The quality of placebos used in randomized, controlled trials of lumbar and pelvic joint thrust manipulation-a systematic review

Background context: Spinal manipulative therapy (SMT) has been attributed with substantial non-specific effects. Accurate assessment of the non-specific effects of SMT relies on high-quality studies with low risk of bias that compare with appropriate placebos.

Purpose: This review aims to characterize the types and qualities of placebo control procedures used in controlled trials of manually applied, lumbar and pelvic (LP)-SMT, and to evaluate the assessment of subject blinding and expectations.

Study design: This is a systematic review of randomized, placebo-controlled trials.

Methods: We searched MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials, Index to Chiropractic Literature, and relevant bibliographies. We included randomized, placebo or sham-controlled trials where the index treatment was manually applied LP-SMT. There were no restrictions on the type of condition being investigated. Two independent reviewers selected the studies, assessed study quality, and extracted the data. Relevant data were the type and quality of placebo control(s) used, the assessment of blinding and expectations, and the results of those assessments.

Results: Twenty-five randomized, placebo-controlled trials were included in this review. There were 18 trials that used a sham manual SMT procedure for their placebo control intervention; the most common approach was with an SMT setup but without the application of any thrust. One small pilot study used an unequivocally indistinguishable placebo, two trials used placebos that had been validated as inert a priori, and eight trials reported on the success of subject blinding. Risk of bias was high or unclear, for all included studies.

Conclusions: Imperfect placebos are ubiquitous in clinical trials of LP-SMT, and few trials have assessed for successful subject blinding or balanced expectations of treatment success between active and control group subjects. There is thus a strong potential for unmasking of control subjects, unequal non-specific effects between active and control groups, and non-inert placebos in existing trials. Future trials should consider assessing the success of subject blinding and ensuring inertness of their placebo a priori, as a minimum standard for quality.