Your Medical Chart Might be Biased

by Danielle Ofri
Slate Magazine

We’ve known for a long time that there are racial disparities in health outcomes. For about as long, we’ve tried to chalk this fact up to recalcitrant socioeconomic issues. But a cover story in the New York Times Magazine pointed out why this is too easy a dismissal. Economic issues can not explain all of the disparity: At any given level of income, maternal and fetal outcomes are worse for African American women and their babies than for their white counterparts. This holds true even for the most affluent groups who all, presumably, had access to excellent medical care.

The conclusion of many experts is that just living as a black person in America creates a “toxic level of physiological stress” that increases the risk of conditions like hypertension and pre-eclampsia, among others. We’ve known about toxic stress, but we’ve always lumped it in with poverty, not necessarily with race This article made me rethink the role we medical professionals play in creating this toxic environment, and consider what we can do to alleviate it.

A recent paper caught my eye because it captured one of the more subtle aspects of the brew: how we write about patients in the chart. Mary Catherine Beach and her colleagues at Johns Hopkins University were curious about whether our choice of language transmits bias from one medical professional to another. The researchers created a hypothetical case of an African American man with sickle cell disease, a condition that typically requires opiate medications for control of the painful flares. They wrote two versions of the medical chart, one with neutral language and one with language—taken from real charts— that could be viewed as more stigmatizing. Medical students and residents were randomized to read one of the charts and then asked about their attitude toward the patient and how much pain medication they would prescribe.

Those trainees who read the chart with the more stigmatizing language exhibited more negative attitudes toward the patient and elected to give less aggressive pain treatment. This result is probably not surprising—we know that black patients tend to receive lower rates of pain treatment. But what is intriguing is how subtle the differences in language were between the two charts. In the first chart, the patient was described as a “28-year old man with sickle cell disease” and in the second chart as a “28-year old sickle cell patient.” Before the symptoms occurred the patient “spent yesterday afternoon with friends” vs. “was hanging out with friends outside McDonalds.”

For the physical examination, the doctor observed in the first chart that the patient “is in obvious distress,” and in the second that the patient “appears to be in distress.” A nursing note in the first chart reported that the patient “is not tolerating the oxygen mask and still has 10/10 pain,” and in the second chart that the patient “refuses to wear his oxygen mask and is insisting that his pain is ‘still a 10.’”

The descriptions in the second chart weren’t necessarily inaccurate, but together they subtly paint the patient as a less reliable person, someone who perhaps is trying to game the system for drugs. According to Beach, this type of language not only discredits the patients’ report of pain, but highlights details that reinforce negative stereotypes. Medical charts are the primary means of communication among medical professionals, so this sort of language covertly signals to other members of the team that this is a “low class” person who isn’t trustworthy or deserving.

As soon as Beach put it this way, I could see that our supposedly objective medical records contain racially laden “dog whistles” of the sort that we regularly decry in political speech. In the last two years we’ve gotten more adept at noticing and calling out references to inner cities, illegal aliens, international bankers, Sharia law, and locker-room talk,

We doctors like to think that we treat all our patients equally. We would never think of ourselves as racist or marginalizing. Yet it’s there in our language. During my internship, there were always some patients who elicited a particularly negative response, usually from some combination of drug addiction, homelessness, abrasive personality, and perceived overuse of the system. I’d sometimes hear a staff member mutter about them as “not a citizen.” The message was: don’t expend your efforts on these patients.

Beach and her colleagues demonstrate how the medical chart can be used to communicate this message between staff members. Whether done consciously or unconsciously, this dog-whistle language plants the seeds of doubt in the mind of the next person who reads the chart and can influence the subsequent medical care.

The factors that perpetuate the disparities in medical care are varied. Some require government intervention. Some require shifts in societal attitude. Some require restructuring parts of the medical system. But some are within our power as individuals. One thing we doctors and nurses can do is to take a minute to think about how we are writing. We can choose to write that a patient “declines treatment” rather than “refuses treatment.” We can write that a patient “doesn’t drink alcohol” rather than the patient “denies alcohol use.” We can write that a patient “has pain” rather than that he or she “claims to be in pain.”

The substandard quality of care that some of our patients receive is a stain on the medical profession. Individual doctors and nurses can’t solve it all, but we should be on forefront agitating for change. And we can begin with the words we write in the chart.