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
All rapid responses
We noticed that Box D and the references for the extra examples are not available on your website.
Box D: A trial with a composite endpoint with RRRs on the component endpoints with wide confidence intervals
Consider the results of the CURE trial, in which investigators randomized 12,562 patients with acute coronary syndrome to clopidogrel or placebo and examined the impact on the same composite, cardiovascular death, MI, or stroke 7. Here, while one could interpret the point estimates of the RRR as consistent (23%, 7%, and 14% for MI, cardiovascular death, and stroke, respectively) the range of the confidence intervals should give us pause.
While the point estimate and 95% confidence interval on the RRR leave us reasonably confident of an important treatment impact on MI (23%, 95% CI 11 to 33%), the same is not true of either cardiovascular death (7%, 95% CI –8% to 21%), or stroke (14%, 95% CI –18% to 37%). As a result, the statement that clopidogrel reduced a CEP of cardiovascular death, stroke and MI by 20% (the RRR associated with the CEP) is potentially misleading, and using the CEP as a basis for clinical decision-making is problematic.
These are the missing references (preceded by a w in the web publication now):
1. Moyé L. Introduction to Composite Endpoints; Multiple Analyses and Composite Endpoints. Multiple Analyses in Clinical Trials. Fundamentals for Investigators (Statistics for Biology and Health). New York, NY: Springer- Verlag, 2003:219-265.
2. Freemantle N, Calvert M, Wood J, Eastaugh J, Griffin C. Composite outcomes in randomized trials: greater precision but with greater uncertainty? JAMA 2003;289(19):2554-9.
3. Niewoehner D, Erbland M, Deupree R, Collins D, Gross N, Light R, et al. Effect of systemic glucocorticoids on exacerbations of chronic obstructive pulmonary disease. N Engl J Med 1999;340:1941-1947.
4. Yusuf S, Sleight P, Pogue J, Bosch J, Davies R, Dagenais G. Effects of an angiotensin-converting-enzyme inhibitor, ramipril, on cardiovascular events in high-risk patients. The Heart Outcomes Prevention Evaluation Study Investigators. N Engl J Med 2000;342(3):145-53.
5. Lewis EJ, Hunsicker LG, Clarke WR, Berl T, Pohl MA, Lewis JB, et al. Renoprotective effect of the angiotensin-receptor antagonist irbesartan in patients with nephropathy due to type 2 diabetes. N Engl J Med 2001;345(12): 851-60.
6. Dahlof B, Devereux RB, Kjeldsen SE, Julius S, Beevers G, Faire U, et al. Cardiovascular morbidity and mortality in the Losartan Intervention For Endpoint reduction in hypertension study (LIFE): a randomised trial against atenolol. Lancet 2002;359(9311):995-1003.
7. Yusuf S, Zhao F, Mehta SR, Chrolavicius S, Tognoni G, Fox KK. Effects of clopidogrel in addition to aspirin in patients with acute coronary syndromes without ST-segment elevation. N Engl J Med 2001;345(7):494-502.
Competing interests: None declared
Competing interests: No competing interests
The paper by Montori et al.  concerning composite end-points in clinical trials is to be welcomed. In particular, their recommendations for distinguishing between the acceptable and the unacceptable use of composite end-points are of value both to those involved in the design of clinical trials and, more importantly, to those interpreting these studies. There can be little argument about their conclusions that outcomes with different degrees of clinical importance to patients, those with large variations in frequency, or those with differences in the size of treatment effect should militate against the use of composite end- points.
However, Montori et al. may be being too lenient. Even when composite end-points satisfy their conditions, there are sound reasons to shun this practice. Composite end-points lack clarity – hardly a quality that is in abundance in current large-scale randomised trials – and this surely aggravates the difficulties already present in the interpretation of complex data. For instance, what are doctors or, indeed, patients to make of the primary composite end-point of the CAMELOT study  which included cardiovascular death, non-fatal myocardial infarction, stroke or transient ischaemic attack, resuscitated cardiac arrest, new-onset peripheral vascular disease, hospitalisation for angina, hospitalisation for heart failure and coronary revascularisation? Composite end-points also create yet more opportunities for data manipulation. Consider, for example, the conclusions in the abstract of the LIFE study  which used a composite primary end-point of cardiovascular death, stroke or myocardial infarction: “In hypertensive patients without clinically evident vascular disease, losartan was more effective that atenolol in preventing cardiovascular morbidity and death…” This is clearly misleading. The impression given is that losartan reduces mortality compared with atenolol yet there was no statistical difference between the two treatments in terms of either cardiovascular mortality or overall mortality. Given the willingness of medical researchers to manipulate data for their own ends, [4,5] anything which facilitates this activity is to be deprecated.
The choice of a composite end-point is grounded in expediency. When a treatment fails to decrease mortality – as is frequently the case in large -scale cardiovascular trials – other outcomes are added to disguise the absence of any improvement in survival and to present a more favourable conclusion. On the surface, combining mortality with non-fatal myocardial infarction seems an innocent enough manoeuvre but even the latter outcome is of uncertain clinical importance given the new criteria for diagnosis using small increases in cardiac enzymes . However, in many situations, combining mortality and myocardial infarction falls short of providing sufficient numbers of events and, hence, more outcomes with less clinical importance – for instance, admissions to hospital for heart disease – are included in the composite endpoint. But how far is the expansion of a composite end-point to be taken? Perhaps it would be more pertinent to ask how far those with a vested interest in a particular clinical trial are prepared to go in their efforts to procure favourable results. It seems unlikely either that the practice of using composite end-points or the number of outcomes incorporated within them will diminish. Indeed, the indications for expanding composite end-points now include problems with recruitment to clinical trials: by altering the primary end-point during the study to include more frequently occurring – although clinically less significant – outcomes, the numbers of patients required is decreased. 
Composite end-points are a recipe for confusion and a source of misleading information. If the outcomes which seriously affect the lives of patients occur so infrequently that they are not sufficient to yield a statistically significant difference, then this says a great deal about current medical research and, in particular, about large-scale randomised trials. 
 Montori VM, Permanyer-Miralda G, Ferreira-Gonzalez I, Busse JW, Pacheco-Huergo V, et al. Validity of composite end points in clinical trials. BMJ 2005;330;594-6.
 Nissen SE, Tuzcu EM, Libby P, Thompson PD, Ghali M, et al. Effect of antihypertensive agents on cardiovascular events in patients with coronary disease and normal blood pressure. The CAMELOT study: a randomized controlled trial. JAMA 2004;292;2217-26.
 Devereux RB, Dahlof B, Kjeldsen SE, Julius S, Aurup P, et al. Effects of losartan or atenolol in hypertensive patients without clinically evidence vascular disease: a substudy of the LIFE randomized trial. Ann Intern Med 2003;139;169-77.
 Montori VM, Jaeschke R, Schunemann HJ, Bandari M, Brozek JL, et al. Users’ guide to detecting misleading claims in clinical research reports. BMJ 2004;329;1093-6.
 Penston J. Fiction and fantasy in medical research: the large- scale randomised trial. London: The London Press, 2003.
 Brophy JM. The EUROPA trial: correspondence. Lancet 2003;362;1036.
 The PEACE Trial Investigators. Angiotensin-converting-enzyme inhibition in stable coronary artery disease. N Engl J Med 2004;351;2058- 68.
Competing interests: None declared
Competing interests: No competing interests