Development and internal validation of a machine learning prediction model for low back pain non-recovery in patients with an acute episode consulting a physiotherapist in primary care

The clinical course of acute, subacute and persistent low back pain: a systematic review and meta-analysis

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.