Validity of composite end points in clinical trialsBMJ 2005; 330 doi: https://doi.org/10.1136/bmj.330.7491.594 (Published 10 March 2005) Cite this as: BMJ 2005;330:594
Box A – Desirable features of composite endpoints to the trialist
We note that, from the point of view of the clinical trialist, a priori specification of a CEP as a single primary endpoint provides an elegant solution to a number of problems that arise when a trial attempts to estimate a treatment’s effect on a number of outcomes. These include not only requirements for larger sample sizes, but problems in sample size calculation, stopping rules, and calculation of p-values and confidence intervals. A detailed discussion can be found in Moyé.w1
We also note, however, that virtually all trials using a CEP report treatment effects on several endpoints without adjusting for multiple analyses, many trials now use multiple CEPs, and discrepancies between planned analyses in protocols and analyses in published reports are common.w2 Unquestionably, there are statistical merits of specifying a single primary CEP, and risks of spurious positive findings when investigators report statistical tests on multiple outcomes (including on each of the components of the CEP).
Box B – An example of a composite endpoint with components that occur with different frequency
Consider the following statement: in patients with in-stent stenosis of coronary artery bypass grafts, gamma radiation reduced the CEP of death from cardiac causes, Q-wave MI, and revascularization of the target vessel. This result sounds impressive because it suggests that gamma radiation reduces the incidence of death and MI, as well as the need for revascularization. The trial from which we draw this result randomized 120 patients with in-stent stenosis of a saphenous vein graft to gamma radiation (Iridium-192) or placebo.w3 Of those in the Iridium-192 arm, 32% experienced the primary CEP of death from cardiac causes, Q-wave myocardial infarctions, and revascularization of the target vessel at 12-months, as did 63% in the placebo arm (RRR 50%, 95% CI 25 to 68%).
While this result appears compelling, only 2 patients in the placebo arm (3.3%) and 1 patient in the Iridium-192 arm (1.7%) sustained an MI (risk difference 1.7% 95% CI -5.9 to 9.9%). The story is similar for cardiac death, which occurred in 4 patients (7%) in each arm (risk difference 0%, 95% CI -10.3% to 10.3%). Revascularizations constituted the overwhelming majority of the events -- 37 (62%) in the placebo arm, and 17 (28%) in the radiated group. Because of the very large discrepancy in the frequency of the more important and less important endpoints in this trial, the most reasonable conclusion is that the intervention reduced the relative risk of revascularization of the target vessel by 54% (95% CI 29 to 71%), a risk difference of 33% (95% CI 16 to 49%). The trial provides essentially no information about the effect of the intervention on MI or death.
Box C– Example of a trial where biology argues against the expectation of similar RRR across component endpoints.
The Irbesartan Diabetic Nephropathy Trial w4 randomized 1715 hypertensive patients with nephropathy and type 2 diabetes to irbesartan, amlodipine, or placebo. The primary endpoint was the composite of a doubling of the baseline serum creatinine concentration, the onset of end-stage renal disease (serum creatinine > 6.0 mg/dL, initiation of dialysis, or transplantation), or death from any cause. It is extremely plausible that, for two of these three components, doubling of creatinine and crossing the creatinine threshold of 6.0 mg/dL, any treatment effects would be similar – indeed, one would be very surprised if results showed otherwise. On the other hand, there are many contributors to all-cause mortality aside from renal failure (including, for instance, cardiac disease) and it might well be that treatments have different effects on these contributors. Thus, the biological rationale that the treatments would have similar impacts on all three components is weak. The relatively weak biological rationale increases our reluctance to base treatment decisions on the composite, as opposed to its components. Indeed, in this instance, irbesartan lowered the incidence of both doubling of creatinine and end stage renal disease, but without apparent impact on all-cause mortality
FIGURE A Irbesartan Diabetic Nephropathy Study 5
- This Week In The BMJ Published: 10 March 2005; BMJ 330 doi:10.1136/bmj.330.7491.0-e
- ResearchProblems with use of composite end points in cardiovascular trials: systematic review of randomised controlled trialsPublished: 12 April 2007; BMJ 334 doi:10.1136/bmj.39136.682083.AE
- RESEARCHProblems with use of composite end points in cardiovascular trials: systematic review of randomised controlled trialsPublished: 02 April 2007; BMJ doi:10.1136/bmj.39136.682083.AE
- ResearchDefinition, reporting, and interpretation of composite outcomes in clinical trials: systematic reviewPublished: 18 August 2010; BMJ 341 doi:10.1136/bmj.c3920
- Research Methods & Reporting Published: 22 November 2013; BMJ 347 doi:10.1136/bmj.f6782
- Analysis Published: 26 April 2007; BMJ 334 doi:10.1136/bmj.39169.447488.94
- Editorial Published: 12 April 2007; BMJ 334 doi:10.1136/bmj.39176.461227.80
- Editorial Published: 18 August 2010; BMJ 341 doi:10.1136/bmj.c3529
- RESEARCHPharmacological prevention of serious anaphylactic reactions due to iodinated contrast media: systematic reviewPublished: 31 July 2006; BMJ doi:10.1136/bmj.38905.634132.AE
- ResearchPharmacological prevention of serious anaphylactic reactions due to iodinated contrast media: systematic reviewPublished: 28 September 2006; BMJ 333 doi:10.1136/bmj.38905.634132.AE
- Analysis Published: 07 February 2019; BMJ 364 doi:10.1136/bmj.l341
- Clinical Review Published: 05 November 2019; BMJ 367 doi:10.1136/bmj.l5887
- US government sues Yale’s teaching hospital for age discrimination in testing older doctorsBMJ February 18, 2020, 368 m639; DOI: https://doi.org/10.1136/bmj.m639
- Royal College of Pathologists: president elect is barred after misconduct allegations are upheldBMJ February 18, 2020, 368 m638; DOI: https://doi.org/10.1136/bmj.m638
- Measuring multimorbidity beyond counting diseases: systematic review of community and population studies and guide to index choiceBMJ February 18, 2020, 368 m160; DOI: https://doi.org/10.1136/bmj.m160
- UK deal over inclisiranBMJ February 18, 2020, 368 m579; DOI: https://doi.org/10.1136/bmj.m579
- Matt Morgan: When nothing is the right thing to sayBMJ February 18, 2020, 368 m574; DOI: https://doi.org/10.1136/bmj.m574
- Identifying adverse events: reflections on an imperfect gold standard after 20 years of patient safety research
- Benefits and harms of intensive glycemic control in patients with type 2 diabetes
- The problem with composite indicators
- PREDIMED trial of Mediterranean diet: retracted, republished, still trusted?
- Composite measures of healthcare quality: sensible in theory, problematic in practice
- A randomised controlled trial assessing the efficacy of an electronic discharge communication tool for preventing death or hospital readmission
- Temporal Trends in Use of Composite End Points in Major Cardiovascular Randomized Clinical Trials in Prominent Medical Journals
- Impact of sleep disordered breathing on short-term post-operative outcome after elective coronary artery bypass graft surgery: a prospective observational study
- Health Status Outcomes in Patients With Acute Myocardial Infarction After Rehospitalization
- Selection of the primary end point in an observational cohort study
- Glycemic Control for Patients With Type 2 Diabetes Mellitus: Our Evolving Faith in the Face of Evidence
- Clinical Relevance of Rehospitalizations for Unstable Angina and Unplanned Revascularization Following Acute Myocardial Infarction
- Progress: the ROPAC multinational registry advances our understanding of an important outcome in pregnant women with heart disease
- Concordance of effects of medical interventions on hospital admission and readmission rates with effects on mortality
- Dangers of non-specific composite outcome measures in clinical trials
- Usefulness of Exercise Test and Myocardial Perfusion-Gated Single Photon Emission Computed Tomography to Improve the Prediction of Major Events
- Comparison of the Prognosis of Spontaneous and Percutaneous Coronary Intervention-Related Myocardial Infarction
- Use of troponin to diagnose periprocedural myocardial infarction: effect on composite endpoints in the British Bifurcation Coronary Study (BBC ONE)
- Use and Interpretation of Composite End Points in Orthopaedic Trials
- Weighting Components of Composite End Points in Clinical Trials: An Approach Using Disability-Adjusted Life-Years
- Cardiopulmonary exercise testing for the selection of patients undergoing surgery for lung cancer: friend or foe?
- Interpreting composite outcomes in trials
- Definition, reporting, and interpretation of composite outcomes in clinical trials: systematic review
- Trial and Error: How to Avoid Commonly Encountered Limitations of Published Clinical Trials
- What has the Cochrane Collaboration ever done for newborn infants?
- Outcome Measures and Implications for Sample-Size Calculations
- Randomized Trial of Reamed and Unreamed Intramedullary Nailing of Tibial Shaft Fractures
- Intensive glucose control did not prevent important complications in type 2 diabetes
- Primary Prevention of Cardiovascular Mortality and Events With Statin Treatments: A Network Meta-Analysis Involving More Than 65,000 Patients
- Hospital mortality: when failure is not a good measure of success
- Periprocedural Bleeding and 1-Year Outcome After Percutaneous Coronary Interventions: Appropriateness of Including Bleeding as a Component of a Quadruple End Point
- The Problem With Composite End Points in Cardiovascular Studies: The Story of Major Adverse Cardiac Events and Percutaneous Coronary Intervention
- COPD dashboard: about official Documents, Authors, Science, Health status and BOARDs
- Waking up from the DREAM of preventing diabetes with drugs
- Composite and surrogate outcomes in randomised controlled trials
- Problems with use of composite end points in cardiovascular trials: systematic review of randomised controlled trials
- Obesity and migraine: a population study.
- Pharmacological prevention of serious anaphylactic reactions due to iodinated contrast media: systematic review
- MOSES: Superiority or Noninferiority? * Response:
- How should clinicians interpret results reflecting the effect of an intervention on composite end points: should I dump this lump?