Dealing With Competing Risks in Clinical Trials

News

  • Author: James F. Troendle, Eric S. Leifer and Lauren Kunz
  • Date: 26 April 2019
  • Copyright: Image copyright of Patrick Rhodes

In a paper published in Statistics in Medicine, the authors have investigated different primary efficacy analysis approaches for a 2‐armed randomized clinical trial when interest is focused on a time to event primary outcome that is subject to a competing risk. They extend the work of Friedlin and Korn (2005) by considering estimation as well as testing and by simulating the primary and competing events' times from both a cause‐specific hazards model as well as a joint subdistribution–cause‐specific hazards model. They show that the cumulative incidence function can provide useful prognostic information for a particular patient but is not advisable for the primary efficacy analysis. Instead, it is preferable to fit a Cox model for the primary event which treats the competing event as an independent censoring.

The paper is available via the link below and the authors explain their findings in further detail below:

Dealing with competing risks in clinical trials: How to choose the primary efficacy analysis?

James F. Troendle, Eric S. Leifer and Lauren Kunz

Statistics in Medicine, Volume 37, Issue 19, 30 August 2018, pages 2787-2796

thumbnail image: Dealing With Competing Risks in Clinical Trials

A time-to-event clinical trial studies whether an intervention prevents or delays a particular clinical event of interest. This event of interest is called the primary event. Sometimes there is a competing event which may occur before the primary event and which precludes the later occurrence of the primary event. For example, the Women's Health Initiative Estrogen+Progesterone and Estrogen-only trials randomized post-menopausal women to receive either hormone replacement therapy or placebo. The primary event was non-fatal heart attack or coronary death and the competing event was non-coronary death. It was hypothesized that hormone replacement therapy would be beneficial for the primary event, but would not impact the competing event.

A key question is how to statistically test whether the intervention is superior to the placebo for the primary event. Two popular methods for doing this are the cause specific hazard (CSH) and the cumulative incidence function (CIF). The CSH and CIF methods differ in how they handle a patient who experiences the competing event before the primary event. In the CSH method, such a patient is ``censored" or lost to follow-up when the competing event occurs. From the statistical standpoint, such censoring is no different than if the patient decided to no longer participate in the trial. In the CIF method, the occurrence of the competing event is explicitly accounted for.

The CIF is preferable to the CSH for estimating the probability that a future patient will experience the primary event by a certain time, such as one year from diagnosis. Indeed, since a patient who has the competing event cannot subsequently have the primary event, the competing event probability needs to be accounted for in the primary event probability. However, the CSH is preferable for determining the superiority of the intervention for the primary event relative to the placebo. This is because the intervention can appear to the CIF as superior for the primary event only because it is worse for the competing event, with more intervention patients having the competing event which precludes the primary event.

An important caveat to using the CSH is that the primary and competing events must be somewhat unrelated, such as for the Women's Health Initiative. If the two events are related, such as heart attack death and heart failure death, neither the CIF nor the CSH should be used to statistically test for superiority of the intervention. In such an instance, two options to consider are a composite event analysis or a bivariate analysis. The composite event analysis is a single statistical test of whether the intervention significantly delays the occurrence of either the primary or competing event. The bivariate analysis has two statistical tests. The first test is whether the intervention significantly delays the primary event. The second test is whether the intervention is no worse than the placebo for delaying the competing event. Both conclusions need to be established for the bivariate analysis to declare the superiority of the intervention. While the composite and bivariate analyses shift the focus away from the primary event alone, they are worth considering when an alternative is needed to the CSH.

Related Topics

Related Publications

Related Content

Site Footer

Address:

This website is provided by John Wiley & Sons Limited, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ (Company No: 00641132, VAT No: 376766987)

Published features on StatisticsViews.com are checked for statistical accuracy by a panel from the European Network for Business and Industrial Statistics (ENBIS)   to whom Wiley and StatisticsViews.com express their gratitude. This panel are: Ron Kenett, David Steinberg, Shirley Coleman, Irena Ograjenšek, Fabrizio Ruggeri, Rainer Göb, Philippe Castagliola, Xavier Tort-Martorell, Bart De Ketelaere, Antonio Pievatolo, Martina Vandebroek, Lance Mitchell, Gilbert Saporta, Helmut Waldl and Stelios Psarakis.