Moving Beyond the Hazard Ratio in Quantifying the Between-Group Difference in Survival Analysis

Abstract

In a longitudinal clinical study to compare two groups, the primary end point is often the time to a specific event (eg, disease progression, death). The hazard ratio estimate is routinely used to empirically quantify the between-group difference under the assumption that the ratio of the two hazard functions is approximately constant over time. When this assumption is plausible, such a ratio estimate may capture the relative difference between two survival curves. However, the clinical meaning of such a ratio estimate is difficult, if not impossible, to interpret when the underlying proportional hazards assumption is violated (ie, the hazard ratio is not constant over time). Although this issue has been studied extensively and various alternatives to the hazard ratio estimator have been discussed in the statistical literature, such crucial information does not seem to have reached the broader community of health science researchers. In this article, we summarize several critical concerns regarding this conventional practice and discuss various well-known alternatives for quantifying the underlying differences between groups with respect to a time-to-event end point. The data from three recent cancer clinical trials, which reflect a variety of scenarios, are used throughout to illustrate our discussions. When there is not sufficient information about the profile of the between-group difference at the design stage of the study, we encourage practitioners to consider a prespecified, clinically meaningful, model-free measure for quantifying the difference and to use robust estimation procedures to draw primary inferences.

The Future of Clinical Trials in Cardiovascular Medicine

“…Other novel statistical approaches have been proposed as alternatives to hazard ratios in clinical trials, including the use of restricted mean survival time, which may be a more meaningful metric of benefit. [94][95][96] Although these approaches may have some benefit in superiority trials, because they allow for the estimation of benefit in terms of time rather than reduction in risk, they may be more useful as way to design safety and noninferiority trials, and have recently been proposed as a alternative to standard approaches that can reduce sample size for a cardiovascular safety noninferiority trial, 97 which have been criticized for being too large, long, and costly. June 21, 2016…”

The Myth of Making Inferences for an Overall Treatment Efficacy with Data from Multiple Comparative Studies Via Meta-Analysis

“…With the mixture population model approach, we cannot obtain a weighted average of the study-specific hazard functions for each group due to the fact that the hazard function is not a probability. On the other hand, an alternative summary measure such as the event rate or the restricted mean survival time at a specific follow-up time point can be considered for each group [20,21]. We may then similarly obtain an estimate based on an interpretable mixture of these study-specific event rates (or restricted mean event times) across all studies to construct a group contrast measure for a target mixture population.…”

Melphalan, prednisone, and thalidomide vs melphalan, prednisone, and lenalidomide (ECOG E1A06) in untreated multiple myeloma

“…17 Group differences in time to event distributions were also measured using restricted mean survival time representing the area under the survival curves when cutting follow-up at a certain time. 18 Response evaluation was based on the International Myeloma Working Group response criteria. 19 Per-protocol response rate was adjudicated centrally, and because of a high number of unevaluable cases using perprotocol criteria, a second adjudicated response by principal investigator and coordinating center was also evaluated using International Myeloma Working Group criteria.…”