Methodology in clinical back pain trials.

Criteria are presented to assist in the critical appraisal of investigations into the efficacy of therapy for low-back pain. Potential sources of statistical noise and experimental bias are identified in the selection of subjects, the application of therapy, and the evaluation of outcomes. When these criteria were applied to a series of recent articles, published in this journal, only two valid randomized controlled trials and four properly controlled cohort studies were found out of 147 reports.