included data from Dickens 2000 to be indirect because of the high proportion of participants with major depression. Such participants may not represent most people with low back pain.

We did not downgrade any comparison for publication bias. We were unable to formally investigate publication bias using funnel plots or statistical tests because the analyses included fewer than 10 studies. However, we considered the influence of unpublished data by including conference abstracts and searching trial registries. While we did not find evidence of publication bias, we are unable to rule it out. It is worth mentioning that while all SNRI studies for non‐specific low back pain were industry‐sponsored and results were favourable, the effects were small and may be deemed as clinically unimportant. As such, we considered that unpublished negative studies would minimally impact these results.

Potential biases in the review process

We considered a range of biases during the review process and attempted to reduce them in several ways. Updated methods for this version of the review were established a priori, and important differences between the current and previous versions have been highlighted in Differences between protocol and review. We conducted a comprehensive search of three databases and three clinical trial registries to identify eligible studies. While it is possible that some studies were missed, we consider this risk to be minimal.

Antidepressant classes were grouped based on World Health Organization (WHO) Anatomical Therapeutic Chemical (ATC) classification codes. We included studies of bupropion and trazodone, both classified under N06AX 'other antidepressants'. Our analyses of other antidepressants may therefore represent a mechanistically heterogeneous group of medicines.

We prespecified a between‐group difference of 10 points on a 0 to 100 scale as the minimal clinically important effect for pain. During the conduct of the review, we decided not to use this threshold to interpret results because it fails to acknowledge patient preferences, health status, risk of harms, and treatment costs and accessibility (Abdel Shaheed 2023; Hansford 2023).

We planned to investigate important heterogeneity through subgroup analyses, and test the robustness of effects with sensitivity analyses. However, the number of included studies for each comparison was insufficient for these analyses to be meaningful.

One of the included studies was authored by a member of this review team (Urquhart 2018). As such, there may have been a risk of bias in terms of the review and appraisal of this study. We minimised this risk by allocating data extraction and risk of bias and GRADE assessments to members of the review team who were not involved in the conduct of the study (MF, AC).

Agreements and disagreements with other studies or reviews

Our ability to compare our findings with those of the previous version of this review is limited due to the use of updated methodology and study eligibility criteria. Urquhart 2008 did not identify studies of SNRIs, whereas the current review included six studies of the SNRI duloxetine that provided moderate‐certainty evidence of small effects on low back pain and disability. In line with our results, Urquhart 2008 found no evidence that TCAs and SSRIs improved low back pain intensity, although GRADE was not available to judge the certainty of the evidence. The risk of bias assessments differed for the same studies included across both reviews, likely due to the use of different assessment criteria.

The network meta‐analysis from Birkinshaw 2023 demonstrated moderate‐certainty evidence of small to moderate effects of the SNRI duloxetine, and moderate‐certainty evidence of small effects of the SNRI milnacipran, for people with chronic pain. Similarly, our analyses demonstrated moderate‐certainty evidence of benefits for duloxetine for people with non‐specific low back pain at short‐term follow‐up, although we deemed these effects to be small. While we found positive effects for milnacipran in spine‐related leg pain at short‐term follow‐up, the certainty of the evidence was very low. Birkinshaw 2023 also found very low‐certainty evidence for harm outcomes across all antidepressants, inhibiting them from drawing conclusions on potential harms. This is largely in agreement with the findings of our review, except for SNRIs, which were associated with a small risk of adverse events based on moderate‐certainty evidence.

In 2021, two systematic reviews examined the effects of antidepressants on non‐specific low back pain (Ferraro 2021), and non‐specific low back pain and sciatica (Ferreira 2021). Our review identified just two studies not included in those reviews. Despite minor methodological differences (both reviews included data from both phases of cross‐over studies' analyses), the estimates of effects and certainty of evidence ratings were comparable.

Authors' conclusions

Implications for practice.

Moderate‐certainty evidence suggests that for people with non‐specific low back pain, serotonin and norepinephrine reuptake inhibitors (SNRIs) probably have small effects on pain intensity, trivial effects on disability, and are probably associated with adverse effects. Moderate‐certainty evidence suggests that tricyclic antidepressants (TCAs) probably have little to no effect on pain intensity, but probably have a small effect on disability in people with non‐specific low back pain. For both SNRIs and TCAs, effects were observed in people with chronic symptoms (≥ three months) up to 14 weeks following a treatment course of at least seven weeks. It remains unclear whether the duration of symptoms or medicine dose modify the treatment effect. These findings may not apply to older people, or those with significant comorbidities, as these populations were generally excluded from studies. There is insufficient evidence to support or refute the use of selective serotonin reuptake inhibitors (SSRIs), TeCAs (tetracyclic antidepressants), or other antidepressants in non‐specific low back pain, or any antidepressant class in spine‐related leg pain. Critically, these findings do not imply that severely depressed people with non‐specific low back pain or spine‐related leg pain should not be treated with antidepressants. The findings of this review might be considered in future updates to clinical practice guidelines for low back pain and spine‐related leg pain.

Implications for research.

While we can be moderately confident of our estimates for SNRIs and TCAs on pain and disability for non‐specific low back pain in the short term, there is a chance that the true effects are substantially different. As such, there remains a need to conduct placebo‐controlled randomised trials for these drug classes. For spine‐related leg pain, there is a lack of high‐quality studies for any antidepressant class, warranting adequately‐powered, well‐designed trials in this population. Although the true effects of SNRIs and TCAs for spine‐related leg pain are likely to be substantially different from the positive estimates we observed, these classes might be prioritised for future investigations.

Our review highlighted several limitations that must be addressed in future trials. Parallel designs are more robust than cross‐over designs, given the conditions and intervention in question. For example, the variable courses of low back pain and spine‐related leg pain may result in period effects (Kongsted 2015), and carryover effects are possible following antidepressant administration. When cross‐over designs are used, trialists should fully report first phase data to allow their inclusion in meta‐analyses (Higgins 2023).

Careful selection and comprehensive description of trial participants is warranted. For spine‐related leg pain specifically, participants might be selected based on whether they have somatic‐referred pain, or radicular pain with or without radiculopathy (Schmid 2023). It is critical that all future trials include detailed descriptions of participant characteristics, including those that stratify health opportunities (Karran 2024). This will improve our understanding of to whom the trial results apply.

Future trials should measure outcomes included in the core outcome set for low back pain to allow comparison and synthesis (Chiarotto 2018). Consistent time points for outcome assessment (e.g. three, six, nine, and 12 months post‐randomisation) will facilitate this. It is important that follow‐up extends beyond the treatment period, as some antidepressant effects may only become apparent after cessation (e.g. withdrawal symptoms) (Horowitz 2023).

Several measures are needed to reduce attrition bias. To allow unbiased estimation of treatment effects, missing outcome data should be minimised by continued follow‐up of participants who cease treatment due to adverse events or lack of efficacy. Simple imputation methods, such as 'last observation carried forward', are likely to introduce bias and should be avoided (Moore 2012; Sterne 2019). When outcome imputation is necessary, trialists should clearly report their assumptions for the relationship between missingness in the outcome and its true value, and investigate these using sensitivity analyses (Sterne 2019).

Trial reports should adhere to CONSORT guidance to ensure complete, clear, and transparent information on the study's methodology and findings (Moher 2010). Better reporting of additional treatments, advice, or other interventions received by trial participants is necessary to enable assessment of co‐intervention bias (Abdel Shaheed 2021). Critically, antidepressant harms must be measured, analysed, and reported in line with published guidance (Cornelius 2022; Junqueira 2023; Phillips 2022).

There are also several considerations for the planning of future systematic reviews. Individual participant data should be sought to allow more robust investigations of potential effect modifiers than what is possible with aggregate data (Riley 2010). Variables of interest include age, symptom duration, depressive symptoms, and medicine dose. More comprehensive evaluations of the potential harms associated with antidepressants are also needed. Given the difficulty of assessing rare or long‐term events in reviews of randomised trials, there may be value in including evidence from alternate data sources, including observational studies, case reports, and adverse event reporting systems (Peryer 2023; Qureshi 2022). Furthermore, it may be reasonable to conduct reviews that focus only on antidepressant harms, regardless of the health condition, as the adverse effect profile may be sufficiently similar in different populations (Mayo‐Wilson 2023; Peryer 2023).

What's new

Date	Event	Description
10 March 2025	New search has been performed	This review has been updated to include results of a new search on 14 November 2024. The review now includes 26 randomised controlled trials with 2932 participants, a total of 16 new studies and 2243 participants.
10 March 2025	New citation required and conclusions have changed	The review now includes moderate‐certainty evidence that: SNRIs probably have small effects on pain intensity and trivial effects on disability in low back pain patients at short‐term, but are probably associated with adverse effects; TCAs probably have small effects on disability in low back pain patients at short‐term. The review author team has been updated to include Michael C Ferraro, Giovanni Ferreira, Michael A Wewege, Christina Abdel Shaheed, Adrian C Traeger, Eric J Visser, James H McAuley and Aidan G Cashin. Willem JJ Assendelft, Martin Roland and Maurits W van Tulder did not contribute to this version of the review. Several changes to methods were made for this version of the review. We included studies of participants with spine‐related leg pain (including somatic‐referred pain, radicular pain with radiculopathy, or radicular pain without radiculopathy), and analysed these studies separately. We used updated guidance for the RoB 1 assessment. We assessed the certainty of evidence using GRADE.

Open in a new tab

History

Protocol first published: Issue 3, 1999 Review first published: Issue 1, 2008

Date	Event	Description
8 September 2010	Amended	Contact details updated.
19 February 2009	New search has been performed	The literature search was updated on 11th November 2008. No new studies were identified. However, two studies that were classified in the first review as 'waiting for assessment' were examined. While the study by Atkinson (2007) has been included in the review, the negative trial by Khoromi (2007) has been excluded. The addition of the study by Atkinson (2007) has not changed the conclusions of the original review.
19 June 2008	Amended	Converted to new review format.

Open in a new tab

Acknowledgements

Editorial and peer‐reviewer contributions

Cochrane Musculoskeletal supported the authors in the development of this review update.

The following people conducted the editorial process for this article:

Sign‐off Editor (final editorial decision): Rachelle Buchbinder, Head, Musculoskeletal Health and Wiser Health Care Units, School of Public Health and Preventive Medicine, Monash University;
Managing Editor (selected peer reviewers, provided editorial guidance to authors, edited the article): Luisa Fernandez Mauleffinch, Cochrane Central Editorial Service;
Editorial Assistant (conducted editorial policy checks, collated peer‐reviewer comments and supported editorial team): Jacob Hester, Cochrane Central Editorial Service;
Copy Editor (copy editing and production): Faith Armitage; Cochrane Central Production Service;
Peer‐reviewers (provided comments and recommended an editorial decision): Zachary A Cupler, Butler VA Health Care System, USA (clinical/content review); Brian Duncan (consumer review); Jennifer Hilgart, Cochrane (methods review); and Jo Platt, Central Editorial Information Specialist (search review). One additional peer reviewer provided clinical/content peer review but chose not to be publicly acknowledged.

Appendices

Appendix 1. Search strategies

Databases

Cochrane Central Register of Controlled Trials (CENTRAL)

#1 MeSH descriptor: [Antidepressive Agents] explode all trees

#2 MeSH descriptor: [Antidepressive Agents, Second‐Generation] explode all trees

#3 antidepressants

#4 (anti‐depressant* or antidepressant*)

#5 MeSH descriptor: [Serotonin Uptake Inhibitors] explode all trees

#6 MeSH descriptor: [Citalopram] explode all trees

#7 MeSH descriptor: [Serotonin Antagonists] explode all trees

#8 MeSH descriptor: [Fluoxetine] explode all trees

#9 SSRI*

#10 selective serotonin reuptake inhibitors

#11 (Citalopram or Celexa)

#12 (Fluoxetine or Prozac)

#13 (Paroxetine or Paxil or Seroxat)

#14 (Sertraline or Zoloft or Lustral)

#15 (Escitalopram or Lexapro or Cipralex)

#16 (Fluvoxamine or Luvox)

#17 (Desvenlafaxine or Pristiq)

#18 (Venlafaxine or Effexor)

#19 (Duloxetine or Cymbalta)

#20 (Milnacipran or Ixel or Savella)

#21 (Reboxetine or Edronax)

#22 (Viloxazine or Vivalan)

#23 MeSH descriptor: [Antidepressive Agents, Tricyclic] explode all trees

#24 tricyclic antidepressants

#25 tricyclic antidepressant*

#26 MeSH descriptor: [Amitriptyline] explode all trees

#27 (Amitriptyline or Elavil or Endep)

#28 (Clomipramine or Anafranil)

#29 (Desipramine or Norpramin or Pertofrane)

#30 (Dosulepin or Dothiepin or Prothiaden)

#31 (Doxepin or Adapin or Sinequan)

#32 (Imipramine or Tofranil)

#33 (Lofepramine or Feprapax or Gamanil or Lomont)

#34 (Nortriptyline or Pamelor)

#35 (Protriptyline or Vivactil)

#36 (Trimipramine or Surmontil)

#37 (Amoxapine or Asendin)

#38 (Loxapine or Loxpac or Loxitane)

#39 (Maprotiline or Deprilept or Ludiomil or Psymion)

#40 (Mazindol or Mazanor or Sanorex)

#41 (Mianserin or Bolvidon or Norval or Tolvon)

#42 (Mirtazapine or Remeron or Avanza or Zispin)

#43 (Setiptiline or Tecipul)

#44 MeSH descriptor: [Monoamine Oxidase Inhibitors] explode all trees

#45 Monoamine oxidase Inhibitors

#46 monoamine oxidase inhibitor*

#47 (Isocarboxazid or Marplan or Moclobemide)

#48 (Aurorix or Manerix or Phenelzine)

#49 (Nardil or Selegiline or L‐Deprenyl)

#50 (Eldepryl or Zelapar or Emsam)

#51 (Trancypromine or Parnate)

#52 MeSH descriptor: [Phenelzine] explode all trees

#53 MeSH descriptor: [Moclobemide] explode all trees

#54 MeSH descriptor: [Selegiline] explode all trees

#55 MeSH descriptor: [Tranylcypromine] explode all trees

#56 MeSH descriptor: [Isocarboxazid] explode all trees

#57 serotonin‐norepinephrine reuptake inhibitor*

#58 norepinephrine reuptake inhibitor*

#59 serotonin‐noradrenaline reuptake inhibitor*

#60 noradrenaline reuptake inhibitor*

#61 (Duloxetine or Cymbalta or Milnacipram or Ixel or Savella)

#62 (Ruboxetine or Edronax or Viloxazine or Vivalan)

#63 MeSH descriptor: [Tryptophan] explode all trees

#64 L‐tryptophan

#65 #1 or #2 or #3 or #4 or #5 or #6 or #7 or #8 or #9 or #10 or #11 or #12 or #13 or #14 or #15 or #16 or #17 or #18 or #19 or #20 or #21 or #22 or #23 or #24 or #25 or #26 or #27 or #28 or #29 or #30 or #31 or #32 or #33 or #34 or #35 or #36 or #37 or #38 or #39 or #40 or #41 or #42 or #43 or #44 or #45 or #46 or #47 or #48 or #49 or #50 or #51 or #52 or #53 or #54 or #55 or #56 or #57 or #58 or #59 or #60 or #61 or #62 or #63 or #64

#66 MeSH descriptor: [Back Pain] explode all trees

#67 MeSH descriptor: [Low Back Pain] explode all trees

#68 dorsalgia

#69 coccyx or coccydynia or spondylosis or sciatica

#70 (lumb* NEAR3 pain):ti,ab,kw

#71 MeSH descriptor: [Spine] explode all trees

#72 MeSH descriptor: [Spinal Diseases] explode all trees

#73 (lumbago or discitis)

#74 spinal fusion

#75 facet NEAR joint*

#76 MeSH descriptor: [Intervertebral Disc] explode all trees

#77 postlaminectomy

#78 arachnoiditis

#79 failed NEAR back

#80 MeSH descriptor: [Cauda Equina] explode all trees

#81 lumb* NEAR vertebra*

#82 stenosis NEAR (spine or root or spinal)

#83 slipped NEAR (disc* or disk*)

#84 degenerat* NEAR (disc* or disk*)

#85 herniat* NEAR (disc* or disk*)

#86 displac* NEAR (disc* or disk*)

#87 prolaps* NEAR (disc* or disk*)

#88 back NEAR pain

#89 back disorder*

#90 MeSH descriptor: [Sciatic Neuropathy] explode all trees

#91 #66 or #67 or #68 or #69 or #70 or #71 or #72 or #73 or #74 or #75 or #76 or #77 or #78 or #79 or #80 or #81 or #82 or #83 or #84 or #85 or #86 or #87 or #88 or #89 or #90

#92 #91 AND #65 in Trials

MEDLINE

1 dorsalgia.tw,kf.

2 exp Back Pain/

3 (backache or back‐ache).tw,kf.

4 exp Low Back Pain/

5 (back adj3 pain).tw,kf.

6 (lumb* adj3 pain).tw,kf.

7 coccyx.tw,kf.

8 coccydynia.tw,kf.

9 sciatica.tw,kf.

10 exp sciatic neuropathy/

11 spondylosis.tw,kf.

12 lumbago.tw,kf.

13 back disorder$.tw,kf.

14 or/1‐13

15 exp Antidepressive Agents/

16 exp Antidepressive Agents, Second‐Generation/

17 antidepressants.tw,kf.

18 (anti‐depressant$ or antidepressant$).tw,kf.

19 exp Serotonin Uptake Inhibitors/ or exp Citalopram/ or exp Serotonin Antagonists/ or exp Fluoxetine/ or SSRI$.tw,kf.

20 selective serotonin reuptake inhibitors.tw,kf.

21 (Citalopram or Celexa).tw,kf.

22 (Fluoxetine or Prozac).tw,kf.

23 (Paroxetine or Paxil or Seroxat).tw,kf.

24 (Sertraline or Zoloft or Lustral).tw,kf.

25 (Escitalopram or Lexapro or Cipralex).tw,kf.

26 (Fluvoxamine or Luvox).tw,kf.

27 (Desvenlafaxine or Pristiq).tw,kf.

28 (Venlafaxine or Effexor).tw,kf.

29 (Duloxetine or Cymbalta).tw,kf.

30 (Milnacipran or Ixel or Savella).tw,kf.

31 (Reboxetine or Edronax).tw,kf.

32 (Viloxazine or Vivalan).tw,kf.

33 exp Antidepressive Agents, Tricyclic/

34 tricyclic antidepressants.tw,kf.

35 tricyclic antidepressant$.tw,kf.

36 exp Amitriptyline/

37 (Amitriptyline or Elavil or Endep).tw,kf.

38 (Clomipramine or Anafranil).tw,kf.

39 (Desipramine or Norpramin or Pertofrane).tw,kf.

40 (Dosulepin or Dothiepin or Prothiaden).tw,kf.

41 (Doxepin or Adapin or Sinequan).tw,kf.

42 (Imipramine or Tofranil).tw,kf.

43 (Lofepramine or Feprapax or Gamanil or Lomont).tw,kf.

44 (Nortriptyline or Pamelor).tw,kf.

45 (Protriptyline or Vivactil).tw,kf.

46 (Trimipramine or Surmontil).tw,kf.

47 (Amoxapine or Asendin).tw,kf.

48 (Loxapine or Loxpac or Loxitane).tw,kf.

49 (Maprotiline or Deprilept or Ludiomil or Psymion).tw,kf.

50 (Mazindol or Mazanor or Sanorex).tw,kf.

51 (Mianserin or Bolvidon or Norval or Tolvon).tw,kf.

52 (Mirtazapine or Remeron or Avanza or Zispin).tw,kf.

53 (Setiptiline or Tecipul).tw,kf.

54 exp Monoamine Oxidase Inhibitors/

55 monoamine oxidase inhibitors.tw,kf.

56 Monoamine oxidase Inhibitor$.tw,kf.

57 (Isocarboxazid or Marplan).tw,kf.

58 (Aurorix or Manerix or Phenelzine).tw,kf.

59 (Nardil or Selegiline or L‐Deprenyl).tw,kf.

60 (Eldepryl or Zelapar or Emsam).tw,kf.

61 (Trancypromine or Parnate or Moclobemide).tw,kf.

62 exp Phenelzine/

63 exp Moclobemide/

64 exp Selegiline/

65 exp Tranylcypromine/

66 exp Isocarboxazid/

67 serotonin‐norepinephrine reuptake inhibitor$.tw,kf.

68 norepinephrine reuptake inhibitor$.tw,kf.

69 serotonin‐noradrenaline reuptake inhibitor$.tw,kf.

70 noradrenaline reuptake inhibitor$.tw,kf.

71 (Duloxetine or Cymbalta or Milnacipram or Ixel or Savella).tw,kf.

72 (Ruboxetine or Edronax or Viloxazine or Vivalan).tw,kf.

73 exp Tryptophan/

74 L‐tryptophan.tw,kf.

75 or/15‐74

76 randomized controlled trial.pt.

77 controlled clinical trial.pt.

78 randomized.ab.

79 placebo.ab.

80 drug therapy.fs.

81 randomly.ab.

82 trial.ab.

83 groups.ab.

84 76 or 77 or 78 or 79 or 80 or 81 or 82 or 83

85 exp animals/ not humans.sh.

86 84 not 85

87 14 and 75 and 86

Embase

1 dorsalgia.tw.

2 back pain.tw.

3 3 exp LOW BACK PAIN/

4 exp BACKACHE/

5 (lumb$ adj3 pain).tw.

6 coccyx.tw.

7 coccydynia.tw.

8 sciatica.tw.

9 sciatica/

10 exp ISCHIALGIA/

11 spondylosis.tw.

12 lumbago.tw.

13 back disorder$.tw.

14 or/1‐13

15 exp antidepressive agents/

16 antidepressants.tw.

17 (anti‐depressant$ or antidepressant$).tw.

18 exp Serotonin Antagonists/ or SSRI$.tw.

19 selective serotonin reuptake inhibitors.tw.

20 (Citalopram or Celexa).tw.

21 (Fluoxetine or Prozac).tw.

22 (Paroxetine or Paxil or Seroxat).tw.

23 (Sertraline or Zoloft or Lustral).tw.

24 (Escitalopram or Lexapro or Cipralex).tw.

25 (Fluvoxamine or Luvox).tw.

26 (Desvenlafaxine or Pristiq).tw.

27 (Venlafaxine or Effexor).tw.

28 (Duloxetine or Cymbalta).tw.

29 (Milnacipran or Ixel or Savella).tw.

30 (Reboxetine or Edronax).tw.

31 (Viloxazine or Vivalan).tw.

32 tricyclic antidepressant$.tw.

33 amitriptyline plus chlordiazepoxide/ or amitriptyline plus perphenazine/

34 (Amitriptyline or Elavil or Endep).tw.

35 (Clomipramine or Anafranil).tw.

36 (Desipramine or Norpramin or Pertofrane).tw.

37 (Dosulepin or Dothiepin or Prothiaden).tw.

38 (Doxepin or Adapin or Sinequan).tw.

39 (Imipramine or Tofranil).tw.

40 (Lofepramine or Feprapax or Gamanil or Lomont).tw.

41 (Nortriptyline or Pamelor).tw.

42 (Protriptyline or Vivactil).tw.

43 (Trimipramine or Surmontil).tw.

44 (Amoxapine or Asendin).tw.

45 (Loxapine or Loxpac or Loxitane).tw.

46 (Maprotiline or Deprilept or Ludiomil or Psymion).tw.

47 (Mazindol or Mazanor or Sanorex).tw.

48 (Mianserin or Bolvidon or Norval or Tolvon).tw.

49 (Mirtazapine or Remeron or Avanza or Zispin).tw.

50 (Setiptiline or Tecipul).tw.

51 monoamine oxidase inhibitors.tw.

52 Monoamine oxidase Inhibitor$.tw.

53 (Isocarboxazid or Marplan or Moclobemide).tw.

54 (Trancypromine or Parnate).tw.

55 (Eldepryl or Zelapar or Emsam).tw.

56 (Nardil or Selegiline or L‐Deprenyl).tw.

57 (Aurorix or Manerix or Phenelzine).tw.

58 serotonin‐norepinephrine reuptake inhibitor$.tw.

59 norepinephrine reuptake inhibitor$.tw.

60 serotonin‐noradrenaline reuptake inhibitor$.tw.

61 noradrenaline reuptake inhibitor$.tw.

62 (Duloxetine or Cymbalta or Milnacipram or Ixel or Savella).tw.

63 (Ruboxetine or Edronax or Viloxazine or Vivalan).tw.

64 L‐tryptophan.tw.

65 or/15‐64

66 Randomized controlled trial/

67 Controlled clinical study/

68 random$.ti,ab.

69 randomization/

70 intermethod comparison/

71 placebo.ti,ab.

72 (compare or compared or comparison).ti.

73 ((evaluated or evaluate or evaluating or assessed or assess) and (compare or compared or comparing or comparison)).ab.

74 (open adj label).ti,ab.

75 ((double or single or doubly or singly) adj (blind or blinded or blindly)).ti,ab.

76 double blind procedure/

77 parallel group$1.ti,ab.

78 (crossover or cross over).ti,ab.

79 ((assign$ or match or matched or allocation) adj5 (alternate or group$1 or intervention$1 or patient$1 or subject$1 or participant$1)).ti,ab.

80 (assigned or allocated).ti,ab.

81 (controlled adj7 (study or design or trial)).ti,ab.

82 (volunteer or volunteers).ti,ab.

83 human experiment/

84 trial.ti.

85 or/66‐84

86 (random$ adj sampl$ adj7 (cross section$ or questionnaire$1 or survey$ or database$1)).ti,ab. not (comparative study/ or controlled study/ or randomi?ed controlled.ti,ab. or randomly assigned.ti,ab.)

87 Cross‐sectional study/ not (randomized controlled trial/ or controlled clinical study/ or controlled study/ or randomi?ed controlled.ti,ab. or control group$1.ti,ab.)

88 (((case adj control$) and random$) not randomi?ed controlled).ti,ab.

89 (Systematic review not (trial or study)).ti.

90 (nonrandom$ not random$).ti,ab.

91 Random field$.ti,ab.

92 (random cluster adj3 sampl$).ti,ab.

93 (review.ab. and review.pt.) not trial.ti.

94 we searched.ab. and (review.ti. or review.pt.)

95 update review.ab.

96 (databases adj4 searched).ab.

97 (rat or rats or mouse or mice or swine or porcine or murine or sheep or lambs or pigs or piglets or rabbit or rabbits or cat or cats or dog or dogs or cattle or bovine or monkey or monkeys or trout or marmoset$1).ti. and animal experiment/

98 Animal experiment/ not (human experiment/ or human/)

99 or/86‐98

100 85 not 99

101 14 and 65 and 100

Clinical trial registries

ClinicalTrials.gov

Condition/disease: as per table

Intervention/treatment: as per table

Study status: All studies

Sex: All

Age: Adult and older adults

Sex: All studies

WHO ICTRP (advanced search)

Condition AND intervention (as per table)

Recruitment status: all

EUCTR

Condition AND intervention (as per table)

Condition	Intervention
Back pain/sciatica	agomelatine
	amineptine
	amitriptyline
	amoxapine
	bupropion
	butriptyline
	citalopram
	clomipramine
	desipramine
	desvenlafaxine
	dibenzepin
	dimetacrine
	dosulepin
	doxepin
	duloxetine
	escitalopram
	etoperidone
	fluoxetine
	fluvoxamine
	gepirone
	imipramine
	imipramine oxide
	isocarboxazid
	levomilnacipran
	lofepramine
	maprotiline
	mianserin
	milnacipran
	mirtazapine
	moclobemide
	nefazodone
	nortriptyline
	oxitriptan
	paroxetine
	phenelzine
	protriptyline
	reboxetine
	sertraline
	setiptiline
	tianeptine
	tranylcypromine
	trazodone
	trimipramine
	tryptophan
	venlafaxine
	vilazodone
	viloxazine
	vortioxetine
Table: Clinical trial registry search terms for conditions and interventions

Open in a new tab

Data and analyses

Comparison 1. All antidepressant classes versus placebo for non‐specific low back pain and spine‐related leg pain.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Non‐specific low back pain: pain intensity (0 to 100) at immediate‐term follow‐up (≤ 4 weeks)	6		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.1.1 SNRI	4	1424	Mean Difference (IV, Random, 95% CI)	‐6.12 [‐8.42, ‐3.82]
1.1.2 TCA	2	78	Mean Difference (IV, Random, 95% CI)	1.20 [‐12.36, 14.76]
1.2 Non‐specific low back pain: pain intensity (0 to 100) at short‐term follow‐up (> 4 to 16 weeks)	11		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.2.1 SNRI	4	1415	Mean Difference (IV, Random, 95% CI)	‐5.25 [‐7.17, ‐3.34]
1.2.2 SSRI	3	199	Mean Difference (IV, Random, 95% CI)	1.20 [‐4.90, 7.30]
1.2.3 TCA	4	417	Mean Difference (IV, Random, 95% CI)	‐2.00 [‐7.25, 3.24]
1.2.4 TeCA	1	52	Mean Difference (IV, Random, 95% CI)	‐4.50 [‐17.59, 8.59]
1.2.5 Other antidepressants	1	39	Mean Difference (IV, Random, 95% CI)	‐5.40 [‐23.08, 12.28]
1.3 Non‐specific low back pain: pain intensity (0 to 100) at mid‐term follow‐up (> 16 to 52 weeks)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.3.1 TCA	1	146	Mean Difference (IV, Random, 95% CI)	‐8.20 [‐16.28, ‐0.12]
1.4 Non‐specific low back pain: disability (RMDQ, 0 to 24) at short‐term follow‐up (> 4 to 16 weeks): SNRI	4		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.4.1 SNRI	4	1348	Mean Difference (IV, Random, 95% CI)	‐0.91 [‐1.30, ‐0.51]
1.5 Non‐specific low back pain: disability (ODI, 0 to 100) at short‐term follow‐up (> 4 to 16 weeks): SSRI	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.5.1 SSRI	1	92	Mean Difference (IV, Random, 95% CI)	‐2.20 [‐8.11, 3.71]
1.6 Non‐specific low back pain: disability (RMDQ, 0 to 24) at short‐term follow‐up (> 4 to 16 weeks): TCA	3	330	Mean Difference (IV, Random, 95% CI)	‐1.76 [‐2.70, ‐0.82]
1.7 Non‐specific low back pain: disability (RMDQ, 0 to 24) at mid‐term follow‐up (> 16 to 52 weeks): TCA	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.7.1 TCA	1	146	Mean Difference (IV, Random, 95% CI)	‐1.20 [‐2.73, 0.33]
1.8 Non‐specific low back pain: depressive symptoms (BDI, 0 to 63) at short‐term follow‐up (> 4 to 16 weeks): SNRI	2	613	Mean Difference (IV, Random, 95% CI)	‐0.09 [‐1.20, 1.02]
1.9 Non‐specific low back pain: depressive symptoms (MADRS, 0 to 60) at short‐term follow‐up (> 4 to 16 weeks): SSRI	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.9.1 SSRI	1	92	Mean Difference (IV, Random, 95% CI)	‐0.10 [‐3.64, 3.44]
1.10 Non‐specific low back pain: depressive symptoms (BDI, 0 to 63) at short‐term follow‐up (> 4 to 16 weeks): TCA	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.10.1 TCA	2	194	Mean Difference (IV, Random, 95% CI)	‐1.59 [‐2.95, ‐0.23]
1.11 Non‐specific low back pain: depressive symptoms (BDI, 0 to 63) at mid‐term follow‐up (> 16 to 52 weeks): TCA	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.11.1 TCA	1	118	Mean Difference (IV, Random, 95% CI)	‐0.93 [‐3.44, 1.58]
1.12 Non‐specific low back pain: depressive symptoms (BDI, 0 to 63) at short‐term follow‐up (> 4 to 16 weeks): other antidepressants	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.12.1 Other antidepressants	1	40	Mean Difference (IV, Random, 95% CI)	2.21 [‐4.00, 8.42]
1.13 Non‐specific low back pain: HRQoL (EQ‐5D, 0 to 1) at short‐term follow‐up (> 4 to 16 weeks): SNRI	4		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.13.1 SNRI	4	1401	Mean Difference (IV, Random, 95% CI)	0.03 [0.01, 0.06]
1.14 Non‐specific low back pain: HRQoL (SMD) at short‐term follow‐up (> 4 to 16 weeks): TCA	2		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.14.1 TCA	2	194	Std. Mean Difference (IV, Random, 95% CI)	0.07 [‐0.43, 0.58]
1.15 Non‐specific low back pain: HRQoL (EQ‐5D, 0 to 100 VAS) at mid‐term follow‐up (> 16 to 52 weeks): TCA	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.15.1 TCA	1	118	Mean Difference (IV, Random, 95% CI)	3.80 [‐1.77, 9.37]
1.16 Spine‐related leg pain: pain intensity (0 to 100) at immediate‐term follow‐up (≤ 4 weeks)	3		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.16.1 TCA	3	155	Mean Difference (IV, Random, 95% CI)	‐15.18 [‐30.97, 0.61]
1.17 Spine‐related leg pain: pain intensity (0 to 100) at short‐term follow‐up (> 4 to 16 weeks)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.17.1 SNRI	1	11	Mean Difference (IV, Random, 95% CI)	‐46.10 [‐89.29, ‐2.91]
1.17.2 TCA	1	60	Mean Difference (IV, Random, 95% CI)	‐23.00 [‐32.12, ‐13.88]
1.18 Spine‐related leg pain: pain intensity (0 to 100) at mid‐term follow‐up (> 16 to 52 weeks)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.18.1 TCA	1	60	Mean Difference (IV, Random, 95% CI)	‐27.00 [‐36.12, ‐17.88]
1.19 Spine‐related leg pain: disability (SMD) at immediate‐term follow‐up (≤ 4 weeks)	2		Std. Mean Difference (IV, Random, 95% CI)	Subtotals only
1.19.1 TCA	2	122	Std. Mean Difference (IV, Random, 95% CI)	‐0.60 [‐1.09, ‐0.10]
1.20 Spine‐related leg pain: disability (ODI, 0 to 100) at short‐term follow‐up (> 4 to 16 weeks)	2		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.20.1 SNRI	1	11	Mean Difference (IV, Random, 95% CI)	‐4.40 [‐20.25, 11.45]
1.20.2 TCA	1	60	Mean Difference (IV, Random, 95% CI)	‐13.00 [‐19.42, ‐6.58]
1.21 Spine‐related leg pain: disability (ODI, 0 to 100) at mid‐term follow‐up (> 16 to 52 weeks)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.21.1 TCA	1	60	Mean Difference (IV, Random, 95% CI)	‐20.00 [‐27.74, ‐12.26]
1.22 Spine‐related leg pain: depressive symptoms (HADS‐D, 0 to 21) at immediate‐term follow‐up (≤ 4 weeks)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
1.22.1 TCA	1	62	Mean Difference (IV, Fixed, 95% CI)	0.23 [‐1.51, 1.97]
1.23 Spine‐related leg pain: depressive symptoms (BDI, 0 to 63) at short‐term follow‐up (> 4 to 16 weeks)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.23.1 SNRI	1	11	Mean Difference (IV, Random, 95% CI)	‐1.70 [‐13.48, 10.08]
1.24 Spine‐related leg pain: HRQoL (SF‐36, 0 to 100) at short‐term follow‐up (> 4 to 16 weeks)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
1.24.1 SNRI	1	11	Mean Difference (IV, Random, 95% CI)	‐2.70 [‐12.42, 7.02]
1.25 Non‐specific low back pain and spine‐related leg pain: total adverse events	13		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.25.1 SNRI	5	1510	Risk Ratio (M‐H, Random, 95% CI)	1.17 [1.07, 1.27]
1.25.2 SSRI	2	107	Risk Ratio (M‐H, Random, 95% CI)	1.83 [0.14, 24.19]
1.25.3 TCA	7	474	Risk Ratio (M‐H, Random, 95% CI)	1.76 [0.79, 3.90]
1.25.4 TeCA	1	52	Risk Ratio (M‐H, Random, 95% CI)	0.93 [0.79, 1.09]
1.26 Non‐specific low back pain and spine‐related leg pain: serious adverse events	7		Peto Odds Ratio (Peto, Fixed, 95% CI)	Subtotals only
1.26.1 SNRI	5	1510	Peto Odds Ratio (Peto, Fixed, 95% CI)	1.75 [0.79, 3.89]
1.26.2 TCA	1	142	Peto Odds Ratio (Peto, Fixed, 95% CI)	6.64 [0.41, 106.72]
1.26.3 Other antidepressants	1	42	Peto Odds Ratio (Peto, Fixed, 95% CI)	0.90 [0.16, 4.96]
1.27 Non‐specific low back pain and spine‐related leg pain: withdrawals due to adverse events	17		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only
1.27.1 SNRI	6	1543	Risk Ratio (M‐H, Random, 95% CI)	2.36 [1.66, 3.36]
1.27.2 SSRI	2	139	Risk Ratio (M‐H, Random, 95% CI)	3.54 [0.78, 16.02]
1.27.3 TCA	8	655	Risk Ratio (M‐H, Random, 95% CI)	1.81 [0.89, 3.69]
1.27.4 TeCA	1	69	Risk Ratio (M‐H, Random, 95% CI)	10.91 [1.48, 80.66]
1.27.5 Other antidepressants	2	96	Risk Ratio (M‐H, Random, 95% CI)	3.45 [0.59, 20.10]

Open in a new tab

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Afilal 2020.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 90‐day follow‐up Number of study centres: 1 Country: Morocco Study setting: Hospital Rheumatology Department Study dates: May 2019 to November 2019
Participants	General characteristics Number randomised: 62 (clomipramine = 31; placebo = 31) Mean age: clomipramine = 54.81 (SD 11.65) years; placebo = 55.74 (SD 11.35) years Age range: not reported Sex and gender: clomipramine = 30 (96.8%) female, 1 (3.2%) male; 28 (90.3%) female, 3 (9.7%) male Condition: chronic lumbar radicular pain Mean duration of symptoms: clomipramine = 52.96 (SD 38.2) months; placebo = 51.77 (SD 62.2) months Depression: not specified Characteristics that stratify health opportunities Level of country income: lower‐middle Race and ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria men and women aged 20 to 80 years chronic lumbar radicular pain (defined as pain lumbosacral radicular radiating into the leg below the knee, whose diagnosis has been confirmed by sectional imaging (CT or MRI)) symptoms present for greater than 3 months VAS pain ≥ 6/10 pain not improved by NSAIDs, analgesics and physical treatment Exclusion criteria specific spinal pathological entity such as infection, inflammatory or tumour pathology, or fractures spinal surgery within the preceding 4 weeks coexisting major disease for which clomipramine may be inappropriate: glaucoma, urinary problems, recent myocardial infarction, convulsions current use of other antidepressants current use of recommended full and regular doses of another analgesic allergy or contraindication for experimental drugs pregnancy, planning or trying to become pregnant, or breastfeeding
Interventions	Intervention: intravenous clomipramine Target dose: 75 mg/day Run in: 25 mg/day for 1 day, 50 mg/day for 1 day Treatment period: 90 days (10‐day intravenous therapy with an additional 80 days of open‐label oral clomipramine therapy in the intravenous clomipramine arm only) Comparator: inert placebo Rescue analgesia: not reported Concomitant medications: intravenous paracetamol administered to both groups intravenously at 3 g/day for 10 days, parecoxib administered to both groups intravenously for 3 days (dose not reported)
Outcomes	Pain intensity (10‐cm VAS) Mean baseline score: clomipramine = 6.58 (SD 0.56); placebo = 6.64 (SD 0.66) Time points measured: baseline, day 5, day 10, day 90^a Roland Morris Disability Questionnaire Mean baseline score: clomipramine = 17.12 (SD 3.78); placebo = 15.12 (SD 3.87) Time points measured: baseline, day 5, day 10, day 90^a Hospital Anxiety and Depression Scale Mean baseline score: clomipramine = 8.83 (SD 3.67); placebo = 5.96 (SD 3.77) Time points measured: baseline, day 5, day 10, day 90^a Adverse events Time points measured: not reported
Notes	Funding source: not reported Declarations of interest: none declared ^aDay 90 outcomes not eligible for inclusion in analyses All outcome data included in analyses were provided by the study authors.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generated using a computer random number generator.
Allocation concealment (selection bias)	Unclear risk	While opaque allocation envelopes were used, it is unclear whether they were sequentially numbered.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Reported as 'double‐blind'; active and placebo solutions were indistinguishably clear and colourless and there was no indication that blinding was broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants were likely to have been blinded and there was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	Low risk	Complete outcome data provided with no dropouts reported.
Selective reporting (reporting bias)	High risk	The trial was retrospectively registered.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Alcoff 1982.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 8‐week follow‐up Number of study centres: 2 Country: USA Study setting: primary care Study dates: 1 September 1980 to 1 March 1981
Participants	General characteristics Number randomised: 50 (imipramine = 28; placebo = 22) Mean age: imipramine = 29.2 years; placebo = 33.8 years Age range: not reported Sex and gender: imipramine = 14 (50%) female, 14 (50%) male; placebo = 12 (54.5%) female, 10 (45.5%) male Condition: low back pain Mean duration of symptoms: not reported Depression: imipramine = 6 (21%); placebo = 4 (18%) (clinical diagnosis) Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: imipramine = 2 (7.1%) Black participants, 4 (14.3%) Hispanic participants, 22 (78.6%) White participants; placebo = 4 (18.2%) Black participants, 0 (0%) Hispanic participants, 18 (81.8%) White participants Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria experiencing low back pain for at least 6 weeks if a first episode, or two or more prior episodes lasting at least 2 weeks with a current episode of a minimum of 2 weeks' duration Exclusion criteria not reported
Interventions	Intervention: oral imipramine hydrochloride Target dose: 150 mg/day Run in: 75 mg/day for 3 days Treatment period: 8 weeks Comparator: inert placebo Rescue analgesia: not reported Concomitant medications: not reported
Outcomes	Pain intensity (Short Back Pain Questionnaire) Mean baseline score: not reported Time points measured: baseline, week 2, week 4, week 8 Beck Depression Questionnaire Mean baseline score: not reported Time points measured: baseline, week 8
Notes	Funding source: "This study was supported by the Bureau of Medicine and Surgery, Department of the Navy, Clinical Investigation Program #0‐08‐1461" Declarations of interest: not reported We attempted to contact the study authors to request point estimates and measures of variance for eligible outcomes but current contact details could not be found.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Described as randomised but insufficient information about the sequence generation process reported.
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not described.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Both participants and research staff described as blinded; active and placebo drugs were identical in size, colour and taste, and no indication that the blinding could have been broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants likely to have been blinded and no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	2/28 (7%) participants in imipramine group and 2/22 (9%) participants in placebo group did not provide outcome data. While missing data were minimal, reasons for missingness differed across groups (those on imipramine discontinued due to moving from the study area and no reason was provided for those on placebo). Methods to impute missing data were not reported. The amount of missing outcome data is enough to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	High risk	No study protocol or registration; outcomes have been reported incompletely so that they cannot be entered in a meta‐analysis (point estimates and measures of variance not provided).
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Atkinson 1998.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 8‐week follow‐up Number of study centres: not reported Country: USA Study setting: primary care and orthopaedic clinics Study dates: September 1990 to December 1993
Participants	General characteristics Number randomised: 78 (nortriptyline = 38; placebo = 40) Mean age: nortriptyline = 45.79 (SD 10.59) years; placebo = 47.13 (SD 10.65) years Age range: not reported Sex and gender: nortriptyline = 0 (0%) female, 38 (100%) male; placebo = 0 (0%) female, 40 (100%) male Gender: not reported Condition: low back pain Mean duration of symptoms: nortriptyline = 16.04 (SD 11.25) years; placebo = 13.58 (SD 12.94) years Depression: non‐depressed participants specifically recruited Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: nortriptyline = 32 (84%) White participants; placebo = 29 (73%) White participants Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: nortriptyline = mean 13.79 (SD 3.05) years; placebo = mean 15.14 (SD 2.72) years Socioeconomic status: not reported Social capital: not reported Inclusion criteria age 21–65 years low back pain (at T‐6 or below) present on a daily basis for the preceding 6 months or longer able to understand the study measures (based on an 8th grade literacy level) Exclusion criteria major coexisting medical illness (e.g. chronic obstructive pulmonary disease or cardiac disease, which might confound assessment of function or health‐ related quality of life) coexisting orthopaedic or pain disorder other than those related to the lumbar lesion current mood disorder (e.g. major depression, dysthymia, or bipolar mood disorder) history of a psychoactive substance use disorder within the preceding 12 months.
Interventions	Intervention: oral nortriptyline Target dose: plasma concentration of 50–150 ng/mL Run in: 25 mg/day for 3 days; 50 mg/day for 4 days; 75 mg/day for 3 days; 100 mg/day for 4 days Treatment period: 8 weeks Comparator: inert placebo Rescue analgesia: not reported Concomitant medications: ongoing use of non‐opioids (e.g. aspirin, NSAIDs) permitted
Outcomes	Pain intensity (Descriptor Differential Scale) Mean baseline score: nortriptyline = 11.66 (SD 4.83); placebo = 9.95 (SD 3.94) Time points measured: baseline and weekly until week 8 Beck Depression Inventory Mean baseline score: nortriptyline = 8.34 (SD 6.51); placebo = 8.5 (SD 7.67) Time points measured: baseline and weekly until week 8 Quality of Well‐Being Scale Mean baseline score: nortriptyline = 0.614 (SD 0.65); placebo = 0.616 (SD 0.71) Time points measured: baseline and weekly until week 8
Notes	Funding source: "This work was supported in part by the United States Department of Veterans Affairs, and by the National Institutes of Health Grant MO1‐RR00827" Declarations of interest: not reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generated using a random number table.
Allocation concealment (selection bias)	Low risk	Randomisation schedule held by a research pharmacist not involved in other aspects of the study. The code for participant assignment was held by the research pharmacy until completion of data analysis.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants and study physician described as blinded. Drug and placebo were administered in identical capsules with dose adjustments made in the placebo group to match the active group, and no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants were likely to have been blinded and there was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	10/38 (26%) participants in the nortriptyline group and 11/40 (28%) participants in the placebo group withdrew from the study but the number of withdrawals that continued to provide outcome data was not reported. While all 78 randomised participants were included in ITT analysis, last‐observation‐carried‐forward was used as the method of imputation. It is likely that participants who withdrew did not continue to provide outcome data, and an inappropriate method of imputation was used, both likely to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	Unclear risk	No study protocol or registration; insufficient information available to assess selective reporting.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but use of co‐interventions only reported for completer sample.

Open in a new tab

Atkinson 1999.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 3 Study duration: 8‐week follow‐up Number of study centres: not reported Country: USA Study setting: primary care and orthopaedic clinics Study dates: July 1994 to December 1996
Participants	General characteristics Number randomised: 103 (maprotiline = 33; paroxetine = 34; placebo = 36) Mean age: 49.2 (SD 9.4) years Age range: not reported Sex and gender: 38 female (36.9%); 65 male (63.1%) Condition: low back pain Mean duration of symptoms: 14.5 (SD 11.1) years Depression: participants with depression excluded Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: 87 White (84.5%) Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: mean 14.9 (SD 2.7) years Socioeconomic status: 'middle class'; mean annual family income USD 450,000 Social capital: not reported Inclusion criteria ages 21 to 65 low back pain (at T‐6 or below) present on a daily basis for the preceding six months or longer able to understand the study measures (based on an 8th grade literacy level) Exclusion criteria major coexisting medical illness (e.g. chronic obstructive pulmonary disease or cardiac disease) coexisting orthopaedic or pain disorder other than those related to the lumbar lesion current mood disorder (e.g. major depression, dysthymia, or bipolar mood disorder) history of a psychoactive substance use disorder within the preceding 12 months
Interventions	Intervention 1: oral maprotiline Target dose: 150mg/day Run in: 50 mg/daily for 3 days; 100 mg/day for 3 days Intervention 2: oral paroxetine Target dose: 30 mg/day Run in: 10 mg/day for 3 days; 20 mg/day for 3 days Treatment period: 8 weeks Comparator: active placebo (diphenhydramine up to 37.5 mg/day) Rescue analgesia: not reported Concomitant medications: opioid analgesics not permitted; use of non‐opioids (e.g. aspirin, NSAIDs) permitted
Outcomes	Pain intensity (Descriptor Differential Scale) Mean baseline score: maprotiline = 12.21 (SD 4.52); paroxetine = 10.54 (SD 3.71); placebo = 10.53 (SD 3.74) ('completer sample' only) Time points measured: baseline, week 8 Beck Depression Inventory Mean baseline score: 6.1 (SD 5.3) (all arms) Time points measured: baseline, week 8
Notes	Funding source: "This work was supported in part by the United States Department of Veterans Affairs, and by the National Institutes of Health Grant MO1‐RR00827" Declarations of interest: not reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generated using a random number table.
Allocation concealment (selection bias)	Low risk	Randomisation schedule held by a research pharmacist not involved in other aspects of the study. The code for participant assignment was held by the research pharmacy until completion of data analysis.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Both participants and research staff were blinded but blinding success tests performed at study exit indicated unsuccessful blinding of both participants and personnel (significant differences in proportions of correct guesses between active and placebo groups).
Blinding of outcome assessment (detection bias) Self‐reported outcomes	High risk	Self‐reported outcome assessment likely to have been unblinded for some participants (inadequate blinding as indicated by blinding success test).
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	13/33 (39%) participants in the maprotiline group, 12/34 (35%) participants in the paroxetine group, and 4/32 (11%) participants in the placebo group withdrew from the protocol and did not provide outcome data for the main (completer) analysis. Last‐observation‐carried‐forward used as method of imputation for ITT analysis. The amount of missing outcome data, imbalance in missing outcome data between intervention groups, and inappropriate method of imputation is enough to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	High risk	No study protocol or registration; outcomes have been reported incompletely so that they cannot be entered in a meta‐analysis (point estimates and measures of variance not reported for ITT sample).
Other bias	Unclear risk	No early stopping or unplanned interim analyses; use of co‐interventions during treatment period only reported for completer sample.

Open in a new tab

Atkinson 2007.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 7 Study duration: 12‐week follow‐up Number of study centres: not reported Country: USA Study setting: primary care clinics Study dates: July 1999 to December 2004
Participants	General characteristics Number randomised: 121 (desipramine = 52; fluoxetine = 43; placebo = 26) Mean age: 46.4 (SD 10.2) years Age range: not reported Sex and gender: 47 (38.8%) female, 74 (61.2%) male Condition: low back pain Mean duration of symptoms: not reported Depression: participants with major depression at time of screening or within preceding 12 months were excluded Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: 84 (69.4%) White participants Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: high school educated = 69 (57%) Socioeconomic status: 'predominantly middle income (US $45,000 annually)' Social capital: not reported Inclusion criteria 21 to 65 years of age low back pain (T‐6 or below, secondary to degenerative disc disease) present "on a daily basis" for the previous 6 months or longer English speaking, literate, and able to understand and communicate with the study team not a present candidate for back surgery CYP‐450 2D6 extensive metabolizer phenotype Exclusion criteria major coexisting medical illness which might confound assessment of function or life quality or other conditions for which classical antidepressants would be inappropriate another coexisting chronic pain problem significant coexisting musculoskeletal disorder, sciatica as a primary problem, or back pain due to systemic disorders history within the previous 12 months of alcohol or other substance‐use disorder, major depression, or dysthymia lifetime history of bipolar disorder or psychosis dementia known allergy to the study drugs use of psychoactive agents that would need to be continued during the study, or other agents that might interact with the study drugs before treatment with study agents CYP‐450 2D6 poor metabolizer phenotype
Interventions	Intervention 1: oral desipramine (low, medium & high arms ‐ subsequently combined for analyses) Target dose: low = 50 ng/mL; medium = 110 ng/mL; high = 150 ng/mL Intervention 2: oral fluoxetine (low, medium & high arms ‐ subsequently combined for analyses) Target dose: low = 100 ng/mL; medium = 200 ng/mL; high = 400 ng/mL Run in: 4‐week phase to reach target plasma concentrations Treatment period: 12 weeks Comparator: active placebo (benztropine mesylate 0.5 mg/day) Rescue analgesia: not reported Concomitant medications: opioid analgesics not permitted within 3 weeks before and during study, use of non‐opioids (e.g. aspirin, NSAIDs) permitted
Outcomes	Pain intensity (Descriptor Differential Scale) Baseline score: 9.4 (SD 4.0) Time points measured: not reported, data available for baseline & week 12 Roland and Morris Disability Questionnaire Baseline score: not reported Time points measured: not reported Beck Depression Inventory Baseline score: 5.7 (SD 4.7) (all arms) Time points measured: baseline, 'each study visit'
Notes	Funding source: "Supported by the Department of Veterans Affairs and National Institutes of Health grant MO 1 RR00827" Declarations of interest: not reported We contacted the authors of this study to request disability scores for the fluoxetine versus placebo comparison. The authors responded but reported the required data were not available.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generated using a computer random number generator.
Allocation concealment (selection bias)	Low risk	Randomisation schedule held by a research pharmacist not involved in other aspects of the study. The code for participant assignment was held by the research pharmacy until completion of data analysis.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Both participants and research staff were blinded, equivalent active and placebo pills were ingested, and blinding success tests were performed at trial exit, indicating successful blinding.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants were blinded, and a blinding success test was performed at trial exit, indicating successful blinding.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	38/121 (31%) of participants withdrew from the trial; number of withdrawals that continued to provide outcome data was not reported. While all 121 randomised participants were included in ITT analysis, last‐observation‐carried‐forward was used as the method of imputation. It is likely that participants who withdrew did not continue to provide outcome data, and an inappropriate method of imputation was used, both likely to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	High risk	No study protocol or registration; pain intensity was measured at weeks 1 and 4 post‐randomisation but not reported.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Dickens 2000.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 8‐week follow‐up Number of study centres: 1 Country: United Kingdom Study setting: rheumatology outpatient clinic Study dates: not reported
Participants	General characteristics Number randomised: 98 (baseline characteristics reported for n = 92) Mean age: paroxetine = 44 (SD 9.7) years; placebo = 46 (SD 10.6) years Age range: not reported Sex and gender: paroxetine = 23 (52%) female, 21 (48%) male; placebo = 27 (56%) female; 21 (44%) male Condition: non‐specific low back pain Mean duration of symptoms: not reported Depression: participants with significant depressive symptoms Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria low back pain for more than 6 months Montgomery‐Åsberg Depression Rating Scale score ≥ 16 Oswestry Disability Index ≥ 30% Exclusion criteria significant physical/mental disorder systemic inflammatory disorder or malignancy contraindications to taking paroxetine taking opiates, corticosteroids, or psychotropic medication other than temazepam up to 20 mg/day recent surgery
Interventions	Intervention: oral paroxetine Target dose: 20 mg/day Run in: pre‐randomisation run in with placebo for 7 days (both arms) Treatment period: 8 weeks Comparator: inert placebo Rescue analgesia: not reported Concomitant medications: combined analgesics (e.g. codeine‐related drugs with acetaminophen‐like drugs), simple analgesics, and NSAIDs permitted
Outcomes	Pain intensity (100‐mm VAS) Mean baseline score: paroxetine = 55.1 (SD 22.8); placebo = 56.1 (SD 21.4) Time points measured: baseline, week 2, week 4, week 8 Oswestry Low Back Pain Disability Questionnaire Mean baseline score: paroxetine = 54.2 (SD 13.7); placebo = 54.7 (SD 10.3) Time points measured: baseline, week 2, week 4, week 8 Montgomery‐Åsberg Depression Rating Scale Mean baseline score: paroxetine = 28.4 (SD 5.3); placebo = 26.2 (SD 5.8) Time points measured: baseline, week 2, week 4, week 8
Notes	Funding source: SmithKline Beecham Declarations of interest: not reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generated using a computer random number generator.
Allocation concealment (selection bias)	Low risk	Study medication kits were sequentially numbered, held and distributed by the hospital pharmacy, and allocated to consecutive participants in strict sequential order.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	While drug and placebo were identical in appearance, only study personnel were reported as blinded.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Unclear risk	It is unclear whether participants were blinded.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	Unclear risk	Withdrawals not reported for individual study arms but only 6/98 (6%) participants were excluded from the ITT analysis. Both multiple imputation and last‐observation‐carried‐forward (LOCF) were used as methods of imputation (only LOCF estimates reported after no difference of imputation method was identified). Because withdrawals were not reported by treatment phase, the effect of incomplete data on effect estimates is unclear.
Selective reporting (reporting bias)	Unclear risk	No study protocol or registration; insufficient information available to assess selective reporting.
Other bias	High risk	No early stopping or unplanned interim analyses but significant differences in analgesic use between groups reported.

Open in a new tab

Goodkin 1990.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 6‐week follow‐up Number of study centres: 2 Country: USA Study setting: university pain clinic and veteran medical centre Study dates: not reported
Participants	General characteristics Number randomised: 42 (trazodone = 22; placebo = 20) Mean age: trazodone = 51.4 (SD 13.1) years; placebo = 56.1 (SD 12.6) years Age range: not reported Sex and gender: trazodone = 8 (36.4%) female, 14 (63.6%) male; placebo = 8 (40%) female, 12 (60%) male Condition: chronic low back pain Mean duration of symptoms: trazodone = 16.7 (SD 14.5) years; placebo = 24.3 (SD 17) years Depression: trazodone = 9 (41%); placebo = 8 (40%) (history of depression) Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: trazodone = 2 (9.1%) Black participants, 2 (9.1%) Hispanic participants, 18 (81.8%) White participants; placebo = 1 (5%) Black participants, 1 (5%) Hispanic participant, 18 (90%) White participants Culture: not reported Language: not reported Occupation: trazodone = 3 (13.6%) unemployed, 19 (86.4%) employed; placebo = 3 (15%) unemployed, 17 (85%) employed Religion: not reported Education: trazodone = mean 11.91 (SD 3.85) years; placebo = mean 12.28 (SD 3.44) years Socioeconomic status: trazodone mean annual income = USD 26,825 (SD 31,149); placebo mean annual income = USD 31,203 (35,003) Social capital: not reported Inclusion criteria 18 years of age or older documented organic changes in the lumbosacral area minimum of one year of continuous back pain or two prior episodes of low back pain at least two weeks in duration with a current episode of at least two weeks Exclusion criteria four or more additional pain sites illiterate suffered myocardial infarction less then six months earlier class IV or worse congestive heart failure treated for or history of cardiac arrhythmia (unless a non‐demand pacemaker was placed) chronic or unstable angina hepatic failure renal failure cancer that might relate to the aetiology of back pain spinal cord or canal abnormalities dementia or other neurological disease that limited capacity to meet assessment demands major psychiatric disorder alcohol or substance abuse/dependence prior to the onset of pain women who planned to become pregnant or refused to use a contraceptive method persons who had a previous therapeutic trial of trazodone or had a history of allergy or sensitivity to trazodone taking more than 200 mg/day of an antidepressant medication
Interventions	Intervention: oral trazodone Target dose: 600 mg/day Run in: 50 mg/day for 3 days; 100 mg/day for 3 days; 150 mg/day for 3 days; 200 mg/day for 3 days; 250 mg/day for 3 days; 300 mg/day for 3 days; 350 mg/day for 3 days; 400 mg/day for 3 days; 450 mg/day for 3 days; 500 mg/day for 3 days; 550 mg/day for 3 days Treatment period: 6 weeks Comparator: inert placebo Rescue analgesia: not reported Concomitant medications: narcotic or NSAIDs permitted at limited doses, other medications (not specified) permitted at baseline doses
Outcomes	Pain intensity (100‐mm VAS) Mean baseline score: trazodone = 6.45 (SD 1.70); placebo = 6.51 (SD 1.49) Time points measured: baseline, week 2, week 4, week 6 Beck Depression Inventory Mean baseline score: trazodone = 16.27 (SD 10.39); placebo = 15.20 (SD 7.01) Time points measured: baseline, week 6
Notes	Funding source: "This work was supported by NIH grants MH18764, MH16744, NIMH Mental Health Clinical Research Center Grant MH41115, grant from the Procter & Gamble Company, a grant from the Stanford University Health Sciences Research and Development Fund, and a grant from the Western Research and Development Office of the Veterans Administration" Declarations of interest: not reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Described as randomised but insufficient information about the sequence generation process reported.
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not described.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Reported as 'double‐blind'; treatments matched and process to unblind participants following treatment described. Blinding reported as successful for research staff and there was no indication that the blinding could have been broken for participants.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants were likely to have been blinded and there was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	3/22 (14%) participants in the trazodone group and 0/20 (0%) participants in the placebo group missing outcome data for pain intensity. No method of imputation described. The imbalance of missing data across groups was enough to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	Unclear risk	No study protocol or registration; insufficient information available to assess selective reporting.
Other bias	Low risk	No early stopping or unplanned interim analyses and only minor between‐group differences in use of co‐interventions.

Open in a new tab

Gould 2020.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 4 Study duration: 12‐week follow‐up Number of study centres: 1 Country: USA Study setting: not reported Study dates: January 2010 to December 2014
Participants	General characteristics Number randomised: 142 (desipramine = 37; placebo = 33; desipramine + cognitive behavioural therapy (CBT) = 34; placebo + CBT = 37) (baseline characteristics reported for n = 141) Mean age: desipramine = 56.5 (SD 11.7) years; placebo = 57.9 (SD 10.9) years; desipramine + CBT = 51.5 (SD 13.7) years; placebo + CBT = 57.8 (SD 9.2) years Age range: not reported Sex and gender: desipramine = 5 (13.5%) female, 32 (86.5%) male desipramine + CBT = 3 (8.1%) female, 34 (91.9%) male placebo = 3 (9.7%) female, 28 (90.3%) male placebo + CBT = 4 (12.1%) female, 29 (87.9%) male Condition: non‐specific low back pain Mean duration of symptoms: not reported Depression: participants with current (1 month) major depression excluded Characteristics that stratify health opportunities Level of country income: high Race: desipramine = 1 (2.7%) Asian participant, 6 (16.2%) Black participants, 4 (10.8%) Multiracial participants, 25 (67.6%) White participants, 1 (2.7%) unknown desipramine + CBT = 1 (2.7%) American Indian or Alaska Native participants, 2 (5.4%) Asian participants, 4 (10.8%) Black participants, 7 (18.9%) Multiracial participants, 1 (2.7%) Pacific Islander participants, 21 (56.8%) White participants, 1 (2.7%) unknown placebo = 1 (3.1%) Asian participant, 5 (15.6%) Black participants, 3 (9.4%) Multiracial participants, 1 (3.1%) Pacific Islander participant, 20 (62.5%) White participants, 2 (6.3%) unknown placebo + CBT = 2 (5.9%) American Indian or Alaska Native participants, 1 (2.9%) Asian participants, 6 (17.6%) Black participants, 2 (5.9%) Multiracial participants, 18 (53.0%) White participants, 5 (14.7%) unknown Ethnicity: desipramine = 6 (16.2%) Hispanic participants desipramine + CBT = 5 (13.5%) Hispanic participants placebo = 4 (12.5%) Hispanic participants placebo + CBT = 10 (30.3%) Hispanic participants Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: desipramine = mean 14.6 (SD 2.5) years desipramine + CBT = mean 15.5 (SD 2.5) years placebo = mean 15.3 (SD 2.2) years placebo + CBT = mean 15.0 (SD 3.1) years Socioeconomic status: not reported Social capital: not reported Inclusion criteria age 18 to 70 non‐specific low back pain (thoracic vertebra T‐6 or below) experience of pain “on a daily basis” ≥ 6 months at a minimum intensity ≥ 4/10 on a 10‐point scale presently not a candidate for back surgery (1 previous back surgery permitted if it was > 5 years ago) agreement to discontinue muscle relaxants, anticonvulsants, antidepressants, and opioids at least 2 weeks before screening and throughout the study if female, not pregnant or lactating, with a negative pregnancy test at screening, and agreement to use reliable contraception for the duration of the study English speaking able to understand the study, and communicate with the team Exclusion criteria current or history of medical illness that would interact with antidepressant medication or confound reports of pain‐related disability (e.g. systolic blood pressure > 160 mm Hg or diastolic > 95 mm Hg, diabetes, renal or hepatic disease, chronic obstructive pulmonary disease, and cancer or class III or IV organic heart disease) significant coexisting orthopaedic or pain problems: sciatica or back pain due to other disorders (e.g. fibromyalgia, vertebral fracture, osteomyelitis, metastatic cancer, rheumatoid arthritis, and spinal stenosis) alcohol or other substance‐use disorders within the past 12 months or positive urine toxicology for a drug of abuse at screening current (1 month) major depression or dysthymia diagnosis; lifetime diagnosis of bipolar disorder (I or II), psychosis, or dementia additional known contraindication of desipramine use (e.g. allergy), or use of other medications or other agents (e.g. herbal supplements that might interact with the study drug use of psychotropics (e.g. antidepressants or anxiolytics) previous use of desipramine for pain treatment use of systemic corticosteroids or corticosteroid injections within 3 months of screening concurrent psychotherapy haematologic abnormality (haemoglobin < 9.4 gm/dL; absolute WBC < 3000/mm, platelets < 100,000) immunosuppression major surgery within the previous 12 months participation in other clinical trials or use of experimental drugs within 2 months of screening
Interventions	Intervention: oral desipramine Target dose: serum concentration of 15 to 65 ng/mL Run in: titrated according to CYP 450 2D6 phenotype; extensive metaboliser phenotype = 10 mg/day for week 1 then weekly increase of 10 mg until reaching 20‐60 mg/day; poor metaboliser phenotype = 10mg/day for weeks 1 & 2 with final dose determined by serum sample of desipramine concentration Treatment period: 12 weeks Comparator: active placebo (benztropine mesylate 0.125 mg/day) Rescue analgesia: not reported Concomitant medications: use of NSAIDs permitted
Outcomes	Pain intensity (Descriptor Differential Scale) Mean baseline score: desipramine = 9.97 (SD 5.05); placebo = 11.96 (SD 4.63); desipramine + CBT = 11.91 (SD 5.07); placebo + CBT = 11.60 (SD 4.37) Time points measured: baseline, week 12 Roland Morris Disability Questionnaire Mean baseline score: desipramine = 10.51 (SD 5.44); placebo = 12.47 (SD 3.75); desipramine + CBT = 11.65 (SD 5.41); placebo + CBT = 12.97 (SD 4.33) Time points measured: baseline, week 12
Notes	Funding source: "This research was supported by the Office of Research and Development, Clinical Sciences Research and Development, Department of Veterans Affairs" Declarations of interest: "The funding organization was not involved in the (1) design and conduct of the study; (2) collection, management, analysis, and interpretation of the data; (3) preparation, review, or approval of the manuscript; or (4) decision to submit the manuscript for publication" Total adverse event outcome data were included from the study trial registry.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generated using a (web‐based) computer random number generator (www.randomizer.org).
Allocation concealment (selection bias)	Low risk	Central allocation at clinical research pharmacy who alone held the allocation key.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants, the study physician, and research personnel were blinded to medication. Active placebo used to ensure blinding and daily doses for both treatments were taken via a single capsule. Dose escalation was determined by the blinded study physician. There was no indication that the blinding could have been broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants were likely to have been blinded and there was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	11/38 (29%) participants in the desipramine group, 16/37 (43%) participants in the desipramine plus CBT group, 9/33 (27%) participants in the placebo group, and 7/34 (21%) participants in the placebo plus CBT group did not complete the study but the number of withdrawals who continued to provide outcome data was not reported. ITT was conducted with last‐observation‐carried‐forward used as the method of imputation. It is likely that participants who withdrew did not continue to provide outcome data, and an inappropriate method of imputation was used, both likely to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	Low risk	All outcomes prespecified in the trial registration were reported in the manuscript.
Other bias	High risk	The trial was stopped early, with only 71% of the planned sample size recruited.

Open in a new tab

Jenkins 1976.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 4‐week follow‐up Number of study centres: 1 Country: United Kingdom Study setting: Royal Air Force in‐patient medical rehabilitation unit Study dates: not reported
Participants	General characteristics Number randomised: 59 (imipramine = 30; placebo = 29) (baseline characteristics reported for n = 44) Mean age: imipramine = 26 years; placebo = 26.7 years Age range: imipramine = 18 to 45 years; placebo = 18 to 49 years Sex and gender: imipramine = 0 (0%) female, 23 (100%) male; placebo = 3 (14.3%) female, 18 (85.7%) male Gender: not reported Condition: low back pain Mean duration of symptoms: not reported Depression: included 15 participants with Beck Depression Inventory score > 14 Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria patients of either sex admitted for in‐patient treatment of low back pain Exclusion criteria pain due to tubercular, neoplastic or metabolic bone disease patients suffering from rheumatoid arthritis patients treated during the previous two weeks with monoamine oxidase inhibitors patients for whom imipramine is contraindicated (e.g. glaucoma, prostatism, advanced renal or liver disease, ischaemic heart disease)
Interventions	Intervention: oral imipramine Target dose: 75 mg/day Run in: none reported Treatment period: 4 weeks Comparator: inert placebo Rescue analgesia: analgesics prescribed when considered 'essential' Concomitant medications: hypnotics (assumed as needed)
Outcomes	Pain intensity (10‐cm VAS) Mean baseline score: imipramine = 4.6; placebo = 4.7 (extracted from graph) Time points measured: baseline, week 1, week 2, week 3, week 4
Notes	Funding source: Geigy Pharmaceuticals Declarations of interest: not reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: ''The treatments were to be randomly allocated". Comment: random allocation used but details of the random sequence generation not reported.
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not described.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Described as double‐blind but there was insufficient information regarding the indistinguishability of trial treatments. While breaking of blinding was reported for the first 6 people, the trial was subsequently restarted, with only naive patients included.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Unclear risk	Self‐report outcome measurement was likely to be influenced by a lack of blinding for some participants.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	7/30 (23%) of the participants in the tofranil group and 8/29 (28%) participants in the placebo group withdrew from treatment and were excluded from the analysis. The amount of missing outcome data and lack of methods to account for missing data are enough to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	High risk	No study protocol or registration; outcomes have been reported incompletely so that they cannot be entered in a meta‐analysis (point estimates and measures of variance not provided) (data included in analyses extracted from figure).
Other bias	Unclear risk	No early stopping or unplanned interim analyses but insufficient information to assess unequal application of co‐interventions.

Open in a new tab

Johnson 2011.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: cross‐over Number of arms: 2 Study duration: 13‐week follow‐up Number of study centres: not reported Country: USA Study setting: not reported Study dates: September 2006 to January 2010
Participants	General characteristics Number randomised: 14 Mean age: 36.93 (SD 13.05) years Age range: not reported Sex and gender: 0 female (0%); 14 male (100%) Condition: chronic low back pain Mean duration of symptoms: not reported Depression: not reported Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria at least 6 months of back pain with no radicular symptoms a minimum 4/10 pain rating for two weeks prior to enrolment no current pain medication use (except acetaminophen) able to comply with study visit schedule and other study requirements capable of performing the experimental tasks Exclusion criteria contraindications for MRI examination (e.g. metallic implants such as pacemakers, surgical aneurysm clips, or known metal fragments embedded in the body) known hypersensitivity to duloxetine or any of the inactive ingredients uncontrolled narrow‐angle glaucoma
Interventions	Intervention: oral duloxetine Target dose: 60 mg/day Run in: 30 mg/day for 1 week Treatment period: 6 weeks each treatment phase Comparator: inert placebo Rescue analgesia: not reported Concomitant medications: acetaminophen permitted
Outcomes	Brief Pain Inventory Mean baseline score: not reported Time points measured: baseline, week 1, week 2, week 6, week 7, week 8, week 12, week 13
Notes	Funding source: "Supported by a research grant from Eli Lilly" Declarations of interest: not reported We contacted the authors of this study to request outcome data from the first phase of cross‐over. The authors responded but reported that the required data were not available.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Described as randomised but insufficient information about the sequence generation process reported.
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not described.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Participant and investigator blinding reported on trial registry, but there was no reported information regarding the indistinguishability of trial treatments.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Unclear risk	It is unclear whether participants were adequately blinded.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	Low risk	No missing outcome data.
Selective reporting (reporting bias)	High risk	Outcomes have been reported incompletely so that they cannot be entered in a meta‐analysis (point estimates and measures of variance not provided).
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Katz 2005.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: cross‐over Number of arms: 2 Study duration: 16‐week follow‐up Number of study centres: 1 Country: USA Study setting: university medical centre Study dates: not reported
Participants	General characteristics Number randomised: 54 (bupropion first phase = 26; placebo first phase = 28) (baseline characteristics reported for n = 44) Mean age: bupropion first phase = 49.8 (SD 10.0) years; placebo first phase = 51.4 (SD 11.4) years Age range: not reported Sex and gender: bupropion first phase = 12 (57.1%) female, 9 (40.9%) male; placebo first phase = 9 (39.1%) female, 14 (60.9%) male Condition: chronic low back pain Mean duration of symptoms: not reported Depression: not specified Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria 18 years of age or older chronic low back pain for 3 months or more Exclusion criteria history of seizure, cranial trauma, or central nervous system disorder history of bulimia or anorexia any significant medical condition other than chronic pain personal or family history of bipolar disorder prior antidepressant treatment except for a single trial of 1 antidepressant medication for chronic low back pain history of excessive alcohol use or any illicit drug use use of any medication that may be associated with lowering of seizure thresholds history of treatment with bupropion or a monoamine oxidase inhibitor pregnancy or lack of adequate birth control in women history of suicide attempt or current suicide intent or plan other pain more severe than chronic low back pain known hypersensitivity to bupropion
Interventions	Intervention: oral bupropion Target dose: 300 mg/day Run in: 150 mg/day for 3 days Treatment period: 7‐week treatment per cross‐over phase with 1‐week washout between phases Comparator: inert placebo Rescue analgesia: not reported Concomitant medications: not reported
Outcomes	Pain intensity (0 to 10 numerical rating scale) Mean baseline score: 5.07 (SD 1.88) Time points measured: baseline, weekly until week 13 Beck Depression Inventory Mean baseline score: not reported Time points measured: not reported
Notes	Funding source: "Supported in part by an investigator‐initiated research grant from GlaxoSmithKline to R.H.D., who has also received research support, consulting fees, or lecture honoraria in the past year from Abbott Laboratories, Eli Lilly & Co., Endo Pharmaceuticals, EpiCept Corporation, NeurogesX, Novartis Pharmaceuticals, Organon, Ortho‐McNeil Pharmaceutical, Pfizer, Purdue Pharma, Ranbaxy Corporation, Reliant Pharmaceuticals, Renovis, and UCB Pharma" Declarations of interest: not reported We contacted the authors of this study to request outcome data from the first phase of cross‐over. The authors responded but reported that the required data were not available.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generated using a computer random number generator.
Allocation concealment (selection bias)	Low risk	Computer‐generated randomisation list; assumed to be central allocation.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants and study personnel described as blinded, drug and placebo as 'matching', and no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants were likely to have been blinded and there was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	10/54 (19%) missing outcome data across both cross‐over phases. ITT analysis assumed to use last‐observation‐carried‐forward method. The amount of missing outcome data for each treatment arm is unclear but the method of imputation is enough to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	Unclear risk	No study protocol or registration; insufficient information available to assess selective reporting.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Khoromi 2007.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: cross‐over Number of arms: 4 Study duration: 38‐week follow‐up (4 x 9‐week treatment phases with 4‐day washouts between) Number of study centres: 1 Country: USA Study setting: NIH Clinical Center Study dates: February 2001 to March 2004
Participants	General characteristics Number randomised: 55 (morphine first phase = 15; nortriptyline first phase = 13; morphine plus nortriptyline first phase = 13; placebo first phase = 14) Median age: 53 years Age range: 19 to 65 years Sex and gender: 25 (45.5%) female, 30 (54.5%) male Condition: chronic lumbar root pain Median duration of symptoms: 5 (range 0.3 to 37) years Depression: people with depression treated with antidepressants within the preceding 6 months & Beck Depression Inventory score ≥ 20 were excluded Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria evidence of lumbar radiculopathy, including pain in one or both buttocks or legs for 3 months or greater for at least 5 days a week and at least one of the following features on the side corresponding to leg pain: sharp and shooting pain below the knee pain evoked by straight leg raising to 60° or less decreased or absent ankle reflex weakness of muscles below the knee sensory loss in L5/S1 distribution electromyographic evidence for L4, L5, or S1 root denervation imaging (MRI, CT/myelogram) evidence of nerve root compression in the lower lumbar region average leg pain of at least 4/10 for the past month on 0‐10 NRS willingness to refrain from making changes in non‐study medications taken for sciatica between 18 and 65 years of age Exclusion criteria serious medical illness involving other organ systems such as unstable angina disease, advanced diabetes, and cancer prostatic disease requiring usage of urological medications pregnancy or lactation history of depression requiring treatment with antidepressants within the 6 months preceding study participation or a score of 20 or greater on the Beck Depression Inventory at the screening visit history of narcotic or alcohol abuse arrow angle glaucoma seizure disorder fibromyalgia pain of greater intensity in any other location than the low back or leg polyneuropathy and peripheral vascular disease associated with symptoms of numbness, or burning pain in the lower extremities allergy to morphine, nortriptyline or benztropine evidence for multi‐somatoform disorder as assessed by the Patient Health Questionnaire‐15 unwillingness to be tapered off of opioids and then maintained drug‐free for two weeks prior to randomisation to study medications for participants on maintenance opioid regimen at the time of study enrolment
Interventions	Intervention 1: oral morphine Target dose: 90 mg/day Run in: 15 mg/day for 3 days increased to 30 mg/day (if 15 mg/day tolerated); 15 mg increase/week weeks 2, 3, 4, 5 Intervention 2: oral nortriptyline Target dose: 100 mg/day Run in: 0 mg/day week 1, 25 mg/day week 2, 50 mg/day week 3, 75 mg/day week 4 Intervention 3: oral morphine plus oral nortriptyline Target dose: 90 mg/day morphine, 100 mg/day nortriptyline Run in: not reported Treatment period: 9 weeks per treatment phase (5 weeks escalation, 2 weeks maintenance, 2 weeks tapering) with 4‐day washout between treatment phases Comparator: active placebo (benztropine 0.25 to 1 g/day) Rescue analgesia: anti‐inflammatory medications and acetaminophen permitted Concomitant medications: changes to analgesic medication regimens, and opioids, SSRIs and tricyclic medications prohibited
Outcomes	Pain intensity (0‐10 numerical scale average leg pain over 24 hours) Mean baseline score: 4.9 (SD 2.43) (completers only, N = 28) Time points measured: baseline, daily during treatment phases (average of 2‐week maintenance phase scores used for primary outcome) Oswestry Low Back Pain Disability Questionnaire Mean baseline score: 30 (SD 15) Time points measured: not reported Beck Depression Inventory Mean baseline score: 8 (SD 6.7) Time points measured: not reported 36‐item Short Form of Health Survey (General Health) Mean baseline score: 68 (SD 20) Time points measured: not reported Adverse effects Time points measured: twice‐weekly throughout study period
Notes	Funding source: "This study was supported by an intramural grant from the National Institute of Dental and Craniofacial Research. MS Contin placebo tablets were a gift from Purdue Pharma" Declarations of interest: not reported We attempted to contact the study authors to request outcome data from first phase of the cross‐over, but we could not find current contact details.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Quote: "Patients were assigned by random numbers within blocks of four to one of four treatment sequences specified by a Latin square." Comment: Random number table probably used.
Allocation concealment (selection bias)	Unclear risk	Quote: "Randomization was performed by the NIH Pharmaceutical Development Service." Comment: Assumed to be central allocation.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants and personnel described as blinded, treatments matched in appearance and active placebo used, but blinding success tests performed indicated unsuccessful blinding of both participants and personnel.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	High risk	Self‐reported outcome assessment likely to have been unblinded (inadequate blinding as indicated by blinding success test).
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	27/55 (49%) of participants failed to complete more than two treatment phases and were excluded from analyses. The amount of missing outcome data and lack of methods to account for missing data are enough to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	Unclear risk	No study protocol or registration; insufficient information available to assess selective reporting.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Konno 2016.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 14‐week follow‐up Number of study centres: 58 Country: Japan Study setting: medical institutions Study dates: May 2013 to July 2014
Participants	Number randomised: 458 (duloxetine = 232; placebo = 226) (baseline characteristics reported for n = 456) Mean age: duloxetine = 60.0 (SD 13.2) years; placebo = 57.8 (SD 13.7) years Age range: not reported Sex/gender: duloxetine = 115 female (50%), 115 male (50%); placebo = 122 female (54.0%), 104 male (46.0%) Condition: chronic low back pain Mean duration of symptoms: duloxetine = 9.8 (SD 10.1) years; placebo = 10.3 (SD 10.6) years Depression: participants with major depressive disorders excluded Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: duloxetine = 230 (100%) participants; placebo = 226 (100%) Asian participants Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria male and female outpatients of age 20 to < 80 years low back pain persisting for at least 6 months used NSAIDs for at least 14 days per month for an average of 3 months before the start of the study and for at least 14 days during the 1‐month period before the start of the study, regardless of dose of NSAIDs and route of administration did not have radiculopathy symptoms or other specific low back diseases had a BPI (Brief Pain Inventory) pain (average pain) of 4 at visit 1 (week ‐1 to ‐2) and visit 2 (week 0) Exclusion criteria history of low back surgery receiving invasive treatment for the relief of low back pain within 1 month before visit 1 requiring crutches or a walker diagnosed as having major depressive disorders according to the Mini International Neuropsychiatric Interview or suicidal tendencies according to the Columbia‐Suicide Severity Rating Scale
Interventions	Intervention: oral duloxetine Target dose: 60 mg/day Run in: 20 mg/day for 1 week; 40 mg/day for 1 week Treatment period: 14 weeks Comparator: inert placebo Rescue analgesia: use of drugs with an analgesic effect permitted for up to 3 consecutive days and for up to a cumulative total of 20 days Concomitant medications: use of analgesics (including NSAIDs) and other therapeutic drugs (including muscle relaxants, antidepressants, sedatives, and benzodiazepines) prohibited
Outcomes	Pain intensity (Brief Pain Inventory) Mean baseline score: duloxetine = 5.14 (SD 1.11); placebo = 5.09 (1.04) Time points measured: baseline, week 14 Roland‐Morris Disability Questionnaire Mean baseline score: duloxetine = 7.59 (SD 4.38); placebo = 7.77 (SD 4.77) Time points measured: baseline, week 14 European Quality of Life Questionnaire‐5 Dimension Mean baseline score: duloxetine = 0.69 (SD 0.11); placebo = 0.69 (SD 0.10) Time points measured: baseline, week 14 36‐Item Short‐Form Health Survey (General Health) Mean baseline score: duloxetine = 58.54 (SD 16.53); placebo = 58.35 (SD 16.78) Time points measured: baseline, week 14
Notes	Funding source: "Shionogi & Co. Ltd., Eli Lilly Japan K.K., and Eli Lilly and Company funds were received in support of this work" Declarations of interest: not reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generated using a computer random number generator (SAS Version 9.1 PLAN procedure).
Allocation concealment (selection bias)	Low risk	Quote: "After allocation, the assignment table was sealed by the investigator in charge of allocation, and remained inaccessible by all involved parties until after finalization of the clinical report." Comment: 'all parties' assumed to include participants and other study personnel.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants and study personnel described as blinded, and the appearance and labelling of study drug was indistinguishable from placebo. There was no indication that the blinding could have been broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants were likely to have been blinded and there was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	23/232 (9.9%) participants in the duloxetine group and 26/226 (11.5%) participants in the placebo group were reported to have discontinued treatment but the proportion of these participants who did not continue to provide outcome data was not reported. Last‐observation‐carried‐forward used in 'full analysis set'. It is likely that participants who withdrew did not continue to provide outcome data and an inappropriate method of imputation was used, both likely to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	Low risk	All outcomes prespecified in the trial registration were reported in the manuscript.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Kurniawati 2020.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 14‐day follow‐up Number of study centres: 1 Country: Indonesia Study setting: Neurology Department of Teaching Hospital Study dates: not reported
Participants	General characteristics Number randomised: 63 (amitriptyline = 33; placebo = 30) Median age: amitriptyline = 42 years; placebo = 37.5 years Age range: amitriptyline = 20 to 49 years; placebo = 20 to 49 years Sex and gender: imipramine = 26 (78.8%) female, 7 (21.2%) male; placebo = 21 (70.0%) female, 9 (30.0%) male Condition: chronic non‐specific low back pain Mean duration of symptoms: not reported Depression: participants with depression excluded Characteristics that stratify health opportunities Level of country income: lower‐middle Race and ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: amitriptyline = 7 (21.2%) junior high school, 26 (78.8%) senior high school; placebo = 10 (30.3%) junior high school, 20 (66.7%) senior high school Socioeconomic status: not reported Social capital: not reported Inclusion criteria* non‐specific chronic lower back pain "not received treatment yet" treatment with analgesic drugs within a week before study enrolment aged 20 to 49 years ≥ 4 VAS score willing to participate in the research by sign informed consent Exclusion criteria* "depression, such as the heart is relatively unknown and noticed by radiographers and cardiologists" allergy to acetaminophen or amitriptyline taking drugs that could interact with acetaminophen or amitriptyline, history of heart disease cardiac arrhythmias, history of liver disease above range transaminase enzymes epilepsy pregnant or breastfeeding gastrointestinal disease *includes uninterpretable criteria
Interventions	Intervention: oral amitriptyline Target dose: 10 mg/day Run in: not reported Treatment period: 14 days Comparator: inert placebo Rescue analgesia: not reported Concomitant medications: participants in both groups received acetaminophen 1500 mg/day
Outcomes	Pain intensity (VAS) Median baseline score: amitriptyline = 5.0 (range 4.2 to 8.5); placebo = 5.2 (range 4.1 to 8.2) Time points measured: baseline, day 14 Adverse events Time points measured: not reported
Notes	Funding source: internal funding Declarations of interest: none declared
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Described as randomised but insufficient information about the sequence generation process reported.
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not described.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Described as double‐blind but there was no reported information regarding the indistinguishability of trial treatments.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Unclear risk	It is unclear whether participants were adequately blinded.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	Unclear risk	No information regarding the number of participants who provided outcome data (pain intensity outcomes).
Selective reporting (reporting bias)	High risk	No study protocol or registration; the number of participants who provided pain intensity outcome data was not provided.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Marks 2014.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 82 days Number of study centres: 1 Country: USA Study setting: University Medical Center Study dates: October 2010 to September 2011
Participants	General characteristics Number randomised: 11 (milnacipran = 7; placebo = 4) Mean age: not reported Age range: not reported Sex and gender: milnacipran = 6 (85.7%) female, 1 (14.3%) male; 2 (50.0%) female, 2 (50.0%) male Condition: chronic radicular pain Mean duration of symptoms: not reported Depression: not specified Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: not reported Ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria chronic (> 6 months) radiologically confirmed radicular pain at least 5 days a week (described as sharp or shooting below the level of the knee associated with lumbar or sacral disc disease without suspicion of recent injury) remote (> 1 year ago) history of surgical intervention (e.g. “failed back syndrome”) permitted if current symptoms met severity criterion participant‐rated radicular pain VAS ≥ 40 mm at screen and baseline visits signed written informed consent to participate in the study Exclusion criteria use of antidepressant or anticonvulsant medication within 4 weeks of screening visit (6 weeks for fluoxetine) received procedural intervention (including but not limited to lumbar epidural steroid injection, facet median nerve block or radiofrequency ablation, spinal cord stimulator) within 3 months of screen history of bipolar disorder or psychosis as confirmed by the Mini‐International Neuropsychiatric Interview drug dependence or substance abuse disorder according to DSM‐IV‐TR criteria (excluding nicotine) within the preceding 6 months
Interventions	Intervention: oral milnacipran Target dose: 100 mg/day (increased to 200 mg/day on day 15 or later at investigator's discretion) Run in: 12.5 mg/day on day 1, 25 mg/day on days 2 to 3, 50 mg/day on days 4 to 7 Treatment period: 82 days (medicine washout days 71 to 82) Comparator: inert placebo Rescue analgesia: not reported Concomitant medications: not reported
Outcomes	Pain intensity (VAS ‐ radicular pain) Mean baseline score: milnacipran = 58.7 (SD 9.4); placebo = 67.7 (SD 26.5) Time points measured: baseline, week 1, week 2, week 4, week 6, week 8, week 10 Oswestry Low Back Pain Disability Questionnaire Mean baseline score: milnacipran = 40.1 (SD 7.5); placebo = 44.7 (SD 11.2) Time points measured: baseline, week 1, week 2, week 4, week 6, week 8, week 10 Beck Depression Inventory Mean baseline score: milnacipran = 13.8 (SD 7.3); placebo = 14.3 (SD 9.3) Time points measured: baseline, week 1, week 2, week 4, week 6, week 8, week 10 36‐Item Short‐Form Health Survey Mean baseline score: milnacipran = 96.8 (SD 6.3); placebo = 101.3 (SD 4.0) Time points measured: baseline, week 1, week 2, week 4, week 6, week 8, week 10
Notes	Funding source: "This study was supported by an investigator‐initiated grant from Forest Laboratories, SAV‐MD‐12" Declarations of interest: "Dr Patkar has served as a consultant to Braeburn, Bristol‐Myers Squibb, Cubist, Otsuka, and Titan; has received grant/research support from EnVivio, Forest, Lundbeck, Shire, Sunovion, and Titan; has received honoraria from Alkermes Otsuka, and Pfizer; and has served on the speakers’ or advisory boards of Alkermes, Otsuka, and Sunovion. Drs Marks and Pae report no conflicts of interest related to the subject of this article"
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Described as randomised but insufficient information about the sequence generation process reported.
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not described.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Described as double‐blind, placebo described as matching, and no indication that the blinding could have been broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants likely to have been blinded and no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	2/7 (29%) of participants in the milnacipran group and 1/4 (25%) of participants in the placebo group discontinued the study and did not provide outcome data. ITT was conducted with last‐observation‐carried‐forward used as the method of imputation. The amount of missing outcome data and inappropriate method of imputation are enough to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	High risk	Outcome assessments obtained at weeks 1, 2, 4, 6, 8, and 10 but only reported for week 10.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

NCT01225068.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 6‐week follow‐up Number of study centres: not reported Country: USA Study setting: Northwestern University Feinberg School of Medicine Study dates: not reported
Participants	Number randomised: 40 (milnacipran = 20; placebo = 20) Mean age: milnacipran = 47.1 (SD 11.6) years; placebo = 48.3 (SD 9.1) years Age range: not reported Sex and gender: milnacipran = 10 (50%) female, 10 (50%) male; placebo = 11 (55%) female, 9 (45%) male Condition: chronic neuropathic low back pain Mean duration of symptoms: not reported Depression: participants with Beck Depression Inventory Score > 30 excluded Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria history of low back pain for a minimum of 6 months with radiation to leg or buttocks over 18 and under 70 years of age visual analogue scale pain score > 50 mm generally stable health willing to abstain from alcohol during the course of the study if female, must be post‐menopausal, or practicing a highly effective method of contraception or abstinence during the course of the study able to read and understand instructions and the questionnaires willing to participate in daily data collection requirements via telephone able understand all aspects of the study, and willing to sign an informed consent form in that regard Exclusion criteria low back pain associated with systemic signs or symptoms (e.g. fever or chills) evidence of rheumatoid arthritis, ankylosing spondylitis, acute vertebral fractures, fibromyalgia, history of surgery or tumour in the back involvement in litigation regarding back pain or other disability claim, or receiving workmen's compensation, or seeking either as a result of low back pain neurological disorder including history of seizures major psychiatric disorder during the past six months active suicidal ideation or recent suicidal behaviour significant other medical disease such as unstable diabetes mellitus, congestive heart failure, coronary or peripheral vascular disease, chronic obstructive lung disease or malignancy significant renal disease or severe renal insufficiency history of, or current, substance abuse/dependence significantly abnormal laboratory values pregnant or lactating any time during the course of the study known sensitivity to milnacipran or other SNRI glaucoma taking any monoamine oxidase inhibitors (MAOI), sibutramine, digoxin, tricyclic antidepressants, other SNRI, opioids Beck Depression Inventory Score > 30 current sleep disorder
Interventions	Intervention: oral milnacipran Target dose: 100 mg/day with option to increase to 200 mg/day after 2 weeks Run in: 'gradual' escalation for 1 week Treatment period: 6 weeks Comparator: inert placebo Rescue analgesia: not reported Concomitant medications: not reported
Outcomes	Pain intensity (0‐100 VAS) Mean baseline score: not reported Time points measured: baseline, week 1, week 2, week 6 Adverse events Time points measured: week 8
Notes	Funding source: not reported Declarations of interest: not reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Described as randomised but insufficient information about the sequence generation process reported.
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not described.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Participant‐ and investigator‐blind reported on trial registry, placebo described as matching, but insufficient information to assess whether the blind could have been broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Unclear risk	Participants were described as blinded, but there was insufficient information to assess whether the blind could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	4/20 (20%) participants in the milnacipran arm and 1/20 (5%) participants in the placebo arm were excluded from the analysis. Per‐protocol analysis used. The difference in amount of missing outcome data across intervention groups and the analysis approach are likely to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	Low risk	All outcomes prespecified in the trial registration were reported.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Pheasant 1983.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: cross‐over Number of arms: 2 Study duration: 16‐week follow‐up Number of study centres: 1 Country: USA Study setting: adult low‐back pain clinic of orthopaedic hospital Los Angeles Study dates: 1 December 1978 to 31 December 1979
Participants	Number randomised: 16 Mean age: 47.2 years Age range: 22 to 68 years Sex and gender: 12 (75%) female, 4 (25%) male Condition: chronic low back pain (> 12 months) Mean duration of symptoms: 9.9 (range 1 to 37) years Depression: not specified Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria Not reported Exclusion criteria eye pain glaucoma conduction defects urinary retention pregnancy
Interventions	Intervention: oral amitriptyline Target dose: 150 mg/day Run in: not reported Treatment period: 6‐week treatment per cross‐over phase with 2‐week washout between phases Comparator: active placebo (atropine 0.2 mg) Rescue analgesia: acetaminophen or aspirin, with or without codeine as required Concomitant medications: not reported
Outcomes	Self‐administered activity questionnaire Mean baseline score: not reported Time points measured: baseline, week 2, week 4, week 6
Notes	Funding source: not reported Declarations of interest: not reported We attempted to contact the study authors to request outcome data from the first phase of cross‐over, but current contact details could not be found.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Described as randomised but insufficient information about the sequence generation process reported.
Allocation concealment (selection bias)	Unclear risk	Quote: "The treatment sequence was known only by the statistician and the pharmacist". Comment: the method of concealment is not described in sufficient detail to allow a definite judgement.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants and examining physician blinded to drug and placebo sequence. Active placebo used and study drugs and placebo were matching in presentation. There was no indication that the blinding could have been broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants likely to have been blinded and no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	7/16 (44%) of all participants excluded from analysis. No method to account for missing outcome data was reported. The amount of missing outcome data is enough to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	Unclear risk	No study protocol or registration; insufficient information available to assess selective reporting.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Pirbudak 2003.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 6‐month follow‐up Number of study centres: not reported Country: Turkey Study setting: not reported Study dates: not reported
Participants	General characteristics Number randomised: 60 (amitriptyline = 30; placebo = 30) Mean age: amitriptyline = 35.7 (SD 7.5) years; placebo = 35.1 (SD 4.4) years Age range: not reported Sex and gender: amitriptyline = 24 female (80.0%), 6 male (20.0%); placebo = 19 female (63.3%), 11 male (36.7%) Condition: acute discogenic back pain Median duration of symptoms: amitriptyline = 3 weeks; placebo = 4 weeks Depression: not specified Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria aged 20 to 48 years disc herniation and unilateral sciatic pain ≤ 3 months duration structural pathology confirmed radiologically Oswestry Disability Index score > 20 Exclusion criteria signs of bilateral root compression undergone surgical treatment neurological deficit and/or previous lumbar spine surgery severe cardiac, pulmonary, haematological, gastrointestinal, kidney disease, glaucoma, urinary retention known allergies to the study drugs
Interventions	Intervention: oral amitriptyline Target dose: 50 mg/day Run in: not reported Treatment period: 6 months Comparator: inert placebo Rescue analgesia: not reported Concomitant medications: both groups received a single dose of 10 mg betamethasone dipropionate, 4 mg betamethasone sodium phosphate and 0.125% bupivacaine administered epidurally
Outcomes	Pain intensity (0‐10 VAS) Mean baseline score: amitriptyline = 8.1 (SD 2.1); placebo = 8.4 (SD 2.6) Time points measured: baseline, week 2, week 6, 3 months, 6 months Oswestry Disability Index Mean baseline score: amitriptyline = 51 (SD 12.7); placebo = 55 (SD 12.5) Time points measured: baseline, week 2, week 6, 3 months, 6 months Adverse events Time points measured: not reported
Notes	Funding source: not reported Declarations of interest: not reported
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Described as randomised but insufficient information about the sequence generation process reported.
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not described.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Described as double‐blind but there was no reported information regarding the indistinguishability of trial treatments.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Unclear risk	It is unclear whether participants were adequately blinded.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	Unclear risk	No missing outcome data specifically reported and the number of participants who provided outcome data for analyses at 3‐ and 6‐month endpoints was not provided.
Selective reporting (reporting bias)	High risk	No study protocol or registration; the number of participants who provided outcome data for analyses at 3‐ and 6‐month endpoints was not provided.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Schliessbach 2018.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: cross‐over Number of arms: 4 Study duration: 2‐hour follow‐up Number of study centres: 1 Country: Switzerland Study setting: University Department of Anesthesiology and Pain Therapy Study dates: July 2010 to April 2014
Participants	General characteristics Number randomised: 50 Mean age: 54.4 (SD 17.3) years Age range: not reported Sex and gender: 32 female (64%), 18 male (36%) Condition: chronic low back pain Mean duration of symptoms: 11.2 (SD 12.3) years Depression: participants with major depression (Beck Depression Inventory short form score > 9) were excluded. Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria 18 to 80 years chronic low back pain of at least 3 months Exclusion criteria pain intensity at rest < 3/10 on the numerical rating scale (NRS) at the time of testing suspected radicular pain (as defined by leg pain associated with an MRI finding of a herniated disc or foraminal stenosis) signs or suspicion of neurological dysfunction at the tested sites pregnancy (as assessed by pregnancy test) breastfeeding ongoing treatment with an antidepressant opioid or anticonvulsant intake of centrally active substances (including drug or alcohol abuse) known allergy or pharmacological contraindications to imipramine or tolterodine multisite or widespread pain, systemic inflammatory or rheumatological disease major depression (Beck Depression Inventory short form score > 9)
Interventions	Interventions Intervention 1: oral imipramine Target dose: 75 mg (single dose) Intervention 2: oral oxycodone Target dose: 15 mg (single dose) Intervention 3: oral clobazam Target dose: 20 mg (single dose) Run in: none Treatment period: 2 hours (1 week washout between treatment phases) Comparator: active placebo (tolterodine 1 mg) Rescue analgesia: not permitted Concomitant medications: not permitted
Outcomes	Pain intensity (0‐10 NRS) Mean baseline score: 5.6 (SD 1.8) Time points measured: baseline, 30 minutes, 60 minutes, 90 minutes, 2 hours Beck Depression Inventory Mean baseline score: 2.3 (SD 2.3) Time points measured: baseline
Notes	Funding source: "Swiss National Science Foundation SNF in the context of the Special Program for University Medicine SPUM 33CM30_140339 to author MC" Declarations of interest: none declared We contacted the authors of this study to request data from the first phase of the cross‐over, but we received no response to our emails.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generated using a computer random number generator.
Allocation concealment (selection bias)	Low risk	Quote: "Randomization is performed by the hospital pharmacy". Comment: central allocation
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants and study personnel described as blinded. Active placebo that was indistinguishable in appearance from the study drug was used. There was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants were likely to have been blinded and there was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	18/35 (51%) of participants in the imipramine group discontinued after the first of 3 cross‐over phases and only 15 participants were included in subsequent phases. No method to account for missing outcome data was reported. The amount of missing outcome data is enough to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	High risk	While all outcomes specified in the trial registration and protocol were reported in the manuscript, the trial commenced (July 2010) before trial registration (August 2010) and publication of the protocol (2015).
Other bias	High risk	The target sample size of 150 participants specified in the trial registration and protocol were not recruited.

Open in a new tab

Schukro 2016.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: cross‐over Number of arms: 2 Study duration: 10‐week follow‐up Number of study centres: 1 Country: Austria Study setting: outpatient clinic of the department of special anaesthesia and pain therapy at a medical university Study dates: May 2010 to September 2013
Participants	General characteristics Number randomised: 41 (duloxetine first phase = 16; placebo first phase = 18; duloxetine second phase = 15; placebo second phase = 11) Mean age: 57.9 (SD 13.4) years Age range: not reported Sex and gender: 21 female (51%) 20 male (49%) Condition: chronic low back pain with a neuropathic component Mean duration of symptoms: 18 (range 6 to 70) months Depression: participants with mild depression present for more than 12 months (≥ 10 points on the Beck Depression Inventory) were excluded Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria 18 to 80 years chronic low back and leg pain (> 6‐month duration) visual analogue scale (VAS) score > 5 cm on 10‐cm VAS scale painDETECT score >12 Exclusion criteria prior use of opioids classified by the World Health Organization as level III (e.g. fentanyl, tapentadol, morphine, hydromorphone, buprenorphine, and oxycodone) mild depression present for more than 12 months (defined as ≥ 10 points on the Beck Depression Inventory use of antidepressants or benzodiazepines 6 months before study entry drug abuse pregnancy severe coexisting diseases (such as severe heart failure, severe hypertension, glaucoma, convulsion, and kidney dysfunction)
Interventions	Intervention: oral duloxetine Target dose: 120 mg/day Run in: 30 mg/day to 60 mg/day first week, 60 mg/day to 120 mg/day second week Treatment period: 4‐week treatment per cross‐over phase with 2‐week washout between phases Comparator: inert placebo Rescue analgesia: metamizole up to 3000 mg/day and/or up to 600 mg tramadol/day Concomitant medications: medication that could interfere with pain such as analgesic medication (non‐opioids or opioids), antidepressants, and anticonvulsants discontinued prior to treatment
Outcomes	Pain intensity (10‐cm VAS) Mean baseline score: 6.8 (SD 1.5) Time points measured: baseline, twice daily during final week of 4‐week treatment period 36‐item Short Form Health Survey 2nd Edition Mean baseline score: physical component summary = 28.4 (SD 8.7); mental component summary = 48.9 (SD 11.4) Time points measured: baseline, week 4
Notes	Funding source: quote: "supported by the Medical Scientific Fund of the Mayor of the City of Vienna, Vienna, Austria, which was acquired by Sibylle Pramhas, M.D." Declarations of interest: none declared We contacted the authors of this study to request data from the first phase of the cross‐over, but we received no response to our emails.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generated using a computer random number generator.
Allocation concealment (selection bias)	Low risk	Study drug kits were prepared in a hospital pharmacy and administered according to the randomisation list which was held by an independent research nurse (central allocation).
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants and study physicians were described as blinded, the study drugs and placebo matched in appearance, and there was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants were likely to have been blinded and there was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	16/41 (37%) randomised participants excluded from ITT analysis. The amount of missing outcome data is enough to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	High risk	Beck Depression Inventory was prespecified as an outcome in the trial registration and specified in the manuscript methods but results were not reported.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Skljarevski 2009a.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 4 Study duration: 13‐week follow‐up Number of study centres: not reported Country: USA and Argentina Study setting: not reported Study dates: December 2006 to November 2007
Participants	General characteristics Number randomised: 404 duloxetine 20 mg = 59 duloxetine 60 mg = 116 duloxetine 120 mg = 112 placebo = 117 Mean age: duloxetine 20 mg = 52.9 (SD 12.8) years duloxetine 60 mg = 53.3 (SD 14.7) years duloxetine 120 mg = 54.9 (SD 14.8) years placebo = 54.0 (SD 13.5) years Age range: not reported Sex and gender: duloxetine 20 mg = 36 (61.0%) female, 23 (39.0%) male duloxetine 60 mg = 67 (57.8%) female, 49 male (42.2%) duloxetine 120 mg = 65 (58.0%) female, 47 (42.0%) male placebo = 64 (54.7%) female, 53 (45.3%) male Condition: chronic low back pain Mean duration of symptoms: duloxetine 20 mg = 12.5 (SD 11.7) years duloxetine 60 mg = 10.5 (SD 11.1) years duloxetine 120 mg = 13.9 (SD 13.0) years placebo = 10.3 (SD 9.5) years Depression: participants with major depressive disorder excluded Characteristics that stratify health opportunities Level of country income: high (USA) & upper‐middle (Argentina) Race and ethnicity: duloxetine 20 mg = 7 (11.9%) African participants; 46 (78.0%) Caucasian [understood to mean White] participants; 6 (10.2%) Hispanic participants duloxetine 60 mg = 11 (9.5%) African participants; 91 (78.4) Caucasian participants; 1 (0.9%) East Asian participant; 13 (11.2%) Hispanic participants duloxetine 120 mg = 5 (4.5%) African participants; 92 (82.1%) Caucasian participants; 1 (0.9%) East Asian participant; 11 (9.8%) Hispanic participants; 3 (2.7%) West Asian participants placebo = 10 (8.5%) African participants; 93 (79.5%) Caucasian participants; 1 (0.9%) East Asian participant; 11 (9.4%) Hispanic participants; 1 (0.9%) Native American participant; 1 West Asian (0.9%) participant Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria clinical diagnosis of chronic low back pain pain either restricted to lower back (Class 1) or associated with radiation to the proximal portion of the lower limb only (Class 2 according to the Quebec Task Force on Spinal Disorders) pain present on most days for ≥ 6 months average weekly pain ratings ≥ 4 during the week prior to randomization Exclusion criteria clinical or radiographic evidence of radicular compression or spinal stenosis presence of spondylolisthesis grade 3–4 history of ≥ 1 low back surgery any low back surgery within 12 months invasive procedures to reduce low back pain within 1 month major depressive disorder body mass index > 40 seeking disability compensation related to back pain
Interventions	Intervention 1: oral duloxetine Target dose: 20 mg/day Intervention 2: oral duloxetine Target dose: 60 mg/day Intervention 3: oral duloxetine Target dose: 120 mg/day Run in: 30 mg/day increased by 30 mg/week until target dose reached (60 mg & 120 mg arms only) Treatment period: 13 weeks Comparator: inert placebo Rescue analgesia: 'episodic use' (≤ 3 consecutive days, ≤ 20 total days) of short‐acting analgesics permitted Concomitant medications: continued NSAID therapy permitted at pre‐trial dose and frequency; regular use of antidepressants, anticonvulsants, opioids, muscle relaxants and other analgesics prohibited.
Outcomes	Pain intensity (0‐10 Likert scale, average of weekly 24‐hour pain ratings) Mean baseline score: duloxetine 20 mg = 6.4 (SD 1.4); duloxetine 60 mg = 6.2 (SD 1.4); duloxetine 120 mg = 6.1 (SD 1.5); placebo = 6.2 (SD 1.3) Time points measured: baseline & weekly until week 13 Brief Pain Inventory ‐ Severity* Mean baseline score: duloxetine 20 mg = 6.3 (SD 1.6); duloxetine 60 mg = 5.9 (SD 1.7); duloxetine 120 mg = 6.0 (SD 1.6); placebo = 6.1 (SD 1.5) Time points measured: baseline, week 13 Roland‐Morris Disability Questionnaire Mean baseline score: duloxetine 20 mg = 10.00 (SD 4.82); duloxetine 60 mg = 8.91 (SD 4.56); duloxetine 120 mg = 8.72 (SD 4.82); placebo = 8.32 (SD 4.92) Time points measured: baseline, week 13 Beck Depression Inventory–II Mean baseline score: duloxetine 20 mg = 6.02 (SD 6.33); duloxetine 60 mg = 7.05 (SD 7.17); duloxetine 120 mg = 5.20 (SD 5.00); placebo = 6.15 (SD 7.52) Time points measured: baseline, week 13 EuroQoL Questionnaire–5 Dimension Mean baseline score: not reported Time points measured: baseline, week 13 36‐item Short Form Health Survey Mean baseline score: not reported Time points measured: baseline, week 13 *Brief Pain Intensity ‐ Severity average pain scores used in short‐term analysis for pain intensity (point estimates and measures of variance not reported for trial primary outcome of 0‐10 weekly 24‐hour average pain scores)
Notes	Funding source: "Study design, funding and drugs supplied by Eli Lilly and Company" Declarations of interest: "Authors V. Skljarevski, M. Ossanna, H. Liu‐Seifert, Q. Zhang, A. Chappell, S. Iyengar and M. Detke are or were at the time of submission employees of Eli Lilly and Company and may be minor shareholders" We extracted pain intensity immediate‐term, disability short‐term, depressive symptoms short‐term, and total adverse event data from the study trial registry.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generated using a computer random number generator with an Interactive Voice Response System (Perceptive Informatics Inc., Waltham, MA, USA).
Allocation concealment (selection bias)	Low risk	Interactive Voice Response System (assumed to be centrally controlled).
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants and study personnel reported as blinded. Active and placebo treatments identical in appearance, smell and taste. There was no indication that the blinding could have been broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants were likely to have been blinded and there was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	16/59 (27%) participants in the duloxetine 20 mg group, 36/116 (31%) participants in the duloxetine 60 mg, 50/112 (45%) participants in the duloxetine 120 mg group and 35/117 (30%) participants in the placebo group discontinued treatment but the proportion of these participants who continued to provide outcomes was not reported. The primary ITT analysis excluded only 3/59 (5%) participants in the duloxetine 20 mg group, 8/116 (7%) participants in the duloxetine 60 mg group, 4/112 (4%) participants in the duloxetine 120 mg group and 4/117 (4%) participants in the placebo group, but last‐observation‐carried‐forward was used to impute missing data. While the amount of missing outcome data is unclear, the method of imputation is likely to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	Low risk	All outcomes prespecified in the trial registration were reported in the manuscript or in the trial registry itself. Outcomes missing point estimates or measures of variance reported in trial registry.
Other bias	Low risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Skljarevski 2010a.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 13‐week follow‐up Number of study centres: 18 Country: Brazil, France, Germany, Mexico, Netherlands Study setting: not reported Study dates: January 2007 to October 2008
Participants	General characteristics Number randomised: 236 (duloxetine = 115; placebo = 121) Mean age: duloxetine = 51.8 (SD 14.9) years; placebo = 51.2 (SD 13.5) years Age range: not reported Sex and gender: duloxetine = 71 (61.7%) female, 44 (38.3%) male; placebo = 73 (60.3%) female, 48 (39.7%) male Condition: chronic low back pain Mean duration of symptoms: duloxetine = 8.8 (SD 8.8) years; placebo = 9.5 (SD 8.6) years Depression: participants with major depressive disorder excluded Characteristics that stratify health opportunities Level of country income: high (France, Germany, Netherlands); upper‐middle (Brazil, Mexico) Race/ethnicity: duloxetine = 6 (5.2%) African participants, 85 (73.9%) Caucasian [understood to mean White] participants, 23 (20.0%) Hispanic participants, 1 (0.9%) Native American participant; placebo = 6 (5.0%) African participants; 91 (75.2%) Caucasian participants; 2 (1.7%) East Asian participants; 21 (17.4%) Hispanic participants; 1 (0.8%) Native American participant Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria ≥ 18 years of age Low back pain (T‐6 or lower) for most days for the past 6 months or longer weekly mean of 24‐hour average pain score ≥ 4 at baseline Exclusion criteria clinical signs and/or radiographic evidence of spinal stenosis, high‐grade spondylolisthesis (grade 3 and 4), or spinal fracture clinical signs and/or radiographic or electrophysiology evidence of radicular compression history of more than 1 low back surgery low back surgery within 12 months before study entry invasive procedures aimed to reduce low back pain within the past month before study entry participation in previous studies investigating duloxetine hydrochloride treated with monoamine oxidase inhibitor within 14 days of randomization previous diagnosis of psychosis, bipolar disorder, or schizoaffective disorder current major depressive disorder disability compensation or litigation female participants who were pregnant or breast‐feeding body mass index > 40 kg/m² completed daily diaries for < 70% of the days between screening and randomization
Interventions	Intervention: oral duloxetine Target dose: 60 or 120 mg/day (depending on response) Run in: 30 mg/day for 1 week; 60 mg/day for 6 weeks; 'non‐responders' < 30% reduction in mean Brief Pain Inventory scores increased to 120 mg/day Treatment period: 13 weeks Comparator: inert placebo Rescue analgesia: use of short‐acting analgesics (defined as no more than 3 consecutive days or no more than 20 total days) was allowed. Concomitant medications: participants regularly using (≥ 4 days/month for 3 months before study entry) therapeutic doses of NSAIDs or acetaminophen permitted to continue these therapies as long as the doses or frequency were not changed during the study. Use of antidepressants, anticonvulsants, muscle relaxants, or analgesics (other than NSAIDs), and procedures aimed to relieve pain (including acupuncture, chiropractic treatment, and transcutaneous electrical nerve stimulation) were not permitted.
Outcomes	Pain intensity (Brief Pain Inventory 0‐10 Scale, 24‐hour average) Mean baseline score: duloxetine = 5.9 (SD 1.6); placebo = 6.0 (SD 1.7) Time points measured: baseline, week 13 Roland‐Morris Disability Questionnaire‐24 Mean baseline score: not reported Time points measured: baseline, week 13 EuroQoL Questionnaire: 5 Dimensions Mean baseline score: duloxetine = 0.49 (SD 0.31); placebo = 0.49 (SD 0.32) Time points measured: baseline, week 13 36‐Item Short Form Health Status Survey (General Health) Mean baseline score: duloxetine = 16.27 (SD 4.00); placebo = 15.75 (SD 4.03) Time points measured: baseline, week 13 Beck Depression Inventory‐II Mean baseline score: duloxetine = 3.65; placebo = 3.87 Time points measured: baseline, week 13
Notes	Funding source: "Sponsored by Eli Lilly and Company, Indianapolis, IN, USA."; "Corporate/Industry and Foundation funds were received in support of this work" Declarations of interest: "Drs. Skljarevski, Desaiah, Liu‐Seifert, Zhang, Chappell, and Iyengar are employees and stockholders of Eli Lilly and Company. Dr. Detke was a full‐time employee and a major stock holder of Eli Lilly and Company until March 2009 and is currently a full‐time employee and a major stock holder of Medavante Corporation. Dr. Atkinson serves on Lilly Pain Advisory Board. Dr. Backonja serves on Lilly Pain Advisory Board and in addition performed clinical trials and received research funding from Allergan, Astellas, Johnson and Johnson, Lilly, Merck, NeurogesX, and Pfizer." We extracted depressive symptoms and health‐related quality of life short‐term data from the study trial registry.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Described as randomised but insufficient information about the sequence generation process reported.
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not described.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Described as double‐blind but there was no reported information regarding the indistinguishability of trial treatments.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Unclear risk	It is unclear whether participants were adequately blinded.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	31/115 (27%) participants in the duloxetine group and 23/121 participants in the placebo group discontinued treatment but the proportion of these participants who continued to provide outcome data was not reported. 4/115 (3%) participants in the duloxetine group and 5/121 (4%) participants in the placebo group were excluded from the ITT analysis. Last‐observation‐carried‐forward was used as the method of imputation. It is likely that participants who withdrew did not continue to provide outcome data, and an inappropriate method of imputation was used, both likely to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	Low risk	All outcomes prespecified in the trial registration were reported in the manuscript or in the trial registry itself. Missing point estimates or measures of variance for outcomes were reported in the trial registry.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Skljarevski 2010b.

*Study characteristics*
Methods	Methods Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 12‐week follow‐up Number of study centres: 27 Country: Germany, Netherlands, Poland, Russia, Spain, USA Study setting: not reported Study dates: not reported
Participants	General characteristics Number randomised: 401 (duloxetine = 198; placebo = 203) Mean age: duloxetine = 54.9 (SD 13.7) years; placebo = 53.4 (SD 14.2) years Age range: not reported Sex and gender: duloxetine = 118 (59.6%) female, 80 (40.4%) male; placebo = 128 (63.1%) female, 75 (36.9%) male Condition: chronic low back pain Mean duration of symptoms: duloxetine = 8.3 (SD 8.2) years; placebo = 8.7 (SD 9.0) years Depression: participants with major depressive disorder excluded (as determined by the depression module of the Mini‐International Neuropsychiatric Interview) Characteristics that stratify health opportunities Level of country income: high (Germany, Netherlands, Poland, Spain, United States); upper‐middle (Russia) Race and ethnicity: duloxetine = 5 (2.5%) African participants, 189 (95.5%) Caucasian participants [understood to mean White], 4 (2.0%) Hispanic participants; placebo = 5 (2.5%) African participants; 193 (95.1%) Caucasian participants; 4 (2.0%) Hispanic participants; 1 Native American (0.5%) participant Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria male or female outpatients ≥ 18 years of age chronic low back pain as the primary painful condition, present on most days for at least 6 months and a rating of ≥ 4 on the Brief Pain Inventory average pain at visit 1 and visit 2 pain restricted to the low back or associated with radiation to the proximal portion of the lower limb only (class 1 and class 2 according to Quebec Task Force on Spinal Disorders) if female, testing negative for pregnancy at visit 1 sufficient education to understand the study procedures and communicate well with the site personnel Exclusion criteria > 1 low back surgery and 1 of these surgeries occurred within 12 months of study entry symptoms, signs, or imaging records consistent with radicular compression or spinal stenosis invasive procedures aimed to reduce low back pain within month before study entry participated in previous studies investigating duloxetine hydrochloride major depressive disorder as determined using the depression module of the Mini‐International Neuropsychiatric Interview at suicidal risk treated with monoamine oxidase inhibitor within 14 days of randomization previous diagnosis of psychosis, bipolar disorder, or schizoaffective disorder under evaluation for disability compensation or involved in a related litigation pregnant or breastfeeding females body mass index > 40 kg/m² requires daily use of nonsteroidal anti‐inflammatory drugs, acetaminophen, opioid analgesics, or other protocol‐excluded medication, for the duration of the study, as judged by the investigator
Interventions	Intervention: oral duloxetine Target dose: 60 mg/day Run in: none Treatment period: 12 weeks Comparator: inert placebo Rescue analgesia: episodic use of short‐acting analgesics including ibuprofen, acetaminophen and naproxen (not > 3 consecutive days and not > 20 total days during the treatment period) Concomitant medications: not reported
Outcomes	Pain intensity (Brief Pain Inventory 0‐10 Scale, 24‐hour average) Mean baseline score: duloxetine = 5.8 (SD 1.4); placebo = 5.8 (SD 1.4) Time points measured: baseline, week 3, week 6, week 9, week 12 Roland‐Morris Disability Questionnaire‐24 Mean baseline score: duloxetine = 9.6 (SD 4.6); placebo = 9.3 (SD 4.8) Time points measured: baseline, week 12 European Quality of Life Questionnaire–5 Dimension Mean baseline score: duloxetine = 0.52 (SD 0.27); placebo = 0.57 (SD 0.24) Time points measured: baseline, week 12 36‐Item Short Form Health Survey (General Health) Mean baseline score: duloxetine = 53.54 (SD 19.31); placebo = 53.11 (SD 22.12) Time points measured: baseline, week 12
Notes	Funding source: "Supported by Eli Lilly and Company, Indianapolis, Indiana" Declarations of interest: quote: "Drs Skljarevski and Desaiah, Ms Zhang, and Ms Alaka are employees of Eli Lilly and Company and hold company stocks. Drs Palacios, Miazgowski, and Patrick were study investigators and received funding from Eli Lilly and Company, Indianapolis, Indiana." Total adverse event outcome data were included from the study trial registry.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Described as randomised but insufficient information about the sequence generation process reported.
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not described.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Described as double‐blind but there was no reported information regarding the indistinguishability of trial treatments.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Unclear risk	It is unclear whether participants were adequately blinded.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	51/198 (26%) participants in the duloxetine group and 47/203 (23%) participants in the placebo group discontinued treatment but the proportion of these participants who continued to provide outcome data was not reported. No participants were excluded from the ITT analysis. Baseline‐observation‐carried‐forward used for the primary analysis only, with last‐observation‐carried‐forward used for all other analyses. It is likely that participants who withdrew did not continue to provide outcome data, and an inappropriate method of imputation was used for most analyses, both likely to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	Low risk	All outcomes prespecified in the trial registration were reported in the manuscript or in the trial registry itself. Missing point estimates or measures of variance for outcomes were reported in the trial registry.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Treves 1991.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 3 Study duration: 10‐day follow‐up Number of study centres: not reported Country: France Study setting: Hospital Rheumatology clinic Study dates: not reported
Participants	Number randomised: 68 (clomipramine morning = 25; clomipramine evening = 27; placebo = 16) Mean age: 45.6 years Age range: 19 to 76 years Sex and gender: 35 female (51.4%); 33 male (48.5%) Condition: low back pain and sciatica Mean duration of symptoms: not reported Depression: 47% of participants had clinical depression Characteristics that stratify health opportunities Level of country income: high Race: not reported Ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria low back pain or sciatica Exclusion criteria inflammatory, infectious or malignant conditions
Interventions	Intervention 1: intravenous clomipramine morning Target dose: 75 mg/day Intervention 2: intravenous clomipramine evening Target dose: 75 mg/day Run in: 'progressive dose' with 75 mg/day from day 3 Treatment period: 10 days Comparator: inert placebo Rescue analgesia: not reported Concomitant medications: not reported
Outcomes	Pain intensity (Verbal Rating Scale) Mean baseline score: not reported Time points measured: baseline, day 10
Notes	Funding source: Ciba Geigy Laboratories supplied medicines and placebo Declarations of interest: not reported We attempted to contact the study authors to request point estimates and measures of variance for pain intensity, but current contact details could not be found.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Described as randomised but insufficient information regarding generation of the random sequence reported.
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not described.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Described as double‐blind but there was no reported information regarding the indistinguishability of trial treatments.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Unclear risk	It is unclear whether participants were adequately blinded.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	Low risk	No missing outcome data.
Selective reporting (reporting bias)	High risk	Outcomes have been reported incompletely so that they cannot be entered in a meta‐analysis (point estimates and measures of variance not provided).
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Urquhart 2018.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 2 Study duration: 6‐month follow‐up Number of study centres: not reported Country: Australia Study setting: hospital, medical, and allied health clinics; community Study dates: 30 April 2012 to 1 June 2016
Participants	General characteristics Number randomised: 146 (amitriptyline = 72; placebo = 74) Mean age: amitriptyline 53.5 (SD 14.2) years; placebo = 56 (SD 13.2) years Age range: not reported Sex and gender: amitriptyline = 28 (38.9%) female; 44 (61.1%) male; placebo = 28 (37.8%) female, 46 (62.2%) male Condition: chronic non‐specific low back pain Mean duration of symptoms: amitriptyline = 13.3 (SD 12.6) years; 15.2 (SD 13.2) years Diagnostic criteria: clinical; pain below the costal margin and above the gluteal folds, without a specific cause Characteristics that stratify health opportunities Level of country income: high Race: not reported Ethnicity: not reported Culture: not reported Language: not reported Occupation: amitriptyline = 41 (58%) in paid employment; placebo 36 (51%) in paid employment Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria 18 to 75 years of age chronic non‐specific low back pain for > 3 months Exclusion criteria pathological entity major coexisting illness that might confound function or for which amitriptyline may be contraindicated another significant musculoskeletal condition history of psychosis current or previously diagnosed depression with or without the use of medication prior or current use of antidepressants current use of opioids contraindication or allergy to amitriptyline pregnancy, planning or trying to become pregnant or breastfeeding inability to give informed consent
Interventions	Intervention: oral amitriptyline Target dose: 25 mg/day Run in: none Treatment period: 6 months Comparator: active placebo (benztropine mesylate 1 mg/day) Rescue analgesia: not reported Concomitant medications: nonopioid analgesics and NSAIDs permitted
Outcomes	Pain intensity (100‐mm VAS) Mean baseline score: amitriptyline = 39.8 (SD 20.5); placebo = 43.4 (SD 21.0) Time points measured: baseline, 3 months, 6 months Roland Morris Disability Questionnaire Mean baseline score: amitriptyline = 7.54 (SD 4.37); placebo = 8.15 (SD 4.54) Time points measured: baseline, 3 months, 6 months Beck Depression Inventory Mean baseline score: amitriptyline = 10.5 (SD 6.71); placebo = 11.2 (SD 8.63) Time points measured: baseline, 3 months, 6 months EuroQol VAS (EQ‐5D‐5L) Mean baseline score: amitriptyline = 69.3 (SE 1.8); placebo = 71.3 (SE 1.7) Time points measured: baseline, 3 months, 6 months
Notes	Funding source: quote: "National Health and Medical Research Council (NHMRC, Australia, ID 1024401). Drs Urquhart, Wluka, and Wang are recipients of NHMRC Career Development Fellowships (Clinical Level 1 No. 1011975; Clinical Level 2 No. 1063574; Clinical Level 1 No. 1065464, respectively)." Declarations of interest: none declared Health‐related quality of life short‐term data provided by study authors.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generated using a computer random number generator.
Allocation concealment (selection bias)	Low risk	Central, pharmacy‐controlled randomisation. Allocation schedule prepared by a statistician not involved in trial conduct.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants and study personnel reported as blinded in study protocol. Active placebo was used, reported as identical to the study drug. There was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants were likely to have been blinded and there was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	High risk	14/72 (19%) and 11/72 (15%) of participants in the duloxetine group had missing outcome data at 3 and 6 months, respectively; and 16/74 (22%) and 17/74 (23%) of participants in the placebo group had missing outcome data at 3 and 6 months, respectively. All participants included in ITT analysis, with multiple imputation used to handle missing data. Missing outcome data balanced in numbers across intervention groups at 3 months but not 6 months; similar reasons for missing data across groups. The amount and between‐group imbalance in missing outcome data are enough to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	Low risk	All outcomes prespecified in the trial registration were reported in the manuscript, except for DDS scores. These data were available from the corresponding author on request.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

Vanelderen 2015.

*Study characteristics*
Methods	Study design: randomised controlled trial Study grouping: parallel Number of arms: 3 Study duration: 14‐day follow‐up Number of study centres: 1 Country: Belgium Study setting: tertiary multidisciplinary pain clinic Study dates: September 2011 to August 2013
Participants	Number randomised: 60 (amitriptyline = 20; minocycline = 20; placebo 20) Mean age: amitriptyline = 50 (SE 3) years; minocycline = 47 (SE 3) years; placebo = 51 (SE 3) years Age range: not reported Sex and gender: amitriptyline = 7 female (35%), 13 male (65%); minocycline = 9 (45%) female, 11 male (55%); placebo = 11 female (55%), 9 male (45%) Condition: lumbar radicular neuropathic pain Mean duration of symptoms: amitriptyline = 3.2 (SE 0.6) months; minocycline = 2.5 (SE 0.3) months; placebo = 2.8 (SE 0.4) months Depression: participants with depression excluded Characteristics that stratify health opportunities Level of country income: high Race and ethnicity: not reported Culture: not reported Language: not reported Occupation: not reported Religion: not reported Education: not reported Socioeconomic status: not reported Social capital: not reported Inclusion criteria 18 to 80 years of age neuropathic lumbar radicular pain caused by lumbar disc herniation, spinal canal stenosis, or failed back surgery syndrome (epidural fibrosis) confirmed by computer tomography or magnetic resonance imaging level of pathology on imaging studies correlates with dermatomal distribution in the leg Exclusion criteria diabetic, alcoholic, or drug‐induced polyneuropathies depression or psychiatric comorbidity affecting pain sensation use of antidepressants fibromyalgia and chronic fatigue syndrome pregnancy spinal cord damage malignancy allergy to minocycline or amitriptyline
Interventions	Intervention 1: oral amitriptyline Target dose: 25 mg/day Intervention 2: oral minocycline Target dose: 100 mg/day Run in: not reported Treatment period: 2 weeks Comparator: inert placebo Rescue analgesia: 50mg tramadol, maximum 3 times/day with 6 hours between doses Concomitant medications: only paracetamol or NSAIDs permitted during the trial period if dose was stabilised for at least 1 week before enrolment
Outcomes	Pain intensity (11‐point numeric rating scale, average leg pain over 24 hours) Mean baseline score: amitriptyline = 6.9 (SE 0.4); minocycline = 7.1 (SE 0.4); placebo = 7.4 (SE 0.3) Time points measured: baseline, day 7, day 14 Adverse events Time points measured: day 7, day 14
Notes	Funding source: "Support was provided solely from institutional and/or departmental sources" Declarations of interest: "The authors declare no competing interests"
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random sequence generated using a computer random number generator.
Allocation concealment (selection bias)	Low risk	Study medication kits were prepared by a hospital pharmacist who alone held the allocation key. The schedule safeguarded until the end of the study.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Participants and study personnel described as blinded. The trial medications were identical in appearance. There was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Self‐reported outcomes	Low risk	Participants were likely to have been blinded and there was no indication that blinding could have been broken.
Blinding of outcome assessment (detection bias) Assessor‐reported outcomes	Low risk	No assessor‐reported outcomes of interest to this review.
Incomplete outcome data (attrition bias)	Unclear risk	3/20 (15%) participants in the amitriptyline group and 3/20 (15%) participants in the placebo discontinued treatment but appear to be included in the ITT analysis. The method to account for missing data was not reported. It is unclear whether the amount of missing outcome data and method to account for missing data were enough to have induced clinically relevant bias in the observed effect size.
Selective reporting (reporting bias)	High risk	The day 7 endpoint was specified as one of the primary endpoints in the trial registry but numerical data were not reported.
Other bias	Unclear risk	No early stopping or unplanned interim analyses but no information to assess unequal application of co‐interventions.

Open in a new tab

CT: computed tomography; ITT: intention‐to‐treat analysis; MRI: magnetic resonance imaging; NRS: numerical rating scale; NSAID: non‐steroidal anti‐inflammatory drug; SD: standard deviation; SNRI: serotonin and norepinephrine reuptake inhibitor; SSRI: selective serotonin reuptake inhibitor; USD: US dollars; VAS: visual analogue scale

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Brannan 2005	Ineligible patient population: major depressive disorder with comorbid pain, separate data for low back pain patients not reported.
Chrzanowski 1977	Ineligible study design: not a randomised controlled trial.
Davidson 1994	Ineligible study design: not a randomised controlled trial.
DeLomba 2021	Ineligible patient population: chronic musculoskeletal pain, separate data not reported for low back pain patients.
EUCTR2013‐004564‐55‐RO	Ineligible intervention: combination therapy including an antidepressant and an anticonvulsant.
Gaynor 2011	Ineligible patient population: major depressive disorder with comorbid pain, separate data for low back pain patients not reported.
Gourlay 1986	Ineligible patient population: mixed chronic pain, separate data for low back pain patients not reported.
Hameroff 1982	Ineligible patient population: mixed chronic pain, separate data for low back pain patients not reported.
Hameroff 1984	Ineligible patient population: mixed chronic pain, separate data for low back pain patients not reported.
Isaac 2003	Ineligible patient population: mixed chronic pain
Johansson 1979	Ineligible patient population: mixed chronic pain, separate data for low back pain patients not reported.
Jørgensen 1984	Ineligible patient population: mixed chronic pain
Krell 2005	Ineligible study design: not a randomised controlled trial
Loldrup 1989	No outcomes of interest
McQuay 1992	Ineligible patient population: mixed chronic pain, separate data for low back pain patients not reported.
McQuay 1993	Ineligible patient population: mixed chronic pain, separate data for low back pain patients not reported.
NCT01119924	Ineligible patient population: depression patients with mixed pain
NCT01288937	Ineligible patient population: idiopathic neuropathic pain
Pilowsky 1982	Ineligible patient population: chronic pain
Pilowsky 1990	Ineligible patient population: mixed chronic pain, separate data for low back pain patients not reported.
Skljarevski 2009b	Ineligible study design: extension study of a randomised controlled trial
Skljarevski 2010c	Ineligible study design: pooled analysis of randomised controlled trials
Skljarevski 2010d	Ineligible study design: open‐label extension study
Skljarevski 2012	Ineligible study design: pooled analysis of randomised controlled trials
Songer 1996	Ineligible study design: not a randomised controlled trial
Storch 1986	Ineligible study design: not a randomised controlled trial
Tavafian 2014	Ineligible intervention: antidepressant not tested
Tsuji 2017	Ineligible study design: secondary analysis of a randomised controlled trial
Tyrer 1996	Ineligible patient population: mixed chronic pain, separate data for low back pain patients not reported.
Urquhart 2021	Ineligible study design: secondary analysis of a randomised controlled trial
Usha Rani 1996	Ineligible patient population: chronic rheumatic pain, separate data for low back pain patients not reported.
Vijayalakshmi 2016	Ineligible patient population: neuropathic pain, separate data for radiculopathy patients not reported.
Ward 1984	Ineligible comparator: comparative effectiveness study of two antidepressants
Yucel 2005	Ineligible patient population: neuropathic pain
Zitman 1990	Ineligible patient population: mixed chronic pain, separate data for low back pain patients not reported.

Open in a new tab

Characteristics of studies awaiting classification [ordered by study ID]

NCT03249558.

Methods	Parallel randomised controlled trial
Participants	Patients with chronic neck or back pain for at least 3 months
Interventions	Intervention: oral duloxetine Target dose: 60 mg/day Treatment period: 10 weeks Comparator: inert placebo Concomitant medications: morphine 60 mg/day
Outcomes	Visual analogue scale (VAS)
Notes	Despite several attempts to contact study authors, it is unclear whether separate data for participants with low back pain are available.

Open in a new tab

Characteristics of ongoing studies [ordered by study ID]

ACTRN12624000919516.

Study name	Duloxetine for chronic sciatica (DREAM): an adaptive randomised placebo‐controlled trial
Methods	Parallel randomised controlled trial
Participants	Adults with radiating pain into one leg in a dermatomal distribution, of at least three months duration and moderate intensity, with evidence of nerve root involvement
Interventions	Intervention: oral duloxetine Target dose: 60 mg/day Treatment period: 12 weeks (plus 2‐week taper) Comparator: inert placebo
Outcomes	Leg pain intensity, 0 to 10 numerical rating scale Disability, Roland Morris Disability Questionnaire Low back pain intensity, 0 to 10 numerical rating scale Health‐related quality of life, EQ‐5D‐5L Depressive symptoms, Patient Health Questionnaire Adverse events
Starting date	1 October 2024 (anticipated)
Contact information	Dr Giovanni Ferreira Institute for Musculoskeletal Health, The University of Sydney, Level 10N KGV Building, Missensden Road, Camperdown, NSW, 2050, Australia P: +61 02 86276681 E: giovanni.ferreira@sydney.edu.au
Notes

Open in a new tab

Differences between protocol and review

Since the last version of this review (Urquhart 2008), we revised the methods to conform with updated conduct and reporting standards of systematic reviews, as recommended by the Cochrane Handbook for Systematic Reviews of Interventions, the Cochrane Methodological Expectations of Cochrane Intervention Reviews (MECIR), and Cochrane Musculoskeletal group guidance. The revised methods were prospectively established in a protocol before the conduct of the review and approved by Cochrane Musculoskeletal editors.

Differences between the current version and the previous version of this review

Participants: we included participants with spine‐related leg pain (including somatic‐referred pain, radicular pain with radiculopathy, or radicular pain without radiculopathy) as well as low back pain. The title has been updated from 'Antidepressants for non‐specific low back pain' to 'Antidepressants for low back pain and spine‐related leg pain' to reflect this change.

Searches: we did not search PsycINFO but added searches of three trial registries.

Outcomes: we removed overall improvement, return to work, and physiological outcomes as outcomes, but we included total adverse events, serious adverse events, and withdrawals due to adverse events.

Assessment of risk of bias: we assessed risk of bias using updated criteria for the RoB 1 tool (Higgins 2017).

Assessment of the certainty of the evidence: we assessed the certainty of evidence using the GRADE framework which was unavailable in the previous version of this review.

Analysis: primary analyses were separated by antidepressant class, and by prespecified timepoints of interest.

Differences between the current version of this review and the updated protocol

Measures of treatment effect: we planned to calculate the number needed to treat for an additional beneficial outcome (NNTB) and the number needed to treat for an additional harmful outcome (NNTH) for dichotomous outcomes. Following updated editorial guidance, we did not calculate these measures but have reported anticipated absolute effects in the summary of findings tables. We also prespecified a between‐group difference of 10 points on a 0 to 100 scale as the minimal clinically important effect for pain. However, as we developed the review, we decided not to use this threshold, due to its limitations, and only describe the size of the effect.

Contributions of authors

Conceived the review: MF, DU, JM, AC

Drafted the protocol: MF

Developed and ran search strategy: MF (with support from Cochrane Musculoskeletal)

Obtained studies: MF

Selected studies for inclusion: MF and GF, MW, CAS, AT, JH, or AC

Extracted data from studies: MF and GF, MW, CAS, AT, or AC

Evaluated risk of bias: MF and GF, MW, AT, or AC

GRADE judgements: MF and AC

Entered data into RevMan: MF

Carried out the analysis: MF

Interpreted the analysis: MF, DU, GF, MW, CAS, AT, JH, EV, JH, AC

Drafted the review: MF

Reviewed and approved final version: MF, DU, GF, MW, CAS, AT, JH, EV, JM, AC

Some authors involved in the previous published version of this review in 2008 are no longer included on the author byline: Willem JJ Assendelft, Martin Roland, Maurits W van Tulder. Some of the content retained in this review reflects their contributions.

Sources of support

Internal sources

Australian Government Research Training Program, Australia

Stipend for MF
Neuroscience Research Australia PhD Pearl supplementary scholarship, Australia

Stipend for MF
National Health and Medical Research Council/Medical Research Future Fund Career Development Fellowship (Clinical Level 2 APP1142809), Australia

Salary for DU
National Health and Medical Research Council Investigator Grant Fellowship (APP2009808), Australia

Salary for GF
Postgraduate Scholarship, National Health and Medical Research Council, Australia

Stipend for MW
PhD Supplementary Scholarship from Neuroscience Research Australia, Australia

Stipend for MW
Investigator Grant, National Health and Medical Research Council, Australia

Salary for CAS
National Health and Medical Research Council, Australia

Grant support for AT
University of Notre Dame , Australia

Salary for EV
Joondalup Health Campus, Australia

Salary for EV
National Health and Medical Research Council Investigator Grant, Australia

Salary for JM
National Health and Medical Research Council Investigator Grant, paid by University of New South Wales , Australia

Salary for AC
Amsterdam UMC, University of Amsterdam, department of Public and Occupational Health, Amsterdam, Netherlands

Salary for JH

External sources

No external sources of support, Other

Not applicable

Declarations of interest

MF is a sign‐off editor for Cochrane but was not involved in the editorial process for this review. He receives a stipend from an Australian Research Training Program grant and Neuroscience Research Australia.

DU reports being an author on a study (funded by National Health and Medical Research Council, Australia, ID 1024401) that was eligible for inclusion in this review but was not involved in selection, data extraction, and assessment of risk of bias and the certainty of evidence involving this study. DU is an editor for Cochrane but was not involved in the editorial process for this review. DU receives salary support from the National Health and Medical Research Council of Australia and Medical Research Future fund grants.

GF reports being an author of an ongoing study that was eligible for inclusion in this review but was not involved in selection and data extraction involving this study. GF received salary support from a National Health and Medical Research Council of Australia grant.

MW received salary support from a National Health and Medical Research Council of Australia grant, Neuroscience Research Australia, and University of New South Wales.

CAS received salary support from a National Health and Medical Research Council of Australia grant and is employed by the University of Sydney.

AT receives grant support from the National Health and Medical Research Council of Australia.

JH is an editor for Cochrane but was not involved in the editorial process for this review.

EV is an investigator on a National Health and Medical Research Council of Australia grant.

JM receives salary support from a National Health and Medical Research Council of Australia grant.

AC receives salary support from a National Health and Medical Research Council of Australia grant and is employed by the University of New South Wales.

New search for studies and content updated (conclusions changed)

References

References to studies included in this review

Afilal 2020 {published and unpublished data}

Afilal S. Efficacy of clomipramine for chronic lumbar radicular pain a randomized clinical trial (Medical specialisation dissertation). Rabat (Morocco): Mohammed V University Faculty of Medicine and Pharmacy, 2020. [Google Scholar]
Afilal SA, Fellous S, Ilham A, Abouqal R, Hmamouchi I, Alami N, et al. Efficacy of clomipramine for chronic lumbar radicular pain a randomized clinical trial. Annals of the Rheumatic Diseases 2020;79:59. [DOI: 10.1136/annrheumdis-2020-eular.4373] [DOI] [Google Scholar]
PACTR202001764151121. Efficacy of clomipramine for chronic lumbar radicular pain: a randomized clinical trial. trialsearch.who.int/Trial2.aspx?TrialID=PACTR202001764151121 (first posted 6 November 2019).

Alcoff 1982 {published data only}

Alcoff J, Jones E, Rust P, Newman R. Controlled trial of imipramine for chronic low back pain. Journal of Family Practice 1982;14:841-6. [PubMed] [Google Scholar]

Atkinson 1998 {published data only}

Atkinson HJ, Slater MA, Williams RA, Zisook S, Patterson TL, Grant I, et al. A placebo-controlled randomized clinical trial of nortriptyline for chronic low back pain. Pain 1998;76:287-96. [DOI] [PubMed] [Google Scholar]

Atkinson 1999 {published data only}

Atkinson JH, Slater MA, Wahlgren DR, Williams RA, Zisook S, Pruitt SD, et al. Effects of noradrenergic and serotonergic antidepressants on chronic low back pain intensity. Pain 1999;83:137-45. [DOI] [PubMed] [Google Scholar]

Atkinson 2007 {published data only (unpublished sought but not used)}

Atkinson JH, Slater MA, Capparelli EV, Wallace MS, Zisook S, Abramson I, et al. Efficacy of noradrenergic and serotonergic antidepressants in chronic back pain. Journal of Clinical Psychopharmacology 2007;27(2):135-42. [DOI] [PubMed] [Google Scholar]
NCT00018200. Effect of antidepressants on back pain. clinicaltrials.gov/study/NCT00018200 (first posted 23 June 2005).

Dickens 2000 {published data only (unpublished sought but not used)}

Dickens C, Jayson M, Sutton C, Creed F. The relationship between pain and depression in a trial using paroxetine in sufferers of chronic low back pain. Psychosomatics 2000;41:490-9. [DOI] [PubMed] [Google Scholar]
GlaxoSmithKline. A double-blind study comparing the efficacy and tolerability of paroxetine and placebo in hospital out-patients with depressive symptoms associated with chronic back pain. www.gsk.com/media/1543/depression_701_full.pdf 2001.

Goodkin 1990 {published data only}

Goodkin K, Gullion CM, Agras WS. A randomized, double-blind, placebo-controlled trial of trazodone hydrochloride in chronic low back pain syndrome. Journal of Clinical Psychopharmacology 1990;10:269-78. [PubMed] [Google Scholar]

Gould 2020 {published data only (unpublished sought but not used)}

Gould HM, Atkinson JH, Chircop-Rollick T, D'Andrea J, Garfin S, Patel SM, et al. A randomized placebo-controlled trial of desipramine, cognitive behavioral therapy, and active placebo therapy for low back pain. Pain 2020;161(6):1341-9. [DOI: 10.1097/j.pain.0000000000001834] [DOI] [PubMed] [Google Scholar]
NCT00964886. Efficacy of antidepressants in chronic back pain. clinicaltrials.gov/study/NCT00964886 (first posted 24 August 2009).

Jenkins 1976 {published data only}

Jenkins DG, Ebbutt AF, Evans CD. Tofranil in the treatment of low back pain. Journal of International Medical Research 1976;4:28-40. [PubMed] [Google Scholar]

Johnson 2011 {published and unpublished data}

Chatterjee N, Noor N, Crowell A, Mackey S. Duloxetine and placebo alter different gray matter regions in chronic low back pain patients. Journal of Pain 2010;11(4 Suppl 1):S40. [Google Scholar]
Johnson K, Chatterjee N, Noor N, Crowell A, McCue R, Mackey S. Effects of duloxetine and placebo in patients with chronic low back pain. Journal of Pain 2011;12(4 Suppl 1):49. [Google Scholar]
NCT00388414. Imaging study of chronic low back pain in patients taking pain medication. clinicaltrials.gov/study/NCT00388414 (first posted 13 October 2010).

Katz 2005 {published data only}

Katz J, Pennella-Vaughan J, Hetzel RD, Kanazi GE, Dworkin RH. A randomised, placebo-controlled trial of bupropion sustained release in chronic low back pain. Journal of Pain 2005;6(10):656-61. [DOI] [PubMed] [Google Scholar]

Khoromi 2007 {published data only}

Khoromi S, Cui L, Nackers L, Max MB. Morphine, nortriptyline and their combination vs. placebo in patients with chronic lumbar root pain. Pain 2007;130(1-2):66-75. [DOI: 10.1016/j.pain.2006.10.029] [DOI] [PMC free article] [PubMed] [Google Scholar]

Konno 2016 {published and unpublished data}

Konno S, Oda N, Ochiai T, Alev L. Randomized, double-blind, placebo-controlled phase III trial of duloxetine monotherapy in Japanese patients with chronic low back pain. Spine 2016;41(22):1709–17. [DOI: 10.1097/BRS.0000000000001707] [DOI] [PMC free article] [PubMed] [Google Scholar]
NCT01855919. A study of duloxetine (LY248686) in participants with chronic low back pain. clinicaltrials.gov/study/NCT01855919 (first posted 14 May 2013).
jRCT2080222083. A phase 3 clinical trial of duloxetine in patients with chronic low back pain. jrct.niph.go.jp/en-latest-detail/jRCT2080222083 (first posted 15 May 2013).

Kurniawati 2020 {published data only}

Kurniawati N, Suharjanti I, Hamdan M, Rahardjo JE, Ardiansyah D, Hidayati HB. Efficacy of acetaminophen plus amitriptyline compared to acetaminophen to reduce pain intensity in nonspecific chronic lower back pain. Indian Journal of Forensic Medicine & Toxicology 2020;14(2):1819-25. [DOI: 10.37506/ijfmt.v14i2.3202] [DOI] [Google Scholar]

Marks 2014 {published and unpublished data}

Marks DM, Pae CU, Patkar AA. A double-blind, placebo-controlled, parallel-group pilot study of milnacipran for chronic radicular pain (sciatica) associated with lumbosacral disc disease. Primary Care Companion for CNS Disorders 014;16(4):26268. [DOI: 10.4088/PCC.14m01658] [DOI] [PMC free article] [PubMed] [Google Scholar]
NCT01777581. Milnacipran for lumbosacral radicular pain. clinicaltrials.gov/study/NCT01777581 (first posted 24 January 2013).

NCT01225068 {unpublished data only}

NCT01225068. Effect of milnacipran in chronic neuropathic low back pain. clinicaltrials.gov/ct2/show/NCT01225068 (first posted 19 October 2010).

Pheasant 1983 {published data only}

Pheasant H, Bursk A, Goldfarb J, Azen SP, Weiss JN, Borelli L. Amitriptyline and chronic low-back pain. A randomized double-blind crossover study. Spine 1983;8:552-7. [DOI] [PubMed] [Google Scholar]

Pirbudak 2003 {published data only}

Pirbudak L, Karakurum G, Satana T, Karadasli H, Topalhan M, Oner U, et al. Epidural steroid injection and amitriptyline in the management of acute low back pain originating from lumbar disc herniation [Dysk hernysy kaynakli akut dyskojenyk bel adrilarinda epydural steroyd enjeksyyonu ve amytryptlyn kullanimi]. Artroplasty Artroskopyk Cerrahy [Journal of Arthroplasty & Arthroscopic Surgery] 2003;14(2):89-93. [Google Scholar]

Schliessbach 2018 {published data only (unpublished sought but not used)}

NCT01179828. Linking altered central pain processing and genetic polymorphism to drug efficacy in chronic low back pain (prediction). clinicaltrials.gov/study/NCT01179828 (first posted 10 August 2010).
Schliessbach J, Siegenthaler A, Bütikofer L, Limacher A, Juni P, Vuilleumier PH, et al. Effect of single-dose imipramine on chronic low-back and experimental pain. PLoS One 2018;13(5):e0195776. [DOI: 10.1371/journal.pone.0195776] [DOI] [PMC free article] [PubMed] [Google Scholar]

Schukro 2016 {published data only (unpublished sought but not used)}

2009-012713-22. Randomized double-blind study comparing the efficacy of duloxetine with placebo in patients with chronic low back pain. clinicaltrialsregister.eu/ctr-search/search?query=2009-012713-22 (first posted 14 October 2010).
NCT01166048. Comparison of the efficacy of duloxetine with placebo in patients with chronic low back pain with a radicular component. clinicaltrials.gov/study/NCT01166048 (first posted 19 July 2010).
Schukro RP, Oehmke MJ, Geroldinger A, Heinze G, Kress HG, Pramhas S. Efficacy of duloxetine in chronic low back pain with a neuropathic component: a randomized, double-blind, placebo-controlled crossover trial. Anesthesiology 2016;124(1):150-8. [DOI: 10.1097/ALN.0000000000000902] [DOI] [PubMed] [Google Scholar]

Skljarevski 2009a {published and unpublished data}

NCT00408876. Duloxetine versus placebo in chronic low back pain. clinicaltrials.gov/study/NCT00408876 (first posted 6 December 2006).
Skljarevski V, Ossanna M, Liu-Seifert H, Zhang Q, Chappell A, Iyengar S, et al. A double-blind, randomized trial of duloxetine versus placebo in the management of chronic low back pain. European Journal of Neurology 2009;16(9):1041-8. [DOI] [PubMed] [Google Scholar]

Skljarevski 2010a {published and unpublished data}

EUCTR2006-003484-31. Effect of duloxetine 60 mg to 120 mg once daily in patients with chronic low back pain. clinicaltrialsregister.eu/ctr-search/search?query=2006-003484-31 (first posted 4 October 2006).
NCT00424593. Duloxetine versus placebo in chronic low back pain. clinicaltrials.gov/study/NCT00424593 (first posted 17 January 2007).
Skljarevski V, Desaiah D, Liu-Seifert H, Zhang Q, Chappell AS, Detke MJ, et al. Efficacy and safety of duloxetine in patients with chronic low back pain. Spine 2010;35(13):E578-85. [DOI: 10.1097/BRS.0b013e3181d3cef6] [DOI] [PubMed] [Google Scholar]
Skljarevski V, Desaiah D, Liu-Seifert H, Zhang Q, Chappell AS, Detke MJ, et al. Efficacy of duloxetine in chronic low back pain. European Journal of Neurology 2008;15(Suppl 3)(320):P2447. [DOI] [PubMed] [Google Scholar]

Skljarevski 2010b {published and unpublished data}

EUCTR2009-012713-22. Randomized double-blind study comparing the efficacy of duloxetine with placebo in patients with chronic low back pain. clinicaltrialsregister.eu/ctr-search/trial/2009-012713-22/AT (first posted 14 April 2010).
NCT00767806. A study for patient with chronic low back pain. clinicaltrials.gov/study/NCT00767806 (first posted 6 October 2008).
NL-OMON32370. Effect of duloxetine 60 mg once daily versus placebo in patients with chronic low back pain. onderzoekmetmensen.nl/en/trial/32370 (first posted 26 June 2008).
Skljarevski V, Zhang S, Desaiah D, Alaka KJ, Palacios S, Miazgowski T, et al. Duloxetine versus placebo in patients with chronic low back pain: a 12-week, fixed-dose, randomized, double-blind trial. Journal of Pain 2010;11(12):1282-90. [DOI: 10.1016/j.jpain.2010.03.002] [DOI] [PubMed] [Google Scholar]
Skljarevski V, Zhang S, Desaiah D, Palacios S, Miazgowski T, Patrick K. (248) Efficacy and safety of duloxetine 60 mg once-daily in patients with chronic low back pain. Journal of Pain 2010;11:s38. [DOI] [PubMed] [Google Scholar]
Skljarevski V, Zhang S, Desaiah D, Palacios S, Miazgowski T, Patrick K. Effect of duloxetine 60 mg once daily versus placebo in patients with chronic low back pain: a 12-week, randomized, double-blind trial. Pain Medicine 2010;11:322. [DOI] [PubMed] [Google Scholar]

Treves 1991 {published data only}

Treves R, Montane De La Roque P, Dumond JJ, Bertin P, Arnaud M, Desproges-Gotteron R. Prospective study of the analgesic action of clomipramine versus placebo in refractory low back pain and sciatica (68 cases) [Étude prospective de l'action antalgique de la clomipramine versus placebo dans les lombosciatalgies rebelles (68 cas)]. Revue du Rhumatisme et des Maladies Osteo-articulaires 1991;58:549-52. [PubMed] [Google Scholar]

Urquhart 2018 {published and unpublished data}

ACTRN12612000131853. Is amitriptyline effective in the management of chronic low back pain? [Is low dose amitriptyline more effective than an active placebo in the management of chronic low back pain? A double-blind, randomized, placebo-controlled trial]. anzctr.org.au/Trial/Registration/TrialReview.aspx?id=361972&isReview=true (first posted 25 January 2012).
Urquhart D, Wluka A, Van Tulder M, Heritier S, Forbes A, Fong C, et al. Efficacy of low-dose amitriptyline for chronic low back pain: a double-blind, randomised controlled trial. Osteoarthritis and Cartilage 2018;26:s416. [Google Scholar]
Urquhart DM, Wluka AE, Sim MR, Tulder M, Forbes A, Gibson SJ, et al. Is low-dose amitriptyline effective in the management of chronic low back pain? Study protocol for a randomised controlled trial. Trials 2016;17(514):1-8. [DOI: 10.1186/s13063-016-1637-1] [DOI] [PMC free article] [PubMed] [Google Scholar]
Urquhart DM, Wluka AE, Van Tulder M, Heritier S, Forbes A, Fong C, et al. Efficacy of low-dose amitriptyline for chronic low back pain: a randomized clinical trial. JAMA Internal Medicine 2018;178(11):1474-81. [DOI: 10.1001/jamainternmed.2018.4222] [DOI] [PMC free article] [PubMed] [Google Scholar]

Vanelderen 2015 {published data only}

Gelijkens V, Van Zundert J, De Vooght P, Vander Laenen M, Heylen R, Vanelderen, P. The effectiveness of amitriptyline in the treatment of subacute lumbar radicular pain. European Journal of Anaesthesiology 2014;52:232. [Google Scholar]
NCT01869907. Effect of minocycline on pain caused by nerve damage [Effect of minocycline on neuropathic pain]. clinicaltrials.gov/study/NCT01869907 (first posted 2 June 2013).
Vanelderen P, Van Zundert J, Kozicz T, Puylaert M, De Vooght P, Mestrum R, et al. Effect of minocycline on lumbar radicular neuropathic pain: a randomized, placebo-controlled, double-blind clinical trial with amitriptyline as a comparator. Anesthesiology 2015;122(2):399-406. [DOI: 10.1097/ALN.0000000000000508] [DOI] [PubMed] [Google Scholar]

References to studies excluded from this review

Brannan 2005 {published data only}

Brannan SK, Mallinckrodt CH, Brown EB, Wohlreich MM, Watkin JG, Schatzberg AF. Duloxetine 60 mg once-daily in the treatment of painful physical symptoms in patients with major depressive disorder. Journal of Psychiatric Research 2005;39(1):43-53. [DOI] [PubMed] [Google Scholar]

Chrzanowski 1977 {published data only}

Chrzanowski J. Imipramine in the treatment of chronic sciatic pains [Imipramina w leczeniu przewlekłej rwy kulszowej]. Wiadomosci Lekarskie 1977;30(18):1439-40. [PubMed] [Google Scholar]

Davidson 1994 {published data only}

Davidson JR, France RD. Bupropion in chronic low back pain. Journal of Clinical Psychiatry 1994;55(8):362. [PubMed] [Google Scholar]

DeLomba 2021 {published data only}

DeLomba W, Bernstein M, Basso K, Suárez E, McLean S, Beaudoin F. 231 The tolerability and effectiveness of duloxetine for the prevention of persistent musculoskeletal pain after trauma and injury: a pilot three-group randomized controlled trial. Annals of Emergency Medicine 2021;78:S94. [Google Scholar]
Strauss DH, Santhanam DR, McLean SA, Beaudoin FL. Study protocol for a randomised, double-blind, placebo-controlled clinical trial of duloxetine for the treatment and prevention of musculoskeletal pain: altering the transition from acute to chronic pain (ATTAC pain). BMJ Open 2019;9(3):e025002. [DOI: 10.1136/bmjopen-2018-025002] [DOI] [PMC free article] [PubMed] [Google Scholar]

EUCTR2013‐004564‐55‐RO {unpublished data only}

EUCTR2013-004564-55-RO. A randomized, multicenter, double-blind, placebo controlled, parallel study to examine the efficacy and safety of a combination treatment in subjects with chronic low back pain [A randomized, multicenter, double-blind, placebo controlled, parallel study to assess the efficacy and safety of the combined administration of sertraline and lamotrigine in subjects with chronic low back pain]. trialsearch.who.int/Trial2.aspx?TrialID=EUCTR2013-004564-55-RO (first posted 30 May 2014).

Gaynor 2011 {published data only}

Gaynor PJ, Gopal M, Zheng W, Martinez JM, Robinson MJ, Hann D. Duloxetine versus placebo in the treatment of major depressive disorder and associated painful physical symptoms: a replication study. Current Medical Research and Opinion 2011;27(10):1859-67. [DOI: 10.1185/03007995.2011.609540] [DOI] [PubMed] [Google Scholar]

Gourlay 1986 {published data only}

Gourlay GK, Cherry DA, Cousins MJ, Love BL, Graham JR, McLachlan MO. A controlled study of a serotonin reuptake blocker, zimelidine, in the treatment of chronic pain. Pain 1986;25(1):35-52. [DOI: 10.1016/0304-3959(86)90006-0] [DOI] [PubMed] [Google Scholar]

Hameroff 1982 {published data only}

Hameroff SR, Cork RC, Scherer K. Doxepin effects on chronic pain, depression and plasma opioids. Journal of Clinical Psychiatry 1982;43(8):22-6. [PubMed] [Google Scholar]

Hameroff 1984 {published data only}

Hameroff SR, Weiss JL, Lerman JC, Cork RC, Watts KS, Crago BR, et al. Doxepin's effects on chronic pain and depression: a controlled study. Journal of Clinical Psychiatry 1984;45(3):47-53. [PubMed] [Google Scholar]

Isaac 2003 {published data only}

Isaac M. Double-blind study of venlafaxine in chronic pain. In: 156th Annual Meeting of the American Psychiatric Association; 2003 May 17-22; San Francisco (CA). 2003.

Johansson 1979 {published data only}

Johansson F, Von Knorring L. A double-blind controlled study of a serotonin uptake inhibitor (zimelidine) versus placebo in chronic pain patients. Pain 1979;7(1):69-78. [DOI: 10.1016/0304-3959(79)90108-8] [DOI] [PubMed] [Google Scholar]

Jørgensen 1984 {published data only}

Jørgensen B, Nørrelund N, Bech P, Jakobsen K. Pain and depressive symptoms in general practice. A controlled study of Anafranil (clomipramine) versus placebo [En koolrolleret undersegelse af anafranil (klomipramin) versus placebo]. Ugeskrift for Læger 1984;146(38):2868-72. [PubMed] [Google Scholar]

Krell 2005 {published data only}

Krell HV, Leuchter AF, Cook IA, Abrams M. Evaluation of reboxetine, a noradrenergic antidepressant, for the treatment of fibromyalgia and chronic low back pain. Psychosomatics 2005;46(5):379-84. [DOI] [PubMed] [Google Scholar]

Loldrup 1989 {published data only}

Loldrup D, Langemark M, Hansen HJ, Olesen J, Bech P. Clomipramine and mianserin in chronic idiopathic pain syndrome. A placebo controlled study. Psychopharmacology 1989;99(1):1-7. [DOI: 10.1007/BF00634443] [DOI] [PubMed] [Google Scholar]

McQuay 1992 {published data only}

McQuay JH, Carroll D, Glynn CJ. Low dose amitriptyline in the treatment of chronic pain. Anaesthesia 1992;47(8):646-52. [DOI: 10.1111/j.1365-2044.1992.tb02383.x] [DOI] [PubMed] [Google Scholar]

McQuay 1993 {published data only}

McQuay HJ, Carroll D, Glynn CJ. Dose-response for analgesic effect of amitriptyline in chronic pain. Anaesthesia 1993;48(4):281-5. [DOI: 10.1111/j.1365-2044.1993.tb06943.x] [DOI] [PubMed] [Google Scholar]

NCT01119924 {unpublished data only}

NCT01119924. A clinical study for assessment of the efficacy and safety of mirtazapine (Smilon) in depression patients with pain. clinicaltrials.gov/study/NCT01119924 (first posted 7 May 2010).

NCT01288937 {unpublished data only}

NCT01288937. A placebo controlled, randomized, double blind trial of milnacipran for the treatment of idiopathic neuropathy pain. clinicaltrials.gov/study/NCT01288937 (first posted 2 February 2011).

Pilowsky 1982 {published data only}

Pilowsky I, Hallett EC, Bassett DL, Thomas PG, Penhall RK. A controlled study of amitriptyline in the treatment of chronic pain. Pain 1982;14:169-79. [DOI: 10.1016/0304-3959(82)90097-5] [DOI] [PubMed] [Google Scholar]

Pilowsky 1990 {published data only}

Pilowsky I, Barrow GC. A controlled study of psychotherapy and amitriptyline used individually and in combination in the treatment of chronic intractable, 'psychogenic' pain. Pain 1990;40(1):3-19. [DOI: 10.1016/0304-3959(90)91045-K] [DOI] [PubMed] [Google Scholar]

Skljarevski 2009b {published data only}

Skljarevski V, Zhang S, Chappell AS, Detke MJ, Murray I, Backonja M. Maintenance of effect of duloxetine in patients with chronic low back pain. European Journal of Neurology 2009;16(Suppl 3):92. [DOI] [PubMed] [Google Scholar]

Skljarevski 2010c {published data only}

Skljarevski V, Bair MJ, Ossanna MJ, Frakes E, Zhang S, Alaka K. OMERACT responder analysis of patients treated with duloxetine for chronic low back pain. Arthritis and Rheumatism 2010;62(10):S73-4. [Google Scholar]

Skljarevski 2010d {published data only}

Skljarevski V, Zhang S, Chappell AS, Walker DJ, Murray I, Backonja M. Maintenance of effect of duloxetine in patients with chronic low back pain: a 41-week uncontrolled, dose-blinded study. Pain Medicine 201;11(5):648-57. [DOI: 10.1111/j.1526-4637.2010.00836.x] [DOI] [PubMed] [Google Scholar]

Skljarevski 2012 {published data only}

Skljarevski V, Liu P, Zhang S, Ahl J, Martinez JM. Efficacy and safety of duloxetine in patients with chronic low back pain who used versus did not use concomitant nonsteroidal anti-inflammatory drugs or acetaminophen: a post hoc pooled analysis of 2 randomized, placebo-controlled trials. Pain Research and Treatment 2012;2012:296710. [DOI: 10.1155/2012/296710] [DOI] [PMC free article] [PubMed] [Google Scholar]

Songer 1996 {published data only}

Songer DA, Schulte H. Venlafaxine for the treatment of chronic pain. American Journal of Psychiatry 1996;153(3):737. [DOI] [PubMed] [Google Scholar]

Storch 1986 {published data only}

Storch H, Steck P. Anti-depressive medication in the treatment of low back pain: a controlled trial [Begleitende thymoleptische therapie im rahmen einer kontrollierten studie mit maprotilin (ludiomil) bei der behandlung von kreuzschmerzen]. Nervenarzt 1982;53(8):445-50. [PubMed] [Google Scholar]

Tavafian 2014 {published data only}

Tavafian SS, Jamshidi AR, Mohammad K. Treatment of low back pain: randomized clinical trial comparing a multidisciplinary group-based rehabilitation program with oral drug treatment up to 12 months. International Journal of Rheumatic Diseases 2014;17:159-64. [DOI] [PubMed] [Google Scholar]

Tsuji 2017 {published data only}

Tsuji T, Itoh N, Ochiai T, Konno S. Response to duloxetine in chronic low back pain: exploratory post hoc analysis of a Japanese Phase III randomized study. Journal of Pain Research 2017;10:2157-68. [DOI: 10.2147/JPR.S138172] [DOI] [PMC free article] [PubMed] [Google Scholar]

Tyrer 1996 {published data only}

Tyrer S, Lauder J, Michie M, Griffiths H, Charlton J. The effect of trazodone in depressed and non-depressed patients in chronic pain. European Psychiatry 1996;11(4):359s. [Google Scholar]

Urquhart 2021 {published data only}

Urquhart DM, Wluka AE, Heritier S, Fong C, Wang Y, Cicuttini FM. Effect of low-dose amitriptyline on low back pain with a neuropathic component: a post hoc analysis. Journal of Pain 2010;21(6):899-902. [DOI: 10.1016/j.spinee.2021.01.012] [DOI] [PubMed] [Google Scholar]

Usha Rani 1996 {published data only}

An evaluation of antidepressants in rheumatic pain conditions. Usha Rani P, Naidu MU, Prasad VB, Ramesh Kumar Rao T, Shobha JC. Anesthesia & Analgesia 1996;83(2):371-5. [DOI] [PubMed] [Google Scholar]

Vijayalakshmi 2016 {published data only}

Vijayalakshmi 2016. Efficacy and safety of duloxetine in patients with neuropathic pain. International Journal of Pharmtech Research 2016;9(4):48-53. [Google Scholar]

Ward 1984 {published data only}

Ward 1984. Antidepressants in concomitant chronic back pain and depression: doxepin and desipramine compared. Journal of Clinical Psychiatry 1984;45(3):54-9. [PubMed] [Google Scholar]

Yucel 2005 {published data only}

Yucel A, Ozyalcin S, Talu GK, Kiziltan E, Yucel B, Andersen OK, et al. The effect of venlafaxine on ongoing and experimentally induced pain in neuropathic pain patients: a double blind, placebo controlled study. Journal of Pain 2005;1(9):407-16. [DOI: 10.1016/j.ejpain.2004.09.009] [DOI] [PubMed] [Google Scholar]

Zitman 1990 {published data only}

Zitman FG, Linssen ACG, Edelbroek PM, Stijnen T. Low dose amitriptyline in chronic pain: the gain is modest. Pain 1990;42(1):35-42. [DOI: 10.1016/0304-3959(90)91089-2] [DOI] [PubMed] [Google Scholar]

References to studies awaiting assessment

NCT03249558 {unpublished data only}

NCT03249558. Effect of combined morphine and duloxetine on chronic pain (duloxetine). clinicaltrials.gov/study/NCT03249558 (first posted 11 August 2017).

References to ongoing studies

ACTRN12624000919516 {unpublished data only}

ACTRN12624000919516. Duloxetine for chronic sciatica (DREAM): an adaptive randomised placebo-controlled trial [DREAM: an adaptive randomised placebo-controlled trial of duloxetine compared to placebo for reducing leg pain in people with chronic sciatica]. anzctr.org.au/Trial/Registration/TrialReview.aspx?ACTRN=12624000919516 (first posted 29 July 2024).

Additional references

Abdel Shaheed 2021

Abdel Shaheed C, Blyth F, Furmage AM, Stanaway F. Getting the “balance” right in clinical trials. BMJ 2021;375:n2869. [DOI: 10.1136/bmj.n2869] [DOI] [PubMed] [Google Scholar]

Abdel Shaheed 2023

Abdel Shaheed C, Mathieson S, Wilson R, Furmage AM, Maher CG. Who should judge treatment effects as unimportant? Journal of Physiotherapy 2023;69(3):133-5. [DOI: 10.1016/j.jphys.2023.04.001] [DOI] [PubMed] [Google Scholar]

Bagg 2020

Bagg MK, O’Hagan E, Zahara P, Wand BM, Hübscher M, Moseley GL, et al. Systematic reviews that include only published data may overestimate the effectiveness of analgesic medicines for low back pain: a systematic review and meta-analysis. Journal of Clinical Epidemiology 2020;124:149-59. [DOI: 10.1016/j.jclinepi.2019.12.006] [DOI] [PubMed] [Google Scholar]

Birkinshaw 2023

Birkinshaw H, Friedrich CM, Cole P, Eccleston C, Serfaty M, Stewart G, et al. Antidepressants for pain management in adults with chronic pain: a network meta‐analysis. Cochrane Database of Systematic Reviews 2023, Issue 5. Art. No: CD014682. [DOI: 10.1002/14651858.CD014682.pub2] [DOI] [PMC free article] [PubMed] [Google Scholar]

Bogduk 2009

Bogduk N. On the definitions and physiology of back pain, referred pain, and radicular pain. Pain 2009;147:17-9. [DOI: 10.1016/j.pain.2009.08.020] [DOI] [PubMed] [Google Scholar]

Center for Scientific Integrity 2018

The Center for Scientific Integrity. Retraction Watch database. Available from: http://retractiondatabase.org/.

Chiarotto 2018

Chiarotto A, Boers M, Deyo RA, Buchbinder R, Corbin TP, Costa LO, et al. Core outcome measurement instruments for clinical trials in nonspecific low back pain. Pain 2018;159(3):481-95. [DOI: 10.1097/j.pain.0000000000001117] [DOI] [PMC free article] [PubMed] [Google Scholar]

Chou 2017

Chou R, Deyo R, Friedly J, Skelly A, Hashimoto R, Weimer M, et al. Systemic pharmacologic therapies for low back pain: a systematic review for an American College of Physicians clinical practice guideline. Annals of Internal Medicine 2017;166(7):480-93. [DOI: 10.7326/M16-2458] [DOI] [PubMed] [Google Scholar]

Cornelius 2022

Cornelius VR, Phillips R. Improving the analysis of adverse event data in randomized controlled trials. Journal of Clinical Epidemiology 2022;144:185-92. [DOI: 10.1016/j.jclinepi.2021.12.023] [DOI] [PubMed] [Google Scholar]

Deeks 2023

Deeks JJ, Higgins JP, Altman DG, editor(s). Chapter 10: Analysing data and undertaking meta-analyses. In: Higgins JP, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al, editor(s). Cochrane Handbook for Systematic Reviews of Interventions Version 6.4 (updated August 2023). Cochrane, 2023. Available from https://training.cochrane.org/handbook/archive/v6.4.

Dieleman 2016

Dieleman JL, Baral R, Birger M, Bui AL, Bulchis A, Chapin A, et al. US spending on personal health care and public health, 1996-2013. JAMA 2016;316(24):2627-46. [DOI: 10.1001/JAMA.2016.16885] [DOI] [PMC free article] [PubMed] [Google Scholar]

Dionne 2008

Dionne CE, Dunn KM, Croft PR, Nachemson AL, Buchbinder R, Walker BF, et al. A consensus approach toward the standardization of back pain definitions for use in prevalence studies. Spine 2008;33(1):95-103. [DOI: 10.1097/BRS.0b013e31815e7f94] [DOI] [PubMed] [Google Scholar]

Dworkin 2008

Dworkin RH, Turk DC, Wyrwich KW, Beaton D, Cleeland CS, Farrar JT, et al. Interpreting the clinical importance of treatment outcomes in chronic pain clinical trials: IMMPACT recommendations. Journal of Pain 2008;9(2):105-21. [DOI: 10.1016/j.jpain.2007.09.005] [DOI] [PubMed] [Google Scholar]

FDA 2020

United States Food and Drug Administration. Title 21 - Food and Drugs Chapter I - Food and Drug Administration Department of Health and Human Services Subchapter D - Drugs for Human Use. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm?fr=312.32 (accessed 3 September 2021).

Ferraro 2021

Ferraro MC, Bagg MK, Wewege MA, Cashin AG, Leake HB, Rizzo RR, et al. Efficacy, acceptability, and safety of antidepressants for low back pain: a systematic review and meta-analysis. Systematic Reviews 2021;10(62):1-13. [DOI: 10.1186/s13643-021-01599-4] [DOI] [PMC free article] [PubMed] [Google Scholar]

Ferreira 2021

Ferreira GE, McLachlan AJ, Lin CW, Zadro JR, Abdel Shaheed C, O’Keeffe M, et al. Efficacy and safety of antidepressants for the treatment of back pain and osteoarthritis: systematic review and meta-analysis. BMJ 2021;372:m4825. [DOI: 10.1136/bmj.m4825] [DOI] [PMC free article] [PubMed] [Google Scholar]

Ferreira 2023

Ferreira ML, Luca K, Haile LM, Steinmetz JD, Culbreth GT, Cross M, et al. Global, regional, and national burden of low back pain, 1990–2020, its attributable risk factors, and projections to 2050: a systematic analysis of the Global Burden of Disease Study 2021. Lancet Rheumatology 2023;5(6):e316–29. [DOI: 10.1016/S2665-9913(23)00098-X] [DOI] [PMC free article] [PubMed] [Google Scholar]

Finnerup 2016

Finnerup NB, Haroutounian S, Kamerman P, Baron R, Bennett DL, Bouhassira D, et al. Neuropathic pain: an updated grading system for research and clinical practice. Pain 2016;157(8):1599-606. [DOI: 10.1097/j.pain.0000000000000492] [DOI] [PMC free article] [PubMed] [Google Scholar]

Gouveia 2017

Gouveia N, Rodrigues A, Ramiro S, Eusébio M, Machado PM, Canhao H, et al. The use of analgesic and other pain‐relief drugs to manage chronic low back pain: results from a national survey. Pain Practice 2017;17(3):353-65. [DOI: 10.1111/papr.12455] [DOI] [PubMed] [Google Scholar]

GRADEpro GDT [Computer program]

GRADEpro GDT. Hamilton (ON): McMaster University (developed by Evidence Prime), accessed 14 November 2024. Available at https://www.gradepro.org.

Hansford 2023

Hansford HJ, Jones MD, Cashin AG, Ostelo RW, Chiarotto A, Williams SA, et al. The smallest worthwhile effect on pain intensity of nonsteroidal anti-inflammatory drugs and exercise therapy for acute and chronic low back pain: a benefit-harm trade-off study. Journal of Physiotherapy 2023;69(4):240-8. [DOI: 10.1016/j.jphys.2023.08.006] [DOI] [PubMed] [Google Scholar]

Harmer 2017

Harmer CJ, Duman RS, Cowen PJ. How do antidepressants work? New perspectives for refining future treatment approaches. Lancet Psychiatry 2017;4(45):409-18. [DOI: 10.1016/S2215-0366(17)30015-9] [DOI] [PMC free article] [PubMed] [Google Scholar]

Harrisson 2020

Harrisson SA, Ogollah R, Dunn KM, Foster NE, Konstantinou K. Prevalence, characteristics, and clinical course of neuropathic pain in primary care patients consulting with low back-related leg pain. Clinical Journal of Pain 2020;36(11):813-24. [DOI: 10.1097/AJP.0000000000000879] [DOI] [PubMed] [Google Scholar]

Health and Social Care Information Centre 2016

Prescribing and Medicines Team Health and Social Care Information Centre. Prescriptions dispensed in the community: England 2005-2015. https://digital.nhs.uk/data-and-information/publications/statistical/prescriptions-dispensed-in-the-community/prescriptions-dispensed-in-the-community-statistics-for-england-2005-2015 (accessed 1 November 2021).

Henssler 2024

Henssler J, Schmidt Y, Schmidt U, Schwarzer G, Bschor T, Baethge C. Incidence of antidepressant discontinuation symptoms: a systematic review and meta-analysis. Lancet Psychiatry 2024;11(7):526-35. [DOI: 10.1016/S2215-0366(24)00133-0] [DOI] [PubMed] [Google Scholar]

Higgins 2017

Higgins JP, Altman DG, Sterne JA. Chapter 8: Assessing risk of bias in included studies. In: Higgins JP, Churchill R, Chandler J, Cumpston MS, editor(s). Cochrane Handbook for Systematic Reviews of Interventions Version 5.2.0 (updated June 2017). Cochrane, 2017. Available from https://training.cochrane.org/handbook/archive/v5.2.

Higgins 2023

Higgins JP, Eldridge S, Li T, editor(s). Chapter 23: Including variants on randomized trials. In: Higgins JP, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al, editor(s). Cochrane Handbook for Systematic Reviews of Interventions Version 6.4 (updated August 2023). Cochrane, 2023. Available from https://training.cochrane.org/handbook/archive/v6.4.

Hirschfeld 2005

Hirschfeld RM, Mallinckrodt C, Lee TC, Detke MJ. Time course of depression-symptom improvement during treatment with duloxetine. Depression and Anxiety 21;4:170-7. [DOI: 10.1002/DA.20071] [DOI] [PubMed] [Google Scholar]

Hong 2013

Hong J, Reed C, Novick D, Happich M. Costs associated with treatment of chronic low back pain: an analysis of the UK general practice research database. Spine 2013;38(1):75-82. [DOI: 10.1002/art.34347] [DOI] [PubMed] [Google Scholar]

Horowitz 2023

Horowitz MA, Framer A, Hengartner MP, Sørensen A, Taylor D. Estimating risk of antidepressant withdrawal from a review of published data. CNS Drugs 2023;37:143-57. [DOI: 10.1007/s40263-022-00960-y] [DOI] [PMC free article] [PubMed] [Google Scholar]

IASP 2012

IASP. Classification of Chronic Pain. Seattle, WA: IASP Press, 2012. [Google Scholar]

Ivanova 2011

Ivanova JI, Birnbaum HG, Schiller M, Kantor E, Johnstone BM, Swindle RW. Real-world practice patterns, health-care utilization, and costs in patients with low back pain: the long road to guideline-concordant care. Spine Journal 2011;11(7):622-32. [DOI: 10.1016/j.spinee.2011.03.017] [DOI] [PubMed] [Google Scholar]

James 2023

James RJ, Ferguson E. Depression, cognition, and pain: exploring individual, cultural and country-level effects across Europe. Journal of Pain 2023;24(6):1104-15. [DOI: 10.1016/j.jpain.2023.03.006] [DOI] [PubMed] [Google Scholar]

Junqueira 2023

Junqueira DR, Zorzela L, Golder S, Loke Y, Gagnier JJ, Julious SA, et al. CONSORT Harms 2022 statement, explanation, and elaboration: updated guideline for the reporting of harms in randomised trials. BMJ 2023;381:e073725. [DOI: 10.1136/bmj-2022-073725] [DOI] [PubMed] [Google Scholar]

Karran 2024

Karran EL, Cashin AG, Barker T, Boyd MA, Chiarotto A, Mohabir V, et al. Development of recommendations for a minimum dataset for Identifying Social factors that Stratify Health Opportunities and Outcomes (ISSHOOs) in pain research. MethodsX 2024;12:102496. [DOI: 10.1016/j.mex.2023.102496] [DOI] [PMC free article] [PubMed] [Google Scholar]

Knezevic 2021

Knezevic NN, Candido KD, Vlaeyen JW, Van Zundert J, Cohen SP. Low back pain. Lancet 2021;398(10294):78-92. [DOI: 10.1016/S0140-6736(21)00733-9] [DOI] [PubMed] [Google Scholar]

Koes 2006

Koes BW, Van Tulder MW, Thomas, S. Diagnosis and treatment of low back pain. BMJ 2006;332(7555):1430-4. [DOI: 10.1136/bmj.332.7555.1430] [DOI] [PMC free article] [PubMed] [Google Scholar]

Koes 2018

Koes BW, Backes D, Bindels PJ. Pharmacotherapy for chronic non-specific low back pain: current and future options. Expert Opinion on Pharmacotherapy 2018;19(6):537-45. [DOI: 10.1080/14656566.2018.1454430] [DOI] [PubMed] [Google Scholar]

Kongsted 2015

Kongsted A, Kent P, Hestbaek L, Vach W. Patients with low back pain had distinct clinical course patterns that were typically neither complete recovery nor constant pain. A latent class analysis of longitudinal data. Spine Journal 2015;15(5):885-94. [DOI: 10.1016/j.spinee.2015.02.012] [DOI] [PubMed] [Google Scholar]

Konstantinou 2013

Konstantinou K, Hider SL, Jordan JL, Lewis M, Dunn KM, Hay EM. The impact of low back-related leg pain on outcomes as compared with low back pain alone: a systematic review of the literature. Clinical Journal of Pain 2013;29(7):644-54. [DOI: 10.1097/AJP.0b013e31826f9a52] [DOI] [PubMed] [Google Scholar]

Kremer 2018

Kremer M, Yalcin I, Goumon Y, Wurtz X, Nexon L, Daniel D, et al. A dual noradrenergic mechanism for the relief of neuropathic allodynia by the antidepressant drugs duloxetine and amitriptyline. Journal of Neuroscience 2018;38(46):9934-54. [DOI: 10.1523/JNEUROSCI.1004-18.2018] [DOI] [PMC free article] [PubMed] [Google Scholar]

Lachin 2016

Lachin JM. Fallacies of last observation carried forward analyses. Clinical Trials 2016;13(2):161-8. [DOI: 10.1177/1740774515602688] [DOI] [PMC free article] [PubMed] [Google Scholar]

Mafi 2013

Mafi JN, McCarthy EP, Davis RB, Landon BE. Worsening trends in the management and treatment of back pain. JAMA Internal Medicine 2013;173(17):1573-81. [DOI: 10.1001/jamainternmed.2013.8992] [DOI] [PMC free article] [PubMed] [Google Scholar]

Mathieson 2018

Mathieson S, Valenti L, Maher CG, Britt H, Li Q, McLachlan AJ, et al. Worsening trends in analgesics recommended for spinal pain in primary care. European Spine Journal 2018;27(5):1136-45. [DOI: 10.1007/s00586-017-5178-4] [DOI] [PubMed] [Google Scholar]

Mayo‐Wilson 2023

Mayo-Wilson E, Qureshi R, Li T. Conducting separate reviews of benefits and harms could improve systematic reviews and meta-analyses. Systematic Reviews 2023;12(67):1-5. [DOI: 10.1186/s13643-023-02234-0] [DOI] [PMC free article] [PubMed] [Google Scholar]

McCleane 2008

McCleane G. Antidepressants as analgesics. CNS Drugs 2008;22:139-56. [DOI: 10.2165/00023210-200822020-00005] [DOI] [PubMed] [Google Scholar]

Micó 2006

Micó JA, Ardid D, Berrocoso E, Eschalier A. Antidepressants and pain. Trends in Pharmacological Sciences 2006;27(7):348-54. [DOI: 10.1016/j.tips.2006.05.004] [DOI] [PubMed] [Google Scholar]

Mika 2013

Mika J, Zychowska M, Makuch W, Rojewska E, Przewlocka B. Neuronal and immunological basis of action of antidepressants in chronic pain–clinical and experimental studies. Pharmacological Reports 2013;65(6):1611-21. [DOI: 10.1016/S1734-1140(13)71522-6] [DOI] [PubMed] [Google Scholar]

Moher 2010

Moher D, Hopewell S, Schulz KF, Montori V, Gøtzsche PC, Devereaux PJ, et al. CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials. BMJ 2010;340:c869. [DOI: 10.1136/bmj.c869] [DOI] [PMC free article] [PubMed] [Google Scholar]

Moore 2012

Moore RA, Straube S, Eccleston C, Derry S, Aldington D, Wiffen P, et al. Estimate at your peril: imputation methods for patient withdrawal can bias efficacy outcomes in chronic pain trials using responder analyses. Pain 2012;153(2):265-8. [DOI: 10.1016/j.pain.2011.10.004] [DOI] [PubMed] [Google Scholar]

NICE 2016

National Institute for Health and Care Excellence (NICE). Low back pain and sciatica in over 16s: assessment and management; 30 November 2016. Available at https://www.nice.org.uk/guidance/ng59. [PubMed]

NICE 2022

National Institute for Health and Care Excellence (NICE). Depression in adults: treatment and management; 29 June 2022. Available at https://www.nice.org.uk/guidance/ng222. [PubMed]

O'Neill 2014

O'Neill J, Tabish H, Welch V, Petticrew M, Pottie K, Clarke M, et al. Applying an equity lens to interventions: using PROGRESS ensures consideration of socially stratifying factors to illuminate inequities in health. Journal of Clinical Epidemiology 2014;67(1):56-64. [DOI: 10.1016/j.jclinepi.2013.08.005] [DOI] [PubMed] [Google Scholar]

Oliveira 2018

Oliveira CB, Maher CG, Pinto RZ, Traeger AC, Lin CW, Chenot JF, et al. Clinical practice guidelines for the management of non-specific low back pain in primary care: an updated overview. European Spine Journal 2018;27(11):2791-803. [DOI: 10.1007/s00586-018-5673-2] [DOI] [PubMed] [Google Scholar]

Orhurhu 2019

Orhurhu V, Olusunmade M, Akinola Y, Urits I, Orhurhu MS, Viswanath O, et al. Depression trends in patients with chronic pain: an analysis of the nationwide inpatient sample. Pain Physician 2019;22:E487-94. [PubMed] [Google Scholar]

Ostelo 2008

Ostelo RW, Deyo RA, Stratford P, Waddell G, Croft P, Von Korff M, et al. Interpreting change scores for pain and functional status in low back pain: towards international consensus regarding minimal important change. Spine 2008;33(1):90-4. [DOI: 10.1097/BRS.0B013E31815E3A10] [DOI] [PubMed] [Google Scholar]

Page 2020

Page MJ, Moher D, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. PRISMA 2020 explanation and elaboration: updated guidance and exemplars for reporting systematic reviews. BMJ 2021;372:n160. [DOI: 10.1136/bmj.n71] [DOI] [PMC free article] [PubMed] [Google Scholar]

Page 2023

Page MJ, Higgins JP, Sterne JA. Chapter 13: Assessing risk of bias due to missing results in a synthesis. In: Higgins JP, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al, editor(s). Cochrane Handbook for Systematic Reviews of Interventions Version 6.4 (updated August 2023). Cochrane, 2023. Available from https://training.cochrane.org/handbook/archive/v6.4.

Paoli 1960

Paoli F, Dacourt G, Corsa P. Preliminary note on the action of imipramine in painful conditions [Note préliminaire sur l’action de l’imipramine dans les états douloureux]. Revue Neurologique 1960;2:503-4. [PubMed] [Google Scholar]

Peryer 2023

Peryer G, Golder S, Junqueira D, Vohra S, Loke YK. Chapter 19: Adverse effects. In: Higgins JP, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al, editor(s). Cochrane Handbook for Systematic Reviews of Interventions Version 6.4 (updated August 2023). Cochrane, 2023. Available from https://training.cochrane.org/handbook/archive/v6.4.

Phillips 2022

Phillips R, Cro S, Wheeler G, Bond S, Morris TP, Creanor S, et al. Visualising harms in publications of randomised controlled trials: consensus and recommendations. BMJ 2022;377:e068983. [DOI: 10.1136/bmj-2021-068983] [DOI] [PMC free article] [PubMed] [Google Scholar]

Price 2024

Price MR, Mead KE, Cowell DM, Troutner AM, Barton TE, Walters SA, et al. Medication recommendations for treatment of lumbosacral radiculopathy: a systematic review of clinical practice guidelines. PM&R 2024;16(10):1128-42. [DOI: 10.1002/pmrj.13142] [DOI] [PubMed] [Google Scholar]

Qaseem 2017

Qaseem A, Wilt TJ, McLean RM, Forciea MA. Noninvasive treatments for acute, subacute, and chronic low back pain: a clinical practice guideline from the American College of Physicians. Annals of Internal Medicine 2017;166(7):514-30. [DOI: 10.7326/M16-2367] [DOI] [PubMed] [Google Scholar]

Qureshi 2022

Qureshi R, Mayo-Wilson E, Li T. Harms in systematic reviews paper 1: an introduction to research on harms. Journal of Clinical Epidemiology 2022;143:186-96. [DOI: 10.1016/j.jclinepi.2021.10.023] [DOI] [PMC free article] [PubMed] [Google Scholar]

Rayner 2016

Rayner L, Hotopf M, Petkova H, Matcham F, Simpson A, McCracken LM. Depression in patients with chronic pain attending a specialised pain treatment centre: prevalence and impact on health care costs. Pain 2016;157(7):1472-9. [DOI: 10.1097/j.pain.0000000000000542] [DOI] [PMC free article] [PubMed] [Google Scholar]

RevMan 2024 [Computer program]

Review Manager (RevMan). Version 8.11.0. The Cochrane Collaboration, 2024. Available at https://revman.cochrane.org.

Riley 2010

Riley RD, Lambert PC, Abo-Zaid G. Meta-analysis of individual participant data: rationale, conduct, and reporting. BMJ 2010;340:c221. [DOI: 10.1136/bmj.c221] [DOI] [PubMed] [Google Scholar]

Riley 2011

Riley RD, Higgins JP, Deeks JJ. Interpretation of random effects meta-analyses. BMJ 2011;342(7804):964-7. [DOI: 10.1136/bmj.d549] [DOI] [PubMed] [Google Scholar]

Santesso 2020

Santesso N, Glenton C, Dahm P, Garner P, Akl EA, Alper B, et al. GRADE guidelines 26: informative statements to communicate the findings of systematic reviews of interventions. Journal of Clinical Epidemiology 2020;119:126-35. [DOI: 10.1016/j.jclinepi.2019.10.014] [DOI] [PubMed] [Google Scholar]

Schmid 2023

Schmid AB, Tampin B, Baron R, Finnerup NB, Hansson P, Hietaharju A, et al. Recommendations for terminology and the identification of neuropathic pain in people with spine-related leg pain. Outcomes from the NeuPSIG working group. Pain 2023;164(8):1693-704. [DOI: 10.1097/j.pain.0000000000002919] [DOI] [PMC free article] [PubMed] [Google Scholar]

Schünemann 2023

Schünemann HJ, Higgins JP, Vist GE, Glasziou P, Akl EA, Skoetz N, et al. Chapter 14: Completing ‘Summary of findings’ tables and grading the certainty of the evidence. In: Higgins JP, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al, editor(s). Cochrane Handbook for Systematic Reviews of Interventions Version 6.4 (updated August 2023). Cochrane, 2023. Available from https://training.cochrane.org/handbook/archive/v6.4.

Sterne 2019

Sterne JA, Savović J, Page MJ, Elbers RG, Blencowe NS, Boutron I, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ 2019;366:l4898. [DOI: 10.1136/bmj.l4898] [DOI] [PubMed] [Google Scholar]

Urits 2019

Urits I, Peck J, Orhurhu MS, Wolf J, Patel R, Orhurhu V, et al. Off-label antidepressant use for treatment and management of chronic pain: evolving understanding and comprehensive review. Current Pain and Headache Reports 2019;23(66):1-10. [DOI: 10.1007/s11916-019-0803-z] [DOI] [PubMed] [Google Scholar]

Vos 2020

Vos T, Lim SS, Abbafati C, Abbas KM, Abbasi M, Abbasifard M, et al. Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet 2020;396(10258):1204-22. [DOI: 10.1016/S0140-6736(20)30925-9] [DOI] [PMC free article] [PubMed] [Google Scholar]

WHO 2022

World Health Organization Collaborating Centre for Drug Statistics Methodology. ATC/DDD Index 2022. https://www.whocc.no/atc_ddd_index/ (accessed 10 February 2022).

References to other published versions of this review

Urquhart 2008

Urquhart D, Hoving JL, Assendelft WJJ, Roland M, Tulder MW. Antidepressants for non-specific low back pain. Cochrane Database of Systematic Reviews 2008, Issue 1. Art. No: CD001703. [DOI: 10.1002/14651858.CD001703.pub3] [DOI] [PMC free article] [PubMed] [Google Scholar]

Van Tulder 1999

Van Tulder MW, Hienkens E, Roland M, Assendelft WJ. Antidepressants for non‐specific low‐back pain. Cochrane Database of Systematic Reviews 1999, Issue 7. Art. No: CD001703. [DOI: 10.1002/14651858.CD001703] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0001] Afilal S. Efficacy of clomipramine for chronic lumbar radicular pain a randomized clinical trial (Medical specialisation dissertation). Rabat (Morocco): Mohammed V University Faculty of Medicine and Pharmacy, 2020. [Google Scholar]

[CD001703-bib-0002] Afilal SA, Fellous S, Ilham A, Abouqal R, Hmamouchi I, Alami N, et al. Efficacy of clomipramine for chronic lumbar radicular pain a randomized clinical trial. Annals of the Rheumatic Diseases 2020;79:59. [DOI: 10.1136/annrheumdis-2020-eular.4373] [DOI] [Google Scholar]

[CD001703-bib-0003] PACTR202001764151121. Efficacy of clomipramine for chronic lumbar radicular pain: a randomized clinical trial. trialsearch.who.int/Trial2.aspx?TrialID=PACTR202001764151121 (first posted 6 November 2019).

[CD001703-bib-0004] Alcoff J, Jones E, Rust P, Newman R. Controlled trial of imipramine for chronic low back pain. Journal of Family Practice 1982;14:841-6. [PubMed] [Google Scholar]

[CD001703-bib-0005] Atkinson HJ, Slater MA, Williams RA, Zisook S, Patterson TL, Grant I, et al. A placebo-controlled randomized clinical trial of nortriptyline for chronic low back pain. Pain 1998;76:287-96. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0006] Atkinson JH, Slater MA, Wahlgren DR, Williams RA, Zisook S, Pruitt SD, et al. Effects of noradrenergic and serotonergic antidepressants on chronic low back pain intensity. Pain 1999;83:137-45. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0007] Atkinson JH, Slater MA, Capparelli EV, Wallace MS, Zisook S, Abramson I, et al. Efficacy of noradrenergic and serotonergic antidepressants in chronic back pain. Journal of Clinical Psychopharmacology 2007;27(2):135-42. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0008] NCT00018200. Effect of antidepressants on back pain. clinicaltrials.gov/study/NCT00018200 (first posted 23 June 2005).

[CD001703-bib-0009] Dickens C, Jayson M, Sutton C, Creed F. The relationship between pain and depression in a trial using paroxetine in sufferers of chronic low back pain. Psychosomatics 2000;41:490-9. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0010] GlaxoSmithKline. A double-blind study comparing the efficacy and tolerability of paroxetine and placebo in hospital out-patients with depressive symptoms associated with chronic back pain. www.gsk.com/media/1543/depression_701_full.pdf 2001.

[CD001703-bib-0011] Goodkin K, Gullion CM, Agras WS. A randomized, double-blind, placebo-controlled trial of trazodone hydrochloride in chronic low back pain syndrome. Journal of Clinical Psychopharmacology 1990;10:269-78. [PubMed] [Google Scholar]

[CD001703-bib-0012] Gould HM, Atkinson JH, Chircop-Rollick T, D'Andrea J, Garfin S, Patel SM, et al. A randomized placebo-controlled trial of desipramine, cognitive behavioral therapy, and active placebo therapy for low back pain. Pain 2020;161(6):1341-9. [DOI: 10.1097/j.pain.0000000000001834] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0013] NCT00964886. Efficacy of antidepressants in chronic back pain. clinicaltrials.gov/study/NCT00964886 (first posted 24 August 2009).

[CD001703-bib-0014] Jenkins DG, Ebbutt AF, Evans CD. Tofranil in the treatment of low back pain. Journal of International Medical Research 1976;4:28-40. [PubMed] [Google Scholar]

[CD001703-bib-0015] Chatterjee N, Noor N, Crowell A, Mackey S. Duloxetine and placebo alter different gray matter regions in chronic low back pain patients. Journal of Pain 2010;11(4 Suppl 1):S40. [Google Scholar]

[CD001703-bib-0016] Johnson K, Chatterjee N, Noor N, Crowell A, McCue R, Mackey S. Effects of duloxetine and placebo in patients with chronic low back pain. Journal of Pain 2011;12(4 Suppl 1):49. [Google Scholar]

[CD001703-bib-0017] NCT00388414. Imaging study of chronic low back pain in patients taking pain medication. clinicaltrials.gov/study/NCT00388414 (first posted 13 October 2010).

[CD001703-bib-0018] Katz J, Pennella-Vaughan J, Hetzel RD, Kanazi GE, Dworkin RH. A randomised, placebo-controlled trial of bupropion sustained release in chronic low back pain. Journal of Pain 2005;6(10):656-61. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0019] Khoromi S, Cui L, Nackers L, Max MB. Morphine, nortriptyline and their combination vs. placebo in patients with chronic lumbar root pain. Pain 2007;130(1-2):66-75. [DOI: 10.1016/j.pain.2006.10.029] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0020] Konno S, Oda N, Ochiai T, Alev L. Randomized, double-blind, placebo-controlled phase III trial of duloxetine monotherapy in Japanese patients with chronic low back pain. Spine 2016;41(22):1709–17. [DOI: 10.1097/BRS.0000000000001707] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0021] NCT01855919. A study of duloxetine (LY248686) in participants with chronic low back pain. clinicaltrials.gov/study/NCT01855919 (first posted 14 May 2013).

[CD001703-bib-0022] jRCT2080222083. A phase 3 clinical trial of duloxetine in patients with chronic low back pain. jrct.niph.go.jp/en-latest-detail/jRCT2080222083 (first posted 15 May 2013).

[CD001703-bib-0023] Kurniawati N, Suharjanti I, Hamdan M, Rahardjo JE, Ardiansyah D, Hidayati HB. Efficacy of acetaminophen plus amitriptyline compared to acetaminophen to reduce pain intensity in nonspecific chronic lower back pain. Indian Journal of Forensic Medicine & Toxicology 2020;14(2):1819-25. [DOI: 10.37506/ijfmt.v14i2.3202] [DOI] [Google Scholar]

[CD001703-bib-0024] Marks DM, Pae CU, Patkar AA. A double-blind, placebo-controlled, parallel-group pilot study of milnacipran for chronic radicular pain (sciatica) associated with lumbosacral disc disease. Primary Care Companion for CNS Disorders 014;16(4):26268. [DOI: 10.4088/PCC.14m01658] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0025] NCT01777581. Milnacipran for lumbosacral radicular pain. clinicaltrials.gov/study/NCT01777581 (first posted 24 January 2013).

[CD001703-bib-0026] NCT01225068. Effect of milnacipran in chronic neuropathic low back pain. clinicaltrials.gov/ct2/show/NCT01225068 (first posted 19 October 2010).

[CD001703-bib-0027] Pheasant H, Bursk A, Goldfarb J, Azen SP, Weiss JN, Borelli L. Amitriptyline and chronic low-back pain. A randomized double-blind crossover study. Spine 1983;8:552-7. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0028] Pirbudak L, Karakurum G, Satana T, Karadasli H, Topalhan M, Oner U, et al. Epidural steroid injection and amitriptyline in the management of acute low back pain originating from lumbar disc herniation [Dysk hernysy kaynakli akut dyskojenyk bel adrilarinda epydural steroyd enjeksyyonu ve amytryptlyn kullanimi]. Artroplasty Artroskopyk Cerrahy [Journal of Arthroplasty & Arthroscopic Surgery] 2003;14(2):89-93. [Google Scholar]

[CD001703-bib-0029] NCT01179828. Linking altered central pain processing and genetic polymorphism to drug efficacy in chronic low back pain (prediction). clinicaltrials.gov/study/NCT01179828 (first posted 10 August 2010).

[CD001703-bib-0030] Schliessbach J, Siegenthaler A, Bütikofer L, Limacher A, Juni P, Vuilleumier PH, et al. Effect of single-dose imipramine on chronic low-back and experimental pain. PLoS One 2018;13(5):e0195776. [DOI: 10.1371/journal.pone.0195776] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0031] 2009-012713-22. Randomized double-blind study comparing the efficacy of duloxetine with placebo in patients with chronic low back pain. clinicaltrialsregister.eu/ctr-search/search?query=2009-012713-22 (first posted 14 October 2010).

[CD001703-bib-0032] NCT01166048. Comparison of the efficacy of duloxetine with placebo in patients with chronic low back pain with a radicular component. clinicaltrials.gov/study/NCT01166048 (first posted 19 July 2010).

[CD001703-bib-0033] Schukro RP, Oehmke MJ, Geroldinger A, Heinze G, Kress HG, Pramhas S. Efficacy of duloxetine in chronic low back pain with a neuropathic component: a randomized, double-blind, placebo-controlled crossover trial. Anesthesiology 2016;124(1):150-8. [DOI: 10.1097/ALN.0000000000000902] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0034] NCT00408876. Duloxetine versus placebo in chronic low back pain. clinicaltrials.gov/study/NCT00408876 (first posted 6 December 2006).

[CD001703-bib-0035] Skljarevski V, Ossanna M, Liu-Seifert H, Zhang Q, Chappell A, Iyengar S, et al. A double-blind, randomized trial of duloxetine versus placebo in the management of chronic low back pain. European Journal of Neurology 2009;16(9):1041-8. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0036] EUCTR2006-003484-31. Effect of duloxetine 60 mg to 120 mg once daily in patients with chronic low back pain. clinicaltrialsregister.eu/ctr-search/search?query=2006-003484-31 (first posted 4 October 2006).

[CD001703-bib-0037] NCT00424593. Duloxetine versus placebo in chronic low back pain. clinicaltrials.gov/study/NCT00424593 (first posted 17 January 2007).

[CD001703-bib-0038] Skljarevski V, Desaiah D, Liu-Seifert H, Zhang Q, Chappell AS, Detke MJ, et al. Efficacy and safety of duloxetine in patients with chronic low back pain. Spine 2010;35(13):E578-85. [DOI: 10.1097/BRS.0b013e3181d3cef6] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0039] Skljarevski V, Desaiah D, Liu-Seifert H, Zhang Q, Chappell AS, Detke MJ, et al. Efficacy of duloxetine in chronic low back pain. European Journal of Neurology 2008;15(Suppl 3)(320):P2447. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0040] EUCTR2009-012713-22. Randomized double-blind study comparing the efficacy of duloxetine with placebo in patients with chronic low back pain. clinicaltrialsregister.eu/ctr-search/trial/2009-012713-22/AT (first posted 14 April 2010).

[CD001703-bib-0041] NCT00767806. A study for patient with chronic low back pain. clinicaltrials.gov/study/NCT00767806 (first posted 6 October 2008).

[CD001703-bib-0042] NL-OMON32370. Effect of duloxetine 60 mg once daily versus placebo in patients with chronic low back pain. onderzoekmetmensen.nl/en/trial/32370 (first posted 26 June 2008).

[CD001703-bib-0043] Skljarevski V, Zhang S, Desaiah D, Alaka KJ, Palacios S, Miazgowski T, et al. Duloxetine versus placebo in patients with chronic low back pain: a 12-week, fixed-dose, randomized, double-blind trial. Journal of Pain 2010;11(12):1282-90. [DOI: 10.1016/j.jpain.2010.03.002] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0044] Skljarevski V, Zhang S, Desaiah D, Palacios S, Miazgowski T, Patrick K. (248) Efficacy and safety of duloxetine 60 mg once-daily in patients with chronic low back pain. Journal of Pain 2010;11:s38. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0045] Skljarevski V, Zhang S, Desaiah D, Palacios S, Miazgowski T, Patrick K. Effect of duloxetine 60 mg once daily versus placebo in patients with chronic low back pain: a 12-week, randomized, double-blind trial. Pain Medicine 2010;11:322. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0046] Treves R, Montane De La Roque P, Dumond JJ, Bertin P, Arnaud M, Desproges-Gotteron R. Prospective study of the analgesic action of clomipramine versus placebo in refractory low back pain and sciatica (68 cases) [Étude prospective de l'action antalgique de la clomipramine versus placebo dans les lombosciatalgies rebelles (68 cas)]. Revue du Rhumatisme et des Maladies Osteo-articulaires 1991;58:549-52. [PubMed] [Google Scholar]

[CD001703-bib-0047] ACTRN12612000131853. Is amitriptyline effective in the management of chronic low back pain? [Is low dose amitriptyline more effective than an active placebo in the management of chronic low back pain? A double-blind, randomized, placebo-controlled trial]. anzctr.org.au/Trial/Registration/TrialReview.aspx?id=361972&isReview=true (first posted 25 January 2012).

[CD001703-bib-0048] Urquhart D, Wluka A, Van Tulder M, Heritier S, Forbes A, Fong C, et al. Efficacy of low-dose amitriptyline for chronic low back pain: a double-blind, randomised controlled trial. Osteoarthritis and Cartilage 2018;26:s416. [Google Scholar]

[CD001703-bib-0049] Urquhart DM, Wluka AE, Sim MR, Tulder M, Forbes A, Gibson SJ, et al. Is low-dose amitriptyline effective in the management of chronic low back pain? Study protocol for a randomised controlled trial. Trials 2016;17(514):1-8. [DOI: 10.1186/s13063-016-1637-1] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0050] Urquhart DM, Wluka AE, Van Tulder M, Heritier S, Forbes A, Fong C, et al. Efficacy of low-dose amitriptyline for chronic low back pain: a randomized clinical trial. JAMA Internal Medicine 2018;178(11):1474-81. [DOI: 10.1001/jamainternmed.2018.4222] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0051] Gelijkens V, Van Zundert J, De Vooght P, Vander Laenen M, Heylen R, Vanelderen, P. The effectiveness of amitriptyline in the treatment of subacute lumbar radicular pain. European Journal of Anaesthesiology 2014;52:232. [Google Scholar]

[CD001703-bib-0052] NCT01869907. Effect of minocycline on pain caused by nerve damage [Effect of minocycline on neuropathic pain]. clinicaltrials.gov/study/NCT01869907 (first posted 2 June 2013).

[CD001703-bib-0053] Vanelderen P, Van Zundert J, Kozicz T, Puylaert M, De Vooght P, Mestrum R, et al. Effect of minocycline on lumbar radicular neuropathic pain: a randomized, placebo-controlled, double-blind clinical trial with amitriptyline as a comparator. Anesthesiology 2015;122(2):399-406. [DOI: 10.1097/ALN.0000000000000508] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0054] Brannan SK, Mallinckrodt CH, Brown EB, Wohlreich MM, Watkin JG, Schatzberg AF. Duloxetine 60 mg once-daily in the treatment of painful physical symptoms in patients with major depressive disorder. Journal of Psychiatric Research 2005;39(1):43-53. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0055] Chrzanowski J. Imipramine in the treatment of chronic sciatic pains [Imipramina w leczeniu przewlekłej rwy kulszowej]. Wiadomosci Lekarskie 1977;30(18):1439-40. [PubMed] [Google Scholar]

[CD001703-bib-0056] Davidson JR, France RD. Bupropion in chronic low back pain. Journal of Clinical Psychiatry 1994;55(8):362. [PubMed] [Google Scholar]

[CD001703-bib-0057] DeLomba W, Bernstein M, Basso K, Suárez E, McLean S, Beaudoin F. 231 The tolerability and effectiveness of duloxetine for the prevention of persistent musculoskeletal pain after trauma and injury: a pilot three-group randomized controlled trial. Annals of Emergency Medicine 2021;78:S94. [Google Scholar]

[CD001703-bib-0058] Strauss DH, Santhanam DR, McLean SA, Beaudoin FL. Study protocol for a randomised, double-blind, placebo-controlled clinical trial of duloxetine for the treatment and prevention of musculoskeletal pain: altering the transition from acute to chronic pain (ATTAC pain). BMJ Open 2019;9(3):e025002. [DOI: 10.1136/bmjopen-2018-025002] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0059] EUCTR2013-004564-55-RO. A randomized, multicenter, double-blind, placebo controlled, parallel study to examine the efficacy and safety of a combination treatment in subjects with chronic low back pain [A randomized, multicenter, double-blind, placebo controlled, parallel study to assess the efficacy and safety of the combined administration of sertraline and lamotrigine in subjects with chronic low back pain]. trialsearch.who.int/Trial2.aspx?TrialID=EUCTR2013-004564-55-RO (first posted 30 May 2014).

[CD001703-bib-0060] Gaynor PJ, Gopal M, Zheng W, Martinez JM, Robinson MJ, Hann D. Duloxetine versus placebo in the treatment of major depressive disorder and associated painful physical symptoms: a replication study. Current Medical Research and Opinion 2011;27(10):1859-67. [DOI: 10.1185/03007995.2011.609540] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0061] Gourlay GK, Cherry DA, Cousins MJ, Love BL, Graham JR, McLachlan MO. A controlled study of a serotonin reuptake blocker, zimelidine, in the treatment of chronic pain. Pain 1986;25(1):35-52. [DOI: 10.1016/0304-3959(86)90006-0] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0062] Hameroff SR, Cork RC, Scherer K. Doxepin effects on chronic pain, depression and plasma opioids. Journal of Clinical Psychiatry 1982;43(8):22-6. [PubMed] [Google Scholar]

[CD001703-bib-0063] Hameroff SR, Weiss JL, Lerman JC, Cork RC, Watts KS, Crago BR, et al. Doxepin's effects on chronic pain and depression: a controlled study. Journal of Clinical Psychiatry 1984;45(3):47-53. [PubMed] [Google Scholar]

[CD001703-bib-0064] Isaac M. Double-blind study of venlafaxine in chronic pain. In: 156th Annual Meeting of the American Psychiatric Association; 2003 May 17-22; San Francisco (CA). 2003.

[CD001703-bib-0065] Johansson F, Von Knorring L. A double-blind controlled study of a serotonin uptake inhibitor (zimelidine) versus placebo in chronic pain patients. Pain 1979;7(1):69-78. [DOI: 10.1016/0304-3959(79)90108-8] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0066] Jørgensen B, Nørrelund N, Bech P, Jakobsen K. Pain and depressive symptoms in general practice. A controlled study of Anafranil (clomipramine) versus placebo [En koolrolleret undersegelse af anafranil (klomipramin) versus placebo]. Ugeskrift for Læger 1984;146(38):2868-72. [PubMed] [Google Scholar]

[CD001703-bib-0067] Krell HV, Leuchter AF, Cook IA, Abrams M. Evaluation of reboxetine, a noradrenergic antidepressant, for the treatment of fibromyalgia and chronic low back pain. Psychosomatics 2005;46(5):379-84. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0068] Loldrup D, Langemark M, Hansen HJ, Olesen J, Bech P. Clomipramine and mianserin in chronic idiopathic pain syndrome. A placebo controlled study. Psychopharmacology 1989;99(1):1-7. [DOI: 10.1007/BF00634443] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0069] McQuay JH, Carroll D, Glynn CJ. Low dose amitriptyline in the treatment of chronic pain. Anaesthesia 1992;47(8):646-52. [DOI: 10.1111/j.1365-2044.1992.tb02383.x] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0070] McQuay HJ, Carroll D, Glynn CJ. Dose-response for analgesic effect of amitriptyline in chronic pain. Anaesthesia 1993;48(4):281-5. [DOI: 10.1111/j.1365-2044.1993.tb06943.x] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0071] NCT01119924. A clinical study for assessment of the efficacy and safety of mirtazapine (Smilon) in depression patients with pain. clinicaltrials.gov/study/NCT01119924 (first posted 7 May 2010).

[CD001703-bib-0072] NCT01288937. A placebo controlled, randomized, double blind trial of milnacipran for the treatment of idiopathic neuropathy pain. clinicaltrials.gov/study/NCT01288937 (first posted 2 February 2011).

[CD001703-bib-0073] Pilowsky I, Hallett EC, Bassett DL, Thomas PG, Penhall RK. A controlled study of amitriptyline in the treatment of chronic pain. Pain 1982;14:169-79. [DOI: 10.1016/0304-3959(82)90097-5] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0074] Pilowsky I, Barrow GC. A controlled study of psychotherapy and amitriptyline used individually and in combination in the treatment of chronic intractable, 'psychogenic' pain. Pain 1990;40(1):3-19. [DOI: 10.1016/0304-3959(90)91045-K] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0075] Skljarevski V, Zhang S, Chappell AS, Detke MJ, Murray I, Backonja M. Maintenance of effect of duloxetine in patients with chronic low back pain. European Journal of Neurology 2009;16(Suppl 3):92. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0076] Skljarevski V, Bair MJ, Ossanna MJ, Frakes E, Zhang S, Alaka K. OMERACT responder analysis of patients treated with duloxetine for chronic low back pain. Arthritis and Rheumatism 2010;62(10):S73-4. [Google Scholar]

[CD001703-bib-0077] Skljarevski V, Zhang S, Chappell AS, Walker DJ, Murray I, Backonja M. Maintenance of effect of duloxetine in patients with chronic low back pain: a 41-week uncontrolled, dose-blinded study. Pain Medicine 201;11(5):648-57. [DOI: 10.1111/j.1526-4637.2010.00836.x] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0078] Skljarevski V, Liu P, Zhang S, Ahl J, Martinez JM. Efficacy and safety of duloxetine in patients with chronic low back pain who used versus did not use concomitant nonsteroidal anti-inflammatory drugs or acetaminophen: a post hoc pooled analysis of 2 randomized, placebo-controlled trials. Pain Research and Treatment 2012;2012:296710. [DOI: 10.1155/2012/296710] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0079] Songer DA, Schulte H. Venlafaxine for the treatment of chronic pain. American Journal of Psychiatry 1996;153(3):737. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0080] Storch H, Steck P. Anti-depressive medication in the treatment of low back pain: a controlled trial [Begleitende thymoleptische therapie im rahmen einer kontrollierten studie mit maprotilin (ludiomil) bei der behandlung von kreuzschmerzen]. Nervenarzt 1982;53(8):445-50. [PubMed] [Google Scholar]

[CD001703-bib-0081] Tavafian SS, Jamshidi AR, Mohammad K. Treatment of low back pain: randomized clinical trial comparing a multidisciplinary group-based rehabilitation program with oral drug treatment up to 12 months. International Journal of Rheumatic Diseases 2014;17:159-64. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0082] Tsuji T, Itoh N, Ochiai T, Konno S. Response to duloxetine in chronic low back pain: exploratory post hoc analysis of a Japanese Phase III randomized study. Journal of Pain Research 2017;10:2157-68. [DOI: 10.2147/JPR.S138172] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0083] Tyrer S, Lauder J, Michie M, Griffiths H, Charlton J. The effect of trazodone in depressed and non-depressed patients in chronic pain. European Psychiatry 1996;11(4):359s. [Google Scholar]

[CD001703-bib-0084] Urquhart DM, Wluka AE, Heritier S, Fong C, Wang Y, Cicuttini FM. Effect of low-dose amitriptyline on low back pain with a neuropathic component: a post hoc analysis. Journal of Pain 2010;21(6):899-902. [DOI: 10.1016/j.spinee.2021.01.012] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0085] An evaluation of antidepressants in rheumatic pain conditions. Usha Rani P, Naidu MU, Prasad VB, Ramesh Kumar Rao T, Shobha JC. Anesthesia & Analgesia 1996;83(2):371-5. [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0086] Vijayalakshmi 2016. Efficacy and safety of duloxetine in patients with neuropathic pain. International Journal of Pharmtech Research 2016;9(4):48-53. [Google Scholar]

[CD001703-bib-0087] Ward 1984. Antidepressants in concomitant chronic back pain and depression: doxepin and desipramine compared. Journal of Clinical Psychiatry 1984;45(3):54-9. [PubMed] [Google Scholar]

[CD001703-bib-0088] Yucel A, Ozyalcin S, Talu GK, Kiziltan E, Yucel B, Andersen OK, et al. The effect of venlafaxine on ongoing and experimentally induced pain in neuropathic pain patients: a double blind, placebo controlled study. Journal of Pain 2005;1(9):407-16. [DOI: 10.1016/j.ejpain.2004.09.009] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0089] Zitman FG, Linssen ACG, Edelbroek PM, Stijnen T. Low dose amitriptyline in chronic pain: the gain is modest. Pain 1990;42(1):35-42. [DOI: 10.1016/0304-3959(90)91089-2] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0090] NCT03249558. Effect of combined morphine and duloxetine on chronic pain (duloxetine). clinicaltrials.gov/study/NCT03249558 (first posted 11 August 2017).

[CD001703-bib-0091] ACTRN12624000919516. Duloxetine for chronic sciatica (DREAM): an adaptive randomised placebo-controlled trial [DREAM: an adaptive randomised placebo-controlled trial of duloxetine compared to placebo for reducing leg pain in people with chronic sciatica]. anzctr.org.au/Trial/Registration/TrialReview.aspx?ACTRN=12624000919516 (first posted 29 July 2024).

[CD001703-bib-0092] Abdel Shaheed C, Blyth F, Furmage AM, Stanaway F. Getting the “balance” right in clinical trials. BMJ 2021;375:n2869. [DOI: 10.1136/bmj.n2869] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0093] Abdel Shaheed C, Mathieson S, Wilson R, Furmage AM, Maher CG. Who should judge treatment effects as unimportant? Journal of Physiotherapy 2023;69(3):133-5. [DOI: 10.1016/j.jphys.2023.04.001] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0094] Bagg MK, O’Hagan E, Zahara P, Wand BM, Hübscher M, Moseley GL, et al. Systematic reviews that include only published data may overestimate the effectiveness of analgesic medicines for low back pain: a systematic review and meta-analysis. Journal of Clinical Epidemiology 2020;124:149-59. [DOI: 10.1016/j.jclinepi.2019.12.006] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0095] Birkinshaw H, Friedrich CM, Cole P, Eccleston C, Serfaty M, Stewart G, et al. Antidepressants for pain management in adults with chronic pain: a network meta‐analysis. Cochrane Database of Systematic Reviews 2023, Issue 5. Art. No: CD014682. [DOI: 10.1002/14651858.CD014682.pub2] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0096] Bogduk N. On the definitions and physiology of back pain, referred pain, and radicular pain. Pain 2009;147:17-9. [DOI: 10.1016/j.pain.2009.08.020] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0097] The Center for Scientific Integrity. Retraction Watch database. Available from: http://retractiondatabase.org/.

[CD001703-bib-0098] Chiarotto A, Boers M, Deyo RA, Buchbinder R, Corbin TP, Costa LO, et al. Core outcome measurement instruments for clinical trials in nonspecific low back pain. Pain 2018;159(3):481-95. [DOI: 10.1097/j.pain.0000000000001117] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0099] Chou R, Deyo R, Friedly J, Skelly A, Hashimoto R, Weimer M, et al. Systemic pharmacologic therapies for low back pain: a systematic review for an American College of Physicians clinical practice guideline. Annals of Internal Medicine 2017;166(7):480-93. [DOI: 10.7326/M16-2458] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0100] Cornelius VR, Phillips R. Improving the analysis of adverse event data in randomized controlled trials. Journal of Clinical Epidemiology 2022;144:185-92. [DOI: 10.1016/j.jclinepi.2021.12.023] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0101] Deeks JJ, Higgins JP, Altman DG, editor(s). Chapter 10: Analysing data and undertaking meta-analyses. In: Higgins JP, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al, editor(s). Cochrane Handbook for Systematic Reviews of Interventions Version 6.4 (updated August 2023). Cochrane, 2023. Available from https://training.cochrane.org/handbook/archive/v6.4.

[CD001703-bib-0102] Dieleman JL, Baral R, Birger M, Bui AL, Bulchis A, Chapin A, et al. US spending on personal health care and public health, 1996-2013. JAMA 2016;316(24):2627-46. [DOI: 10.1001/JAMA.2016.16885] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0103] Dionne CE, Dunn KM, Croft PR, Nachemson AL, Buchbinder R, Walker BF, et al. A consensus approach toward the standardization of back pain definitions for use in prevalence studies. Spine 2008;33(1):95-103. [DOI: 10.1097/BRS.0b013e31815e7f94] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0104] Dworkin RH, Turk DC, Wyrwich KW, Beaton D, Cleeland CS, Farrar JT, et al. Interpreting the clinical importance of treatment outcomes in chronic pain clinical trials: IMMPACT recommendations. Journal of Pain 2008;9(2):105-21. [DOI: 10.1016/j.jpain.2007.09.005] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0105] United States Food and Drug Administration. Title 21 - Food and Drugs Chapter I - Food and Drug Administration Department of Health and Human Services Subchapter D - Drugs for Human Use. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm?fr=312.32 (accessed 3 September 2021).

[CD001703-bib-0106] Ferraro MC, Bagg MK, Wewege MA, Cashin AG, Leake HB, Rizzo RR, et al. Efficacy, acceptability, and safety of antidepressants for low back pain: a systematic review and meta-analysis. Systematic Reviews 2021;10(62):1-13. [DOI: 10.1186/s13643-021-01599-4] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0107] Ferreira GE, McLachlan AJ, Lin CW, Zadro JR, Abdel Shaheed C, O’Keeffe M, et al. Efficacy and safety of antidepressants for the treatment of back pain and osteoarthritis: systematic review and meta-analysis. BMJ 2021;372:m4825. [DOI: 10.1136/bmj.m4825] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0108] Ferreira ML, Luca K, Haile LM, Steinmetz JD, Culbreth GT, Cross M, et al. Global, regional, and national burden of low back pain, 1990–2020, its attributable risk factors, and projections to 2050: a systematic analysis of the Global Burden of Disease Study 2021. Lancet Rheumatology 2023;5(6):e316–29. [DOI: 10.1016/S2665-9913(23)00098-X] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0109] Finnerup NB, Haroutounian S, Kamerman P, Baron R, Bennett DL, Bouhassira D, et al. Neuropathic pain: an updated grading system for research and clinical practice. Pain 2016;157(8):1599-606. [DOI: 10.1097/j.pain.0000000000000492] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0110] Gouveia N, Rodrigues A, Ramiro S, Eusébio M, Machado PM, Canhao H, et al. The use of analgesic and other pain‐relief drugs to manage chronic low back pain: results from a national survey. Pain Practice 2017;17(3):353-65. [DOI: 10.1111/papr.12455] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0111] GRADEpro GDT. Hamilton (ON): McMaster University (developed by Evidence Prime), accessed 14 November 2024. Available at https://www.gradepro.org.

[CD001703-bib-0112] Hansford HJ, Jones MD, Cashin AG, Ostelo RW, Chiarotto A, Williams SA, et al. The smallest worthwhile effect on pain intensity of nonsteroidal anti-inflammatory drugs and exercise therapy for acute and chronic low back pain: a benefit-harm trade-off study. Journal of Physiotherapy 2023;69(4):240-8. [DOI: 10.1016/j.jphys.2023.08.006] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0113] Harmer CJ, Duman RS, Cowen PJ. How do antidepressants work? New perspectives for refining future treatment approaches. Lancet Psychiatry 2017;4(45):409-18. [DOI: 10.1016/S2215-0366(17)30015-9] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0114] Harrisson SA, Ogollah R, Dunn KM, Foster NE, Konstantinou K. Prevalence, characteristics, and clinical course of neuropathic pain in primary care patients consulting with low back-related leg pain. Clinical Journal of Pain 2020;36(11):813-24. [DOI: 10.1097/AJP.0000000000000879] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0115] Prescribing and Medicines Team Health and Social Care Information Centre. Prescriptions dispensed in the community: England 2005-2015. https://digital.nhs.uk/data-and-information/publications/statistical/prescriptions-dispensed-in-the-community/prescriptions-dispensed-in-the-community-statistics-for-england-2005-2015 (accessed 1 November 2021).

[CD001703-bib-0116] Henssler J, Schmidt Y, Schmidt U, Schwarzer G, Bschor T, Baethge C. Incidence of antidepressant discontinuation symptoms: a systematic review and meta-analysis. Lancet Psychiatry 2024;11(7):526-35. [DOI: 10.1016/S2215-0366(24)00133-0] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0117] Higgins JP, Altman DG, Sterne JA. Chapter 8: Assessing risk of bias in included studies. In: Higgins JP, Churchill R, Chandler J, Cumpston MS, editor(s). Cochrane Handbook for Systematic Reviews of Interventions Version 5.2.0 (updated June 2017). Cochrane, 2017. Available from https://training.cochrane.org/handbook/archive/v5.2.

[CD001703-bib-0118] Higgins JP, Eldridge S, Li T, editor(s). Chapter 23: Including variants on randomized trials. In: Higgins JP, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al, editor(s). Cochrane Handbook for Systematic Reviews of Interventions Version 6.4 (updated August 2023). Cochrane, 2023. Available from https://training.cochrane.org/handbook/archive/v6.4.

[CD001703-bib-0119] Hirschfeld RM, Mallinckrodt C, Lee TC, Detke MJ. Time course of depression-symptom improvement during treatment with duloxetine. Depression and Anxiety 21;4:170-7. [DOI: 10.1002/DA.20071] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0120] Hong J, Reed C, Novick D, Happich M. Costs associated with treatment of chronic low back pain: an analysis of the UK general practice research database. Spine 2013;38(1):75-82. [DOI: 10.1002/art.34347] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0121] Horowitz MA, Framer A, Hengartner MP, Sørensen A, Taylor D. Estimating risk of antidepressant withdrawal from a review of published data. CNS Drugs 2023;37:143-57. [DOI: 10.1007/s40263-022-00960-y] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0122] IASP. Classification of Chronic Pain. Seattle, WA: IASP Press, 2012. [Google Scholar]

[CD001703-bib-0123] Ivanova JI, Birnbaum HG, Schiller M, Kantor E, Johnstone BM, Swindle RW. Real-world practice patterns, health-care utilization, and costs in patients with low back pain: the long road to guideline-concordant care. Spine Journal 2011;11(7):622-32. [DOI: 10.1016/j.spinee.2011.03.017] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0124] James RJ, Ferguson E. Depression, cognition, and pain: exploring individual, cultural and country-level effects across Europe. Journal of Pain 2023;24(6):1104-15. [DOI: 10.1016/j.jpain.2023.03.006] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0125] Junqueira DR, Zorzela L, Golder S, Loke Y, Gagnier JJ, Julious SA, et al. CONSORT Harms 2022 statement, explanation, and elaboration: updated guideline for the reporting of harms in randomised trials. BMJ 2023;381:e073725. [DOI: 10.1136/bmj-2022-073725] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0126] Karran EL, Cashin AG, Barker T, Boyd MA, Chiarotto A, Mohabir V, et al. Development of recommendations for a minimum dataset for Identifying Social factors that Stratify Health Opportunities and Outcomes (ISSHOOs) in pain research. MethodsX 2024;12:102496. [DOI: 10.1016/j.mex.2023.102496] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0127] Knezevic NN, Candido KD, Vlaeyen JW, Van Zundert J, Cohen SP. Low back pain. Lancet 2021;398(10294):78-92. [DOI: 10.1016/S0140-6736(21)00733-9] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0128] Koes BW, Van Tulder MW, Thomas, S. Diagnosis and treatment of low back pain. BMJ 2006;332(7555):1430-4. [DOI: 10.1136/bmj.332.7555.1430] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0129] Koes BW, Backes D, Bindels PJ. Pharmacotherapy for chronic non-specific low back pain: current and future options. Expert Opinion on Pharmacotherapy 2018;19(6):537-45. [DOI: 10.1080/14656566.2018.1454430] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0130] Kongsted A, Kent P, Hestbaek L, Vach W. Patients with low back pain had distinct clinical course patterns that were typically neither complete recovery nor constant pain. A latent class analysis of longitudinal data. Spine Journal 2015;15(5):885-94. [DOI: 10.1016/j.spinee.2015.02.012] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0131] Konstantinou K, Hider SL, Jordan JL, Lewis M, Dunn KM, Hay EM. The impact of low back-related leg pain on outcomes as compared with low back pain alone: a systematic review of the literature. Clinical Journal of Pain 2013;29(7):644-54. [DOI: 10.1097/AJP.0b013e31826f9a52] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0132] Kremer M, Yalcin I, Goumon Y, Wurtz X, Nexon L, Daniel D, et al. A dual noradrenergic mechanism for the relief of neuropathic allodynia by the antidepressant drugs duloxetine and amitriptyline. Journal of Neuroscience 2018;38(46):9934-54. [DOI: 10.1523/JNEUROSCI.1004-18.2018] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0133] Lachin JM. Fallacies of last observation carried forward analyses. Clinical Trials 2016;13(2):161-8. [DOI: 10.1177/1740774515602688] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0134] Mafi JN, McCarthy EP, Davis RB, Landon BE. Worsening trends in the management and treatment of back pain. JAMA Internal Medicine 2013;173(17):1573-81. [DOI: 10.1001/jamainternmed.2013.8992] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0135] Mathieson S, Valenti L, Maher CG, Britt H, Li Q, McLachlan AJ, et al. Worsening trends in analgesics recommended for spinal pain in primary care. European Spine Journal 2018;27(5):1136-45. [DOI: 10.1007/s00586-017-5178-4] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0136] Mayo-Wilson E, Qureshi R, Li T. Conducting separate reviews of benefits and harms could improve systematic reviews and meta-analyses. Systematic Reviews 2023;12(67):1-5. [DOI: 10.1186/s13643-023-02234-0] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0137] McCleane G. Antidepressants as analgesics. CNS Drugs 2008;22:139-56. [DOI: 10.2165/00023210-200822020-00005] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0138] Micó JA, Ardid D, Berrocoso E, Eschalier A. Antidepressants and pain. Trends in Pharmacological Sciences 2006;27(7):348-54. [DOI: 10.1016/j.tips.2006.05.004] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0139] Mika J, Zychowska M, Makuch W, Rojewska E, Przewlocka B. Neuronal and immunological basis of action of antidepressants in chronic pain–clinical and experimental studies. Pharmacological Reports 2013;65(6):1611-21. [DOI: 10.1016/S1734-1140(13)71522-6] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0140] Moher D, Hopewell S, Schulz KF, Montori V, Gøtzsche PC, Devereaux PJ, et al. CONSORT 2010 explanation and elaboration: updated guidelines for reporting parallel group randomised trials. BMJ 2010;340:c869. [DOI: 10.1136/bmj.c869] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0141] Moore RA, Straube S, Eccleston C, Derry S, Aldington D, Wiffen P, et al. Estimate at your peril: imputation methods for patient withdrawal can bias efficacy outcomes in chronic pain trials using responder analyses. Pain 2012;153(2):265-8. [DOI: 10.1016/j.pain.2011.10.004] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0142] National Institute for Health and Care Excellence (NICE). Low back pain and sciatica in over 16s: assessment and management; 30 November 2016. Available at https://www.nice.org.uk/guidance/ng59. [PubMed]

[CD001703-bib-0143] National Institute for Health and Care Excellence (NICE). Depression in adults: treatment and management; 29 June 2022. Available at https://www.nice.org.uk/guidance/ng222. [PubMed]

[CD001703-bib-0144] O'Neill J, Tabish H, Welch V, Petticrew M, Pottie K, Clarke M, et al. Applying an equity lens to interventions: using PROGRESS ensures consideration of socially stratifying factors to illuminate inequities in health. Journal of Clinical Epidemiology 2014;67(1):56-64. [DOI: 10.1016/j.jclinepi.2013.08.005] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0145] Oliveira CB, Maher CG, Pinto RZ, Traeger AC, Lin CW, Chenot JF, et al. Clinical practice guidelines for the management of non-specific low back pain in primary care: an updated overview. European Spine Journal 2018;27(11):2791-803. [DOI: 10.1007/s00586-018-5673-2] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0146] Orhurhu V, Olusunmade M, Akinola Y, Urits I, Orhurhu MS, Viswanath O, et al. Depression trends in patients with chronic pain: an analysis of the nationwide inpatient sample. Pain Physician 2019;22:E487-94. [PubMed] [Google Scholar]

[CD001703-bib-0147] Ostelo RW, Deyo RA, Stratford P, Waddell G, Croft P, Von Korff M, et al. Interpreting change scores for pain and functional status in low back pain: towards international consensus regarding minimal important change. Spine 2008;33(1):90-4. [DOI: 10.1097/BRS.0B013E31815E3A10] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0148] Page MJ, Moher D, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. PRISMA 2020 explanation and elaboration: updated guidance and exemplars for reporting systematic reviews. BMJ 2021;372:n160. [DOI: 10.1136/bmj.n71] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0149] Page MJ, Higgins JP, Sterne JA. Chapter 13: Assessing risk of bias due to missing results in a synthesis. In: Higgins JP, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al, editor(s). Cochrane Handbook for Systematic Reviews of Interventions Version 6.4 (updated August 2023). Cochrane, 2023. Available from https://training.cochrane.org/handbook/archive/v6.4.

[CD001703-bib-0150] Paoli F, Dacourt G, Corsa P. Preliminary note on the action of imipramine in painful conditions [Note préliminaire sur l’action de l’imipramine dans les états douloureux]. Revue Neurologique 1960;2:503-4. [PubMed] [Google Scholar]

[CD001703-bib-0151] Peryer G, Golder S, Junqueira D, Vohra S, Loke YK. Chapter 19: Adverse effects. In: Higgins JP, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al, editor(s). Cochrane Handbook for Systematic Reviews of Interventions Version 6.4 (updated August 2023). Cochrane, 2023. Available from https://training.cochrane.org/handbook/archive/v6.4.

[CD001703-bib-0152] Phillips R, Cro S, Wheeler G, Bond S, Morris TP, Creanor S, et al. Visualising harms in publications of randomised controlled trials: consensus and recommendations. BMJ 2022;377:e068983. [DOI: 10.1136/bmj-2021-068983] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0153] Price MR, Mead KE, Cowell DM, Troutner AM, Barton TE, Walters SA, et al. Medication recommendations for treatment of lumbosacral radiculopathy: a systematic review of clinical practice guidelines. PM&R 2024;16(10):1128-42. [DOI: 10.1002/pmrj.13142] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0154] Qaseem A, Wilt TJ, McLean RM, Forciea MA. Noninvasive treatments for acute, subacute, and chronic low back pain: a clinical practice guideline from the American College of Physicians. Annals of Internal Medicine 2017;166(7):514-30. [DOI: 10.7326/M16-2367] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0155] Qureshi R, Mayo-Wilson E, Li T. Harms in systematic reviews paper 1: an introduction to research on harms. Journal of Clinical Epidemiology 2022;143:186-96. [DOI: 10.1016/j.jclinepi.2021.10.023] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0156] Rayner L, Hotopf M, Petkova H, Matcham F, Simpson A, McCracken LM. Depression in patients with chronic pain attending a specialised pain treatment centre: prevalence and impact on health care costs. Pain 2016;157(7):1472-9. [DOI: 10.1097/j.pain.0000000000000542] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0157] Review Manager (RevMan). Version 8.11.0. The Cochrane Collaboration, 2024. Available at https://revman.cochrane.org.

[CD001703-bib-0158] Riley RD, Lambert PC, Abo-Zaid G. Meta-analysis of individual participant data: rationale, conduct, and reporting. BMJ 2010;340:c221. [DOI: 10.1136/bmj.c221] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0159] Riley RD, Higgins JP, Deeks JJ. Interpretation of random effects meta-analyses. BMJ 2011;342(7804):964-7. [DOI: 10.1136/bmj.d549] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0160] Santesso N, Glenton C, Dahm P, Garner P, Akl EA, Alper B, et al. GRADE guidelines 26: informative statements to communicate the findings of systematic reviews of interventions. Journal of Clinical Epidemiology 2020;119:126-35. [DOI: 10.1016/j.jclinepi.2019.10.014] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0161] Schmid AB, Tampin B, Baron R, Finnerup NB, Hansson P, Hietaharju A, et al. Recommendations for terminology and the identification of neuropathic pain in people with spine-related leg pain. Outcomes from the NeuPSIG working group. Pain 2023;164(8):1693-704. [DOI: 10.1097/j.pain.0000000000002919] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0162] Schünemann HJ, Higgins JP, Vist GE, Glasziou P, Akl EA, Skoetz N, et al. Chapter 14: Completing ‘Summary of findings’ tables and grading the certainty of the evidence. In: Higgins JP, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al, editor(s). Cochrane Handbook for Systematic Reviews of Interventions Version 6.4 (updated August 2023). Cochrane, 2023. Available from https://training.cochrane.org/handbook/archive/v6.4.

[CD001703-bib-0163] Sterne JA, Savović J, Page MJ, Elbers RG, Blencowe NS, Boutron I, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ 2019;366:l4898. [DOI: 10.1136/bmj.l4898] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0164] Urits I, Peck J, Orhurhu MS, Wolf J, Patel R, Orhurhu V, et al. Off-label antidepressant use for treatment and management of chronic pain: evolving understanding and comprehensive review. Current Pain and Headache Reports 2019;23(66):1-10. [DOI: 10.1007/s11916-019-0803-z] [DOI] [PubMed] [Google Scholar]

[CD001703-bib-0165] Vos T, Lim SS, Abbafati C, Abbas KM, Abbasi M, Abbasifard M, et al. Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet 2020;396(10258):1204-22. [DOI: 10.1016/S0140-6736(20)30925-9] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0166] World Health Organization Collaborating Centre for Drug Statistics Methodology. ATC/DDD Index 2022. https://www.whocc.no/atc_ddd_index/ (accessed 10 February 2022).

[CD001703-bib-0167] Urquhart D, Hoving JL, Assendelft WJJ, Roland M, Tulder MW. Antidepressants for non-specific low back pain. Cochrane Database of Systematic Reviews 2008, Issue 1. Art. No: CD001703. [DOI: 10.1002/14651858.CD001703.pub3] [DOI] [PMC free article] [PubMed] [Google Scholar]

[CD001703-bib-0168] Van Tulder MW, Hienkens E, Roland M, Assendelft WJ. Antidepressants for non‐specific low‐back pain. Cochrane Database of Systematic Reviews 1999, Issue 7. Art. No: CD001703. [DOI: 10.1002/14651858.CD001703] [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Antidepressants for low back pain and spine‐related leg pain

Michael C Ferraro

Donna M Urquhart

Giovanni E Ferreira

Michael A Wewege

Christina Abdel Shaheed

Adrian C Traeger

Jan L Hoving

Eric J Visser

James H McAuley

Aidan G Cashin

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Summary of findings 1. Summary of findings table ‐ Serotonin and norepinephrine reuptake inhibitors compared to placebo for adults with non‐specific low back pain and spine‐related leg pain.

Summary of findings 2. Summary of findings table ‐ Selective serotonin reuptake inhibitors compared to placebo for adults with non‐specific low back pain and spine‐related leg pain.

Summary of findings 3. Summary of findings table ‐ Tricyclic antidepressants compared to placebo for adults with non‐specific low back pain and spine‐related leg pain.

Summary of findings 4. Summary of findings table ‐ Tetracyclic antidepressants compared to placebo for adults with non‐specific low back pain and spine‐related leg pain.

Summary of findings 5. Summary of findings table ‐ Other antidepressants compared to placebo for adults with non‐specific low back pain and spine‐related leg pain.

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Summary of findings and assessment of the certainty of the evidence

Results

Description of studies

Results of the search

1.

Included studies

Study design

Country of origin and number of centres

Study funding and author declarations of interest

Participants

Interventions and comparators

Outcomes

Primary outcomes

Secondary outcomes

Excluded studies

Risk of bias in included studies

2.

Allocation

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Effects of interventions

Non‐specific low back pain