Preoperative opioid use and postoperative return to work following spinal surgery in workers’ compensation settings: a systematic review and meta-analysis

Abstract

Background:

Opioid use prior to spinal surgery is common among patients with workers’ compensation (WC) claims. Extended opioid use for pain management in this population is associated with several adverse outcomes including delayed return to work (RTW).

Objective:

This systematic review and meta-analysis aim to assess the evidence on the association of preoperative opioid use with stable RTW and RTW within 1-year after spinal surgery.

Material and methods:

The authors searched MEDLINE, Embase, PsycINFO, Emcare, CINAHL Plus, Scopus, and Web of Science from inception to 14 January 2023. The authors included studies that compared any preoperative opioid use with no opioid use, and those that enabled a comparison of different durations of preoperative opioid use. The primary outcome was stable RTW after spinal surgery. Secondary outcomes were RTW within 1-year after surgery and cost of WC claims. A random effect model was assumed to pool the effect estimate. The GRADE approach was applied to evaluate the certainty of evidence.

Results:

From 2589 records, 10 studies were included, and of these, nine were considered for quantitative synthesis. All studies were observational with eight retrospective cohort and two case–control studies. Five studies each investigated cervical and lumbar disorders. With moderate certainty evidence, the odds of postoperative stable RTW reduced by half (OR: 0.51, 95% CI: 0.43–0.59; 5549 participants) in patients using opioids preoperatively. Similarly, moderate certainty evidence from 2348 participants demonstrated that the odds of RTW within 1-year after surgery were reduced by more than half in patients with preoperative opioid prescriptions (OR: 0.46, 95% CI: 0.36–0.59).

Conclusions:

This systematic review and meta-analysis shows that preoperative opioid use is associated with a reduction in odds of postoperative RTW by half in patients with WC-funded spinal surgery.

Introduction

Highlights

• Opioid use before spinal surgery in patients funded by workers’ compensation schemes is common.

• This review summarised the association of preoperative opioid use and return to work following spinal surgery from 10 observational studies of patients with workers’ compensation claims.

• Moderate certainty evidence showed the odds of postoperative return to work is reduced by half in patients with a history of preoperative opioid use.

Spinal surgeries such as fusion, decompression, and discectomy are one of the common treatment approaches for patients with back injuries or spinal disorders such as degenerative disc disease (DDD), herniated discs, and spinal stenosis. The aim of these surgeries is to alleviate pain, correct deformity, and improve functional abilities of patients when conservative treatment fails^1–3. Despite the wider use of spinal surgeries as an intervention in patients with cervical and back-related disorders, the effect of these surgeries on postoperative outcomes is debated^4–6. Some studies have suggested that spinal surgery can improve pain, quality of life, and postoperative functional outcomes in selected patient populations^1,7, while others suggest surgery does not offer significant benefit and are associated with poor postoperative outcomes such as prolonged disability, delayed return to work (RTW), and work loss^4,8. Complicating interpretation of these studies, there have been certain preoperative characteristics predictive of poor outcomes such as poor health status and extended opioid use preoperatively⁹.

Returning to work is an essential aspect of recovery for most patients, as it enables them to resume their daily activities, function independently, and regain their financial independence. The growing body of evidence shows that patients with workers’ compensation (WC) claims have worse postoperative outcomes including RTW when compared with those not compensated^10,11. This distinct population may have various predisposing factors for poor postoperative outcomes. Extended preoperative opioid use was one of the factors associated with worse outcomes in patients with spinal surgeries^9,12. Previous studies have shown that WC patients were more likely to have prolonged opioid use^13,14. Prolonged use of opioids before surgery can worsen postoperative outcomes including delayed recovery, extended work disability, and increased healthcare utilisation and costs^9,15.

This systematic review and meta-analysis aims to evaluate the evidence on the association of preoperative opioid use and postoperative work outcomes after spinal surgery in patients funded by WC schemes. The review further aims to answer two specific research questions:

1. does preoperative opioid use affect both the short-term and long-term RTW status after surgery?

2. does the impact on RTW differ by the duration of preoperative opioid use and location of spinal (lumbar or cervical) surgery?

Therefore, this study synthesises the available evidence and quantifies the effect of preoperative opioid use on RTW outcomes in WC population who underwent spinal surgeries.

Material and methods

This systematic review and meta-analysis reported in line with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses)¹⁶, (Supplemental Digital Content 2, http://links.lww.com/JS9/B649) and AMSTAR-2 (Assessing the methodological quality of systematic reviews) guidelines (AMSTAR-2 checklists, Supplemental Digital Content 3, http://links.lww.com/JS9/B650)¹⁷. The protocol was prospectively registered in PROSPERO (International Prospective Register of Systematic Reviews) with a registration number: CRD42023391209.

Eligibility criteria

Studies were considered if they assessed preoperative opioid use and its association with postoperative work outcomes in patients (i) with back disorders (such as cervical/lumbar DDD, radiculopathy, spondylolisthesis, disc herniation, lumbar sprain or stenosis), (ii) who underwent spinal surgeries (cervical/lumbar fusion, decompression, or discectomy), and (iii) funded by WC schemes. Studies were eligible if they were randomised control trials, quasi-randomised trials, cohort (prospective or retrospective), case–control or cross-sectional studies. We included studies that compared any preoperative opioid use with no opioid use, as well as those that enabled a comparison of different durations of preoperative opioid use (e.g. less than 3 months compared with 3 months or more). Literature were not restricted by year, language, or country of study. Studies that did not measure stable RTW or RTW within 1-year after surgery were excluded. Qualitative studies, case-series, case-reports, editorials, conference proceedings and meeting abstracts, commentaries, expert opinions and studies assessed as poor quality were not eligible for inclusion.

Information sources and search strategy

We searched MEDLINE, Embase, PsycINFO, and Emcare via the OVID platform, CINAHL Plus, Scopus and Web of Science from inception to 14 January 2023. Appropriate search terms including keywords and Medical Subject Headings (MeSH) related to preoperative opioid use and postoperative work outcomes following spinal surgery were developed using PICO (Population, Intervention, Context, and Outcome) structure (Supplementary file search strategy Supplemental Digital Content 3, http://links.lww.com/JS9/B650, Supplemental Digital Content 4, http://links.lww.com/JS9/B651). Hand searching of the reference lists of retrieved articles was conducted to identify potential studies that may have been missed in initial searches.

Selection process

All identified citations from each database were imported to Endnote 9 referencing software²¹ to identify and remove duplicates. After duplicate removal, articles were imported to the Covidence platform²² for screening. Two reviewers (Y.G.T. and S.T.) independently screened records by title and abstract for full-text retrieval. Retrieved articles were then assessed for eligibility independently by two reviewers (Y.G.T. and S.T.). Any disagreements between reviewers were resolved by discussion or the involvement of the other authors (A.C., S.G., and S.N.).

Data extraction

Using a data extraction form developed by the authors, two reviewers (Y.G.T. and S.T.) independently extracted the following information: First author last name, year, country, study title, objective, data sources, study design, data analysis method, sample size, sex/age distribution, nature and location of injury/disease, type of surgery, exposure (preoperative opioid information such as duration of use, oral morphine equivalent daily dose), outcome (stable RTW and RTW within 1-year after surgery), effect estimate with 95% CI and statistically adjusted covariates and potential confounders in the multivariable regression models. Discrepancies between the two reviewers were resolved by consensus.

Exposure definitions

Exposure to opioids was considered as either exposure (opioid use) versus no exposure (no opioid use) or ≥3 months use versus <3 months use and >1-year use versus ≤1-year use based on how the studies defined the two groups. The exposure and comparator definitions have been modified from the prospectively registered protocol to include studies that compared prolonged opioid use for >1-year and ≤1-year and to also consider the ±3-month window.

Outcome definitions

The primary outcome was stable RTW, defined as returning to work within 2 years and maintaining work function for 3 months or longer in the following year. This outcome aimed to evaluate sustained work function after surgery to ensure consistencies in outcome measurement, and focus on longer-term functional outcomes postsurgery. Secondary outcomes were RTW within 1-year after surgery, and cost of WC claims. RTW within 1-year outcome was included to assess if there was a difference between shorter-term and longer-term work outcomes after surgery. Cost of claims was included as it is an important metric utilised by WC schemes to monitor claim outcomes and WC scheme performance.

Study risk of bias assessment and certainty of evidence

The methodological quality of each study was independently assessed by two reviewers (Y.G.T. and S.T.) using Joanna Briggs Institute (JBI) quality appraisal tools specific for cohort and case–control studies²³. Disagreement was resolved by consensus. We classified studies as low risk of bias (RoB)/high quality if no or only one item on a quality check-list was assessed as ‘Unclear’ or ‘No’, moderate RoB/quality if two or three items on a check-list were assessed as ‘Unclear’ or ‘No’ and high RoB/low quality if four or more items on a check-list were assessed as ‘Unclear’ or ‘No’. We excluded studies that were assessed as having high RoB/low quality.

We evaluated the certainty of evidence using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework²⁴. Certainty of evidence was assessed for potential downgrading if a serious flaw was observed in the domains of limitations in study design, inconsistency, imprecision, or indirectness of the evidence. We aimed to assess publication bias (using funnel plot techniques and Egger’s regression test) for potential downgrading. However, publication bias was not assessed as the number of included studies was inadequate and below the recommended threshold to detect publication bias in these methods²⁵. We assessed studies for large effect sizes, by converting reported odds ratios to relative risk²⁶. Plausible residual confounding and dose-response relationships were also considered for potential upgrading of the evidence²⁶.

Data synthesis and meta-analysis

A narrative summary and characteristics of included studies were developed by the first author. Individual study information including methodology, participant characteristics, and main findings were summarised and are provided in Table 1. We compared the stable RTW and RTW within 1-year outcomes by opioid exposure duration where the studies allowed comparison. We employed the inverse variance statistical technique and random effect modelling to pool the effect estimate of individual studies. Heterogeneity among the studies was assessed using the I ² statistic. RevMan version 5.4.1 software³⁴ was used to conduct the meta-analysis and generate the forest plots of individual studies effect size and the pooled estimate. Subgroup comparisons were made by bodily location of surgery and duration of opioid exposure/comparators to determine whether these factors affected the relationship between preoperative opioid use and work outcomes. Sensitivity analyses were conducted to assess the robustness of synthesised effect estimates by excluding studies with larger sample sizes, to examine whether these studies may have dominated the overall pooled estimate.

Table 1.

Preoperative opioid use and postoperative stable RTW outcomes among patients underwent spinal surgery in WC setting.

References	Method (design, data sources, and analysis)	Major back disorder, surgery type	Preoperative opioid exposure status	Sample characteristics (sample size, sex, age)	Effect estimate on stable RTW: OR (95% CI)	Findings
Anderson et al.27	• Case–control study • WC claims between 1993 and 2013 • Multivariable logistic regression	• Degenerative disk disease (DDD) and discogenic LBP • Lumbar fusion • Concurrent decompression (n=660)	Exposure Opioids > 1-year Comparator No opioids or opioids ≤ 1-year	N=1037 F= 53 (34.04%) Age >50 before index fusion= 259 (24.98%)	0.46 (0.31–0.69)	Prolonged opioid use before lumbar fusion negatively affects the likelihood of RTW after fusion
Anderson et al.28	• Case–control study • WC claims between 1993 and 2013 • Multivariable logistic regression	• Spondylolisthesis • Lumbar fusion • Concurrent decompression (n=514)	Exposure Opioids > 1-year Comparator No opioids or opioids ≤ 1-year	N=686 F= 209 (30.47%) Age >50 before index fusion= 186 (27.11%)	0.41 (0.21-0.80)	Prolonged preoperative opioid use before lumbar fusion negatively affect RTW after fusion
Faour et al.29	• Retrospective cohort study • WC claims between 1993 and 2011 • Multivariable logistic regression	• Cervical radiculopathy and DDD • Cervical fusion (single level) • Concurrent decompression=NR	Exposure Opioids Comparator No opioids	N=3170 F=NR Age >50 before index fusion=NR	0.42 (0.34–0.52)	Preoperative opioid use negatively affected RTW status of patients underwent cervical fusion for DDD and radiculopathy
		• Cervical radiculopathy • Cervical fusion (single level) • Concurrent decompression=NR	Exposure Opioids, n=305 Comparator No opioids, n=1622	N=1927 F= 662(34.3%) Age >50 before fusion=230 (11.9%)	0.45 (0.35–0.58)
		• Cervical DDD • Cervical fusion (single level) • Concurrent decompression=NR	Exposure Opioids, n=77 Comparator No opioids, n=204	N=281 F= 121 (43.1%) Age >50 before fusion=58 (20.6%)	0.40 (0.22–0.73)
Faour et al.30	• Retrospective cohort study • WC claims between 1993 and 2011 • Multivariable logistic regression	• Cervical radiculopathy • Cervical fusion (single level) • Concurrent decompression=NR	Exposure Opioids, n=305 Comparator No opioids, n=1622	N=1927 F= 662 (34.3%) Age >50 before fusion=230 (11.9%)	0.5 (0.38–0.65)	Preoperative opioid use is associated with worse functional outcomes, mainly RTW status, after single level cervical fusion for radiculopathy
Faour et al.31	• Retrospective cohort study • WC claims between 1993 and 2011 • Multivariable logistic regression	• Cervical DDD • Cervical fusion (single level) • Concurrent decompression=NR	Exposure Opioids (OMEDD 60.75 ±75.8 mg) n=77 Comparator No opioids, n=204	N=281 F= 121 (43.1%) Age >50 before fusion=58 (20.6%)	0.46 (0.25–0.85)	The preoperative use of opioids for management of discogenic neck pain is a negative predictor of successful after fusion
Faour et al.18	• Retrospective cohort study • WC claims between 1993 and 2011 • Multivariable logistic regression	• Cervical radiculopathy • Cervical fusion (single level) • Concurrent decompression=NR	Exposure Opioids ≥3 months (n=172) Comparator Opioids <3 months (STO) n=133	N=305 F=NR Age >50 before index fusion=NR	0.43 (0.27–0.68)	Prolonged preoperative opioid use was associated with poor RTW outcomes after cervical fusion
			Exposure Opioids ITO (3–6 months) n=51 OMEDD=61.3 mg Comparator Opioids < 3 months (STO) OMEDD=74.1 mg n=133		0.49 (0.25–0.94)
			Exposure Opioids LTO (>6 months) (n=121) OMEDD=56.6 mg Comparator Opioids < 3 months (STO) OMEDD=74.1 mg n=133		0.40 (0.24–0.68)
Faour et al.32	• Retrospective cohort study • WC claims between 1993 and 2011 • Multivariable logistic regression	• Cervical radiculopathy and DDD • Cervical fusion (multilevel) • Concurrent decompression=NR	Exposure Opioids Comparator No opioids	N=2133 F=NR Age >50 before index fusion=NR	0.60 (0.48–0.75)	Stable RTW status within 3 years after multilevel cervical fusion was negatively affected by preoperative opioid use
		• Cervical radiculopathy • Cervical fusion (multilevel) • Concurrent decompression=NR	Exposure Opioids, n=212 Comparator No opioids, n=856	N=1068 F=335 (33.2%) Age >55 before fusion=82 (7.7%)	0.56 (0.40–0.78)
		• Cervical DDD • Cervical fusion (multilevel) • Concurrent decompression=NR	Exposure Opioids, n=115 Comparator No opioids, n=326	N=441 F=169 (38.2%) Age >55 before fusion=54 (12.2%)	0.49 (0.30–0.81)
McMillan et al.33	• Retrospective cohort study • WC claims between 2008 and 2016 • Multivariate multinomial logistic regression	• Chronic low back pain • Lumbar fusion (single + multilevel) • Concurrent decompression-NR	Exposure Opioids n=463 Comparator No opioids n=411	N=874 F=272(31.12%) Mean age was 45.6±9.6 years	*3.08 (1.50–6.32) on no work capacity High dose-OMEDD (>40 mg) 0.32 (0.16–0.65) (OR converted to predict work capacity)	Preoperative opioid prescriptions are one of the predictors of reduced work capacity after lumbar fusion
O’Donnell et al.19	• Retrospective cohort study • WC claims between 2005 and 2012 • ANOVA with Post-hoc and Multivariable logistic regression	• Lumbar disk herniation with Radiculopathy • Lumbar discectomy (single level)	Exposure Opioids • STO (less than 14 days) n=126 • MTO (14–90 days) n=315 • LTO (>90 days) n=279 Comparator No opioids, n=566	N=1286 F=309 (24.03%) Mean age in yrs. Preoperative opioids STO=40.0 ±10.9, MTO=39.4±9.7 and LTO=39.1±9.3 no opioids 40.9±10.2	0.54(0.39–0.75)	Preoperative opioid use was a negative predictor of RTW rates after LD in WC patients. No significant difference in RTW rates between no preoperative opioids and STO group as well as between the STO and MTO groups
Tye et al.20	• Retrospective cohort study • WC claims between 1993 and 2013 • Multivariable logistic regression	• Degenerative lumbar stenosis • Lumbar decompression (single + multilevel)	Exposure Opioids ≥3 months n=80 Comparator Opioids <3 months n=60	N=140 F= 41 (29.29%) Age at fusion Opioids ≥3 months 47.2±10.7 yrs. Opioids <3 months 52.1±8.8 yrs.	0.35 (0.13–0.89)	Prolonged preoperative opioid use was associated with poor RTW outcomes after lumbar decompression

F, female; ITO, Intermediate-term opioid use; LD, lumbar discectomy; LTO, long-term opioid use; N, sample size; NR, not reported; OMEDD, oral morphine equivalent daily dose; OR, odds ratio; RTW, return to work; STO, short-term opioid use; WC, workers compensation.

^aOR is converted to odds of substantial working capacity (less than 10 working days wage replacement in the 3-month period and certification of full or partial capacity) to have relatively similar outcome measure for the meta-analysis.

We transformed the reported outcome measures in the individual studies to uniform outcome measures to enable meta-analysis. For one study, which reported odds of reduced work capacity, we converted the outcome to represent the odds of work capacity, to ensure the direction of effect was consistent with the remaining studies³³. We removed the potential sample duplicates in the meta-analysis by excluding studies where their samples overlapped in other studies as a whole or in part^18,29. To avoid double counting of participants, as most were drawn from the same source population, only six effect estimates from five unique studies were pooled together for the meta-analyses of the primary outcome. Furthermore, three studies were not included in the overall pooled estimate due to the difference in the duration of exposure or comparator^20,27,28. Subgroup analysis occurred where studies had uniformity in their exposure and comparator. One study compared preoperative opioids between ≥3 months versus <3 months duration while the comparator in the other studies was no opioid prescriptions²⁰. Thus, this study was removed from the meta-analysis. We tested the impact of this removal and the pooled estimate did not change meaningfully either in subgroup or sensitivity analyses. One study was added twice in the meta-analysis since it included two unique samples with different effect estimates for each (patients with cervical radiculopathy and DDD who underwent multilevel cervical fusion) in the same article³².

Result

Study selection

The literature search identified 2589 records, of which 724 were duplicates and removed before screening. From the 1865 records screened by title and abstract, 1750 were excluded. The full-text for 115 potentially relevant studies were retrieved and assessed for eligibility. Finally, 10 studies were included for review. Hand searching of reference lists did not result in any additional records (Fig. 1).

Figure 1.

PRISMA flow diagram of the article screening and selection process.

Study characteristics

The selected study characteristics and findings on the relationship between preoperative opioid use and postoperative RTW outcomes in a WC setting are presented in Table 1. Nine of the studies were conducted in Ohio state, USA^{18–20,27–32} while the other study was from the state of Victoria, Australia³³. All studies used retrospective observational [retrospective cohort (n=8) and case–control (n=2)] study designs. No randomised control trials or prospective cohort studies were identified. All included studies used accepted WC claim data for participants with cervical and lumbar morbidities who underwent spinal surgeries. Postoperative work outcomes were measured utilising the WC wage replacement payments for work incapacity during the postsurgery period. Studies from the USA measured stable RTW as returning to work within 2 years and maintaining work status for 6 months in the next year^{18–20,27–32}. The Australian study measured RTW as having substantial work capacity, measured by having fewer than 10 days of wage replacement over a 3-month period and a certificate of full or partial capacity by a treating health practitioner between 24 and 27 months of postsurgery³³.

Five of the studies^18,29–32 included participants with cervical morbidities (cervical radiculopathy and DDD) who underwent cervical fusion (four single level fusion^18,29–31 and one multilevel fusion³²). Five of the included studies were on lower back morbidities with lumbar surgeries (three lumbar fusion^27,28,33, one lumbar discectomy¹⁹, and one lumbar decompression²⁰.

Risk of bias in studies

All of the articles were observational studies with low or moderate risk of bias. Eight of the 10 studies were rated as having low risk of bias^{18–20,29–33} while two were rated as moderate risk of bias^27,28. One study was excluded from the review as it was rated as high risk of bias³⁵.

Confounding and covariates

All of the included studies adjusted for covariates and potential confounders in their multivariable regression analyses (Supplementary Table 1 for adjusted covariates and confounders, Supplemental Digital Content 4, http://links.lww.com/JS9/B651). The most commonly corrected variables in the regression models were demographic characteristics (e.g. sex and older age), duration of time from injury to surgery, approach of the surgery (e.g. type of fusion, graft type, and instrumentation during surgery), presurgery health service use (imaging, chiropractic care, and physical therapy), presurgery psychiatric morbidities/mental health evaluations, out of work status, receiving disability benefits, and legal representation before the surgery^{18–20,27–33}.

Results of syntheses/summary of evidence

Primary outcome - Stable RTW

The odds of stable RTW were ≥50% lower with preoperative opioid exposure, and the result was consistent across different subgroups. The meta-analysis demonstrated minimal or no observed heterogeneity within included studies with I ²=0 and statistically significant overall pooled estimate across total or various subgroups with P-value <0.001.

Patients with versus without preoperative opioid prescriptions

Six studies assessed the association between preoperative opioids (when compared to no preoperative opioids) and postoperative stable RTW. All the studies showed opioid prescriptions before surgery were associated with lower odds of stable RTW after surgery. Meta-analysis showed that the pooled effect of preoperative opioid use reduced the odds of stable RTW by half (OR: 0.51, 95% CI: 0.43–0.59, six effect measures from five studies with 5549 participants) (Fig. 2). The evidence for this association was considered to be of moderate certainty, using the GRADE method. Specific to the subgroups by location of the spinal disorder and surgical approach, moderate certainty evidence showed preoperative opioids reduced odds of postoperative stable RTW by half in patients with cervical disorders who underwent fusion (OR: 0.51, 95% CI: 0.43–0.62, four effect measures from three studies, 3717 participants). Similarly, moderate certainty evidence showed odds of stable RTW reduced by more than half in patients with lumbar surgeries (OR:0.47, 95% CI: 0.34–0.64, three studies with 1972 participants) (Table 2). From the effect estimates, stable RTW was slightly lower in lumbar surgeries than cervical fusion subgroups. However, the subgroup differences were not statistically significant (P=0.64) (Fig. 3).

Figure 2.

Forest plot comparison on preoperative opioids (opioids versus no opioids, ≥3 months versus < 3 months and >1-year versus ≤1-year) association with postoperative stable RTW in patients with spinal surgeries.

Table 2.

Summary of findings and certainty of evidence for stable RTW and RTW within 1-year after surgery.

	Summary of findings	Odds ratio
	No of participants (No of studies)	95% CI	Certainty of evidence
Stable RTW
Spinal surgeries (both cervical and lumbar groups)
Preoperative opioid vs no opioids/<3 months	5689 (seven effect estimates from six studies)	0.50 (0.43–0.59)	Moderate
Preoperative opioid vs no opioids	5549 (six effect estimates from five studies)	0.51 (0.43–0.59)	Moderate
Preoperative opioid for ≥3 months vs <3 months	445 (two studies)	0.41 (0.27–0.63)	Moderate
Cervical fusion subgroup
Preoperative opioid vs no opioids	3717 (four studies)	0.51 (0.43–0.62)	Moderate
Lumbar surgeries (fusion, discectomy, and decompression) subgroup
Preoperative opioid Vs no opioids/<3 months	1972 (three studies)	0.47 (0.34–0.64)	Moderate
Preoperative opioid for >1-year vs ≤1-year	1723 (two studies)	0.45 (0.32–0.63)	Low
RTW within 1-year after surgery
Spinal surgeries (both cervical and lumbar groups)
Preoperative opioid Vs no opioids/<3 months	2348 (three studies)	0.46 (0.36–0.59)	Moderate
Preoperative opioid for ≥3 months vs <3 months	445 (two studies)	0.38 (0.25–0.57)	Moderate
Cervical fusion subgroup
Preoperative opioid vs no opioids	2208 (two studies)	0.48 (0.37–0.62)	Moderate

RTW, return to work.

Figure 3.

Forest plot of subgroup comparison (cervical and lumbar surgeries) between preoperative opioids Vs no/<3-month opioids on postoperative stable RTW in patients with spinal surgeries.

Pooled estimates including all studies with a comparator (nonexposure) of no or less than 3 months of preoperative opioid prescriptions were computed (OR:0.50, 95% CI: 0.43–0.59), seven effect measures from six studies and 5689 samples (Supplementary Figure 1, Supplemental Digital Content 4, http://links.lww.com/JS9/B651). No difference was observed between the two pooled estimates when the Tye et al.²⁰ study with a relatively small sample and small weighted proportion was added in the latter. For the sake of uniformity in the comparator, and as opioid duration in the exposure was not precisely reported for the majority of studies, this study was removed from the main analysis.

Patients with preoperative opioid prescriptions for ≥3 months versus <3 months

There was moderate certainty evidence from two studies and 445 participants of a lower odds of stable RTW in workers who received preoperative opioid prescriptions for ≥3 months than those with <3 months (OR: 0.41, 95% CI: 0.27–0.63) (Fig. 2) (Table 2).

Patients with preoperative opioid prescriptions for >1-year versus ≤1-year

Similarly, two studies examined whether postoperative stable RTW differed between participants with a preoperative opioid prescription for more than 1-year when compared to those with less than or equal to 1-year. Low certainty evidence from these studies showed receipt of preoperative opioids for more than 1-year was associated with reduced postoperative stable RTW. The odds of stable RTW was reduced by greater than half in the preoperative opioid prescription for more than 1-year group (OR: 0.45, 95% CI: 0.32–0.63, two studies with 1723 participants) (Table 2) (Fig. 2).

Secondary outcome

RTW within 1-year after surgery

Three of the included studies assessed the secondary outcome of this review, that is RTW within 1-year after surgery in compensated patients with spinal surgeries (Fig. 4).

Figure 4.

Forest plot of comparison on preoperative opioids (opioids versus no opioids and ≥3 months versus < 3 months) association with RTW within 1-year after spinal surgery.

Patients with versus without preoperative opioid prescriptions

Three studies with a total of 2348 participants provided moderate certainty evidence that the odds of RTW within 1-year after spinal (cervical and lumbar) surgery were reduced by more than half for participants with preoperative opioid prescription (OR: 0.46, 95% CI: 0.36–0.59) (Table 2). When the Tye et al.²⁰ study was removed from the meta-analysis, preoperative opioid prescriptions reduced odds of RTW within 1-year after surgery by more than half when compared with no preoperative opioid prescriptions (OR: 0.48, 95% CI: 0.37–0.62), two studies and 2208 participants (Fig. 4). The pooled effect size did not seem to change significantly and this could show the robustness of the effect estimate from the included studies.

Patients with preoperative opioid prescriptions for ≥3 months versus <3 months

Two studies including a total of 445 participants provided moderate certainty evidence that the odds of RTW within 1-year after surgery was lower in participants with preoperative opioid prescriptions for ≥3 months than those with <3 months (OR: 0.38, 95% CI: 0.25–0.57) (Fig. 4, Table 3).

Table 3.

Preoperative opioid use and medical cost in workers compensation setting.

References	Medical cost
Faour et al. a 18	Total medical costs per claim in groups with • Opioids for <3 months: USD 86 925±70 750, • Opioids for 3–6 months: USD 106 057±66 856 • Opioids for >6 months: USD 120 069±75 729$
O’Donnell et al.19	Total medical costs in groups with • Opioids >90 days: USD 64 635±64 236 • Opioids for 14–90 days: USD 51 192±54 480 • Opioids for 14 days: USD 49 418±49 168 • No opioid: USD 39 135±40 878
Tye et al.20	Total medical costs in groups with: • Opioids for <3 months is USD 77 592.7± 65 784.2. • Opioids for ≥3 months USD 148 571.1±109 843.1

Subgroups either by exposure duration or location of disorder/surgery showed the magnitude of association between preoperative opioids and adverse RTW outcomes were not significantly different from the total or other subgroups. However, worse RTW outcomes (stable RTW and RTW within 1-year after surgery) were observed in studies that compared preoperative opioid prescriptions of more than 1-year versus less than or equal to 1-year and more than 3 months versus less than 3 months opioid prescriptions.

Cost of claims

Three studies reported and compared the costs of medical expenditure by preoperative opioid use duration. These studies revealed that the cost of medical claims increased significantly with the duration of preoperative opioid prescriptions. None of the included studies reported total cost of claims including other benefits such as disability and income support payments. Patients with long-term preoperative opioid prescriptions had higher medical costs compared with those with short-term preoperative opioid use. Faour et al.¹⁸ reported that the medical cost was USD 120 069±75 729 and USD 86 925±70 750 in those with >6 month and <3-month opioid prescriptions, respectively. Similarly, the total medical cost reported in by O’Donnell 2018 et al. was USD 64, 635 and USD 51,192 for groups with >3 months and 2 weeks-3 months opioid prescriptions, respectively¹⁹. However, the studies only compared medical costs of claims by duration of opioid prescription and did not report the costs contributed due to opioids (Table 3).

Certainty of evidence

The certainty of evidence for stable RTW outcomes was moderate across subgroups of any preoperative opioid prescriptions versus no opioids, ≥3 months versus <3 months and surgical types/locations. However, the certainty of evidence for preoperative opioid prescriptions >1-year compared with ≤1-year was rated as low since the evidence was downgraded for RoB. This is due to the longer duration of nonexposure and likely differences in the baseline population, potentially as patients prescribed for >1-year may have more serious pain and be more impaired than participants in other studies. The certainty of evidence for RTW within 1-year after surgery was moderate in all subgroup analysis. The body of evidence in those work outcomes rated as moderate were upgraded by one level due to controlling for plausible residual confounding (adjusted relationship was lower than the unadjusted effect in the included studies). There was only one subgroup where an outcome was downgraded. The included studies have the acceptable precision, consistency, directness and no serious limitation in the study design for the rest of subgroups. Publication bias was not assessed for potential downgrading as the number of included studies were less than 10 considered the minimum to assesses publication bias via a funnel plot or more advanced regression-based assessments such as Egger’s test²⁵.

Discussion

This systematic review and meta-analyses assessed and summarised the evidence on the relationship between preoperative opioid use and postoperative RTW outcomes of spinal surgeries funded by WC systems. The review demonstrated that preoperative opioid use was associated with a nearly 50% reduction in the odds of stable RTW and RTW within 1-year after surgery. The review attempted to examine differences on short-term and long-term outcomes and whether the adverse work outcomes differ by duration of preoperative opioid prescriptions or location of spinal surgeries. Preoperative opioids were found to be consistently associated with reduced short-term and long-term work outcomes after surgery, regardless of the duration of use and location of the surgery. These findings add further evidence demonstrating that preoperative opioid use is a predictor for impaired postoperative work function in spinal surgeries.

Subgroup analysis showed RTW was consistently reduced across all subgroups of preoperative opioid prescriptions. Subgroup analysis found no association with bodily location and RTW outcomes. Similarly, subgroup analysis by duration of preoperative opioid use showed all subgroups were associated with decreased RTW outcomes. Worse RTW outcomes were observed with longer durations of preoperative opioids, that is in studies that compared more than 1-year versus less than or equal to 1-year preoperative opioid prescriptions and more than or equal to 3 months versus less than 3 months preoperative opioid prescriptions. This may suggest a dose response relationship between opioid prescriptions and RTW outcomes. The odds of RTW were lower in the groups with extended preoperative opioid prescriptions than those groups with shorter durations. However, from the given studies, it may not be possible to draw a clear conclusion of a dose response relationship between extended opioids and RTW outcomes since the effect size estimates were generated from various comparator durations ranging from no opioids, <3 months to ≤1-year. Stronger inferences of a dose response relationship could be made if the studies used uniform comparator definitions.

Three of the included studies also showed the overall medical costs for patients who underwent spinal surgeries was increased with increased preoperative opioid duration^18–20. These studies showed that spinal surgeries of patients with preoperative opioid use had a significantly higher medical cost. Tye et al.²⁰ showed patients who remained on preoperative opioid therapy longer than 3 months cost the WC scheme on average USD$70 979 more than patients with less than 3 months of preoperative opioid therapy. The studies did not differentiate between the contribution of opioids to the expenditure and other expenses, and only examined whether differences exist on the total medical expenditure among those received opioids (with different duration). So, the expenditure may not be entirely attributed to the cost of the opioids alone. A prior study has showed that opioids and surgeries are independent predictors of high claim cost but the cost of opioids and costs associated with the surgical procedure accounted for only a small proportion of the ultimate claim cost³⁶. Similarly, another study showed opioids are a major driver of medical costs while the cost of the medication itself is a small proportion of the overall medical cost³⁷. Thus, opioids may be considered a surrogate for unmeasured characteristics associated with greater medical cost, as those patients receiving opioids appear to have a different clinical progression and high volume of healthcare utilisation that results in higher medical costs.

It is widely stated in the literature that prolonged preoperative opioid use is generally associated with a higher risk of short-term and long-term adverse postoperative outcomes such as hyperalgesia, poor patient satisfaction, increased surgical site infection, longer hospital stay, higher readmission rate, and adverse drug events after surgery^15,38–40. This systematic review also shows the adverse impact of preoperative opioids extends to work outcomes in addition to these clinical outcomes. To the best of our knowledge and literature search, this is the first systematic review to evaluate and quantify the relationship of preoperative opioid use and postoperative work outcomes following spinal surgery in distinct WC population. This review expands the knowledge base and understanding of the impacts of preoperative opioid use on postoperative long-term recovery outcomes such as resuming work function and maintaining stable RTW which are one of the ultimate treatment goals.

Limitation of included studies and this review

This systematic review has several strengths with unique contribution to the literature by providing important evidence evaluation and summary on the relationship between preoperative opioid use and postoperative long-term recovery indicators, that is RTW following spinal surgery in distinct WC population. However, the review has limitations most notably with the inherent limitation of included studies. The exposure measurement and duration definitions had some variation across the studies despite most of the studies being conducted in the same setting. Furthermore, the opioid duration was unclear or inconsistent in the majority of studies (in five of the studies, duration of preoperative opioid exposure is not clear while their comparators are no opioid exposure), which may limit more objective comparison guided by the specific duration of use. O’Donnell et al.¹⁹ defined short-term, intermediate-term, and long-term opioid duration as less than 14 days, 14–90 days, and >90 days, respectively. However, in Faour et al.¹⁸ the opioid duration definition was reported as less than 3 months, between 3 and 6 months and >6 months were defined as short-term, intermediate-term, and long-term opioid duration, respectively. This might be due a lack of consistent definitions of opioid use durations in WC settings. However, in this systematic review, duration of opioid use was reclassified to uniform durations. Thus, the risk of misclassification on exposure duration is mitigated in this review.

The included articles adjusted for multiple covariates and potential confounding variables in their analyses. This would enable the pooled estimate on preoperative opioids’ association with RTW outcomes to be more reliable and reflect the true magnitude of association, with minimal confounding effect on the relationship. Moreover, the included studies had similar source populations (from a WC setting), reducing the heterogeneity of the study population, with the relative homogeneity of studies in this review suggesting summary effect size may be more robust. Although heterogeneity of studies is expected in meta-analysis⁴¹, a pooled estimate from a highly heterogeneous population is a limitation in many meta-analyses⁴²; however, caution may also be required to interpret effect estimates from a small number of relatively similar studies. Most studies from the US utilised a large dataset but similar data source for their sample recruitment. Studies from these similar settings may have less generalisability.

The review was limited to studies with a focus on distinct WC population setting as previous studies have shown high opioid use in this population. However, findings may vary in studies in the general population. We believe the evidence in the WC setting may reflect nations that have an established statutory WC scheme (such as USA, Canada, and Australia), and may not be generalisable to settings with different clinical contexts and population. Furthermore, the evidence largely arises from a single setting in the US, which could also raise issues on geographical representativeness. However, the studies included data from a large cohort of WC claimants across 8 to 20-year time frames, which provides a unique opportunity of a temporally representative sample. Moreover, this meta-analysis tried to scrutinise a unique set of participants to avoid sample duplicates for a pooled effect estimate. The pooled effect estimate did not show significant changes by subgroup and sensitivity analyses across the short-term and long-term work outcomes after spinal surgery. This further demonstrates the relative robustness of pooled results of opioids’ adverse association with work outcomes, along with the consistency and relative precision of effect size in individual studies.

Implications of the review

This systematic review and meta-analysis provide important insights into the relationship between opioid use and work outcomes in a WC setting. It highlights the adverse associations of preoperative opioid prescriptions with postoperative work outcomes in patients who underwent spinal surgeries, where the necessity of the procedure for most back disorders by itself is still controversial^4–6. The findings support the need to look for alternative pain management strategies as well as promoting a conservative approach to use opioids as the mainstay of pain management in this population. Further prospective studies might be helpful to generate robust evidence with more geographical representation and address the limitations identified in the existing literature.

Conclusions

Our systematic review and meta-analysis found that preoperative opioid use is associated with decreased odds of RTW after spinal surgery in WC-funded workers. Moderate certainty evidence showed the odds of postoperative stable RTW and RTW within 1-year after surgery is reduced by half in patients with a history of preoperative opioid use. However, the evidence is limited by the observational and retrospective nature of included studies. Future well-designed studies are needed to confirm these findings and identify the potential mechanisms underlying this association. In the meantime, efforts should be made to optimise preoperative pain management strategies and minimise opioid use in this population.

Ethical approval

Not applicable.

Consent

Not applicable.

Sources of funding

Yonas Tefera received Monash Graduate Scholarship and Healthy Working Lives Research Group supplemental scholarship for his PhD study. Professor Alex Collie is supported by an Australian Research Council Future Fellowship (Grant #FT190100218). Dr Shannon Gray is supported by an Australian Research Council Discovery Early Career Research Award (#DE220100456). Professor Suzanne Nielsen is supported by a National Health and Medical Research Council Career Development Fellowship (#1163961).

Funders have no involvement in any of the study process and submission of this manuscript.

Authors contribution

Y.G.T.: conceptualised the study, conduct search, screening, extraction, analysis, interpretation, and writing the first draft of the manuscript; A.C.: involved in conceptualisation, screening, interpretation, supervision, and critical revision of the manuscript; S.G. and S.N.: were involved in the screening, interpretation, supervision, and critical revision of the manuscript. All authors approved the final manuscript.

Conflicts of interest disclosure

There are no conflicts of interest.

Research registration unique identifying number (UIN)

ID=CRD42023391209. https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023391209.

Guarantor

Professor Alex Collie, Professor Suzanne Nielsen, Dr Shannon Gray, and Yonas Tefera.

Data availability statement

The data that supports the findings in this review are included in the tables and attached as supplementary files, Supplemental Digital Content 4, http://links.lww.com/JS9/B651.

Provenance and peer review

Not commissioned, externally peer-reviewed.