Table 1

Systematic reviews of homoeopathy with a general scope and of individualised (classical) homoeopathy

AuthorResultsAuthors' conclusions and reviewer's notesQuality assessment*
*Assessment of methodological quality: 1=selection criteria, 2=search strategy, 3=validity assessment of primary studies, 4=presentation of details of primary studies, 5=data synthesis.
RCT=randomised controlled trial.
Reviews with a general scope
Hill1240 RCTs included (1966–89). Half of the trials concluded that homoeopathy was effective, 7 concluded that results were promising but that the studies were underpowered. Of the 3 largest trials, one concluded that homoeopathy was effective while the other 2 found no statistically significant difference between homoeopathy and control.Authors' conclusions: the therapeutic value of homoeopathy cannot be considered to have been demonstrated. Reviewer's notes: the earliest general review of homoeopathy. Few details were provided relating to the search strategy used, and it is possible that this may have been more rigorous than is apparent from the paper.1=fair/poor 2=fair 3=fair 4=fair 5=fair
Kleijnen7,40107 trials met inclusion criteria; 68 were RCTs (1943–90). Of 105 trials with interpretable results, 81 had positive results in favour of homoeopathy and in 24 no positive effects of homoeopathy were found. Positive results were more likely to be found in trials with lower methodological quality.Authors' conclusions: evidence of clinical trials is positive but not sufficient to draw definitive conclusions because of low methodological quality and the unknown role of publication bias. Reviewer's notes: Most of the trials from an earlier review12 are also included in this review.1=fair 2=good 3=good 4=fair 5=fair
Linde13119 RCTs met the inclusion criteria (1966–95. Of these, 30 had inadequate information to allow statistical pooling, leaving 89 RCTs that met all inclusion criteria. The overall OR (all trials) was 2.45 (95% CI 2.05 to 2.93) in favour of homoeopathy. Further analyses showed that trials with better methodological quality were less likely to show positive results in favour of homoeopathy.41Authors' conclusions: results of meta-analysis not compatible with the hypothesis that the clinical effects of homoeopathy are completely due to placebo. Insufficient evidence was found that homoeopathy is clearly efficacious for any single clinical condition. Reviewer's notes: clinically heterogeneous data were combined and the results of the test of heterogeneity were not reported, even though this was carried out. The estimates shown should therefore be interpreted with caution.1=fair 2=good 3=good 4=fair 5=poor
Cucherat1417 comparisons in 16 RCTs met the inclusion criteria (1967–98). 11 of the 17 comparisons (65%) showed statistically significant results in favour of homoeopathy. Overall pooled p value (17 comparisons) was 0.000036. Sensitivity analyses: double blind RCTs only (n=16), p=0.000068. Double blind RCTs of highest methodological quality (n=5), p=0.082. Analysis of the likelihood of publication bias indicated that it was unlikely.Authors' conclusions: there is some evidence that homeopathic treatments are more effective than placebo; however, the strength of this evidence is low because of the low methodological quality of the trials. Studies of high methodological quality were more likely to be negative than the lower quality studies. Reviewer's notes: the pooled p values are based on clinically heterogeneous data. This method should be used and interpreted with caution since it may mask some fundamental differences between studies.1=fair 2=fair 3=fair 4=fair 5=poor
Reviews of individualised (classical) homoeopathy
Linde1532 randomised, quasi-randomised, or double blind design trials met inclusion criteria (1966–98). The methodological quality of trials was variable. 19 placebo controlled trials presented the results in sufficient detail to be included in the meta-analysis. Overall rate ratio (n=19) 1.62 (95% CI 1.17 to 2.23) in favour of homoeopathy. Sensitivity analysis: methodologically best trials (n=6): rate ratio 1.12 (95% CI 0.87 to 1.44).Authors' conclusions: results suggest that individualised homoeopathy has an effect over placebo. However, the evidence is not convincing because of methodological shortcomings of, and inconsistencies between, the trials. Reviewer's notes: the authors have pooled clinically heterogeneous data both for the overall pooling, and for the sensitivity analysis according to methodological quality. Statistical assessments of heterogeneity are not reported. The results should therefore be viewed with caution. There is some overlap between this review and the previous more general paper.13 There is a slight discrepancy between the abstract/main text and tables for the overall rate ratio figures.1=fair 2=good 3=good 4=fair 5=poor
Ernst162 double blind RCTs, 1 unblinded RCT, 3 non-randomised trials met the inclusion criteria (1978–98). 2 trials suggested that homeopathic remedies may be superior to conventional drug therapy; 2 other trials suggested that conventional drug therapy may be superior to homoeopathy. Results of the last two trials suggested no between group differences.Authors' conclusions: all of the included trials had serious methodological flaws. The value of individualised homoeopathy relative to allopathic treatments is therefore unknown. Reviewer's notes: this is the only identified review to address the comparison between homoeopathy and conventional treatments. Assessments of tests of statistical significance for between group comparisons within trials were not presented.1=fair 2=fair 3=fair/poor 4=fair/poor 5=fair