Health Consequences of Exclusion or Underrepresentation of Women in Clinical Studies (I)
Carol S. Weisman and Sandra D. Cassard
This paper considers the health consequences to women as a population as a result of their exclusion from or underrepresentation in clinical studies. (Clinical studies are broadly defined to include observational studies as well as randomized controlled trials of treatment or preventive interventions). The reasons that women, especially women of reproductive age and pregnant women, have been excluded or underrepresented in clinical studies have been amply reviewed.1 Less apparent are the consequences of inadequate studies of women for understanding, preventing, and treating health problems in women. We will describe types of information deficits that exist as a consequence of inadequate research and then examine the evidence of consequences for women's health.
TYPES OF INFORMATION DEFICITS
One type of information deficit occurs for conditions that affect exclusively or primarily women (such as breast cancer and osteoporosis). Failure to conduct sufficient research on these conditions is a special case of exclusion of women from clinical studies and may result in significant gaps in knowledge and in health services for women. If we use as a standard the number of women potentially affected by a condition, then the most obvious information needs pertain to the prevention and treatment of conditions associated with the female reproductive organs and with normal female aging (including menopause). An example of a current information deficit related to aging is the unknown efficacy
of estrogen replacement therapy, calcium supplementation, and exercise in preventing osteoporosis and fractures in postmenopausal women.2
Another type of information deficit occurs for conditions that affect both women and men. These conditions include the leading causes of death in both sexes (cardiovascular disease and lung cancer3) as well as conditions that are more prevalent in women (e.g., depression) or that affect men and women differently (e.g., AIDS). Exclusion of women from clinical studies of this class of conditions, or inclusion of women in numbers too small to detect gender differences or to support subgroup analyses, may result in a "male model" of medical treatment that is inappropriate for women.4 Both biological and psychosocial differences between the sexes may affect etiology, risk factors, disease presentation, disease course, or responses to preventive interventions or treatments. For example, there are numerous information deficits for prevention and treatment of heart disease in women, in part because women were excluded from a number of key trials (e.g., the Lipid Research Clinics Coronary Primary Prevention Trial, the Multiple Risk Factor Intervention Trial, and the Physicians' Health Study). The efficacy of hormone replacement therapy in prevention of heart disease in women is one question requiring further research.5
TYPES OF CONSEQUENCES
If there were overall negative health consequences to women as a result of information deficits, we would expect to see them reflected in gender differences in morbidity and mortality; in gender differences in patterns of diagnosis and treatment for key conditions; in gender differences in survival or outcomes of treatments; or in providers' perceptions that their ability to provide optimal care to women patients is compromised. We will examine each of these briefly.
Morbidity and Mortality
Although average life expectancy for U.S. women is about 7 years longer than for men, women consume more health services (including prescription drugs) than men, and throughout life, women experience more disease and disability than men.6 Further, because women live longer than men, they are more likely to be affected by late-onset diseases such as Alzheimer's disease and osteoporosis.
Mortality trends reveal some interesting gender differences. Although mortality rates from cardiovascular disease have been declining for both sexes over the last two decades, the decline in deaths from ischemic heart disease has been slower for women than for men.7 In addition, some of the key risk factors for heart disease (e.g., smoking, elevated serum cholesterol, obesity) have not
declined as much in women as in men.8 Reflecting women's smoking patterns, age-adjusted incidence of lung cancer increased steadily from 1980 to 1987 in women, while it did not change substantially for men; death rates parallel the incidence rates.9 Also owing to smoking patterns, an increase in deaths from chronic obstructive pulmonary disease in women is expected in the next few decades.10 In 1987, AIDS became the leading cause of death in black women of reproductive age in New York and New Jersey and was expected to be among the five leading causes of death in women in 1991.11 Among the conditions that are unique to women, breast cancer is the only major cause of mortality.12 In recent years, breast cancer mortality has increased, especially in black women.13
Diagnosis and Treatment Patterns
''Residual exclusion" of women from access to medications tested only on men has been observed in the literature; this occurs when providers are reluctant to use drugs in populations for which safety has not been demonstrated during clinical trials.14 Other research suggests that women may be diagnosed later or receive less aggressive treatment than men for specific conditions ("referral bias"). Most noteworthy are studies of access among patients with kidney disease to dialysis and transplantation; of diagnosis of lung cancer by sputum cytology; and of diagnosis and treatment of heart disease.15 Several studies find that women with heart disease may be diagnosed later than men (possibly because tests such as the treadmill exercise test are less effective in women) and that women are less likely than men to have invasive procedures such as coronary angiography, coronary angioplasty, or coronary artery bypass surgery, after relevant covariates such as age and severity are controlled.16 Although frequently attributed to gender bias in physician decision making, less aggressive treatment also could reflect women patients' beliefs and preferences (although this is undocumented).
Lower case survival rates have been observed for women following myocardial infarction and diagnosis of AIDS, perhaps reflecting later diagnosis or less aggressive treatment.17 Another problem is more frequent adverse effects or poorer outcomes for women when drugs or other treatments developed and tested in studies of men are used in women.18 For example, more women than men appear to experience adverse drug effects, possibly owing to failure to study hormone interactions in drug trials (e.g., for antidepressants).19
Higher peri—operative mortality has been observed for women than for men undergoing coronary artery bypass surgery.20 One possibility is that because of
referral bias or difficulties in diagnosis, women receive the procedure at later stages of the disease when their prognosis is poorer. Another possibility is that women's smaller cardiac size or smaller coronary artery diameter may lead to technical problems in surgery. Recent studies report higher mortality for women receiving angioplasty, after controlling for age and severity.21
There are no studies of physicians' or other providers' perceptions of whether and to what extent their care of women patients is impeded by the limitations of clinical studies. However, published practice guidelines and consensus reports provide one indicator of whether the professional community recognizes information gaps affecting the treatment of women.
Two recent publications address key risk factors for heart disease, high blood pressure and high serum cholesterol, and both cite gender-related information deficits. The Fifth Report of the Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure (JNC V) recommends the same approach to management of hypertension in women and men, although it concludes that further study is warranted and identifies some gaps in knowledge (e.g., effects of hormone replacement therapy on blood pressure).22 The recommendation for a similar management approach for women and men has been criticized in part because of missing information on the effects of antihypertensive agents on serum lipids in women.23 With regard to primary prevention of hypertension, however, JNC V does not address gender differences in risk factors (e.g., diet, physical activity, stress levels) or in the efficacy of interventions.24
The report of the 1992 NIH Consensus Development Conference on Triglyceride, High-Density Lipoprotein, and Coronary Heart Disease acknowledges information gaps with regard to women and calls specifically for studies of the effects of estrogen and progesterone use on lipids and on risk of coronary heart disease in women.25
Evidence has been provided that exclusion or underrepresentation of women in clinical studies results in some important information deficits, particularly for the leading causes of morbidity and mortality in women, and that these deficits may adversely affect women's health. These deficits, in fact, may result in a shift in the distribution of risks to women in the general population, including pregnant women, who receive treatments (or are impeded from receiving treatments) that have not been developed or studied in women. This transfer of
risk from women who are potential participants in clinical studies, where informed consent procedures and monitoring provide safeguards, to women in the general population may be unacceptable as a matter of social policy.