A recent article in the Drug Information Journal argued that sloppy study conduct would not affect the rate of false positive conclusions in noninferiority trials. We demonstrate by means of a simulation study that this is not universally true. In fact, whether an analysis is conservative may depend on: 1. The objective of the trial, 2. The population analyzed and the mechanism for imputation of nonevaluable measurements, 3. The mechanism generating nonevaluable measurements, and 4. The selected measure for the treatment effect. We suggest that simulation studies be routinely performed to investigate the effect of these often counteracting influencing factors in practical situations.
Chuang-SteinC.Clinical equivalence—a clarification. Drug Inf J.1999;33:1189–1194.
2.
International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use.Statistical Principles for Clinical Trials. (ICH Topic E9). CPMP/ICH/363/96. Geneva, Switzerland: International Conference on Harmonization; 2001.
3.
Committee for Proprietary Medicinal Products (CPMP).Note for Guidance on Evaluation of New Anti-Bacterial Medicinal Products. CPMP/EWP/558/95. London, United Kingdom: Committee for Proprietary Medicinal Products; 1997.
4.
FarringtonCPManningG.Test statistics and sample size formulae for comparative binomial trials with null hypothesis of non-zero risk difference or non-unity relative risk. Stat Med.1990;9:1447–1454.
