Part 10 : Analysis

Rigorous principles ' shall govern the approaches to analyzing data in the Systolic Hypertension in the Elderly Program (SHEP). In particular, analyses shall adhere to the "intention-to-treat rule," that is, randomized patients shall remain in their assigned treatment group regardless of the treatment actually received. The primary end point in SHEP is the occurrence of fatal or nonfatal stroke. For each person studied, this end point is the first stroke after randomization. The definition of stroke in SHEP is described elsewhere. Therefore, implicitly, all first strokes, regardless of severity, are assigned equal weights. The log-rank test will be used to compare the rates of stroke in the active and placebo treatment arms; it has been chosen because it is the most powerful test of the difference between two treatment groups if their survival (i.e., no event has occurred) curves have proportional hazards; furthermore, the test is generally powerful even if the hazards are not quite proportional. Because the sample size is large enough to ensure that the two treatment arms are balanced with respect to prognostic variables, the primary analysis will not adjust for covariates.