Sex and Racial Bias in Cardiac Care

One of the core principles of SBM, and one of the features that distinguishes it from EBM, is that it considers not just clinical evidence but the full range of scientific evidence in addition to metacognition – principles of logic and critical thinking. Critical thinking includes not just basic logic but also understanding of informal logical fallacies, heuristics, and cognitive biases. These principles apply not just in evaluating the scientific literature, but also in clinical decision-making – the best science does not help if we do not apply it properly when dealing with an individual patient.

This is why a proper medical education should include direct confrontation of the various cognitive biases and how they can adversely affect our clinical decision-making. Most of this occurs as implicit “pearls” of clinical wisdom, but I think it’s better to teach them as explicit formal critical thinking. For example, it’s one thing to teach medical students about the pitfall of giving too much weight to diagnostic “zebras” (rare but enticing diagnoses). It’s another to teach them explicitly about the representativeness heuristic (evaluating probabilities based solely on apparent similarity to type and ignoring the base rate) and all the various forms it may take. Deeper still, it’s important to teach students about the complex relationship between intuitive and analytical thinking in the various stages of decision-making.

Beyond falling for a statistically rare diagnosis, the representativeness heuristic may also cause clinicians to delay or ignore a likely diagnosis because of what are considered “atypical” features (demographic or symptomatic). It is now increasingly recognized, for example, that women receive less aggressive cardiac care simply because they are women. Why does this happen? Doctors learn that a “typical” heart attack patient is male, older, with crushing chest pain, shortness of breath, and diaphoresis (sweating). That is the essence of the representativeness heuristic – thinking about which features are typical, rather than how predictive each feature is. Being female, in the right clinical setting, is not very predictive that a patient is not having a heart attack, at least not enough that standard of care should be ignored or delayed. Yet that is exactly what happens.

Another layer of bias is that the “typical” suite of signs and symptoms of an acute heart attack were developed by studying men, long treated as the default or standard in medicine. Women tend to present a little differently, with a more varied location and type of pain, and may not have chest pain at all. This is considered an “atypical” presentation of a myocardial infarction (MI), but it isn’t. It’s atypical for men, but perfectly typical for women. Even considering this presentation “atypical” is itself a bias.

Yet another layer of bias is the notion that women are “protected” from MI because of their biology. While this is true, relative to men, it has limited relevance when dealing with an individual patient. Statistically women will get an MI 7-10 years later than men, but it is still the leading cause of death in older women. The fact that men have an even higher risk of MI is actually irrelevant. When a patient is presenting with possible symptoms of an MI, their demographic risk relative to another group does not matter. But it creates another bias against accurately assessing the risk of the individual. This is yet another cognitive bias, where we tend to consider magnitude relative to other things rather than intrinsically.

All of these biases have an effect. It is now well-established that women receive less aggressive preventive, diagnostic, and acute cardiac care than men. Adding to this is a recent study looking at the treatment of women presenting to the ER with possible MI:

Women were less likely than men to be triaged as emergent (19.1% versus 23.3%, respectively, P<0.001), to undergo electrocardiography (74.2% versus 78.8%, respectively, P=0.024), or to be admitted to the hospital or observation unit (12.4% versus 17.9%, respectively, P<0.001), but ordering of cardiac biomarkers was similar.

Other studies have looked specifically at the time delay from first symptom to definitive treatment for acute MI. The primary delay seems to be the time from first medical contact to hospital arrival, which was delayed by 28 minutes on average. Essentially women were treated with less urgency when making that first call about their symptoms.

Racial biases also exist in the delivery of medical care. This is one of those phenomena that the more you look for it, the more you find it. In the cardiac treatment study above, they also found:

People of color waited longer for physician evaluation (HR, 0.82 [95% CI, 0.73–0.93]; P<0.001) than White adults after multivariable adjustment, but there were no racial differences in hospital admission, triage level, electrocardiography, or cardiac biomarker testing.

It is reassuring that the treatment was the same, but race did seem to be an independent predictor of the delay in initiation of care. Twenty-five years of research, however, have demonstrated consistent differences in treatment and outcome. The causes are distinct from sex difference, but follow the same themes of systemic and individual biases. One systematic review lists these possible causes:

• Some black patients may decline state of the art treatments or fail to seek out care because they are less informed about symptoms, medications, and treatment related to heart disease than their white counterparts.
• Black patients may be less likely to trust the health care system based on their perception of discrimination and knowledge of historical atrocities like the Tuskegee Experiment of Untreated Syphilis in Negro Males.
• Physicians caring for black patients may be overwhelmed by the volume of patients and the severity of disease at presentation, and underperforming with regards to following evidence-based treatments.
• Unconscious bias with regard to race (“white preference”) and frank racism may result in physicians being selective in applying medical and procedural therapies.
• Black patients may be overrepresented at hospitals that underperform with regards to high-quality, evidence-based cardiac care.
• The physician workforce in the USA does not reflect the nation’s patient population, potentially exacerbating all of the above and depriving many patients of culturally sensitive, compassionate care from individuals with whom they have something in common and whose recommendations they may be more likely to follow.

In order for physicians to be advocates for their patients we need to confront these biases that can lead to worse medical outcomes for our patients. Systemic and societal issue are of course very tricky to deal with and are more of a generational project. Individual clinician biases, however, can yield fairly quickly to data and education. We now have extensive research describing the problem. One solution is to incorporate not just these individual cases, but the deeper principles of critical thinking into standard medical education.