Comparative Effectiveness of Spinal Manipulation for Low Back Pain: A Bayesian Hierarchical Analysis Incorporating Natural History
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Improvement observed during treatment for low back pain (LBP) may reflect natural recovery, contextual response, and treatment-specific effects. We evaluated spinal manipulative therapy (SMT) and common treatment alternatives within the treatment network and against independently estimated natural history.
Methods
We conducted a Bayesian hierarchical evidence synthesis of 133 retained analytic units, analyzing acute and chronic LBP separately. Natural history was estimated from independent external natural-course/prognosis evidence, with posterior uncertainty propagated into the treatment analysis. The network meta-analysis (NMA) compared all treatment categories within each phase; independently estimated natural history provided an external reference for improvement beyond the expected clinical course. The sham category comprised placebo comparators and interventions known to be therapeutically ineffective. Recovery was defined primarily as pain ≤2/10, with alternative definitions examined in sensitivity analyses.
Results
In acute LBP, expected natural recovery was substantial. At 26 weeks, natural-history recovery was 77.3%, compared with 71.1% for SMT. SMT therefore produced −6.2 recoveries per 100 relative to natural history (95% interval −18.6 to +2.6) and +2.7 per 100 relative to sham (−5.7 to +12.5). In chronic LBP, natural-history recovery was 24.1% at 54 weeks. SMT and exercise exceeded natural history by +14.2 (+3.1 to +24.0) and +12.9 (+4.3 to +21.8) recoveries per 100, respectively, and exceeded sham by +12.9 (+2.4 to +22.5) and +11.6 (+3.4 to +20.7) per 100.
Conclusions
The apparent effectiveness of LBP treatment depends strongly on the natural course against which treatment-associated improvement is evaluated. In acute LBP, much of the observed improvement was compatible with expected natural recovery, with little evidence of additional benefit from SMT. In chronic LBP, where expected natural recovery was substantially less favorable, SMT and exercise showed evidence of benefit beyond both natural history and sham.
