STUDY OF MULTI-CENTER CLINICAL TRIALS

点击翻译按钮获取中文摘要

The Multi-Center Clinical Trial (MCCT) is an important tool in the clinical researcher's armamentarium. Multiple sites using a common treatment protocol to assess treatment safety and efficacy characterize the MCCT. If the disease of interest is relatively rare, an MCCT can enroll the desired number of sample patients in a short period of time. Even though centers follow the same clinical protocol, the set of clinical centers may be a significant source of variability. When analyzing the data, assumptions must be made regarding the behavior of the clinical centers set. Specifically, does the set of centers behave as a Fixed or Random effect? If the set of centers is regarded as random and the set number of centers is small, the effect on assessing treatment results may be seriously affected. The purpose of this study was to demonstrate how the variability of the set of clinical centers can affect the interpretation of the effect of treatment. In the first model, we varied the number of centers but held the mean and variance of the treatment effect constant. If five centers were used, treatment effect was statistically significant regardless of whether the set of centers was treated as a random or fixed effect. When the number of centers was decreased from five to two, treatment was statistically significant in the fixed effects model, but non-significant in the random effects model. In the second simulation, we varied the mean difference between treatment group response in each medical center, and treatment was significant in the fixed effects model. The treatment was non-significant in the random effects model, In the third simulation, we varied the variance of treatment group response in each medical center. In the fixed effects models, treatment was significant. In the random effects model, treatment was non-significant. Simulation studies revealed significant differences in treatment effect depending on whether the set of centers was treated as having fixed and random effects. Interpreting the results requires an understanding and justification of the assumptions guiding the data analyses.