META-ANALYSIS: ARE the CLINICAL EFFECTS of HOMEOPATHY PLACEBO EFFECTS?

OBJECTIVE To assess whether the clinical effects of homeopathy reported in randomized, controlled trials is equivalent to that reported for placebo. DATA SOURCES All published reports of controlled clinical trials were collected using multiple sources. There was no language restriction. A previous review using an extensive search strategy identified articles in MEDL1NE and EMBASE computerized medical databases up to 1990. Both databases were also searched from 1990 to 1995 using the medical subject headings (MeSH) homeopathy, homoeopathy, and alternative medicine as were numerous homeopathic and complementary medicine registries including those from the Woodward Foundation (USA), CISCOM (London), AMED (British Library), Homlnform (Glasgow), 1DAG (Netherlands), and CCRH (India). Articles were also identified from conference proceedings, abstracts, individual collections, textbooks, and citation lists from published articles. Attempts were made to contact researchers and manufacturers in homeopathy. STUDY SELECTION Studies were selected if they were randomized, or doubleblind, placebo-controlled trials including individuals either being treated or entered into a preventive trial; if a written report existed (published, unpublished, abstract, thesis, conference proceeding, or monograph); and the report provided sufficient information for outcome rates to be calculated. DATA EXTRACTION Data were extracted on disease treated or prevented, type of homeopathy (classical, clinical, complex, isopathic), remedies used, potency (dilution) used (low, medium, high), language, population, number randomized, study quality, predefined outcome measures, and the authors' report of statistical significance. RESULTS A total of 119 trials met the inclusion criteria for data abstraction and quality assessment; 89 were amenable to metaanalysis. The mean Jadad quality score for the set of trials was 52% of the maximum. All odds ratios were calculated using intention to treat analysis. The overall odds ratio was 2.54 (95% Cl = 2.05 to 2.93) in favor of homeopathy over placebo. The odds ratio for the 26 high quality studies was 1.66 (95% Cl = 1.33 to 2.08). Multiple sensitivity analyses and subgroup analyses did not eliminate the statistical significance of the results. CONCLUSION The results of this meta-analysis revealed that the clinical effects of homeopathy are not entirely explained by the placebo effect.