We propose a new treatment allocation method that assures similar means of continuous prognostic variables between treatment groups in randomized clinical trials. This method is an extension of the minimization method proposed by Pocock and Simon to continuous variables case. There are several measures to be minimized simultaneously, and we make a linear combination comprised of these weighted measures. The sizes of these weights of measures must be determined before implementation of this method. We give some examples to illustrate the achievement of this method and a guideline for the weights.
Many maximum contrast methods have been used to evaluate the dose-response relationship in clinical trials and in toxicological trials. However, the calculation method for p-value and power of such methods in practical form have not been proposed due to the complex property of integration in the multivariate t-distribution with singular correlation matrix. Addressing this issue, this paper presents SAS/IML programs for computing p-value, power and sample size necessary in the application of maximum contrast methods. It was confirmed that the speed of calculation in the proposed programs was far superior to the naive Monte-Carlo simulation keeping the precision as well.
Despite the fact that the left-point imputation for the interval-censored data produces shorter individual survival time than the right-point imputation, we observe that the Kaplan-Meier estimator based on the left-point imputation can sometimes produce greater survival rate than the right-point imputation during certain period of time interval. In practice, if the exact date of occurrence is not known due to interval-censoring, the date of examination when the event was observed is most often recorded as though the exact time of event (right-point imputation). Therefore, we investigated behavior of the deterministic imputations for the Kaplan-Meier estimator and compared with the Turnbull estimator (1976), a standard analysis of interval-censored data, by simulation. The mid-point imputation is shown to have smallest mean squared errors at most range of time under moderate to small censoring for all hazard rates investigated here, although ad hoc estimators of SE underestimate true SE. Under heavy censoring, either the right-, mid-or left-point imputations are shown to have smallest mean squared errors over certain periods. The right-point imputation (KMR) may be acceptable with large percent right-censored under increasing hazard. On the other hand, the Turnbull estimator (TB) is usually not shown to have smallest mean squared errors. We find a systematic bias of TB at tail similar to that of KMR.
In randomized clinical trials, last observation or change from baseline (last observation-baseline) is frequently adopted as an endpoint. In two-arm comparison trials in Japan, a two-sample t-test is often employed in the primary analysis whereas an analysis of covariance (ANCOVA) with baseline covariate assuming equal slopes between the groups is seldom employed. This paper discusses the adoption of the ANCOVA as the primary analysis and recommends its use provided that the assumptions for ANCOVA are satisfied. First, the powers of the t-test and the ANCOVA are presented in order to investigate the extent of how the ANCOVA can increase power. Next, several reasons for not employing the ANCOVA are discussed. Percentage change is also briefly discussed. The power of the ANCOVA with baseline covariate is dependent on the correlation coefficient between baseline and last observation, and on the magnitude relation of variances of baseline and last observation. This paper presents several factors that are responsible for reducing the correlation coefficient and several factors which influence the magnitude relation.