|
Abstract:
|
One prevalent goal within clinical trials is to determine whether or not two drugs interact when applied in combination. Holt, Stamey, Seaman, and Young (2004) developed the likelihood ratio test and proposed a new bootstrap test for drug interaction, assuming that non-interaction is modeled by statistical independence. The methods discussed by Holt, Stamey and Young were specifically targeted to fixed-dose combination studies that involve binary response data. The authors provided a limited comparison of the Type I error and power of the two procedures using Monte Carlo simulation. We will present an expansion of that comparison and provide advice on choosing between the likelihood ratio and bootstrap tests in this context. We consider issues of small samples, unequal sample size, and bias.
|
