Objectives: Use of complementary and alternative medicine (CAM) is growing in all Western countries. The goal of this study was to evaluate quality of randomized controlled trials (RCTs) of CAM interventions for specific diagnoses to inform clinical decision making.

Methods: MEDLINE and related databases were searched for CAM RCTs. Visual review was done of bibliographies, meta-analyses, and CAM journals. Inclusion criteria for review and scoring were blinded RCT, specified diagnosis and intervention, complete study published between January 1, 1966 and July 31, 1998 in an English-language, peer-reviewed journal. Two reviewers independently scored each study.

Results: More than 5,000 trials were found, but only 258 met all study inclusion criteria. The main cause for rejection (> 90%) was that the study was not an RCT or had no blinding. Mean score across 95 diagnosis/intervention categories was 44.7 (S.D. ± 14.3) on a 100-point scale. Ordinary least-squares regression found date of publication, biostatistician as author or consultant, published in one of five widely read English-language medical journals and diagnosis/intervention category of hypertension/relaxation as significant predictors of higher scores.

Conclusions: The overall quality of evidence for CAM RCTs is poor but improving slowly over time, about the same as that of biomedicine. Thus, most services are provided without good evidence of benefit.