Sex Bias in Cardiovascular Care: Title and subTitle BreakShould Women Be Treated More Like Men?

Cardiovascular disease is the principal cause of death for both men and women. However, since 1984 more women than men have died of this disease.¹ Thus, any suggestion that cardiovascular care is biased against women has major public health implications and must be examined carefully. If true, such bias would represent a severe failing of the modern medical system. Since Tobin et al² reported more than a decade ago that women with abnormal nuclear exercise studies were referred far less often than men for cardiac catheterization and coronary artery bypass graft (CABG) surgery, a host of additional studies have confirmed important sex-based differences in many aspects of care for coronary artery disease (CAD).^{3 - 4} Some reports have suggested that these differences are the result of physician bias.

In evaluating possible sex biases in the diagnosis and treatment of CAD, researchers must account for important differences in the clinical epidemiology of the disease. Population data from Framingham and elsewhere show that women have an average of 10 to 15 years more CAD-free life expectancy than men due to a later onset of disease.^{1 ,5} The cause of this protection from early atherosclerosis for women is presumed to be endogenous estrogen, although recent studies have suggested that the understanding of the link between sex hormones and cardiovascular disease is far from complete.^{6 - 7} Therefore, when evaluating a patient with possible CAD symptoms, a clinician must take into account these sex-based differences in age of onset and underlying disease prevalence. These powerful differences in epidemiology make every aspect of diagnostic evaluation more difficult in women.

Framingham data demonstrated that women with CAD most commonly present with angina pectoris initially whereas men first present with myocardial infarction (MI).⁸ Furthermore, in elective presentations women tend to present with atypical forms of angina much more often than men, making correct diagnosis more difficult. Even when women present with typical angina, the underlying prevalence of CAD is lower in women than in men with similar symptoms. For example, a 60-year-old woman with typical angina has approximately a 50% chance of having significant CAD whereas a 60-year-old man with the same symptoms has more than a 90% chance.⁹ For a man with typical angina to have a 50% chance of CAD (similar to the hypothetical 60-year-old woman), he would have to be in his 40s. For a woman with typical angina and a probability of CAD of 90% (similar to the hypothetical 60-year-old man), she would have to be in her late 70s.⁹ Thus, the same symptom complex means something different at different stages of a patient's life for both men and women. The age-related meaning for men is not relevant to women and vice versa.

Even with the aid of noninvasive tests, diagnosing CAD is more difficult in women than men. The reduced sensitivity of stress tests, which is related to the distribution of severe disease in the tested population, and the lower prevalence of disease both reduce the ability of such tests to "rule in" CAD in women. Although the exercise treadmill test has often been regarded as unreliable for diagnosis in women, its prognostic value is actually equivalent to what it is in men after accounting for the lower prevalence of CAD.¹⁰ Stress imaging studies are more accurate for diagnosis of CAD in either sex but do not provide a simple solution to the greater diagnostic challenges in women. For example, in the Women's Ischemia Syndrome Evaluation (WISE) pilot study,¹¹ the overall sensitivity of dobutamine stress echocardiography for detection of angiographically significant CAD in 92 symptomatic women with clinical indications for coronary angiography was only 40% with a specificity of 81%; sensitivity for detection of multivessel disease was 60%. Importantly, the prevalence of at least 50% stenosis of a coronary artery in this group was only 27%. In this same population, the sensitivity of the exercise treadmill test ST segment response was 25% with a specificity of 80%. With such low test accuracy for detection of CAD, it is virtually impossible to reach a high level of diagnostic certainty about the presence or severity of significant CAD.¹² Consequently, the different referral patterns for noninvasive testing between men and women could reflect an inappropriate physician bias and could rather seem much more likely to reflect genuine confusion about the optimal testing strategy to use.

Few studies of sex bias have been able to examine subjective physician decision making directly. An important recent contribution in this area comes from Schulman and colleagues¹³ who used actors to portray patients with CAD symptoms to elicit evidence for physician bias. The probability that these patient actors had CAD according to their symptoms and relevant history was assessed by 720 primary care physicians, of whom 31% were women. Women patient actors were assigned lower probabilities of CAD than men with identical histories (pretest probability 0.64 vs 0.69; P<.001). After adjusting for the pretest probability of CAD, women were 40% less likely to be referred for coronary artery catheterization than men (P = .02). However, after incorporating a race-sex interaction term in the analysis, black women were 60% less likely to be referred for catheterization than men (black or white) or white women. Conversely, neither sex nor race had any effect on referral for stress testing. This pivotal study suggests that at least for 1 major subgroup of women, physician decision making may be inappropriate and further investigations are clearly warranted. In contrast to these findings among primary care physicians, an earlier study found that academic cardiologists were equally accurate in predicting the probability of significant CAD in men or women referred for outpatient evaluation of possible CAD (area under the receiver operating curves, 0.91 for both sexes).¹⁴ Furthermore, although women were referred for catheterization less often than men in this study (18% vs 27%), the difference was completely accounted for by differences in the pretest probability of CAD.

Important care biases also can occur in the management of clinically manifest disease. However, when catheterization is performed in women and CAD is found, sex-based differences in referral for revascularization tend to disappear. In a study of patients referred for cardiac catheterization from 1969 to 1984, 46% of men and 44% of women with significant CAD were referred for CABG surgery.¹⁵ After adjusting for baseline risk of cardiac death, Bickell and colleagues¹⁵ found that men were almost 30% more likely to be referred for CABG surgery when the expected survival benefit from the procedure was minimal relative to medical therapy, whereas there was a trend for more women than men to be referred for CABG surgery in the high-risk and high-benefit subgroup. The patterns suggested that the care of women in this instance actually might be more appropriate.

Despite suggestions to the contrary from earlier studies, more recent studies show that outcome following revascularization is at least as good for women as it is for men. In the Bypass Angioplasty Revascularization Investigation (BARI) trial,¹⁶ for example, women with multivessel CAD undergoing revascularization had similar inhospital mortality and periprocedural MI rates and better adjusted 5-year survival rates than men. Recurrent angina rates at 5 years also were equivalent in men and women.

Acute presentations of CAD herald a 30-day phase of substantially increased risk of mortality and morbidity. Therefore, sex-based differences in the management of acute coronary syndromes deserve particular scrutiny. Maynard and colleagues¹⁷ reported the results of a major multicenter trial focused on emergency department management of acute coronary syndromes. Among 2542 patients with definite acute MI, these investigators found that use of thrombolytic therapy, cardiac catheterization, and revascularization procedures was equivalent in men and women.¹⁷ Among patients with angina without MI (a lower risk group), women underwent fewer invasive procedures. This study did not include the long-term follow-up necessary to judge whether the outcomes of these women were adversely affected.

In this issue of THE JOURNAL, Roger and colleagues¹⁸ provide important new data on the long-term outcomes associated with these different patterns of care. The authors studied 1306 men and 965 women living in Olmsted County, Minnesota, and who presented to an emergency department with symptoms of unstable angina between 1985 and 1992. Patients' sex had no effect on whether they were admitted to the hospital and no effect on the use of resting echocardiography. However, men were 43% more likely to undergo stress testing and 59% more likely to undergo coronary angiography. Although the unadjusted mortality of women was higher, after adjustment for baseline differences in age and other characteristics, men had a nonsignificant (P = .14) trend toward a 2-fold higher risk of death at 30 days and a 23% (P = .07) higher risk of death up to 6 years after the emergency department evaluation. These data are particularly noteworthy because they reflect a geographic population experience and are therefore free of the possible confounding effects of referral bias.

These results complement those of Hochman and colleagues,^{19 - 20} who found that women with acute non–ST segment elevation coronary syndromes have less significant CAD and less multivessel disease than men, despite being older and having a higher prevalence of hypertension, diabetes, and hyperlipidemia. Left ventricular function was also significantly better in women with acute coronary syndromes compared with men.²⁰ In patients with non–ST segment elevation MI, the difference in CAD severity was smaller and 30-day death rates and reinfarction rates were equivalent for men and women in this subgroup.¹⁹ In contrast, among patients with unstable angina without infarction, women had a significantly lower prevalence of obstructive CAD (71% vs 86% for men, P<.001) and a 35% lower adjusted risk of death or MI in the first 30 days after presentation.

Thus, although far from complete, the current picture of sex-based differences in CAD diagnosis and management suggests that although some bias may be present in physician decision making, sex bias is not the major factor explaining the differences in care between men and women. Because sex bias, as an overarching context, carries with it reference to the larger social struggle of women for equality with men in career opportunities, pay, and political power, sex-biased differences in medical care could be viewed as a problem of the male-dominated medical profession using a higher standard of care for men. Simply demonstrating different patterns of care for men and women, however, does not establish that men are getting better care. The challenge in this area is therefore to determine when women or men have been harmed by these differences and to learn from each how to improve the care of the other.