Effectiveness of self-management of low back pain: Systematic review with meta-analysis

Authors


Abstract

Objective

To determine the effectiveness of self-management for nonspecific low back pain (LBP).

Methods

We performed a systematic review searching the Medline, Embase, CINAHL, PsycINFO, LILACS, PEDro, AMED, SPORTDiscus, and Cochrane databases from earliest record to April 2011. Randomized controlled trials evaluating self-management for nonspecific LBP and assessing pain and disability were included. The PEDro scale was used to assess the methodologic quality of included trials. Data were pooled where studies were sufficiently homogenous. Analyses were conducted separately for short- (less than 6 months after randomization) and long-term (at least 12 months after randomization) followup. Six criteria for self-management were used to assess the content of the intervention.

Results

The search identified 2,325 titles, of which 13 original trials were included. Moderate-quality evidence showed that self-management is effective for improving pain and disability for people with LBP. The weighted mean difference at short-term followup for pain was −3.2 points on a 0–100 scale (95% confidence interval [95% CI] −5.1, −1.3) and for disability was −2.3 points (95% CI −3.7, −1.0). The long-term effects were −4.8 (95% CI −7.1, −2.5) for pain and −2.1 (95% CI −3.6, −0.6) for disability.

Conclusion

There is moderate-quality evidence that self-management has small effects on pain and disability in people with LBP. These results challenge the endorsement of self-management in treatment guidelines.

INTRODUCTION

One of the most common and costly musculoskeletal disorders is low back pain (LBP), with treatment costs in excess of £10 billion per year in the UK (1). In Australia, more than $1 billion is spent each year on treatment for LBP (2), and the proportion of the gross national product spent on health costs for this condition is expected to double by 2047 (3). A potential explanation for the large treatment costs is that health care providers often administer passive interventions that provide some immediate pain relief (4), but often LBP recurs (5) and patients return for further treatment, initiating a process of dependence. One potential solution to this problem is to shift from traditional models of care where the patient is a passive recipient of treatment to models where patients are actively involved in the management of their back pain (6).

There is a growing awareness that LBP is a long-term condition and that self-management can decrease the burden of this condition (7). To date, self-management has been described mainly as a model of care where patients use strategies to manage and monitor their own health, retaining a primary role in management, and where they learn skills to be used in the daily management of their health condition (3, 8, 9). As with all complex interventions, the challenge with self-management is defining its core components. Self-management interventions may vary in how they are delivered (e.g., group, individual, mixed), who delivers them (e.g., health care providers, coach, internet), and the setting and components (e.g., educational, physical activity) (9). We have been guided by a recent report outlining the Australian Government's First National Primary Health Care Strategy (3), which has defined 6 core components of effective self-management interventions, including shared responsibility for a plan of care and self-monitoring and management of signs and symptoms, among others.

At present, evidence of effectiveness of self-management of LBP is unclear. Particular problems are inconsistency of how the intervention is named (i.e., self-care, self-management, self-monitoring, self-treatment) and whether interventions fulfill content components of self-management. For instance, May conducted a systematic review in 2010 (8) including trials that evaluated traditional health care provider–delivered treatments defined as physiotherapist-supervised exercises and the delivery of educational booklets. Arguably, trials evaluating interventions that lack core components of self-management potentially underestimate the effectiveness of self-management. The aim of this review was to estimate the effectiveness of self-management for LBP.

Significance & Innovations

  • Self-management is endorsed in most guidelines for the management of low back pain (LBP).

  • This systematic review demonstrates that self-management has only small effects on pain and disability, questioning whether this would be a worthwhile approach in the management of LBP.

  • This is the first meta-analysis of self-management for nonspecific LBP to assess the content validity of self-management programs used in included trials based on a previously proposed set of definition criteria.

MATERIALS AND METHODS

Literature search.

Studies were retrieved from searches of the following databases from the earliest record to April 29, 2011: Medline, Embase, CINAHL, PsycINFO, LILACS, PEDro, AMED, SPORTDiscus, and the Cochrane Register of Clinical Trials. We used the search strategy advocated by the Cochrane Collaboration Back Review Group (10) to locate clinical trials of LBP treatments and then used the keywords “self-management” and “self-care” to identify self-management interventions (see Supplementary Appendix A, available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658).

Eligibility criteria.

Study selection and participants.

Randomized controlled trials were included in the review without language restriction. Trials were eligible if they included participants with nonspecific LBP of any duration. Trials presenting data on people with specific causes for LBP such as infection, neoplasm, fracture, osteoporosis, inflammatory disease, pregnancy, serious nerve root disorders (e.g., cauda equina syndrome), and sciatica were excluded (11).

Interventions and outcome measures.

Many different interventions may be applied as self-management strategies for LBP. Due to inconsistency in how these interventions are named and in implementing the core components of self-management in the literature, we included trials in which at least 1 intervention was indicated by the authors as self-management for LBP by naming the intervention using the terms “self-management” or “self-care.” We chose to define the inclusion criterion in this way because the intention of the intervention is for the participant to manage their own condition with minimal oversight or active involvement of the health care provider. Trials comparing different types of self-management for LBP were also included. When multiple self-management interventions were available within the same trial (12, 13), the one fulfilling more core components of self-management based on the recent report outlining the Australian Government's First National Primary Health Care Strategy (3) was used (e.g., booklet plus reinforcement) (13). Outcomes of interest were pain and disability.

Assessment of trial methodologic quality.

All potentially eligible records were screened for eligibility, and those eligible were assessed for methodologic quality using the PEDro scale (range 0–10) (14, 15). If trials were already listed on the PEDro database (http://www.pedro.org.au/), we adopted these scores.

Two assessors (VCO and RZP) independently rated each study for eligibility and methodologic quality and extracted data. Consensus was initially used to solve disagreements, and a third reviewer (PHF) arbitrated if consensus could not be reached. If there was insufficient information about outcomes, the authors were contacted to provide clarification.

Data extraction.

Baseline demographic data (age, sex, symptom duration, and source of participants) and self-management features (type, amount of support given by providers) were extracted from included trials. Outcome data including mean scores, SDs, and sample sizes were extracted. SDs were not provided in 5 trials, and values were input either based on the average SD of other included trials (13) or derived from standard errors (12, 16) or confidence intervals (CIs) (17, 18) following the Cochrane handbook (19). Outcome data were extracted for 2 time points: short-term (less than 6 months after randomization) and long-term (at least 12 months after randomization) followup assessments (10). When multiple time points were available within the same followup period, the one closer to the end of the intervention for short-term followup and closer to 12 months for long-term followup was used. Pain and disability were transformed to a common 0–100 scale.

Statistical analysis.

Data were pooled in meta-analyses and described as weighted mean differences with 95% CIs. Between-trial heterogeneity was identified using I2 statistics (20). When trials were sufficiently homogeneous (i.e., an I2 less than 50%), pooled effects were calculated using fixed-effects models, whereas random effects were used to estimate pooled effects of heterogeneous trials (i.e., an I2 of 50% or more) (19). Comprehensive Meta-Analysis software, version 2.2.04 (Biostat), was used to conduct all analyses.

The Grading of Recommendations Assessment, Development and Evaluation (GRADE) system (21) was used to summarize the strength of the recommendation for each outcome: 1) high-quality evidence means further research is very unlikely to change our confidence in the estimate of effect; 2) moderate-quality evidence means further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate; 3) low-quality evidence means further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate; and 4) very low-quality evidence means we are very uncertain about the estimate.

Scoring of evidence started at moderate quality, rather than high quality, since we could not adequately test the GRADE criterion publication bias due to the small number of included trials (22). Moreover, the quality of evidence (21) was downgraded one level, from moderate quality, if one of the following was present: low methodologic quality (PEDro score <5), inconsistency of effects among trials, indirectness of participants selected by absence of pathology on magnetic resonance imaging, and imprecision (<300 participants for each outcome) (23). Our research group has used this tool in previously published research (24, 25).

Sensitivity analysis was also carried out to determine the effectiveness of those included trials fulfilling all 6 core components of self-management. Interventions defined as “self-management” or “self-care” were rated for the inclusion of each of the 6 core components advocated in the Australian First National Primary Health Care Strategy report (3) developing patients' 1) knowledge of the condition and management options, 2) shared responsibility for a plan of care, 3) participation in health-promoting activities, 4) self-monitoring and management of signs and symptoms, 5) management of impact on physical functioning and emotional and personal relationships, and 6) confidence in use of support services. Trials were coded independently by 2 assessors (VCO and RZP).

RESULTS

Description of included studies.

Search strategies identified 2,325 titles after removing duplicates. Screening of titles and abstracts identified 154 potentially eligible articles and 13 original trials were included (12, 13, 16–18, 26–35). The main reasons for excluding articles from the 154 potentially eligible articles were: the intervention was not defined using the terms “self-management” or “self-care” (n = 113), the condition was not nonspecific LBP (n = 10), and the study was not randomized (n = 15) (Figure 1). The 13 trials recruited a total of 3,063 participants (61% women) from the community, primary care services, and rehabilitation clinics (Table 1). Trials used written information, discussion sessions, and audiovisual resources (i.e., audiotape, videotape, or web site) as self-management strategies. The amount of support given by health care providers or lay-trained leaders varied between 1 and 13 sessions (see Supplementary Appendix B, available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658). Ten trials compared self-management to minimal interventions: usual care (13, 26, 30, 32, 34), waiting list (27), and written information (12, 16, 28–30, 33, 35). Three trials compared selfmanagement to other interventions: acupuncture (17), massage (17), exercises (18), education (31), and conventional physiotherapy (31).

Figure 1.

Flow chart of exclusion process. LBP = low back pain.

Table 1. Characteristics of included studies*
Author, year (ref.)NAge, yearsWomen, %DurationSourceInterventionOutcome measures and followup evaluationsPEDro score (0–10)
  • *

    LBP = low back pain; HMO = health maintenance organization; SM = self-management intervention; Con = control group; VAS = visual analog scale (included average pain; highest scores indicate worst pain); SIP = Sickness Impact Profile (highest scores indicate worst disabilities); RMDQ = Roland Morris Disability Questionnaire (where 0 = no disability and 23 = worst disabilities); Oswestry = Oswestry Disability Index (where 0 = no disability and 100 = worst disabilities); MVKS = modified von Korff scale (highest scores indicate worst pain/disability); GP = general practitioner.

  • Mean, mean ± SD, or mean (range).

  • Interventions used when multiple self-management interventions were compared within the same trial.

  • §

    All arms of the 3-arm trial comparing self-management to other interventions were included.

Berwick et al, 1989 (12)22234602 weeks to 6 monthsLBP patients contacted by telephone after 2 weeks of the first symptomatic encounter from automated medical record system in an HMOSM: LBP school or LBP school + reinforcement Con: LBP pamphletPain (VAS range 0–10) Disability (SIP range 0–100) Followup: 3, 6, 12, and 18 months6
Cherkin et al, 1996 (13)39142.74887% <6 weeksSuburban primary care clinic in Western WashingtonSM: booklet or booklet + reinforcement Con: usual carePain (VAS range 0–10) Disability (RMDQ range 0–23) Followup: 1, 3, and 7 weeks and 12 months6
Cherkin et al, 2001 (17)§26244.9 ± 11.558>6 weeks (61% >12- month duration)Persistent LBP individuals who visited a primary care physician identified from automated visit data 6 weeks after visit in 1997SM: self-care Con: massage or acupuncturePain (VAS range 0–10) Disability (RMDQ range 0–23) Followup: 4, 10, and 52 weeks7
Chiauzzi et al, 2010 (16)20946.1 ± 1268>3 monthsRecruited through online dissemination and pain treatment clinics in the USSM: web site Con: LBP guide e-mailedPain (VAS range 0–10) Disability (Oswestry range 0–100) Followup: 1, 3, and 6 months6
Damush et al, 2003 (26, 34)21145.5 (18–82)74<3 months (mean 2.6 months)Patients recruited from university-affiliated health centers and emergency departmentsSM: self-management + usual care Con: usual carePain (VAS range 0–10) Disability (RMDQ range 0–23) Followup: 4 and 12 months5
Haas et al, 2005 (27)12077.2 ± 7.784>3 monthsSeniors with chronic LBP from community-based programs offered at 12 locationsSM: self-management Con: waiting listPain (MVKS range 0–100) Disability (MVKS range 0–100) Followup: 6 months6
Lamb et al, 2010 (28)70154.0 ± 1560Mean pain started 13 years agoSubacute and chronic LBP patients from 56 general practices from 7 English regions in the UKSM: cognitive–behavior approach + booklet “The Back Book” Con: booklet “The Back Book”Pain (MVKS range 0–100) Disability (RMDQ range 0–23, MVKS range 0–100) Followup: 3, 6, and 12 months8
Moore et al, 2000 (29) and Strong et al, 2006 (35)22649.5 ± 1154>3 monthsPrimary chronic LBP patients enrolled in Group Health Cooperative, Western WashingtonSM: self-care Con: back pain care bookPain (VAS range 0–10) Disability (RMDQ range 0–23, interference range 0–10) Followup: 3, 6, and 12 months7
Roberts et al, 2002 (30)6439.3 ± 10.335AcuteAcute LBP patients seeking GP care, UKSM: back home leaflet + usual care Con: usual careDisability (Aberdeen LBP Scale range 0–100) Followup: 2nd day and at 2 weeks and 3, 6, and 12 months6
Schenkman et al, 2009 (31)§6143.8 ± 10.764Recurrent LBPRecurrent LBP patients from university outpatient physiotherapy clinicSM: functional movement training Con: 60-minute educational session or conventional physiotherapyPain (VAS range 0–10) Disability (Oswestry range 0–100, RMDQ range 0–23) Followup: 2, 6, and 12 months6
Sherman et al, 2005 (18)§10144 ± 1366>3 months (67% >1-year duration)Patients from Group Health Cooperative, Western WashingtonSM: self-care book Con: exercise or yogaPain (VAS range 0–10) Disability (RMDQ range 0–23) Followup: 12 and 26 weeks8
Von Korff et al, 1998 (33) and Strong et al, 2006 (35)25549.9 ± 11.362>3 monthsPrimary chronic LBP patients enrolled in Group Health Cooperative, Western WashingtonSM: self-management Con: back pain care bookPain (VAS range 0–10) Disability (RMDQ range 0–23, interference range 0–10) Followup: 3, 6, and 12 months6
Von Korff et al, 2005 (32)24049.8 ± 9.463>3 monthsPrimary chronic LBP patients enrolled in Group Health Cooperative in Western WashingtonSM: self-management Con: usual carePain (VAS range 0–10) Disability (RMDQ range 0–23) Followup: 2, 6, 12, and 24 months7

Methodologic quality.

The mean score for methodologic quality using the PEDro scale was 6.5 of 10 (range 5–8). All included studies were randomized, 39% (n = 5) (18, 27, 28, 30, 31) used concealed allocation, and in 92% of trials (n = 12) (12, 13, 16–18, 26–29, 31–34), the groups were judged to be similar at baseline. None of the included trials blinded participants or health care providers, and only 69% (n = 9) (13, 17, 18, 26, 28–32, 34) blinded the assessors. We acknowledge, however, that blinding of participants or health care providers is often not possible in trials evaluating self-management interventions. Loss to followup was inadequate in 4 trials (26, 27, 31, 34, 35), and 46% of trials (n = 6) (13, 26, 30, 31, 34, 35) did not use an intent-to-treat analysis. All of the trials reported between-group differences, point estimates, and variability.

Self-management intervention for LBP.

Self-management versus minimal intervention.

Pooled analyses of 10 original trials reported in 12 articles (12, 13, 16, 26–30, 32–35) with 2,700 participants provided weighted mean differences for pain at short- and long-term followup of −3.2 points (95% CI −5.1, −1.3) and −4.8 points (95% CI −7.1, −2.5) on a 100-point scale, respectively (Figure 2). For disability, the weighted mean differences were −2.3 points (95% CI −3.7, −1.0) and −2.1 points (95% CI −3.6, −0.6), respectively, where negative values favor self-management.

Figure 2.

Meta-analyses for all trials (self-management versus minimal intervention). 95% CI = 95% confidence interval.

According to the GRADE system used in this review to interpret the results, these trials provided moderate-quality evidence that self-management interventions have small but statistically significant effects, compared to minimal interventions, on pain and disability for LBP.

Self-management versus other interventions.

Three 3-arm trials (17, 18, 31) compared self-management to other interventions and findings are reported for each treatment comparison (Figure 3). Two trials (17, 18) reported that self-management was equally or less effective than the comparison interventions of massage, acupuncture, yoga, and exercise. The third trial (31) reported that self-management is more effective than a 60-minute educational session on pain at short-term followup and on disability at short- and long-term followup. Mean differences were −30.0 (95% CI −53.2, −6.8), −11.0 (95% CI −18.2, −4.0), and −13.1 (95% CI −23.0, −3.0), respectively, where negative values favor self-management. This trial also reported that self-management is more effective than physiotherapy on pain (mean difference −21.0; 95% CI −36.6, −5.4) and disability (mean difference −5.1; 95% CI −11.8, −1.6) at short-term followup.

Figure 3.

Effect sizes for individual comparisons (self-management versus other interventions). 95% CI = 95% confidence interval.

Based on the GRADE system, there is low-quality evidence that self-management is not superior to massage, acupuncture, yoga, and exercise in reducing pain or disability. There is also low-quality evidence that self-management is superior to a 60-minute educational session or physiotherapy in reducing pain or disability. Quality of evidence was downgraded one level from moderate quality because there was imprecision of findings (<300 participants for each outcome) (21).

Sensitivity analysis for those self-management interventions fulfilling 6 core components.

Only 8 of the 13 trials evaluated a self-management program that included all 6 core components compared to minimal intervention (Table 2). Interestingly, all 5 noncompliant trials (12, 17, 18, 30, 31) missed the criterion “shared responsibility for negotiating a plan of care” and therefore maintained the traditional dominant role of the health care provider. The 8 trials that contained the 6 core components (13, 16, 26–29, 32–35) were included in the sensitivity analyses. Weighted mean differences for pain at short- and long-term followup were −3.1 (95% CI −5.1, −1.1) and −5.7 (95% CI −8.2, −3.3), respectively, and for disability were −3.0 (95% CI −5.2, −0.8) and −4.7 (95% CI −7.5, −1.9), respectively (Figure 4). The results from the sensitivity analyses were similar to those from the main analyses.

Table 2. Interventions defined as self-management checked for 6-item criteria*
Author, year (ref.)Patient's awarenessShare responsibility negotiatingActivities to promote and protect healthPatients monitoring own symptomsPatients managing social impactPatients' confidence in use of support servicesTotal (0–6)
  • *

    Y = fulfilling item; N = not fulfilling item.

  • Low back pain (LBP) school.

  • LBP school + reinforcement. Interventions used when multiple self-management interventions were compared within the same trial.

  • §

    Booklet.

  • Booklet + reinforcement. Interventions used when multiple self-management interventions were compared within the same trial.

Berwick et al, 1989 (12)YNYNYN3
Berwick et al, 1989 (12)YNYNYY4
Cherkin et al, 1996 (13)§YNYYYY5
Cherkin et al, 1996 (13)YYYYYY6
Cherkin et al, 2001 (17)YNYYYN4
Chiauzzi et al, 2010 (16)YYYYYY6
Damush et al, 2003 (26, 34)YYYYYY6
Haas et al, 2005 (27)YYYYYY6
Lamb et al, 2010 (28)YYYYYY6
Moore et al, 2000 (29) and Strong et al, 2006 (35)YYYYYY6
Roberts et al, 2002 (30)YNYYYY5
Schenkman et al, 2009 (31)NNYYYY4
Sherman et al, 2005 (18)NNYYYY4
Von Korff et al, 1998 (33) and Strong et al, 2006 (35)YYYYYY6
Von Korff et al, 2005 (32)YYYYYY6
Figure 4.

Sensitivity analysis for trials fulfilling 6 criteria (self-management versus minimal intervention). 95% CI = 95% confidence interval.

Pooled data provided moderate-quality evidence for small effects of self-management compared to minimal intervention on pain and disability at short- and long-term followup.

DISCUSSION

Most guidelines (36, 37) emphasize the use of self-management for LBP under the assumption that these approaches are beneficial for patients with LBP. Self-management has also been the cornerstone of the Australian Government's First National Primary Health Care Strategy (3). Prognosis of persistent LBP is not favorable, with more than 50% of cases reporting pain after 12 months from onset (5), and arguably self-management would be an attractive alternative, given its potential role in reducing patients' dependence on the health care system. However, this review provides moderate evidence that the effects of self-management for nonspecific LBP (less than 6 points on a 100-point scale) are only small if compared to minimal intervention, posing the question of whether this would be a worthwhile approach.

Our findings show inconsistent results for the effect of self-management compared to other interventions such as yoga, acupuncture, and physiotherapy (17, 18, 31). Reviews have shown that the effects of self-management are similar to other more costly and intensive interventions. For instance, treatment effects of graded activity are on average <7 points on a 100-point scale compared with minimal intervention (38), and treatment effects of behavioral therapy, spinal manipulative therapy, and nonsteroidal antiinflammatory drugs are on average <10 points on a 100-point scale compared with placebo (4). Arguably, self-management does not have to present larger treatment effects than other more expensive and intensive interventions to be considered worthwhile. However, our results have shown that even when compared to minimal intervention, self-management fails to provide worthwhile effects. Our results are also consistent with those from studies on the effectiveness of self-management for other chronic diseases. For instance, small effects of self-management have been previously reported in studies on chronic obstructive pulmonary disease (39) and coronary heart disease (i.e., effects no larger than 2.6 points on a 100-point scale compared to minimal intervention) (40).

We have included in this review all trials that have explicitly defined one of the interventions as either “self-management” or “self-care.” However, more than one-third of the included trials (39%) (12, 17, 18, 30, 31) were not included in the sensitivity analysis as self-management because they were in fact delivered and/or supervised by health care providers not fulfilling the core components of self-management advocated in the Australian First National Primary Health Care Strategy report (3). A sensitivity analysis was then conducted to assess the impact of the nature of the self-management intervention on its effectiveness. The results have shown that even those interventions (13, 16, 26–29, 32–35) that have conformed to the definition used for self-management in this review (i.e., those in which patients and health care providers shared all aspects of the planning and delivery of management) also showed only trivial effects. These findings suggest that self-management has only modest effects on pain and disability for people with LBP.

One of the trials included in this review has concluded that self-management (compared to usual care) was cost effective for persistent LBP (28), when judged in terms of cost per quality-adjusted life years gained; however, the effect on pain and disability reduction was equally as small as those in the other trials included in this review (e.g., approximately 7% improvement in pain). The question naturally arises as to whether patients would be happy with an effect that small. While we do not have direct data for patient expectations for self-management, a recent survey (41) of patients with chronic LBP has shown that, on average, recipients of care need to see effect sizes (i.e., additional improvement to natural recovery) ranging from 20–30% on pain and disability with interventions such as exercise and antiinflammatory pills to consider them worthwhile. These results suggest that patients are likely to consider the effects of the current versions of self-management not worthwhile and raise questions about its endorsement in treatment guidelines for LBP.

The American College of Physicians and the American Pain Society recommend self-management for patients with LBP as a followup approach for those presenting improvements after conventional treatments (37). The guidelines also state that methods such as e-mail discussion groups, lay person–led groups, videos, and group classes have not been well investigated. Our review has, however, included moderate-quality randomized trials (16, 27, 28, 32, 33), showing that audiovisual resources via a web site providing information on self-management and group discussions have only trivial effects when compared to minimal intervention.

This is the first meta-analysis conducted on the effectiveness of self-management for nonspecific LBP. Although only 13 trials were included in our meta-analysis, our results have provided precise pooled estimates of treatment effects of self-management compared to minimal intervention, with CI limits spanning less than 5 points around the pooled estimate. Moreover, the GRADE system was used for the first time to summarize the strength of the recommendation of self-management for nonspecific LBP.

A potential limitation of the review is restricting trials to those in which at least 1 intervention was named using the terms “self-management” or “self-care.” We used this filter to reduce retrieved trials to a manageable number and because of the lack of clarity on defining self-management interventions in the literature. These inclusion criteria have been used in previous systematic reviews of self-management for other conditions such as the Cochrane review on chronic obstructive pulmonary disease (39) and systematic reviews on cancer (42) and on chronic conditions (6).

While it is clear that the approaches to self-management evaluated in the trials in this review have trivial effects, we caution that it would be premature to completely dismiss self-management of LBP. Instead, we encourage research seeking to further understand its limitations and whether its effectiveness may be increased. For instance, further investigation of self-management on additional outcomes such as self-efficacy may be important and provide a better understanding of its impact on patients' coping skills and LBP. We also need to elucidate which components of a self-management program are and are not effective, since current trials tend to be quite heterogeneous in the content of self-management programs. Qualitative research of patients and providers and trials including process measures may be informative in this regard. The mean ± SD amount of support given by health care providers in the included trials was 4.4 ± 3.5 sessions, with a duration ranging from 1 day to 12 months. Arguably, longer-duration self-management interventions would provide larger treatment effects, which could be perceived as worthwhile effects by recipients of care. More research is clearly required in this complex area if the potential of self-management is to become a reality.

There is moderate-quality evidence that self-management has small effects on pain and disability in people with LBP. Although effective when compared to minimal intervention, we are unsure if self-management provides worthwhile effects in the management of LBP. This result challenges the endorsement of self-management in treatment guidelines.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Mr. Oliveira had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Oliveira, Paulo H. Ferreira, Maher, Pinto, Refshauge, Manuela L. Ferreira.

Acquisition of data. Oliveira, Paulo H. Ferreira, Maher, Pinto, Refshauge, Manuela L. Ferreira.

Analysis and interpretation of data. Oliveira, Paulo H. Ferreira, Maher, Pinto, Refshauge, Manuela L. Ferreira.

Ancillary