Interpretation of Composite Endpoints in Clinical Trials

Please let me know if these references/sites are helpful to you:

1.

Chapter 7 Introduction to Composite Endpoints

https://sph.uth.edu/courses/biometry/lmoye/Webs/PH1835...

A combined or composite endpoint in a clinical trial is a clinically relevant. endpoint that is constructed from combinations of other clinically relevant. endpoints, termed component endpoints or singleton endpoints. Two examples of. singleton endpoints are 1) the …

2. Definition, reporting, and interpretation of composite

www.bmj.com/content/341/bmj.c3920

Design Systematic review of parallel group randomised clinical trials published in ... and interpreting composite endpoints are ... the word about The BMJ. Published in: BMJ · 2010 Authors: Gloria Cordoba · Lisa M Schwartz · Steven Woloshin · Harold T Bae · Peter C Gotzsche Affiliation: Cochrane Collaboration · Dartmouth College About: Clinical trial · Randomized controlled trial · Systematic review · Standardization · Stat…

3. Composite Endpoints in Clinical Trials -

ScienceDirecthttps://www.sciencedirect.com/science/article/pii/S1885585708601164

Composite endpoints are often used in clinical trials, especially in the cardiovascular area. Decreases in

Dennis

Dennis Mazur

Thank you very much for your answer Dr. Mazour. Unfortunately I cannot open the first link (Chapter 7 Introduction to Composite Endpoints), so if you have a pdf could you please send it on my email [email protected], I would appreciate it.

Although, this matter is still not clear for me. For an example, if we analyze in a hypothetical situation a CV trial with these outcomes:

Primary endpoint (mortality, stroke, MI) HR 0,95 (CI 95% 0,85-1,03)

Mortality HR 0,7 (CI 95% 0,6-0,8)

Non fatal MI HR 0,9 (CI 95% 0,9-1,1)

No fatal stroke 1,1 (CI 95% 0,6-1,5)

Can we claim that intervention has an impact on mortality reduction? I am asking this because trial was designed on the basis of primary endpoint (which failed), considering statistical strength of the trial. Since mortality is the rarest event from other parts of the composite endpoints, can we expect that this trial is underpowered for analyzing effect on mortality as a single endpoint?

Marko

Hi Marko,

In pharmaceutical clinical trials, the FDA have published draft guidence that covers the analysis of composite endpoints and their components. See:

https://www.fda.gov/downloads/drugs/guidancecomplianceregulatoryinformation/guidances/ucm536750.pdf

In essence, their view is that the analysis of components should be reported along with the analysis of the composite, but considered descriptive analyses. The FDA would only consider the significance test of the composite endpoint.

You will have specified the endpoint hierarchy a priori. If you had selected the three endpoints as co-primary, then you could have analysed and reported each separately, but would need to adjust the significance level for multiplicity.

In terms of the power of the tests for composite components, if you powered the study based on the composite, then it will most likely be underpowered for all the individual components. This is because the incidence rate of the composite is the incidence of at least one of the components - and so you'd expect the incidence rate of each of the components to be lower. Powering the study based on the composite will mean it is likely underpowered for the components individually.

I hope this helps,

Best regards, Bill