The effect of partial noncompliance on the power of a clinical trial

When designing a clinical trial to test the equality of survival distributions for two treatment groups, the usual assumptions are exponential survival, uniform patient entry, full compliance, and censoring only administratively at the end of the trial. Various authors have presented methods for estimation of sample size or power under these assumptions, some of which allow for an R-year accrual period with T total years of study, T greater than R. The method of Lachin (1981, Controlled Clinical Trials 2, 93-113) is extended to allow for cases where patients enter the trial in a nonuniform manner over time, patients may exit from the trial due to loss to follow-up (other than administrative), other patients may continue follow-up although failing to comply with the treatment regimen, and a stratified analysis may be planned according to one or more prognostic covariates.

In a randomized clinical trial comparing two treatments it can happen that certain patients receive the treatment other than that determined by random allocation. Significance testing is usually performed by 'intention to treat', that is, comparison of groups as determined by random allocation. The difference between the mean values of the outcome variable for these two groups estimates the difference in practical use between two treatment policies, corresponding to a pragmatic approach. An attenuation factor can then be used to obtain an estimate of the underlying difference in effectiveness between the two treatments, corresponding to an explanatory paradigm. Thus two distinct estimates are obtained; in many instances both are valid, but have distinct interpretations. Correspondingly, in planning sample size requirements when deviations from allocated treatment can be anticipated, the target treatment difference may be understood in either an explanatory or a pragmatic sense; for the sample size assessment method it is necessary to take this distinction into account.

A 5.5-fold range in breast cancer incidence rates in 21 countries shows strong correlation with national estimates of per capita intake of dietary fat, but not with other caloric sources (proteins and carbohydrates). It is argued that certain breast cancer and hormone factors may contribute little to the explanation of such international variations in incidence of this neoplasm. It is further argued that experimental studies in animals support a specific role for dietary fat in the promotion of mammary tumors, but the effects of calories alone seem to be largely restricted to tumor initiation. Finally, data from international, migrant-population, and analytic epidemiologic investigations are used to motivate the basic relative risk assumption of study designs thus far proposed for the Women's Health Trial, and some continuing motivations for a dietary intervention (low-fat diet) trial are discussed.