Clinical Trials Do Not Use Random Samples Anymore

Abstract

Current clinical trials do not use random samples anymore. Instead, they use convenience samples from selected hospitals, including only patients with strict characteristics, like cut-off laboratory values. This practice, although it improves the precision of the treatment comparison, raises the risk of non-normal data. This is a problem since the assumption of normality underlies many statistical tests. If this assumption is not satisfied, the logic underlying the distributions of the test statistics used to estimate whether the observed effects are due to chance rather than treatment effect, fails, and, consequently, the resulting p-values are meaningless. Evidence-based medicine is under pressure due to the heterogeneity of current trials (Cleophas 2004; Furberg 2002; Kaaraininen et al. 1991; Cleophas and Cleophas 2003). The possibility of non-normal data cannot be excluded as a contributing cause for this. The current chapter reviews and describes for a non-mathematical readership methods to assess data for compliance with normality, and summarizes solutions for the analysis of non-normal data.