Noninferiority trial designs for odds ratios and risk differences

Department of Epidemiology & Biostatistics, University of California San Francisco, 185 Berry Street, Suite 5700, San Francisco, CA 94107-1762, U.S.A..
Statistics in Medicine (Impact Factor: 2.04). 01/2010; 29(9):982-93. DOI: 10.1002/sim.3846
Source: PubMed

ABSTRACT This study presents constrained maximum likelihood derivations of the design parameters of noninferiority trials for binary outcomes with the margin defined on the odds ratio (ψ) or risk-difference (δ) scale. The derivations show that, for trials in which the group-specific response rates are equal under the point-alternative hypothesis, the common response rate, π(N), is a fixed design parameter whose value lies between the control and experimental rates hypothesized at the point-null, {π(C), π(E)}. We show that setting π(N) equal to the value of π(C) that holds under H(0) underestimates the overall sample size requirement. Given {π(C), ψ} or {π(C), δ} and the type I and II error rates, or algorithm finds clinically meaningful design values of π(N), and the corresponding minimum asymptotic sample size, N=n(E)+n(C), and optimal allocation ratio, γ=n(E)/n(C). We find that optimal allocations are increasingly imbalanced as ψ increases, with γ(ψ)<1 and γ(δ)≈1/γ(ψ), and that ranges of allocation ratios map to the minimum sample size. The latter characteristic allows trialists to consider trade-offs between optimal allocation at a smaller N and a preferred allocation at a larger N. For designs with relatively large margins (e.g. ψ>2.5), trial results that are presented on both scales will differ in power, with more power lost if the study is designed on the risk-difference scale and reported on the odds ratio scale than vice versa.

  • European Journal of Cancer 09/2011; 47. DOI:10.1016/S0959-8049(11)72475-3 · 4.82 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Background: The noninferiority of a novel therapy compared with a standard of care is customarily defined by a noninferiority margin derived from an assessment of what would constitute a clinically relevant decrement in efficacy while preserving some of the treatment effect over placebo. Conundrum: If the one-sided 97.5% CI around the difference in the point estimates of efficacy between the two treatments (investigational drug minus comparator drug) does not extend below the prespecified threshold, noninferiority of the new agent to the comparator is typically concluded. In some cases, the corresponding two-sided 95% CI will fall entirely between zero and the noninferiority delta, technically implying inferiority and noninferiority concurrently. Solution: Stipulating that the upper bound of the two-sided confidence interval reach or exceed zero (as well as fall entirely above the noninferiority limit) to establish statistical noninferiority versus the comparator would avoid paradoxical interpretations.
    Future Virology 11/2012; 7(11):1055-1063. DOI:10.2217/fvl.12.101 · 1.00 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Since the introduction of protease inhibitors and their combination with two nucleoside reverse transcriptase inhibitors in tri-therapy, there has been a continuous improvement in the efficacy of antiretroviral treatments. Such combinations have been rendered even more effective by the introduction of non-nucleoside reverse transcriptase inhibitors and, more recently, integrase inhibitors. This progress has led to a move away from superiority designs towards noninferiority designs for randomized clinical trials for HIV. Noninferiority trials aim to demonstrate that a new regimen is no worse than the current standard. The methodological issues associated with such designs have been discussed, but recent HIV trials provide us with an opportunity to consider the choice of hypotheses. Recent HIV trials have been overpowered, due to the assumption of lower success rates than observed and the enrollment of a large number of patients. The use of stratified statistical methods for primary endpoint analysis, with sample size calculated by classical methods (without stratification), also increases the statistical power. Some HIV trials have a statistical power close to 99%. Surprisingly, the results of some previous studies or phase II trials are not taken into account when designing the corresponding phase III trials. We discuss alternative hypotheses and designs.
    AIDS (London, England) 06/2014; 28(13). DOI:10.1097/QAD.0000000000000369 · 6.56 Impact Factor


Available from