Clinical Trials: Superiority-Testing
One of the flaws of modern statistics is that it can produce statistically significant results even if treatment effects are very small. E.g., a sub-analysis of the SOLVD study1 found symptoms of angina pectoris in 85.3% of the patients on enalapril and in 82.5% of the patients on placebo, difference statistically significant at p<0.01. In a situation like this one has to question about the clinical relevance of the small difference. Another problem of clinical trials is that the statistics is increasingly complex, and that clinicians are at a loss to understand it. This is not, necessarily, a criticism of well-trained and hard-working doctors, but it does have a very dark side. Studies are, generally, accepted if the magic p-values are <0.05, and the disappointment about the small benefit to individual patients comes later. The problem is that a p-value of 0.05 means that the power of finding a true positive effect is only 50%, and, more important, the chance of not finding it is equally 50%. Such a result is hardly acceptable for reliable testing. The objectives of the current study were (1) to give some examples of studies that have been published as unequivocally positive studies, although the treatment effects were substantially smaller than they were expected to be, (2) to introduce superiority-testing as a novel statistical approach avoiding the risk of statistically significant but clinically irrelevant results. Superiority-testing defines a priori in the protocol clinically relevant boundaries of superiority of the new treatment. If the 95% confidence interval of the study result is entirely within the boundary, then superiority is accepted, and we do not have to worry about the p-values anymore.
KeywordsPlacebo Aspirin Oncol Warfarin Malaria
Unable to display preview. Download preview PDF.
- 10.Fetting JH, Siminoff CA, Piantadosi S et al. effects of patients' expectations of benefit with standard breast cancer adjuvant therapy on participants in clinical trials. J Oncol 1990; 8: 1476–82.Google Scholar
- 12.Anonymous. Patients' demands for prescriptions in primary care. Br Med J 1995; 310: 1084–5.Google Scholar