Estimates of subgroup treatment effects in overall nonsignificant trials: To what extent should we believe in them?

In drug development, it sometimes occurs that a new drug does not demonstrate effectiveness for the full study population but appears to be beneficial in a relevant subgroup. In case the subgroup of interest was not part of a confirmatory testing strategy, the inflation of the overall type I error is substantial and therefore such a subgroup analysis finding can only be seen as exploratory at best. To support such exploratory findings, an appropriate replication of the subgroup finding should be undertaken in a new trial. We should, however, be reasonably confident in the observed treatment effect size to be able to use this estimate in a replication trial in the subpopulation of interest. We were therefore interested in evaluating the bias of the estimate of the subgroup treatment effect, after selection based on significance for the subgroup in an overall “failed” trial. Different scenarios, involving continuous as well as dichotomous outcomes, were investigated via simulation studies. It is shown that the bias associated with subgroup findings in overall nonsignificant clinical trials is on average large and varies substantially across plausible scenarios. This renders the subgroup treatment estimate from the original trial of limited value to design the replication trial. An empirical Bayesian shrinkage method is suggested to minimize this overestimation. The proposed estimator appears to offer either a good or a conservative correction to the observed subgroup treatment effect hence provides a more reliable subgroup treatment effect estimate for adequate planning of future studies.     

Some large trial properties of minimum likelihood allocation

The method of minimum likelihood allocation (MLA) for allocating subjects to treatments in a clinical trial amounts to checking at each stage which allocation would lead an outside observer to find the least evidence of a relationship between treatment and factors of prognostic significance, assuming that the observer would use a linear exponential model. One advantage of MLA is that results from game theory and likelihood theory can be used to prove it has desirable long run properties. Two of these demonstrated here are (1) ‘consistency’, in the sense that the average likelihood ratio which measures design imbalance tends to zero, and (2) ‘efficiency’ in the sense that the variance estimates of treatment effects will tend to be minimized in the long run.     

Probability computations for an extension of the geometric distribution

An extension of the stochastic process associated with the geometric distribution is presented. Combinatorial arguments are used to derive probabilities for various events of interest. Probabilities are approximated by evaluating truncated series. Bounds on the errors of approximation are developed. An example is presented and some additional applications are noted.     

Robust inference from multiple test statistics via permutations: a better alternative to the single test statistic approach for randomized trials

Formal inference in randomized clinical trials is based on controlling the type I error rate associated with a single pre‐specified statistic. The deficiency of using just one method of analysis is that it depends on assumptions that may not be met. For robust inference, we propose pre‐specifying multiple test statistics and relying on the minimum p‐value for testing the null hypothesis of no treatment effect. The null hypothesis associated with the various test statistics is that the treatment groups are indistinguishable. The critical value for hypothesis testing comes from permutation distributions. Rejection of the null hypothesis when the smallest p‐value is less than the critical value controls the type I error rate at its designated value. Even if one of the candidate test statistics has low power, the adverse effect on the power of the minimum p‐value statistic is not much. Its use is illustrated with examples. We conclude that it is better to rely on the minimum p‐value rather than a single statistic particularly when that single statistic is the logrank test, because of the cost and complexity of many survival trials. Copyright © 2013 John Wiley & Sons, Ltd.