Abstract
Background: Understanding the clinical course of low back pain is essential to informing treatment recommendations and patient stratification. Our aim was to update our previous systematic review and meta-analysis to gain a better understanding of the clinical course of acute, subacute and persistent low back pain.
Methods: To update our 2012 systematic review and meta-analysis, we searched the Embase, MEDLINE and CINAHL databases from 2011 until January 2023, using our previous search strategy. We included prospective inception cohort studies if they reported on participants with acute (< 6 wk), subacute (6 to less than 12 wk) or persistent (12 to less than 52 wk) nonspecific low back pain at study entry. Primary outcome measures included pain and disability (0–100 scale). We assessed risk of bias of included studies using a modified tool and assessed the level of confidence in pooled estimates using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) tool. We used a mixed model design to calculate pooled estimates (mean, 95% confidence interval [CI]) of pain and disability at 0, 6, 12, 26 and 52 weeks. We treated time in 2 ways: time since study entry (inception time uncorrected) and time since pain onset (inception time corrected). We transformed the latter by adding the mean inception time to the time of study entry.
Results: We included 95 studies, with 60 separate cohorts in the systematic review (n = 17 974) and 47 cohorts (n = 9224) in the meta-analysis. Risk of bias of included studies was variable, with poor study attrition and follow-up, and most studies did not select participants as consecutive cases. For the acute pain cohort, the estimated mean pain score with inception time uncorrected was 56 (95% CI 49–62) at baseline, 26 (95% CI 21–31) at 6 weeks, 22 (95% CI 18–26) at 26 weeks and 21 (95% CI 17–25) at 52 weeks (moderate-certainty evidence). For the subacute pain cohort, the mean pain score was 63 (95% CI 55–71) at baseline, 29 (95% CI 22–37) at 6 weeks, 29 (95% CI 22–36) at 26 weeks and 31 (95% 23–39) at 52 weeks (moderate-certainty evidence). For the persistent pain cohort, the mean pain score was 56 (95% CI 37–74) at baseline, 48 (95% CI 32–64) at 6 weeks, 43 (95% CI 29–57) at 26 weeks and 40 (95% CI 27–54) at 52 weeks (very low–certainty evidence). The clinical course of disability was slightly more favourable than the clinical course of pain.
Interpretation: Participants with acute and subacute low back pain had substantial improvements in levels of pain and disability within the first 6 weeks ( moderate-certainty evidence); however, participants with persistent low back pain had high levels of pain and disability with minimal improvements over time (very low–certainty evidence). Identifying and escalating care in individuals with subacute low back pain who are recovering slowly could be a focus of intervention to reduce the likelihood of transition into persistent low back pain.
Protocol registration: PROSPERO — CRD42020207442
Low back pain is a leading cause of disability worldwide,1,2 affecting about 570 million people globally. About 39% of the adult population will have low back pain in any given year.3 Low back pain is costly; in the United States, health care spending on low back pain was $134.5 billion annually between 1996 and 20164 and is increasing.5 Clinical guidelines recommend triage to identify symptoms that require diagnostic investigation (prevalence of serious pathology < 1% in primary care).6 Nonspecific low back pain, where no specific pathoanatomical cause can be identified, is the most common type of low back pain.7 Management focuses on reducing pain and its consequences through education and reassurance, nonpharmacological treatments (e.g., heat, relative rest, staying active), analgesic medicines (e.g., nonsteroidal anti-inflammatory drugs) and timely review.6
Existing literature suggests that the clinical course of an episode of low back pain is favourable.8,9 However, recurrence is common (about 69% of patients will experience recurrence within 12 months)10 and pain persists for many patients.11 Several studies have shown that acute low back pain is not always associated with a favourable outcome,7 including 2 systematic reviews that showed that, although many patients recover within the first month, low levels of pain and disability often persist.12,13
In 2012, we conducted a meta-analysis (n = 11 166 participants) and concluded that patients with acute or persistent low back pain usually had positive trajectories, with most showing substantial improvement in pain and disability within the first 6 weeks.8 We acknowledged a critical limitation, however, that those with subacute low back pain (i.e., duration 6–12 wk at study entry) were included in the group with persistent low back pain, who typically had been experiencing back pain for more than 12 weeks at study entry. This may have resulted in improved outcomes in the persistent group. In addition, our review did not consider different populations (e.g., by age of cohorts) or confounding factors (e.g., presence of radiculopathy or radicular pain), assess the certainty of the evidence or assess study attrition when assessing risk of bias.
A better understanding of the clinical course of low back pain across various pain durations and populations is important in the early detection of slow recovery and escalation of care at the patient level, and in verifying that recommendations in clinical guidelines to reassure patients about a favourable prognosis are appropriate.14–24 We updated our previous meta-analysis seeking to understand the clinical course of acute (< 6 wk), subacute (6 to less than 12 wk) and persistent (12 to less than 52 wk) low back pain, taking into account age and neuropathic spine-related leg pain.8
Methods
We conducted a systematic review and meta-analysis. The search strategy, data extraction and methodological quality assessment was identical to that of our previous publication,8 except for an additional methodological quality assessment item and a variation to the data analysis (outlined below).
The protocol was prospectively registered with PROSPERO (CRD42020207442). All deviations from that protocol are noted and reporting is in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.25
Data sources and searches
The search strategy from the previous review underwent a Peer Review of Electronic Search Strategies (PRESS) review by an experienced university librarian to confirm that the databases and search terms on each platform were still appropriate.26 With no revisions required, we searched the MEDLINE, Embase and CINAHL databases from 2011 — overlapping with our previous review’s search window by approximately 1 year — until Jan. 16, 2023. We sought cohort studies published in the peer-reviewed literature with no language or age restrictions applied, and searched the reference lists of included studies for additional studies. The search strategy can be found in Appendix 1, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content.
Study selection
Study selection criteria were identical to those of the previous review. We considered prospective inception cohort studies for inclusion if they had a well-defined inception cohort and included participants with acute, subacute or persistent nonspecific low back pain. We defined low back pain as pain or discomfort localized below the costal margins and above the inferior gluteal folds, with or without leg pain. Nonspecific low back pain is a distinct classification of low back pain, where there is no specific identified cause for the low back pain (e.g., cancer, infection, fracture).7
To be included, studies needed to report outcome measures for pain intensity (e.g., a visual analogue scale) or disability (or a self-assessment of physical functioning, such as the Roland Morris Disability Questionnaire), or a global measure of recovery (as per each study’s definition of recovery). These criteria are aligned with core outcome measures for clinical trials of nonspecific low back pain.27,28 Our primary outcomes were pain and disability, while global measure of recovery was a secondary outcome.
We excluded studies reporting participants with low back pain for greater than 12 months as we were interested in capturing the trajectory of low back pain within the first year of pain onset, during which care escalation might be most useful.13 We also excluded retrospective cohorts, experimental or interventional studies, any cohort that was unlikely to follow the normal trajectory (e.g., pregnancy), mixed populations (e.g., studies of patients with either neck pain or low back pain, unless participants with low back pain could be easily differentiated), participants with low back pain specifically recruited with other comorbidities (e.g., osteoarthritis), studies that reported only baseline outcomes or prognostic factors without longitudinal follow-up and studies specifically recruiting only participants with both low back pain and leg pain.
We included studies with participants who had poorly differentiated leg pain along with low back pain, so long as these participants were not specifically sought after, because most studies in our previous review did not exclude those participants. Furthermore, the terminology used to describe spine-related leg pain is inconsistently reported in the literature.29
A team of authors (S.B.W., F.A.B., M.O., M.J.T., B.L. and S.J.S.) reviewed titles and abstracts to exclude irrelevant studies. This screening was conducted independently and in duplicate using Covidence. Reviewer pairs resolved conflicts through discussion. For all studies that passed this first screening, we obtained the full-text article and assessed it for inclusion. The team reviewed full-text studies independently and in duplicate but reviewer pairs were modified. Disagreements were resolved by consensus, or a third reviewer from within the team was consulted to make the final decision. We pooled studies that met the inclusion criteria with the studies from our previous review.
Data extraction
Two reviewers (S.B.W. and one of F.A.B., M.O., M.J.T., B.L. and S.J.S.) extracted data independently and in duplicate using a custom spreadsheet identical to that used in our previous review (Appendix 2, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content). Reviewers resolved discrepancies by consensus. We sought missing data by contacting the corresponding author of each study, with up to 3 follow-up emails if no response was received.
Risk of bias and certainty of evidence
We are unaware of any risk of bias tools designed specifically for systematic reviews of the clinical course of a health condition. Therefore, we adapted the risk of bias assessment tool that we used in our original review,8 itself an adaptation of the methodological criteria outlined by Altman.30 This allowed us to assess bias domains that are specific to clinical course studies, including sampling (2 items), completeness of follow-up (2 items) and outcome reporting (1 item).
For this study, we added an extra measure (study attrition) in our risk of bias assessment, taken from the Quality In Prognosis Studies (QUIPS) tool.31 This addition was a deviation from our registered protocol and was done because we considered it important to capture whether the prognosis outcomes at follow-up were representative of the baseline sample. We applied this additional measure retrospectively to all studies in our previous review so that we captured all 6 measures of bias across all included studies (Appendix 3, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content). Two reviewers (S.B.W., and one of F.A.B., M.O., M.J.T., B.L. and S.J.S.) conducted the risk of bias assessment independently and in duplicate. Discrepancies were resolved by consensus.
To assess the level of confidence in pooled estimates of means and 95% confidence intervals (CIs) of pain and disability scores, we used the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach in accordance with adaptations for prognostic research.32 Our previous review did not include a GRADE assessment and its addition was a deviation from our registered protocol.
Two reviewers (S.B.W. and F.A.B.) conducted the GRADE assessment independently and in duplicate for each outcome measure (pain and disability) for each group (acute, subacute and persistent). Discrepancies were resolved by consensus with a third reviewer (G.L.M.). Appendix 4, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content, provides the GRADE assessment criteria.
Statistical analysis
We conducted meta-analyses on aggregate data, where possible, using pain and disability outcome data. In the absence of mean and standard deviation (SD) in the included studies, we used the median, interquartile range, range and sample size to estimate them where possible, following Wan and colleagues’ methodology.33 We converted pain and disability outcomes to a common 0–100 scale by subtracting the minimum scale value from the score, and dividing this amount by the difference between the maximum and minimum scale value, multiplying the overall result by 100; SDs were increased proportionally.
We classified cohorts into groups by acute (< 6 wk), subacute (6 to less than 12 wk) and persistent low back pain (12 to less than 52 wk), based on the duration of pain (mean or median) at study entry. We classified persistent low back pain as pain for at least 12 weeks (but less than 12 mo), which is well established among clinical guidelines.14 We classified acute and subacute low back pain to be consistent with the definition of acute in the previous review (< 6 wk), and to align with most clinical guidelines.14 The inclusion of a subacute group was a deviation from our registered protocol.
Consistent with our previous review, when means and medians of pain duration were not available, we used the midpoint of the inception time range. Where a cohort spanned multiple groups (e.g., 2–12 mo, both subacute and persistent), we contacted the authors to request that their data be split according to our classifications. Where this was not possible, we used the mean, median or midpoint of the inception time to categorize that cohort.
We modelled pain and disability outcomes as a function of time using Stata 18 software (StataCorp), captured in 2 ways. The first approach was with inception time uncorrected, which captured time since entry into the study. The second approach was with inception time corrected, which captured time since pain onset, calculated by adding the mean or median inception time, with time modelled as a continuous variable.
We fitted a separate meta-regression model to the data for each outcome by patient cohort combination. To account for the nonlinear relationship between pain, disability and time, we used fractional polynomial transformations, which allowed the shape of the fitted line to vary by including functions of time (commonly 2) as predictors of the outcome (Appendix 5, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content). We scaled time (uncorrected) for the acute cohort by adding 1 week to time. We did not use scaling for the subacute and persistent cohorts with inception corrected because time was sufficiently greater than zero to not hinder estimation.
We included random effects for study and for the transformed time coefficients in all models to account for dependence in repeated observations and differences in time course between studies. The random effects were allowed to correlate (using an unstructured variance–covariance matrix) if this further improved model fit and achieved convergence. We chose the best fitting models using the smallest Bayesian Information Criteria closest to 0. We inspected residuals to assess goodness of fit. We weighted means using the inverse of the sum of the study sampling variances extracted for each time point and estimates of residual variance, as is standard in the meta meregress program in Stata 18 (Appendix 6, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content).
We used the regression model to obtain pooled estimates for means and 95% CIs of pain and disability scores at 0, 6, 12, 26 and 52 weeks, although these were not computed if it required extrapolation from the observed data.
To compare the trajectories of pain and disability, we modelled both outcomes simultaneously for all 3 cohorts using the same modelling process except that we included the fixed effect of outcome (pain or disability) in the model and allowed it to interact with time. We considered an interaction of p less than 0.05 as evidence of differing trajectories between the 2 outcome measures.
Because of the heterogeneity we observed on the definition of recovery between studies, we presented the global recovery outcomes (our secondary outcome) for each study descriptively.
Sensitivity analysis
In a deviation from our previous review and registered protocol, we undertook several sensitivity analyses. Because of the high risk of selection bias, we planned sensitivity analyses (for each group of participants with acute, subacute or persistent low back pain) to include only studies that had a high follow-up rate (> 80% of participants).34,35 As it is unclear whether the clinical course of low back pain differs among people with or without radicular pain or radiculopathy, we conducted sensitivity analyses to include only studies that specifically excluded people with radicular pain or radiculopathy. Finally, because of lack of representation of younger (< 18 yr) and older (> 60 yr) participants, we conducted sensitivity analyses to include only studies reporting on participants aged 18–60 years. The modelling process for these subgroups was identical to that of the primary analyses. We considered only the acute and subacute cohorts, given insufficient studies involving cohorts with persistent pain.
Ethics approval
As this study was a systematic review and meta-analysis of published studies, ethics approval was not required.
Results
Our search yielded 28 641 articles and we removed 4891 duplicates. After we screened 23 695 titles and abstracts and 377 full papers, 54 studies met the inclusion criteria (Figure 1). Reference checking yielded 1 additional study but the cohort itself was already included.36 Three studies37–39 were identified in both the original review and the current review (because of overlap in search period); these data were not re-extracted. We extracted data from 52 new articles.36,40–90
Preferred Reporting Items for Systematic Reviews and Meta-Analyses ( PRISMA) flow diagram outlining selection of studies for inclusion. Note: LBP = low back pain.
We included 27 new cohorts in the updated review and 23 new (and 1 updated) cohorts in the pooled analyses (Appendix 7, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content). Once pooled with articles from the previous review, we included 95 articles36–130 in this review, reporting on 60 cohorts (17 974 participants). The pooled analyses include data from 47 cohorts (9224 participants with pain assessment and 8957 participants with disability assessment) (Table 1). Global recovery outcomes were reported in 37 cohorts (13 145 participants).
Characteristics of included studies
Risk of bias and certainty of evidence
The overall risk of bias of included cohorts was variable (Appendix 8, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content). The criterion regarding study attrition was poor; most studies (n = 47, 82.4%) did not report on participants lost to follow-up or, if reported, the participants who were available at follow-up were not representative of the initial study sample. Most studies (n = 31, 54.4%) did not select participants as consecutive cases and participant follow-up rate was poor.
The overall level of confidence in pooled estimates for mean pain and disability scores (and 95% CIs) in both the acute and subacute groups was rated as moderate certainty (GRADE level 3) and for the persistent group as very low certainty (GRADE level 1) (Table 2 and Appendix 9, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content).
Summary of Grading of Recommendations, Assessment, Development and Evaluation (GRADE) assessment32
Clinical course of pain and disability
Pain
Pain outcomes were reported for 43 cohorts, included in the meta-analysis. We classified 31 cohorts as having acute low back pain, 10 cohorts as having subacute low back pain and 3 cohorts as having persistent low back pain. The trajectories of pain for all 3 cohorts are shown in Figure 2 and pooled estimates of mean pain scores and 95% CIs are shown in Table 3.
Trajectory of pain among patients with acute (< 6 wk) (A, D), subacute (6 to less than 12 wk) (B, E) or persistent (12 to less than 52 wk) (C, F) low back pain. The top row of graphs shows pain trajectory for inception time corrected, where time is captured as the time since the onset of low back pain. The bottom row of graphs shows pain trajectory for inception time uncorrected, where time reflects the time since study entry. Green boxes represent the observed mean pain score for each study at each time point. Shaded area represents 95% confidence intervals (CIs).
Pooled estimates of mean pain and disability scores*
Trajectories for the acute low back pain group showed a large reduction in pain initially (baseline to 6 wk: mean reduction 30 out of 100), after which time low-intensity pain persisted (6–26 wk: mean reduction 4 out of 100; 26–52 wk: mean reduction 1 out of 100, with a mean pain score of 21 out of 100 at 52 wk).
The subacute group showed a similar, but less favourable trend to the acute group, with a smaller reduction in pain scores (baseline to 6 wk: mean reduction 34 out of 100; 6–26 wk: no change in mean scores; 26–52 wk: mean reduction 2 out of 100, with a mean pain score of 31 out of 100 at 52 wk).
The persistent group showed greater variability in trajectory and consistently high pain intensity over time (baseline to 6 wk: mean reduction 8 out of 100; 6–26 wk: mean reduction 5 out of 100; 26–52 wk: mean reduction 3 out of 100, with a mean pain score of 40 out of 100 at 52 wk).
Disability
Disability outcomes were reported for 43 cohorts, included in the meta-analysis. We classified 31 cohorts as having acute low back pain, 9 as having subacute low back pain and 4 as having persistent low back pain. The trajectories of disability for all 3 cohorts are shown in Figure 3 and pooled estimates of mean disability scores and 95% CIs are shown in Table 3.
Trajectory of disability in patients with acute (< 6 wk) (A, D), subacute (6 to less than 12 wk) (B, E) and persistent (12 to less than 52 wk) (C, F) low back pain. The top row of graphs shows disability trajectory for inception time corrected, where time is captured as the time since the onset of low back pain. The bottom row of graphs shows disability trajectory for inception time uncorrected, where time reflects the time since study entry. Green boxes represent the observed mean pain score for each study at each time point. Shaded area represents 95% confidence intervals (CIs).
Trajectories for the acute low back pain group showed a large reduction in disability initially (baseline to 6 wk: mean reduction 21 out of 100), after which time low disability persisted (6–26 wk: mean reduction 4 out of 100; 26–52 wk: mean reduction 1 out of 100, with a mean disability score of 17 out of 100 at 52 wk).
The subacute group showed a similar, but less favourable, trend to the acute group, with a smaller reduction in disability scores (baseline to 6 wk: mean reduction 15 out of 100; 6–26 wk: mean reduction 4 out of 100; 26–52 wk: no change in mean scores, with a mean disability score of 25 out of 100 at 52 wk).
The persistent group showed greater variability in trajectory and persisting moderate disability over time (baseline to 6 wk: mean increase 1 out of 100; 6–26 wk: mean increase 1 out of 100; 26–52 wk: no change in mean disability scores, with a mean disability score of 59 out of 100 at 52 wk).
Appendix 10, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content, provides the full table of fitted model results and Appendix 11, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content, provides the spaghetti plots showing pain and disability outcomes for each patient group.
Sensitivity analyses
Because of the small number of included studies reporting high follow-up (> 80% of participants), we had insufficient data points for sensitivity analyses of studies with a high follow-up rate. We were able to complete the remaining 2 sensitivity analyses (exclusion of participants with radiculopathy or radicular pain, and inclusion of participants aged 18–60 yr) in the acute and subacute cohorts. These analyses were not possible for the persistent group.
When we included only studies that specifically excluded radicular pain or radiculopathy, those in both the acute and subacute groups followed a similar pain and disability trajectory to the main analyses (Table 4 and Appendix 12, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content). When we included only studies with participants aged 18–60 years, trajectories for pain and disability also followed a similar trajectory to the main analyses.
Pooled estimates of mean pain and disability scores when excluding cohorts with radiculopathy or radicular pain, and when excluding cohorts with participants younger than 18 years and older than 60 years
Comparison between pain and disability
Among participants with acute low back pain, the course of disability was more favourable than that of pain (uncorrected and corrected p < 0.001). Among those with subacute low back pain, the courses of pain and disability also differed (uncorrected and corrected p < 0.001) with pain tending to increase over time, although the number of observations beyond 26 weeks was small. Among those with persistent low back pain, the course of disability was more favourable than that of pain for the uncorrected time only (uncorrected p < 0.005, corrected p = 0.811) (Appendix 13, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content).
Recovery
Outcomes for recovery from low back pain were reported for 37 cohorts. Studies showed large heterogeneity with regard to definition of recovery and follow-up times. Trajectories of recovery in these cohorts appeared to generally align with the findings of the meta-analyses for pain and disability. Data on recovery rates are reported in Appendix 14, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.230542/tab-related-content.
Interpretation
We found moderate-certainty evidence that the clinical course of low back pain was most favourable in the group with acute low back pain, with a large reduction in pain and disability within the first 6 weeks. After this time, recovery slowed. Moderate-certainty evidence indicated that recovery for those with subacute low back pain also showed significant (but smaller) reductions in pain and disability over the first 6 weeks, with subsequently slowed improvement. The clinical course in the persistent pain group was much less favourable than the other groups, with very low–certainty evidence for minor improvements in pain and disability over time.
To obtain a more precise picture of the clinical course of low back pain, we included a subacute group, which filled a gap in our previous meta-analysis and has important implications.8 In 2012, we had reported that participants with acute and persistent low back pain showed marked improvements in pain and disability within the first 6 weeks, a finding that has guided clinical guidelines internationally.8,14 In separating a subacute group from those with persistent pain, our current analyses suggest that the outcomes for those with persistent low back pain are substantially less favourable than previously thought.
Our findings are consistent with similar work. In a community cohort, De Campos and colleagues131 found that a subsequent new episode of acute low back pain resolved rapidly (about 5 d) among participants who had recently recovered from low back pain (including participants enrolled in a randomized controlled trial to prevent recurrence of low back pain). A systematic review of randomized controlled trials reported a similar pattern of improvement within the first 6 weeks and less pronounced improvement at longer-term follow-up (1 yr).132 This was also shown in a systematic review of emergency department visits among patients with acute low back pain, which reported that most people had an initial reduction in pain intensity, after which symptoms stabilized and persisted up to 6 months later.133 In addition, a 2012 systematic review (search strategy 1990–2010) of cohort studies involving peole visiting primary care for low back pain (< 3 mo) cautioned that, while an initial reduction in pain was common, many still reported pain at longer-term follow-up (65% reported pain at 1 yr), suggesting that intensive follow-up should be directed toward those who have not recovered within 3 months.13 Several studies included in that systematic review were also included in the current meta-analysis.
In our main meta-analysis of cohorts with acute low back pain, 1 study that specifically sought older patients (≥ 60 yr) was an outlier, showing much poorer pain and disability outcomes than the main trajectory trend.74 Most studies excluded participants in this age group, reporting mean ages between 30 and 50 years. However, our sensitivity analyses showed that excluding studies with participants who had radicular pain or radiculopathy, or studies with older (> 60 yr) and younger participants (< 18 yr) did not appear to affect the clinical course. Previous work has shown that people aged 60 years and older are more likely to have disabling and persistent episodes of low back pain than younger people.115,134,135 We had insufficient data in the older age group to support or refute these claims. Furthermore, we found only 2 studies that included people younger than 18 years.96,115 Both recruited people 14 years and older, and reported mean ages of 43.3 and 44.1 years. Therefore, we cannot generalize our findings to populations older than 60 years or younger than 18 years, which identifies a critical evidence gap in the field.
For people with acute, subacute or persistent low back pain, the course of disability was slightly more favourable than that of pain. This finding aligns with key objectives of psychological interventions for persistent low back pain, which aim primarily to improve functioning and quality of life, rather than pain.136
The current understanding that most individuals with a new episode of low back pain get better within 2 weeks may need reconsideration.14 Although most people with acute and subacute low back pain do see improvements early on, our updated meta-analysis shows that many continue to experience ongoing pain and disability. Some benefit may be garnered by providing positive expectations about recovery, but people may also benefit from more realistic expectations.137 Advice might be best focused on the likelihood of symptom recurrence and acknowledging that ongoing symptoms do not necessarily reflect serious pathology.
Our findings also support the need for timely reassessment within the first 12 weeks after an episode of low back pain to identify and escalate care among those recovering slowly. For people who have pain that persists for 12 weeks or more, pain and disability remain high, which highlights the importance of developing better treatments for this group.
Further work is required to increase the certainty of evidence regarding the clinical course of persistent low back pain. The precision and value of clinical course studies for low back pain may be enhanced by pooling trajectories of individual patients with low back pain (i.e., meta-analyses of individual patient data), and by evaluating the stability of trajectories.138 A better understanding of the clinical trajectories of low back pain among older (> 60 yr) and younger (< 18 yr) populations is needed, and more effective treatments for those with persistent low back pain should be developed. A risk-of-bias tool specific to clinical course studies in low back pain is another important gap.
Limitations
Potential limitations of this work include using our own risk-of-bias tool (similar to other clinical course reviews that have used their own tool or modified other tools13,133), which means that all measures of bias may not have been adequately captured. We found variable risk of bias across studies, with overall high attrition and participants in most studies were not recruited as consecutive cases. Unfortunately, we were unable to assess the impact of poor study quality through sensitivity analyses because of low study numbers and, as such, the impact of low-quality studies on the validity of our results is unclear.
In our analyses, to capture time since pain onset, we added mean (or median) inception time to time of study entry. Median time since pain onset is likely smaller than the mean, which may have implications whereby the corrected inception time course may appear more favourable than it really is.
We excluded interventional studies, which could have contributed to our understanding of the clinical course of low back pain. Having said this, the trajectory of low back pain appears to be similar in randomized controlled trials and cohort studies. Finally, definitions of acute low back pain vary between 4 and 12 weeks across guidelines.14 A different definition of acute low back pain may have yielded different results, although the similar trajectories in the acute and subacute groups here suggest otherwise.
Conclusion
Most people with acute and subacute low back pain begin to improve within the first 6 weeks, but many have ongoing pain and disability. Importantly, and in contrast to our previous review, people with persistent low back pain (≥ 12 wk) have ongoing moderate-to-high levels of pain and disability. Identifying and escalating care among people with subacute low back pain who are recovering slowly seems a critical focus of intervention.
Footnotes
Competing interests: Sarah Wallwork has received speaker fees from the Neuro Orthopaedic Institute and Exercise and Sport Science Australia. Felicity Braithwaite has received speaker fees from the Neuro Orthopaedic Institute, San Diego Pain Summit, Medicine Education Science and Health (MESH) . Mary O’Keeffe reports consulting fees from the European Pain Federation EFIC, honoraria from the Physio Network and travel support from INTERACT-EUROPE, PANACEA and Happy. Mervyn Travers has received speaker fees from Aalborg University, Australian Physiotherapy Association, Australian Podiatry Association, Life Ready Physiotherapy, Physio-Formation, Smart Education, O2 Academy, Healthia, Queensland Health, The University of Western Australia, Hemophilia Association Australia, Icelandic Physiotherapy Association and Rural Health West. Belinda Lange is a nonexecutive director of the Australian Physiotherapy Council and the Australasian Institute of Digital Health. G. Lorimer Moseley has received support from the National Health and Medical Research Council of Australia, the Medical Research Future Fund, Reality Health, Connect Health UK, Institutes of Health California, American International Assurance Australia, Workers’ Compensation Boards and professional sporting organisations (International Olympic Committee, English Premier League, La Liga, Australian Football League Clubs, National Basketball League, Athletics Australia, Cricket Australia). Professional and scientific bodies have reimbursed him for travel costs related to presentations of research on pain. He has received speaker fees for lectures on pain and rehabilitation from the Royal Australian College of General Practitioners, Noigroup Australia and MasterSessions. He receives royalties from Noigroup Publications, Orthopedic Physical Therapy Products and Dancing Giraffe Press. He sits on boards with Painaustralia and National Pain Solutions Research Alliance, and is chief executive officer with Pain Revolution. No other competing interests were declared.
This article has been peer reviewed.
Contributors: Sarah Wallwork, Leonardo Costa, Luciola da C. Menezes Costa and G. Lorimer Moseley conceived and designed the study. Sarah Wallwork, Felicity Braithwaite, Mary O’Keeffe, Mervyn Travers, Simon Summers and Belinda Lange acquired the data. Sarah Wallwork, Belinda Chiera and Dana Hince carried out and interpreted the statistical analysis. Sarah Wallwork, Mary O’Keeffe, Dana Hince and G. Lorimer Moseley drafted the manuscript. All of the authors revised the manuscript for critically important intellectual content, approved the final version of the manuscript and agreed to be accountable for all aspects of the work.
Funding: Sarah Wallwork, G. Lorimer Moseley and project costs were supported by an National Health and Medical Research Council Leadership Investigator Grant to G. Lorimer Moseley (ID1178444). Felicity Braithwaite is supported by the John Stuart Colville Fellowship (Arthritis Foundation of South Australia). Mary O’Keeffe is supported by the Marie Curie Fellowship from the European Commission. Leonardo Costa and Luciola da C. Menezes Costa are supported by CNPq research productivity fellowships.
Data sharing: All data from this meta-analysis are available on request from the corresponding author.
- Accepted November 16, 2023.
This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY-NC-ND 4.0) licence, which permits use, distribution and reproduction in any medium, provided that the original publication is properly cited, the use is noncommercial (i.e., research or educational use), and no modifications or adaptations are made. See: https://creativecommons.org/licenses/by-nc-nd/4.0/
References
In this issue
Article tools
Jump to section
Related Articles
Cited By...
- No citing articles found.
More in this TOC Section
Similar Articles
