Correspondence
Enrollment of Women in Cardiovascular Clinical Trials
N Engl J Med 2000; 343:1972-1973December 28, 2000
- Article
To the Editor:
The report by Harris and Douglas (Aug. 17 issue)1 on the enrollment of women in cardiovascular clinical trials funded by the National Heart, Lung, and Blood Institute is predicated on the assumption that the proportions of men and women represented in trials should mirror some underlying ideal — in the authors' case, the difference in the prevalence of cardiovascular disease among men and women in the general population. Using that construct, they examined classes of trials in which women were “underenrolled” (most notably, trials involving heart failure). But then why not characterize women as having been “overenrolled” in all the cardiovascular trials combined?
The authors imply that the proportions of men and women with cardiovascular disease who are enrolled in trials, small and large alike, overall and for particular disorders, should reflect the proportions in the general population. This expectation is unreasonable and unrealistic, because trials involve selected, nonrandom samples of undefined populations.
In any case, representativeness, even if it is regarded as desirable, contributes nothing to the scientific validity of trials. Validity derives from the ability to compare treatments in comparable groups of patients in a trial, not from the representativeness of the persons enrolled.
The difficulty in characterizing populations as being understudied or overstudied according to sex is that, invariably, such characterizations suggest that there is a proper amount of study according to sex.2 The authors suggest that for all cardiovascular trials combined, that amount is in proportion to the difference between men and women with regard to the burden of cardiovascular disease. Their ideal proportions are 51 percent men to 49 percent women. On the basis of differences in mortality from heart disease (1990 rates), the corresponding proportions would be 65 percent to 35 percent, and they would be 71 percent to 29 percent with the use of years of life lost before a person reached the age of 65 years.3 Which of these ratios represents the proper apportionment of effort?
Rates of mortality from heart disease have fallen dramatically since 1950, but the difference in mortality rates between men and women remains largely unchanged. In fact, if anything, the difference has increased. The ratios of the rates of mortality from heart disease among white men to those among white women in 1950 were 2.73, 2.99, and 2.35 for the three decades of life starting at 35 years, as compared with 3.74, 3.40, and 2.69, respectively, in 1990. Could it be that these basic facts explain why the efforts of the federal government to change the proportions of men and women enrolled in cardiovascular trials have been only “moderately successful”?
3 ReferencesCurtis L. Meinert, Ph.D.
Adele Kaplan Gilpin, Ph.D.
Johns Hopkins University, Baltimore, MD 212051
Harris DJ ,Douglas PS . Enrollment of women in cardiovascular clinical trials funded by the National Heart, Lung, and Blood Institute. N Engl J Med 2000;343:475-480
Full Text | Web of Science | Medline2
Meinert CL, Gilpin AK. Estimation of gender bias in clinical trials. Stat Med (in press).
3
National Center for Health Statistics. Health, United States, 1995. Hyattsville, Md.: Public Health Service, 1996. (DHHS publication no. (PHS) 96-1232.)
To the Editor:
We question whether a comparison of the rate of enrollment of women in cardiovascular trials with the prevalence of cardiovascular disease among women in the general population is the most relevant comparison. Harris and Douglas compared the following two proportions: the number of women enrolled in cardiovascular trials divided by the number of men and women enrolled in cardiovascular trials, and the number of women with cardiovascular disease in the population divided by the total number of women in the general population.
To determine whether women are underrepresented in clinical trials, it is more relevant to compare the rate of enrollment of women in cardiovascular trials with the proportion of women with cardiovascular disease among all persons with cardiovascular disease in the general population. For example, a rare disease may affect men and women equally (50 percent of each group) but have a prevalence of 0.1 percent in the general population; if the rate of enrollment of women in trials involving this disease is 20 percent, women are still underrepresented in these trials.
We recomputed the proportions, using data from the American Heart Association1 and the National Center for Health Statistics,2 and compared them with the proportions provided by Harris and Douglas (Table 1Table 1
Enrollment of Women in NHLBI-Funded Cardiovascular Trials and the Proportion of Women with Cardiovascular Disease in the General Population.). Our results show that adequate numbers of women are enrolled in mixed-sex trials of hypertension but not in mixed-sex trials of coronary artery disease or congestive heart failure.2 ReferencesAngela M. Cheung, M.D., Ph.D.
Gary Naglie, M.D.
University Health Network, Toronto, ON M5G 2N2, Canada1
2000 Heart and stroke statistical update. Dallas: American Heart Association, 1999. (See http://www.americanheart.org/statistics/index.html.)
2
National Center for Health Statistics. Health, United States, 1995. Hyattsville, Md.: Public Health Service, 1996. (DHHS publication no. (PHS) 96-1232.)
Author/Editor Response
Dr. Douglas replies:
To the Editor: Drs. Meinert and Gilpin consider it “unreasonable and unrealistic” to expect that the proportions of men and women in the overall patient population be reflected in clinical trials supported by federal dollars. This expectation is reasonable, since the federal government has a fiduciary obligation to serve the U.S. population and a legal obligation not only to consider but also to ensure appropriate proportions in making decisions about research funding. It is also reasonable because, as my colleagues and I showed in our article, this standard has been achieved in hypertension trials. The scientific validity of any individual trial, in which the comparability of study groups is paramount, is not the issue under discussion, but rather the clinical validity of the aggregate of all federally funded trials involving patients with heart disease. Furthermore, 21st-century medical research recognizes that sex is an important element in study design.
The prevalence of disease was not chosen arbitrarily as our standard of success for enrollment. Federal law requires that the enrollment of males and females in federally funded trials be “appropriate to the incidence/prevalence of the disease studied.”1 Aside from legal reasons, the morbidity of chronic diseases such as heart disease is extremely important in assessing the burden of the disease and is more accurately reflected by prevalence than by mortality. Furthermore, consideration of mortality rates only among persons under the age of 65 years excludes much of the U.S. population; overall rates of mortality from cardiovascular diseases in the United States are nearly equal among men and women.
Our calculations of the prevalence of disease were in fact performed as Drs. Cheung and Naglie suggest: the proportional prevalence of disease among women was calculated as the ratio of the prevalence among women to the prevalence in the entire population. This approach required two assumptions, which we acknowledged in our article: that the population was similar among persons 45 to 54 years old, those 55 to 64, and those 65 to 74 and that there were equal numbers of men and women. Our results and theirs are similar with respect to hypertension. In the case of coronary artery disease, the prevalence is higher among men in every age group.2 It is unclear how the authors concluded that the majority of patients with coronary artery disease (51 percent) are actually women.
2 ReferencesPamela S. Douglas, M.D.
University of Wisconsin–Madison Medical School, Madison, WI 537921
National Institutes of Health and Alcohol, Drug Abuse, and Mental Health Administration 2
2000 Heart and stroke statistical update. Dallas: American Heart Association, 1999. (See http://www.americanheart.org/statistics/index.html.)
- Citing Articles (2)
Citing Articles
1
Reshma Jagsi, Amy R. Motomura, Sudha Amarnath, Aleksandra Jankovic, Nathan Sheets, Peter A. Ubel. (2009) Under-representation of women in high-impact published clinical cancer research. Cancer 115:14, 3293-3301
CrossRef2
Eric L. Ding, Sonia R. Nagda. (2009) Women, Contraception, and Consent to Research Participation. Journal of Women's Health 18:4, 439-441
CrossRef
