Endpoint Selection and Relative (Versus Absolute) Risk Reporting
in Published Medication Trials
Michael Hochman, MD
and Danny McCormick, MD, MPH
The Robert Wood Johnson Clinical Scholars® Program, University of California, Los Angeles, and the U.S. Department of Veterans Affairs,
Los Angeles, CA, USA;
Department of Medicine, Cambridge Health Alliance, and Harvard Medical School, Cambridge, MA, USA.
BACKGROUND: The use of surrogate and composite
endpoints, disease-specific mortality as an endpoint,
and relative (rather than absolute) risk reporting in
clinical trials may produce results that are misleading
or difficult to interpret.
OBJECTIVE: To describe the prevalence of these
endpoints and of relative risk reporting in medication
DESIGN AND MAIN MEASURES: We analyzed all
randomized medication trials published in the six highest
impact general medicine journals between June 1, 2008
and September 30, 2010 and determined the percentage
using these endpoints and the percentage reporting
results in the abstract exclusively in relative terms.
KEY RESULTS: We identified 316 medication trials, of
which 116 (37%) used a surrogate primary endpoint
and 106 (34%) used a composite primary endpoint.
Among 118 trials in which the primary endpoint
involved mortality, 32 (27%) used disease-specific
mortality rather than all-cause mortality. Among
157 trials with positive results, 69 (44%) reported
these results in the abstract exclusively in relative
terms. Trials using surrogate endpoints and disease-
specific mortality as an endpoint were more likely to
be exclusively commercially funded (45% vs. 29%,
difference 15% [95% CI 5%–26%], P=0.004, and 39%
vs. 16%, difference 22% [95% CI 6%-37%], P=0.007,
respectively). Trials using surrogate endpoints were
more likely to report positive results (66% vs. 49%,
difference 17% [95% CI 5%–28%], P=0.006) while
those using mortality endpoints were less likely to be
positive (46% vs. 62%, difference −16% [95% CI −27%–
CONCLUSIONS: The use of surrogate and composite
endpoints, endpoints involving disease-specific mor-
tality, and relative risk reporting is common. Articles
should highlight the limitations of these endpoints and
should report results in absolute terms.
KEY WORDS: surrogate endpoints; composite endpoints;
disease-specific mortality; relative risk reduction.
J Gen Intern Med 26(11):1246–52
© Society of General Internal Medicine 2011
Medical interventions are judged as successes or failures in
clinical studies based on their impact on the study endpoints.
Because interventions have a variety of effects, and may
appear effective using one set of endpoints but harmful using
another set, the selection of endpoints is critical.
Recently, there have been growing concerns about the
selection of endpoints and the reporting of results in clinical
trials. In particular, experts have suggested that the use of
and composite endpoints,
the use of disease-specific—rather than all-cause—mortality
as an endpoint,
may mislead readers if the limitations
of these endpoints are not adequately explained. Similar
concerns have also been raised about the reporting of trial
results in relative—rather than absolute—numbers,
Table 1 describes these concerns.
A limited number of studies have examined the prevalence
of the use of surrogate
and composite endpoints
use of disease specific mortality as an endpoint
reporting of results in relative numbers
trials. However, data from these studies are several years old,
typically reflect trials involving treatments for only a limited
spectrum of diseases, and provide only limited details about
factors that may be associated with these endpoints and with
relative risk reporting.
surrogate endpoints, composite endpoints, disease-specific
mortality as an endpoint, and the use of relative risk
reporting in recently published medication trials from the
six highest impact general medicine and internal medicine
journals. We also examine the association between trial
funding source and outcome (i.e. positive vs. negative results)
and these trial characteristics.
Details of the process by which studies were identified have
previously been published.
In summary, we identified all
The views expressed in this article are solely those of the author and do
not necessarily represent those of the Department of Veterans Affairs or
of the Robert Wood Johnson Foundation
Electronic supplementary material The online version of this article
(doi:10.1007/s11606-011-1813-7) contains supplementary material,
which is available to authorized users.
Received April 13, 2011
Revised June 8, 2011
Accepted July 1, 2011
Published online August 13, 2011