Factors predicting outcomes from chronic pain management interventions

Abstract

Chronic pain is the leading cause of years lost to disability worldwide, by a large margin, affecting 20-34% of the world's population. Chronic pain is the target for an increasing number of invasive and expensive treatments, supported by different levels of evidence. At a time when personalised medicine, driven in part by the growth of artificial intelligence, is surging, a scoping review on the factors that affect pain outcomes for procedural interventions is needed. A scoping review is important because placebo controlled trials for the most commonly used treatments consistently show small-to-moderate effect sizes of <0.5 that are often overshadowed by the placebo effect. In this article, personal characteristics, and social and clinical factors that influence surgical and non-surgical procedure pain and functional outcomes are reviewed, their intersectionality is briefly explored, and the evidence base for how dealing with these factors can influence outcomes is outlined.

Introduction

Chronic pain affects 20-34% of the global population.^{1 2} Pain covers the entire spectrum of medicine, with the number of peer reviewed articles published on the topic exceeding those of other specialties. These articles advocate hundreds of treatments for chronic pain, with a disproportionately higher percentage reporting positive results for the subjective numerical pain outcome than those for physical sciences.^{3 4} Despite these advances, the prevalence of chronic pain and pain related disability continues to rise, with pain being the leading cause of disability worldwide.^{5 6}

The number of publications that focus on technical aspects of interventional procedures included in this review, which can be defined as open surgical procedures and minimally invasive procedures that often involve needles or cannulas placed with image or landmark guidance, outnumber those relating to patient selection. This relative lack of research studies has ramifications that extend to access to care (lack of payor authorisation) and clinical practice guidelines. Efficacy studies, which use rigorous selection criteria, support most treatments for chronic pain, but overestimate the benefit in real world populations. Major funding and regulatory organisations, such as the European Medicines Agency, recommend responder analyses, which synchronise with precision medicine, to better assess benefits in specific populations.⁷ Better selection in clinical practice can improve the risk-benefit and cost-effectiveness ratios (ie, the success rates) of expensive and emerging procedures. Unfortunately, a recent rapid recommendation guideline on non-surgical procedures for chronic low back pain failed to consider the effects of patient selection, technique, and lack of viable alternatives.⁸

Heritability accounts for 25-50% of the risk of chronic pain, but can be difficult to distinguish from learnt behaviours (eg, coping skills and lifestyle choices).⁹ Hence the precise contribution of individual components (ie, psychological predisposition, reduced thresholds, and tolerance) cannot be determined at this time. In contrast, phenotyping, which adapts care to observable characteristics, is inexpensive and easy to perform; in fact, phenotyping is sometimes used to select patients for pain alleviating procedures (ie, weight loss or smoking cessation before surgery).^{10 11} Although conceptually appealing, phenotyping also has drawbacks, including the potential to undermine a shared decision model. Another confounding factor in the ability to identify and isolate the effect that specific variables have on outcomes is that many variables overlap and influence each other, a phenomenon known as intersectionality. In this review, our objectives are to outline the evidence behind phenotypic factors associated with outcomes from interventional pain treatments, including quality of life and effectiveness of preprocedural interventions, and to propose opportunities for future research.

Sources and selection criteria

Between October 2024 and March 2025, we searched databases on PubMed, Embase, Ovid, Google Scholar, and Google with key words relating to surgical and non-surgical procedural interventions evaluating outcomes for chronic pain. We prioritised high quality systematic reviews, but included all types of clinical studies, excluding animal research. Online supplemental appendix 1 has a more complete search strategy, including grading of evidence.

Factors affecting pain outcomes

With few exceptions, modifying variables is best conceptualised as a series of mutable points along a continuum. Factors such as mood and job satisfaction are dynamic, disease burden can be mitigated or exacerbated, non-organic signs change over time, and secondary gain(s) can be resolved or emerge anew. Although factors are generally classified as binary in the literature, recognising that most are continuous variables is important, with dose-response curves suggesting poorer outcomes for greater burden (eg, severe psychopathology and obesity have poorer outcomes than milder cases). Modifiable risk factors are those that can be eliminated or controlled for by lifestyle changes or (practical) interventions. In this study, we chose risk factors based predominantly on evidence from clinical studies on variables known to affect outcomes. Table 1 summarises these risk factors. Figure 1 shows how these risks might interact with one another.

Table 1. Risk factors for outcomes after interventional treatment for chronic pain.

Characteristic	Evidence base*	Comments
Age	Moderate evidence as protective factor for acute pain; weak evidence for a mixed-to-negative effect in chronic pain18	Physiological age and functional status, rather than chronological age, correlate better with outcomes. Physiological changes do not occur uniformly in people and can be affected by many other risk factors (eg, smoking or psychopathology)
Sex	Moderate evidence that female sex is a risk factor for chronic pain; no strong evidence for effect on procedural outcomes27 163	Higher risk of persistent postsurgical pain in women
Disease burden	Strong evidence that imaging findings do not strongly correlate with symptom severity. Moderate evidence for an inverse or equivocal relation for pain procedures (greater benefit if less pain or disability), but a direct relation for surgery (greater benefit if greater baseline pain or disability)77 164	The mental health component of disease burden usually has a greater influence on procedural outcomes than anatomical disease or lesion severity (strong evidence that depression is associated with worse outcomes)
Type of injury (eg,neuropathic,nociceptive, or nociplastic)	Variable evidence for procedural efficacy in neuropathic and nociceptive conditions, but strong evidence for poorer outcomes in nociplastic conditions	Outcomes highly dependent on appropriatepatient selection. Most non-surgical interventional literature is limited to short term follow-up periods (ie, 3-6 months)
Non-organic signs	Strong for a moderate effect. Direct correlation between number of signs and poor outcome	As an examination sign not readily explained by pathology (eg, sensory changes not explained by pathology, hyporeflexia), scant evidence exists for an independent effect55,57
Work characteristics	Weak for the physical demands of a job to have a small negative effect	Flexibility and trust in leadership might also increase return-to-work rate62
Compensation status	Strong for a consistently high effect (>2-fold)	Dynamic and responsive secondary gain (workers' compensation) might be more strongly associated with negative pain and functional outcomes69 75
Central sensitisation	Moderate as an independent variable, high for a small-to-moderate association	Central sensitisation might occur after an inciting event (eg, infection) or insidiously48 81 82 165
Interventions to modify preoperative pain and acute postoperative pain	Moderate evidence for decreased risk of persistent postsurgical pain in thoracotomy, breast surgery, and bone marrow harvest with regional anaesthesia. A meta-analysis showed that pregabalin might be helpful in reducing persistent postsurgical pain in cardiac surgery and knee arthroplasty166	Preoperative pain and acute postoperative pain can be minimised by regional anaesthesia, but limited, positive evidence exists for prevention of progression to persistent postsurgical pain
Psychopathology	Moderate-to-strong evidence for increased risk of persistent postsurgical pain with depression or anxiety. Moderate evidence for poorer response in procedures19 167 168	Commonly, psychiatric comorbidities occur with chronic pain
Obesity	Strong evidence for higher risk of complications with surgery and poorer response to procedures109 111	Obesity results in higher weight burden on joints and increased inflammation
Smoking	Conflicting results support slightly poorer outcomes for smokers, although some studies found the opposite result118 120	Most trials did not control for confounding, intersecting variables
Opioid use	Strong evidence for poorer outcomes for surgery128 129 and non-surgical interventional procedures27 127 135	Opioid use might be confounded by other comorbid factors, such as undermanaged psychopathology

^*Online supplemental appendix 1 describes the modified grade criteria used to assess evidence and the limitations in weighing studies that failed to account for intersectionality.

Figure 1. Intersectionality of factors associated with interventional treatment outcome. Examples of factors are: obesity and smoking; diffuse pain and psychiatric variables; dissatisfaction with current employment causing subconscious secondary gain; and association between prescription opioid use and substance misuse. The radial network graph shows connections between factors, with relative strength of relations represented by the thickness of the lines (binary system where a thin line signifies a weak correlation and a thicker line signifies a stronger one). Figure images based on original drawings by Seffrah J Cohen, Brandeis University.

Age

Older adults are more likely to have conditions that cause pain, but the threshold required for noxious stimuli to induce pain increases with age.¹² Elderly individuals tend to report lower acute pain scores¹³ and have a higher tolerance for mild pain,^{12 14} probably as a consequence of age related physiological changes, including decreased peripheral nociceptive fibre density, reduced central nociception, and decreased cortical activation, along with greater protective behavioural factors (ie, greater acceptance and less catastrophising).^{12 14 15}

Not all physiological changes of ageing are favourable. Descending inhibitory pathways degrade, making severe pain less tolerable and increasing the risk of chronic pain.^{15 16} Although the decreased neuroplasticity of older adults might protect against maladaptive somatosensory changes from acute pain, changes that have already occurred are difficult to reverse.^{12 15} Minimal data exist about age as an independent variable for procedural outcomes related to chronic pain. In a small prospective study of 47 patients undergoing CT (computed tomography) guided epidural steroid injections, no correlation was found between age and pain relief.¹⁷ In a secondary analysis of a multisite observational study in 227 patients with chronic knee pain, those who reported a perceived age lower than their chronological age had less pain and disability.¹⁸ A large prospective study (n=346) found that older patients were more likely to have procedural success for non-surgical low back procedures (mean age for procedural success was 54 years v 50 years for failure).¹⁹ A retrospective study (n=160) evaluating spinal fusion for lumbar stenosis found no differences in pain or functional outcomes between the super elderly (≥80 years) and those aged 65-79 years, but a higher rate of postoperative delirium in the cohort aged >80 years.²⁰ The inevitable physiological changes that occur with ageing influence, but do not determine, the perception of pain.

Sex

Chronic pain is more commonly reported in women than men.^{1 21} Factors responsible for the higher chronic pain prevalence in females include biological mechanisms, such as differences in hormone concentrations and neural circuitry (eg, central sensitisation), as well as increased psychosocial risks (eg, comorbid depression).²¹ Several studies have reported a higher risk of persistent postsurgical pain in women for surgeries that include video assisted thoracoscopic surgery, cardiac surgery, lumbar and cervical spine surgery, and orthopaedic trauma.22,25 In a meta-analysis of 10 525 patients receiving video assisted thoracoscopic surgery, a higher risk of persistent postsurgical pain (odds ratio 1.58, 95% confidence interval 1.2 to 1.96) was reported for women.²² For joint replacement, studies have been mixed, with some finding better outcomes and others worse outcomes in females.²⁶ More than 75% of studies have reported differences in outcomes for spinal surgery grouped by sex; most found that females fared worse than males, with the evidence being stronger for poorer short term outcomes.²⁴

A similar trend, however, has not been found for outcomes in non-surgical procedures for chronic pain. Investigators did not report significant differences in responsiveness to procedural interventions between males and females for genicular nerve radiofrequency ablation,²⁷ epidural steroid injections,^{28 29} or lumbar and cervical facet radiofrequency ablation.^{30 31} In a multicentre prospective cohort study of 346 patients undergoing epidural steroid injections, facet procedures, or sacroiliac joint injections, female sex did not predict procedural outcomes.¹⁹ One reason for the differences between surgical and non-surgical interventions in men and women is that the tissue trauma for surgery is many times greater than that for non-surgical interventions, so persistent pain from the intervention itself is less relevant for percutaneous procedures. These findings are in line with higher pain prevalence rates for females for most chronic conditions.¹

Disease burden

Although it seems reasonable that a greater disease burden should produce a higher degree of pain, and that patients with greater pain severity should feel more benefits from interventional procedures, neither premise is proven. Data are mixed about whether a greater severity of lesions or symptoms correlates with outcomes from interventional pain procedures, and an inverse relation is frequently found. In a recent systematic review of 15 studies (median 73 participants per study), no individual prognostic factor (including baseline pain severity, or magnetic resonance imaging (MRI) or electromyography findings) was significantly associated with outcomes after epidural steroid injections.²⁹ In a secondary analysis of a randomised controlled trial conducted in 350 patients undergoing epidural steroid injections, the radiological severity of the lumbar stenosis was not significantly associated with postprocedural outcomes.³² In a retrospective study in >1200 patients receiving epidural steroid injections for sciatica, lower baseline pain scores, lower opioid doses, and unilateral (v bilateral) symptoms were correlated with positive outcomes.²⁸ A prospective study in 346 patients who received injection or radiofrequency ablation for low back pain found inverse relations between baseline pain score, duration of symptoms, and treatment outcomes.¹⁹

The negative relation between duration of symptoms and interventional outcomes has been found for other studies evaluating non-surgical interventions, but only for faster recovery and better short term outcomes for decompression surgery to treat sciatica.^{33 34} In an observational study in 114 patients undergoing transforaminal epidural steroid injections, the size of a herniated disc on MRI had no correlation with procedural outcomes, but patients with an extrusion were more likely to require repeat transforaminal epidural steroid injections.³⁵ Guidelines on lumbar and cervical facet joint pain have not found consistent relations between MRI findings and radiofrequency ablation outcomes.^{36 37}

Among patients who require spinal surgery, the evidence is of low certainty that greater disease severity (ie, pain or disability) is associated with improved postoperative measures. In a systematic review of six studies (n=802), a higher preoperative pain score and lower quality of life score correlated with greater improvements in these same measures postoperatively.³⁸ In another systematic review of 16 studies containing >8300 patients, the evidence was of low certainty that higher preoperative pain scores are associated with greater postoperative improvements in pain after lumbar fusion.³⁹ For total hip arthroplasty but not knee arthroplasty, a correlation seems to exist between greater Kellgren-Lawrence radiographic joint degeneration and improved patient reported surgical outcomes, including pain, function, and health related quality of life, although the results for both are conflicting.^{40 41} These findings are in contrast with multiple studies that showed that lower Kellgren-Lawrence scores were associated with better outcomes from genicular nerve radiofrequency ablation.⁴² One reason for the apparent discrepancy between the surgical and interventional pain literature about the relation between disease burden and outcomes is that when surgery is done to remove a minor abnormality (eg, a small protruding disc compared with a large one), that minor pathology often is not the pain generator.⁴³

Type of injury according to mechanistic categories

Neuropathic pain

Neuropathic pain is caused by a demonstrable lesion or disease of the somatosensory system.⁴⁴ Compared with nociceptive pain, neuropathic pain is associated with a poorer prognosis. For spinal surgery, however, results from decompression surgery for radiculopathy are consistently more favourable than those for axial nociceptive pain secondary to degenerative spondylosis.^{45 46}

Nociceptive pain

Nociceptive pain is the most common type of chronic pain and results from activation of nociceptors after actual or potential non-neural tissue injury. Guidelines and large systematic reviews for many nociceptive pain conditions have found small but consistent benefits for non-surgical interventional procedures, such as radiofrequency ablation and joint injections, that exceed risks of harm in well selected patients (ie, based on prognostic blocks for radiofrequency ablation, and diagnostic criteria for joint injections, without strong risk factors for failure).¹ For joint arthroplasties performed for pain and disability secondary to degeneration, favourable long term outcomes have been found with hip, shoulder, and knee replacements.⁴⁷

Nociplastic pain

Nociplastic pain is pain that results from abnormal processing of sensory information and is characterised by the absence of biomarkers. Because the primary physiological mechanism in nociplastic pain is central sensitisation, peripherally focused treatments, such as interventional procedures or surgery, tend to result in poor outcomes.^{1 48} In a large observational study of >200 patients undergoing total knee arthroplasty, higher preoperative central sensitisation inventory scores were associated with worse outcomes up to two years later.⁴⁹ Higher preoperative central sensitisation inventory scores or a pre-existing nociplastic condition are associated with an increased risk of persistent postsurgical pain and dissatisfaction after joint surgery,^{50 51} and poorer outcomes after injections.⁵² Similar findings have been found in studies⁵³ for lumbar surgery.

Non-organic signs

Originally described as physical examination signs that warrant psychological evaluations and predict failure of surgical treatments, the term non-organic implies physical findings with no observable anatomical cause. ⁵⁴The original five categories were non-anatomical tenderness, pain with sham stimulation, discrepancies in examination findings, regional sensory or motor disturbances that deviate from neuroanatomy, and over-reaction. Although frequently misinterpreted as indicative of malingering, one systematic review found that only 45.5% of 11 studies (n=1883 excluding two duplicate studies) reported an association with secondary gain. This review also reported improvement in non-organic signs with treatment, compliance, or improved self-assessment, suggesting that these signs are dynamic and potentially modifiable.⁵⁵

In another systematic review, most of the 61 studies analysed reported no significant association between psychological distress and non-organic signs, although three of seven studies found an association with the neurotic triad of the Minnesota Multiphasic Personality Inventory, namely depression, hypochondriasis, and hysteria.⁵⁶ The included studies showed that non-organic signs, especially if multiple signs were present, were associated with poor outcomes for both surgery and conservative care. A large, prospective multicentre study (n=318) found a direct correlation between the number of non-organic signs, psychopathology, baseline disease burden, and secondary gain, with poor treatment outcomes for non-surgical low back pain procedures.⁵⁷ In a recent prospective study validating modified non-organic signs in 78 patients with cervical radicular pain, Cohen et al found associations between non-organic signs and psychopathology, baseline disease burden, and negative response to cervical epidural steroid injections.⁵⁸ It is important to recognise that some non-organic signs might represent non-readily observable physical pathology, such as central sensitisation manifesting as superficial tendereness.⁵⁹

Work characteristics

Job satisfaction has been positively associated with walking distance (which is often limited by pain) after limb amputation;⁶⁰ however, in a prospective study evaluating 60 patients after total hip arthroplasty, low job satisfaction was paradoxically associated with return to work.⁶¹ The type of work might also influence the outcomes of surgical pain. A systematic review (n=48 studies for qualitative analysis, 41 for quantitative analysis) found that physically demanding jobs were associated with lower return-to-work rates after total hip arthroplasty, with one of the included studies showing a higher rate in those with higher intention-to-work scores.⁶² For total knee arthroplasty, data are less promising, with some studies reporting higher return-to-work rates,⁶³ others no difference,⁶⁴ and one study reporting lower rates⁶⁵ in less physically demanding jobs. Studies have also reported higher return-to-work rates after joint replacement in those engaged in more flexible job roles and when individuals rated their supervisor as a good leader.^{64 66} Although return to work and pain are not synonymous, because pain is the leading cause of disability worldwide and the main reason for joint replacement, return-to-work rates are often used as an objective surrogate outcome for pain relief. In summary, outside of secondary gain, direct evidence is limited that job satisfaction affects interventional treatment outcomes.

Compensation status and secondary gain

Secondary gain can be internally motivated (ie, being sick to feel less guilty about failing at a job) and a component of any disease. Classic secondary gain is externally motivated and usually conscious (ie, feigning sickness to avoid military service) and might be less common than subconscious alternative motivations. When evaluating literature on secondary gain, one should take heed that the prevalence has not been determined in discrete populations, and remains mostly speculative.

The prevalence of hidden financial motives is difficult to determine. In one cross-sectional study performed in psychiatry patients, although only 6% of participants explicitly reported expectations for secondary gain, when further queried, 42.2% of 166 patients had an expectation of non-medical benefit with treatment, with secondary gain predicting poor outcomes.⁶⁷ Based on comparative data from the New Zealand Accident Compensation Scheme which provides no fault compensation, the estimated prevalence of symptom exaggeration after an unintentional injury is 20-50%.⁶⁸

A meta-analysis of the effect of secondary gain on surgical outcomes (n=129 studies, 20 498 participants) found a strong negative effect on outcomes, with all but five surgeries reporting an adverse influence (odds ratio3.79, 95% confidence interval 3.28 to 4.37).⁶⁹ Although the surgical type was not noted in this meta-analysis, the five most common surgeries all involved pain outcomes. In a subsequent meta-analysis evaluating the effect of compensation status on outcomes of spinal surgery that included satisfaction and return to work (n=31 studies, 3567 participants), Cheriyan et al found a more than twofold greater risk for a negative outcome.⁷⁰ Similarly, negative effects on pain and function have been reported in systematic reviews for shoulder surgery (n=28 studies, 3133 participants, evaluating workers' compensation), and in clinical studies for epidural steroid injections, facet joint risk factor ablation, and pulsed radiofrequency for headaches.2871,74

Not all secondary gain confers the same risk. In a propensity matched study in 51 patients receiving disability compensation and 37 receiving workers' compensation, Gum et al found that although both groups performed more poorly two years after lumbar fusion than matched controls, patients receiving workers' compensation had poorer outcomes.⁷⁵ Among those receiving workers' compensation (n=39 961), a review of data from Denmark’s National Board of Industrial Industryies by Rudbeck et al found lower return-to-work rates within the first year of an initial claim in those with adjudicated claims compared with those with ongoing claims, and higher return-to-work rates in those with recognised claims, regardless of compensation, than in those with rejected or withdrawn claims.⁷⁶

Central sensitisation

Central sensitisation might predict interventional pain outcomes because the primary cause is not regional pathology, but rather the disordered processing of sensory stimuli.⁴⁸ The negative impact of central sensitisation is compounded by the high rates of psychopathology and concomitant pain generators in this population, the low specificity of diagnostic imaging and physical exexamination signs for rheumatological conditions, and procedural overuse.4877,79 For nearly all interventional procedures, the Centers for Medicare and Medicaid Services lists diffuse pain as an exclusion criterion.⁸⁰

In a systematic review on the effect of nervous system sensitisation on musculoskeletal pain outcomes in 13 studies (n=1920, most involved joint replacement), O'Leary et al reported that five exploratory studies found an association between quantitative sensory testing indicative of central sensitisation and poor interventional treatment outcomes, with two studies that used self-reported questionnaires also finding an association; four studies that performed a multivariable analysis were evenly divided on an association.⁸¹ The authors concluded that insufficient evidence existed to support an independent relation between sensitisation and poor outcomes.

A more recent and comprehensive review by Georgopoulos et al reported on 37 studies (five randomised studies, n=3860 participants) where quantitative sensory testing was used before a pain intervention (psychotherapy, pharmacotherapy, integrative treatments, and surgery).⁸² The authors found modest but consistent correlations between baseline quantitative sensory testing measures and follow-up pain and disability. For the prototypical nociplastic condition, fibromyalgia, studies have found a negative effect on outcomes for hip and knee replacement, abdominal surgery, perineural injections, and facet joint radiofrequency ablation,5183,87 with negative radiofrequency ablation outcomes possibly attributable to aberrant responses to diagnostic blocks.⁸⁸

Preoperative and acute postoperative pain

Preoperative pain and severe acute postoperative pain are risk factors for persistent postsurgical pain, defined by ICD-11 (International classification of diseases, 11th revision) as pain persisting for ≥3 months after surgery.⁸⁹ The median incidence of persistent postsurgical pain is 20-30%,⁸⁹ slightly decreasing with time and varying widely by procedure. Studies have found that preoperative pain both at the surgical site and outside the site predicts persistent postsurgical pain.^{90 91} Preoperative pain is complex, however, representing the possibility of pre-existing nociplastic pain but potentially confounded by disease burden, psychopathology, and chronic opioid use. Uncontrolled postoperative pain increases the risk for subsequent persistent postsurgical pain, with severe pain on postoperative day 1 predicting persistent postsurgical pain.⁹²

Psychopathology

Depression, anxiety, and post-traumatic stress disorder often occur with chronic pain, particularly in nociplastic conditions.^{93 94} Investigators have found that depression is associated with a poorer response to interventional procedures, such as epidural steroid injections,²⁹ sacroiliac joint injections,¹⁹ and radiofrequency ablation.¹⁹ Among axis I diagnoses, depression, anxiety disorders, including post-traumatic stress disorder, and general psychological distress are strong predictors of persistent postsurgical pain for the spectrum of surgical procedures.⁹⁵ From a cognitive standpoint, traits such as catastrophising, poor coping, negative outlook, and somatisation are associated with poorer outcomes for patients undergoing spinal cord stimulation, joint replacement, and lumbar surgery.96,98

Considerable intersectionality exists between different variables, such as greater disease burden leading to higher rates of opioid use, maladaptive lifestyle changes, and psychopathology. In a multicentre study in 346 patients undergoing epidural steroid injections, sacroiliac joint injections, or facet interventions, a near inverse linear relation was found between the severity of depression and procedural outcomes, and some psychiatric variables, including sleep, tended to correlate with higher pain scores and longer duration of symptoms.¹⁹ The relation between depression and pain is bidirectional, with more severe depression leading to greater pain and poorer outcomes, and greater pain worsening depressive symptoms.⁹⁹

Sleep disturbances

Sleep disorders are strongly associated with chronic pain, with the relationship being bidirectional.¹⁰⁰ A systematic review that included 18 studies (n=8408, with 12 available for meta-analysis) found a small effect for an association between preoperative sleep disorders and persistent postsurgical pain (r=0.13, 95% confidence interval 0.06 to 0.20).¹⁰¹ Subgroup analysis found an association between studies that used pain intensity as the primary outcome measure but not the presence of pain, and for total knee arthroplasty but not total hip arthroplasty or breast cancer surgery.¹⁰¹ For non-surgical interventions, Cohen et al found that individuals who failed procedures for low back pain scored higher on a preprocedure sleep dysfunction questionnaire, whereas other investigators reported similar results for epidural steroid injections.^{19 102} In summary, the evidence is strong that pre-existing sleep disorders can predispose to failure of pain treatment, with the effects being mediated in part by reduced pain thresholds and tolerance, and a higher rate of diagnosed and undiagnosed psychopathology.¹⁰³

Obesity

Obesity is strongly associated with chronic pain.¹⁰⁴ Obesity carries a dose-dependent risk increase for many pain syndromes, such as a 35% increase in risk for each 5 kg/m² increase in body mass index for knee osteoarthritis¹⁰⁵ and a 26% increased risk for each 3.0 kg/m² increase in body mass index for radiating low back pain.¹⁰⁶ Obesity is associated not only with degenerative arthropathy, but also with a proinflammatory milieu (i.e., increased inflammatory mediators involved in pain), a higher affective-motivational component of pain that is less responsive to procedures, and metabolic alterations that can exacerbate some pain states.¹⁰⁷ Thus, obesity is a risk factor not only for nociceptive pain caused by structural stress, but for other pain conditions, such as neuropathic pain and headache.¹⁰⁸

Data are mixed about the effect of obesity on surgical outcomes for pain. A 2019 systematic review of nine studies (n=10 073, 624 of whom were severely obese) reported lower postoperative Knee Society Objective Scores and Knee Society Functional Scores and higher complication rates after total knee arthroplasty in severely obese patients.¹⁰⁹ For total hip arthroplasty, although some studies showed poorer functional outcomes in individuals who were overweight, most did not.¹¹⁰ Higher rates of complications and a slightly greater risk of treatment failure have been reported in patients with obesity receiving cervical and lumbar spinal surgeries.^{111 112} In the randomized Spine Patient Outcomes Research Trials (SPORT), however, obesity was associated with poorer surgical outcomes than patients with no obesity for some lumbar spine surgeries, but compared with non-operative treatment, pain and functional outcomes were better.¹¹³

Obesity might be associated with a less robust response to non-surgical interventions, such as chiropractic treatment¹¹⁴ and interventional procedures like radiofrequency ablation, joint injections, and epidural steroid injections, and is therefore a relative contraindication for some percutaneous disc procedures.^{19 115 116} Another mediator for the negative effect obesity seems to have on interventional outcomes for chronic pain is the association between obesity and higher pain prevalence rates for most chronic conditions, even in the absence of treatment. Reasons for the discrepancies in findings might include failure to control for the severity of obesity, intersectionality and confounding variables, and differences in outcome measures.

Smoking

Smoking is associated with increased acute postoperative pain. Long term use causes desensitisation of acetylcholine receptors and changes in neuroplasticity, leading to decreased pain thresholds with increased nicotine use and increased pain sensitivity with nicotine withdrawal.¹¹⁷ In a meta-analysis of 33 studies (n=53 362, with nine studies including in quantitative analysis of smoking) evaluating a wide range of procedures (mostly mixed, general surgery and orthopaedic operations), Yang et al found that smoking was a risk factor for increased postoperative pain.¹¹⁸ A review by Liu et al found that smoking predicted moderate-to-severe postoperative pain.¹¹⁹ Although severe acute postoperative pain is associated with the development of persistent postsurgical pain and might be more common in smokers,¹¹⁷ the results are mixed regarding long term pain and functional outcomes. One systematic review based on 40 studies (n=3 037 683) found that smoking was associated with poor clinical outcomes based on qualitative analysis one year after total hip and total knee arthroplasty;¹²⁰ however, a propensity matched study (n=1028) paradoxically found a lower incidence of persistent postsurgical pain in smokers after thoracic surgery.¹²¹

Smokers might also have poorer outcomes after pain procedures, although data are mixed. In a retrospective cohort study performed in 544 patients, Jayabalan et al found that smoking was associated with a higher baseline severity of radicular pain and less relief after lumbar epidural steroid injections.¹²² Smoking was also associated with treatment failure for lumbar epidural steroid injections, lumbar facet radiofrequency ablation, and sacroiliac joint injections in univariable but not multivariable analysis in a prospective study (n=346) by Cohen et al.¹⁹ In contrast, Hughey et al found that smokers undergoing lumbar facet radiofrequency ablation had improved disability six months after the procedure compared with non-smokers.¹²³ Notably, chronic smoking is associated with a variety of psychiatric disorders, sleep disturbances, lower socioeconomic status, and decreased social activities and activities of daily living,^{117 124} all of which are associated with poorer pain outcomes. Although smoking seems to be independently associated with a small negative treatment effect on pain interventions, controlling for observed and unobserved confounding variables can be difficult.

Opioid use

Although opioids are the reference standard for the treatment of acute pain, the evidence in chronic pain is mostly negative. Most studies have found non-superiority or harm in comparison with non-opioid analgesics, and the data on long term pain relief are limited.¹²⁵ Extensive data show that chronic opioid use is associated with poor procedural and non-procedural outcomes. Reasons include greater baseline disease burden, including psychopathology in opioid users, subconscious secondary gain, and the phenomenon known as opioid induced hyperalgesia, which results in pain sensitisation.^{126 127}

A meta-analysis of 45 spinal surgery studies (range 57 to 1 826 868 patients) found that preoperative opioid use resulted in higher postoperative opioid requirements, worse pain and functional outcomes, longer hospital stays, and increased aggregated morbidity and mortality.¹²⁸ In arthroplasties, preoperative opioid use was associated with increased complications and revision surgery, as well as chronic postoperative opioid use. For intermediate term and long term pain and functional outcomes, however, the results are mixed.129,131 For shoulder surgeries, a systematic review of 21 studies (n=257 301) reported a strong association between preoperative and postoperative opioid use (two-thirds of studies), with a weaker association (38% of studies) between opioid use and poor functional outcomes.¹³²

Opioid use also presages poorer results from non-surgical procedures. Opioid use before procedures is associated with poorer outcomes following cervical and lumbar epidural steroid injections^{127 133} and in radiofrequency ablations performed for lumbar and cervical facet pain, sacroiliac joint pain, and knee osteoarthritis.^{27 31 134 135} In one large database review performed in 2887 patients, Southren et al found that receiving periprocedural opioids in opioid naive patients increased the rate of future chronic use after lumbar facet radiofrequency ablation.¹³⁶

The effect of opioids on pain outcomes might be mediated by a bidirectional relationship between opioid use and psychological disease burden. A systematic review and meta-analysis (n=323 515, 10 studies) found that opioid use was significantly associated with an increased incidence of anxiety disorder, depression, and bipolar disorder, with adjusted effect sizes ranging between 1.18 and 1.80).¹²⁶ Furthermore, patients with existing mental health disorders are more likely to be prescribed opioids, as they tend to be less responsive to other pain treatments.¹²⁶

Interventions to mitigate risk factors

Preventive and pre-emptive analgesia

Preventive and pre-emptive analgesia (when an intervention is applied before a known stimulus rather than at any time before, during, or after the stimulus) have been extensively studied in the context of surgery. Pre-emptive analgesia might theoretically prevent persistent postsurgical pain by decreasing pre-stimulus pain and mitigating the bombardment of nociceptive input from surgery. In one systematic review evaluating 15 studies (n=830), pre-emptive analgesia was associated with lower postoperative analgesic requirements and lower pain scores at four hours after operation than the intervention applied after the surgical stimulus, although not between six and 24 hours.¹³⁷ Since severe acute postsurgical pain is a risk factor for persistent postsurgical pain, most studies examining the association have measured postoperative pain over a longer time course (≥24 hours).

A network meta-analysis of double blind studies evaluating 13 preventive analgesia interventions for preventing persistent postsurgical pain (n=107 studies, 13 553 participants) found evidence only for serotonin-norepinephrine reuptake inhibitors (risk ratio 0.34, 95% confidence interval 0.29 to 0.60), neural blockade alone (0.73, 0.61 to 0.87), systemic local anaesthetic (lidocaine 0.69, 0.50 to 0.96), neural blockade with N-methyl-D-aspartate inhibitors (eg, ketamine or dextromethorphan 0.61, 0.46 to 0.81), and neural blockade coupled with a gabapentinoid (0.61, 0.48 to 0.78). For gabapentinoids and N-methyl-D-aspartate antagonists as standalone treatments, the evidence was weak, whereas for non-steroidal anti-inflammatory drugs and other treatments, the evidence was negative or conflicting.¹³⁸ In contrast, an evidence-based narrative review by Chen et al found moderate evidence for epidural analgesia, very weak evidence supporting α₂ agonists (eg, clonidine), intravenous lidocaine, ketamine, and peripheral nerve blocks, and one small, industry-sponsored study supporting peripheral nerve stimulation.¹³⁹ In an earlier narrative review, the authors reported similar findings but also found evidence for continuous or repeated local wound infiltration.¹⁴⁰

Complex regional pain syndrome is often considered a nociplastic pain condition, albeit with physical evidence of disease pathology. The pre-emptive use of vitamin C before high-risk orthopaedic surgery is purported to work via antioxidant effects. One systematic review that included seven placebo controlled and one quasi-experimental study (n=1427) found that doses of 500-1000 mg/day started 42-50 days after the injury, but before distal orthopaedic surgeries, reduced the incidence of complex regional pain syndrome (odds ratio 0.33, 95% confidence interval 0.17 to 0.63), but based on two studies did not decrease pain scores.¹⁴¹ Two large subsequent placebo controlled trials (combined n=621) also found positive effects for prevention of complex regional pain syndrome with vitamin C.^{142 143}

Disability and prehabilitation

A systematic review and meta-analysis evaluating prehabilitation before orthopaedic surgery that included 48 trials (n=3570 participants) reported no benefits for postsurgical pain or quality of life after total knee arthroplasty for up to three months of follow-up, with an uncertain and small benefit for function (standardised mean difference 0.29, 95% confidence interval 0.51 to 0.08).¹⁴⁴ For total hip arthroplasty, the findings at three months were similar (no benefit for pain or quality of life, and uncertain benefit for function). In lumbar spine surgery, a benefit was observed for pain at three months (standardised mean difference –5.93, 95% confidence interval –10.55 to –1.31) but not afterwards, with benefits for function not apparent at six months. In another systematic review of 32 randomised trials (n=1753) examining exercise before or after total hip arthroplasty, the authors found no benefit for function at any time point.¹⁴⁵

Few studies have evaluated the effects of physical therapy or rehabilitation on non-surgical pain outcomes, although one randomised trial performed in 169 patients with cervical radicular pain found that combining physical therapy and pharmacotherapy with cervical epidural steroid injections resulted in better outcomes for up to three months than standalone conservative or interventional treatments (56.9% v 26.8% v 36.7% success rates, respectively).¹⁴⁶

Psychotherapy and education

In a systematic review of 21 studies (n=2288) examining perioperative psychological treatments on various surgical outcomes, Nalinda et al found that psychotherapy (mostly evidence-based cognitive-behavioural therapy) reduced subacute or acute (<3 months) pain (standardised mean difference −0.26, 95% confidence interval −0.48 to −0.04) and disability as well as persistent postsurgical pain (−0.33, −0.61 to −0.06) and chronic disability, although most studies were negative.¹⁴⁷ The authors also found that psychological interventions delivered after surgery and by a psychologist were most effective. An earlier review (n=19 studies, 1893 participants), however, that did not distinguish between treatments found limited evidence for preoperative psychological interventions on pain; anxiety decreased acutely, and long term benefits were seen for mental health quality of life.¹⁴⁸ For preoperative education, a systematic review and meta-analysis of 37 randomised trials (n=3527) found a small positive effect on postoperative pain for up to one week (standardised mean difference −0.23, 95% confidence interval −0.39 to −0.07) but not on function or long-term pain relief.¹⁴⁹

Weight loss

In a narrative review on the effect of weight loss before knee arthroplasty, Godziuk et al found unclear evidence supporting the intervention, with positive studies excluding patients with a body mass index >40.¹⁵⁰ A later review examining preoperative weight loss on knee or hip arthroplasty found equivalent outcomes and a possible increased risk of complications.¹⁵¹ For spinal surgery, one systematic review (four retrospective studies, n=183 570 participants) found lower complication and mortality rates in patients who had previous bariatric surgery compared with obese patients who had not undergone weight loss surgery, whereas another (n=4) reported no difference in complication rates, but a higher incidence of persistent postsurgical pain in patients after bariatric surgery based on the one study not reported in the other review.^{152 153}

Smoking cessation

A systematic review on smoking cessation before surgery (13 studies, two randomised studies, n=1721 participants) found that stopping smoking resulted in higher acute postoperative pain scores and opioid requirements compared with non-smokers, although three studies that grouped patients for length of abstinence found lower pain or opioid requirements, or both, when smoking cessation exceeded three weeks.¹⁵⁴ Clinical guidelines recommend smoking cessation for at least eight weeks to reduce perioperative complications but the guidelines do not address pain.¹⁵⁵

Opioid weaning

Despite the strong association between opioid dose and poorer postoperative pain outcomes, few studies have examined the effect of reducing opioids on persistent postsurgical pain. A case-control study comparing 41 individuals who had successfully (≥50% reduction) been weaned off opioids with an equal number of opioid-dependent patients and opioid-naive patients found that the intervention and opioid-naive groups had better results for measures of quality of life and function after hip or knee arthroplasty than those who remained on opioids.¹⁵⁶ No differences for changes in quality of life or function (including those that measure pain) were found between the intervention and opioid-naive groups at the 6-12 month follow-up, although some follow-up scores favoured the opioid-naive group. In a large database analysis evaluating postoperative opioid use as a surrogate for poor outcomes in patients undergoing spinal fusion, Jain et al divided 17 643 patients into five groups: opioid naive (n=4911), intermittent use (n=9142), chronic use (n=34 840), and two smaller groups where chronic opioid use was discontinued >2 months (n=106) and three months (n=41) before surgery.¹⁵⁷ The authors found a lower incidence of long-term opioid use in those who stopped opioids for >2 months preoperatively (46.2%) compared with chronic opioid users (72.2%), although rates were higher than in opioid-naive patients (3.9%) and intermittent users (10.3%).

Compensation status and secondary gain

For unconscious secondary gain (eg, assuming a sick role to avoid an unpleasant assignment), psychotherapy focusing on insight, identifying legitimate and illegitimate concerns and ways to deal with reinforcers (eg, a difficult boss), and improving coping skills can be helpful. For classic secondary gain, recognising the problem and its root causes (eg, poverty or resentment) and if, possible, dealing with these problems in a compassionate and understanding way, can in some cases alter the trajectory. No studies have evaluated interventions addressing secondary gain, but opinions of experts are mixed, with some recommending treating patients with suspected secondary gain the same as other patients, whereas others recommend education of providers to reduce financial incentives (ie, performing less procedures because patients receive higher payouts for more interventions)^{158 159} (table 2).

Table 2. Evidence base for pretreatment interventions to improve pain related outcomes.

Intervention	Level 2 evidence	Level 3 evidence	Level 4 evidence	Negative or conflicting evidence	Cost effectiveness and comments
Preventive analgesia for chronic pain (>3 months)	Vitamin C for decreased pain in patients with complex regional pain syndrome	To prevent persistent postsurgical pain: serotonin-norpinephrine reuptake inhibitors; regional anaesthesia with or without adjuvants; intravenous lidocaine; local anaesthetic wound infiltration	Ketamine, postoperative local anaesthetic infiltration		Ketamine might work better in those with a strong affective component or baseline sensitisation. Continuous infusion more effective than single shot local wound infiltration. Interventions started preoperatively and continued postoperatively might be more effective than those limited to either the preoperative or postoperative period. Cost effectiveness depends on drug treatment and patient selection138,140
Disability or prehabilitation			Small effect for pain and function for up to six months only for spinal surgery		For most studies on spinal surgery, results favouring intervention were not significant. These measured back, but not leg pain. Cost effective before high risk surgery, although publication bias limits level of certainty144 145 169
Psychotherapy	For mental health quality of life after six months based on two studies		For pain, only for cognitive-behavioural therapy after three months. For acute anxiety, strongest evidence for relaxation techniques		Stronger evidence for preoperative and postoperative treatment, and when given by a psychotherapist. Cost effective for high risk populations, particularly those with high levels of baseline psychopathology and pain related disability147 148 170
Sleep improvement	Evidence for improved postoperative pain up to day 3 after joint replacement with pharmacotherapy and non-pharmacotherapy			Conflicting evidence for analgesic reduction, scant evidence for persistent postsurgical pain	Evidence that preoperative interventions improve sleep. One study found improved depression scores, but not pain or sleep for up to 12 weeks after breast cancer surgery in non-depressed women with perioperative melatonin versus placebo. Cost ineffective for preventing postoperative complications, unclear cost utility on long- term outcomes171 172
Education				Significant benefit only at one week for orthopaedic surgery. Non-significant trend favouring preoperative education for pain and function only for up to three months149	Might be effective for chronic pain in general, but scant or conflicting evidence for acute pain and pain prevention. Lack of high quality evidence on cost effectiveness for pain outcomes, but inexpensive interventions likely to be cost effective173
Smoking cessation		For pain when length of cessation is >3 weeks			Acute cessation results in higher postoperative pain scores. Cost effective in motivated patients for averting postoperative complications that might indirectly affect pain, but direct data are lacking154
Weight loss				No evidence for pain and functional benefit after joint replacement or spinal surgery, with some studies suggesting worse outcomes151,153	For non-interventional treatments, weight loss might provide a small effect compared with minimal care, although the correlation between the amount of weight loss and benefit is weak. Low costs, overall health improvements, and possible beneficial effect on complications and pain outcomes make this cost effective in motivated patients174
Opioid weaning			Indirect evidence for pain outcomes based on non-randomised trials		Long term- opioids used as surrogate for pain, although might not reflect pain in chronic users. Probably cost effective for contained systems (eg, military), but bundled payments (ie, no increased financial reimbursement for visits to implement weaning) make cost utility difficult to prove

Level 1 evidence is absent because no interventions had level 1 evidence.

Future developments

Similar to personalised medicine, screening for pathology and identifying patients suitable to receive interventions can improve risk-benefit and cost-effectiveness ratios. A personalised, mechanism-based approach could dictate which treatments are likely to benefit which patients (ie, identifying biomarkers that could guide treatment and predict outcomes, intravenous ketamine for patients with central sensitisation or a high affective pain component, or pre-emptive nerve blocks in those with severe preoperative pain). Timing might also have a role (ie, epidural analgesia started >24 hours before scheduled amputations might be more likely to prevent chronic pain after amputation, and pre-emptive analgesia might be more effective than preventive measures^{137 160}).

Pragmatically, most studies examining the effect of interventions on outcomes are limited to one treatment group, although combination treatment could provide better outcomes. Enhanced Recovery After Surgery programmes are a multidisciplinary approach aimed at reducing surgical stress and facilitating recovery after surgery.¹⁶¹ Combination treatment improves chronic pain outcomes but can be expensive, increases side effects, and is a greatly under-researched area because of methodological challenges.¹⁶² Finally, large database reviews can determine associations but cannot establish cause-effect relationshipa and have limited use in informing guidelines.¹¹³

Conclusions

Interventions performed for chronic pain show variable results, with more stringent patient selection improving patient outcomes and cost effectiveness. Identifying characteristics associated with patient outcomes can improve the risk-benefit and cost- effectiveness ratios of procedural interventions. Although all factors identified in this review are associated with treatment outcomes, substantial overlap exists between variables so that weighing their influence or establishing a cause-effect relationship is challenging. Most interventions designed to look at modifiable risk factors do not require intensive resources, might be cost effective in a personalised care model (eg, psychotherapy for individuals with psychopathology, smoking cessation for motivated individuals, and education for those engaging in self-management, for alternative treatments, and to align surgical expectations with realistic outcomes), and should be implemented whenever possible. The potential benefit of a successful intervention in high risk patients (ie, stopping opioids or return to work) must be weighed against the likelihood of treatment failure and is ideally decided by personalised medicine in the framework of a shared decision model. With the use of computer models, artificial intelligence, and large scale data, in the future interventional pain doctors might be able to provide precise estimates on the likelihood of success to patients before performing procedures.

Questions for future research.

• How can the weighted value of each risk factor be determined and the degree of intersectionality be better quantified?

• What are the best study designs to determine the effectiveness and cost effectiveness of interventions, and identify risk factors for treatment failure?

• How dependent is cost effectiveness on personalised care models and the identification of phenotypes likely to respond (eg, psychotherapy for motivated patients with psychopathology)?

• What is the best way to conduct inclusive, large scale pragmatic cohort studies that distinguish between associations (through intersecting variables) and true predisposing factors?

• Are multimodal risk factor interventions more effective and cost effective than unimodal interventions?

• Can effective ways to improve interventional treatment outcomes be feasibly implemented?

Patient involvement.

Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. A patient particpated in the peer review process of this manuscript.