Every Patient Tells a Story (32 page)

Physician Bias, Fair and Otherwise

“Doc, my knee, it’s doing this thing again.” Vera Freeman pointed to her red and swollen knee as I entered the small, poorly lit hospital room. She was an attractive young woman, with hair stylishly braided and ornamented with bright beads. “Last night it was just fine,” she reported. “Now just look at it.”

Two weeks earlier, she awakened to find her ankle—not her knee—swollen and painful. She didn’t remember injuring it. “It just blew up,” she said, and when she took it easy for a couple of days, it got better. “But just as soon as it was okay, my wrist swelled up. It was big, and it really hurt. I was getting worried, but it got better too.” The next week, though, her knee began to swell, and she decided to come to the hospital. “It was so weird. It was like I had this swelling that just didn’t know where to light.” She looked at me carefully, to see if I was following her story.

She stayed in the hospital for a couple of days, received some intravenous antibiotics, and was sent home with antibiotics to finish by mouth. She took the pills for a couple of days, but once she felt better, she neglected to take the rest. Now the pain and swelling had wandered back, and she wanted to know why.

Freeman was frank about her history. She had HIV, which had been diagnosed three years before. Otherwise, she thought she was pretty healthy. She did not smoke cigarettes or drink, though she did admit to smoking crack cocaine “occasionally.” She had no children, lived in an apartment with her longtime boyfriend. She had, at times, worked as a prostitute to help her buy crack.

On exam, her dark brown skin felt warm. Moving the joint elicited a sharp cry of pain. As I gently explored the swollen knee, I could feel fluid moving around, like a warm, firm water balloon. The kneecap was separated from the joint it normally covers; I could press it down almost an inch before I felt contact. As I examined her, I assembled a differential diagnosis in my head. A hot, swollen joint is routine in medicine, usually caused by trauma, by gout, by infection. But this “wandering” pain was far from routine. In the textbooks it’s known as a “migratory polyarticular arthritis”—that is, an arthritis that moves from joint to joint—and it is an extraordinary manifestation of just a few pretty ordinary diseases.

It is seen most commonly with gonorrhea (although it is unusual even in this disease), where it is often accompanied by fever and a rash. Lyme disease can also manifest this way, as can viruses like hepatitis and even HIV. But none of those seemed to fit. There were other, less likely possibilities. Rheumatoid arthritis can come on like this, as can lupus.

Justin Thompson, the intern working with me that month, had admitted Freeman for her earlier hospitalization. When I asked him about her, he wearily flipped through a stack of index cards that he pulled from his pockets. “Right. We tapped her knee and cultured her up,” he said, meaning that they’d drawn fluid from her knee, which should offer some clues, and had sent off some of the fluid, as well as her blood and urine, to check for evidence of infection. “I thought it was gonorrhea,” the intern stated flatly. “It’s not the way you usually see it, but gonorrhea can definitely cause this.”

The art of diagnosis can look a lot like profiling. Doctors constantly ask: Is a particular condition more common in men or women? Whites or blacks?
The young or the old? In this way a doctor narrows the possible causes of a given illness in a given patient. Gonorrhea, then, was the most likely diagnosis for this young, sexually active, onetime prostitute. And while none of the tests had confirmed it, none ruled it out either.

But here she was again, knee gigantically swollen—again. This was not part of the disease profile, yet the intern working with me was undeterred. So much so that he had already ordered antibiotics to treat her presumed infection. Because she did not finish her course of antibiotics, the disease had been only partly treated; therefore, all she needed was more antibiotics. “Or maybe her boyfriend was the source,” he said, “and she’s been reexposed since getting treated. Or maybe she’s back on the street.”

These were all reasonable thoughts, but it was clear to me that we needed more evidence to make that diagnosis a second time. I thought we should hold the antibiotics until after we tapped the knee again and repeated her cultures.

I was also interested in the results of the blood tests from Freeman’s earlier hospitalization. I found a computer and tracked down her test results. Lyme was negative; hepatitis, negative; gonorrhea and syphilis, negative. In fact, there was only one set of positive results: the tests for recent strep infection along with several other blood tests consistent with a diagnosis of rheumatic fever. The problem is that rheumatic fever rarely occurs in these days of antibiotics, and when it does it is seen almost exclusively in children. It is practically unheard of for an adult to develop it. Even now that she met some of the criteria for the disease, such a diagnosis was hard to make. She simply didn’t fit the profile.

We went back to the patient. Had she had a sore throat recently? Yes. She’d had a sore throat a few weeks before, but she thought it could have been because of the crack. That convinced me. It now seemed clear that, as unlikely as it might have seemed initially, this young woman had rheumatic fever.

When we went back to the patient, she was dressed and ready to leave. Her knee, which had been red and hot and excruciatingly painful only twenty-four hours earlier, had improved significantly with no intervention.
We scheduled her to see her doctor the following week. As she stood with her bag in hand, I tried to explain rheumatic fever and what it might mean to her, but she wasn’t listening.

“I’m better,” she announced, “so I’m gone.” I gave her her prescriptions and shook her hand, then watched as she limped down the hall, waved gaily from the door, and disappeared.

Recently I caught up with her doctor, who told me that Vera had gotten an echo to look for any signs of damage to her heart or their precious valves that direct the flow of blood through the organ. Everything was completely normal. And it made sense. Cardiac injury is very common in the children who get rheumatic fever; in adults the disease tends to “bite the joints and lick the heart,” causing joint pain but not the more significant cardiac lesions.

What has always stuck in my mind is the intern’s insistence on the diagnosis of gonorrhea even in the face of failed tests for that condition. Was he just being prejudiced against a minority woman with a history of behaviors not sanctioned by the larger society? Possibly, but I think the story is more complicated than that.

At first glance, patients might think that the ideal in diagnosis would be for a doctor to treat (and view) all of their patients identically—to be color, age, gender, and socioeconomically blind. We don’t want our looks to influence our doctor’s objective assessment of our health problems. And yet they must. Illnesses and diseases do not abide by our constitutionally required equal protection. Diseases
do
discriminate on the basis of race, gender, age, and even socioeconomic status.

To take a obvious example: the vast majority of breast cancer patients are women, so it is not wrong for a doctor to automatically drop that diagnosis down in her priority list when confronted with a male patient with a lump on his chest. A less obvious example is prostate cancer: black men are significantly more likely to get this type of cancer than men of other races—four times more likely, in fact, than Korean men, nearly twice as likely as men of European descent. So if a black man comes to a doctor complaining of urinary symptoms, a good doctor will automatically raise her level of suspicion for prostate cancer based solely on the color of the patient’s skin. In fact,
it would be irresponsible of the doctor
not
to take race into account when considering this diagnosis.

Viewed in this light, the bias of the intern clinging to his suspicion of gonorrhea in a woman with a history of drug use and prostitution is not so egregious. Using drugs and having multiple sex partners, after all, are legitimately associated with an increased risk for sexually transmitted infections. What would be egregious is if the intern (or anybody else) settled on a diagnosis of gonorrhea based
only
on the color of the woman’s skin, her clothing, or some other aspect of her appearance or behavior, all of which have nothing to do with one’s risk of gonorrhea.

In other words, patients want doctors to be
legitimately biased
in their thinking and decision-making processes when struggling to find a diagnosis. Doctors should take into account any known associations that might help pin down the cause of an illness. But diagnoses can be missed when doctors apply false generalizations or close off diagnostic possibilities just because they are less likely in a certain group or population (e.g., “This can’t be HIV because the patient is elderly”). Research has shown that medical decision making is shaped by many of the same influences that distort other aspects of human interaction. Indeed, says one group of researchers, “despite their ‘objective’ medical training, physicians remain human actors, socially conditioned to engage in stereotyping, whether consciously or not.” In that respect, medical decision making can be a function of who the patient is as much as what the patient has.

Studies in social science have documented many nonmedical factors that influence medical decisions, including characteristics of the patient such as age, gender, socioeconomic status, race, or ethnicity. These can be important considerations in prioritizing possible diagnoses. But characteristics that have no obvious medical meaning—such as the presence or kind of health insurance, assertive personality type, or even physical attractiveness—have also been shown to play a role in how doctors make decisions about medical diagnosis and care. And even those factors that can affect the probability of disease in some cases, factors such as age and sex, are extraneous in many others.

One of the many careful experiments designed to tease out such influences
illustrates this point. A set of videotaped doctor-patient encounters was created using professional actors. Scripts were created for male “patients” and female “patients” who complained of a set of cardiac symptoms. The scripts and all presenting details were identical in every respect aside from trivial changes in personal pronouns and the like. Two hundred fifty-six doctors practicing in both the United States and the United Kingdom were recruited for the study. They viewed either one or the other video scenario and were then asked a series of questions about what disorder they suspected, what treatments or recommendations they would suggest, and so on. Coronary heart disease (CHD) was chosen because it is the leading killer of both men and women, and although age-specific mortality rates are higher for men than women, twice as many women as men aged forty-five to sixty-four have undetected or “silent” CHD, which suggests that the true incidence between men and woman may be similar. This is a case, in other words, in which doctors should
not
apply a bias in their decision-making processes—here a gender bias.

The study results, however, clearly demonstrated just such a bias. Gender was found to have significant influence on all aspects of doctors’ diagnostic strategies; in each case women received less attention than men presenting with CHD symptoms. Doctors would ask men more questions than women (on average 7 and 5.7 questions, respectively) and perform more extensive examinations for men than women (5.1 compared to 4.3 parts of the body or body systems would be examined, respectively). CHD was mentioned as a possible diagnosis for more men than women (95 and 88 percent, respectively), and doctors had significantly higher certainty of CHD for male than female patients, 57 and 47 percent, respectively, on a scale of 0 (total uncertainty) to 100 percent (total certainty).

The study authors concluded: “Our findings indicate that women presenting with CHD symptoms are disadvantaged in primary care. Doctors provide a less thorough diagnostic search procedure than for men presenting with identical symptoms, and fewer women are given prescriptions appropriate for treating CHD.”

The impact of conscious or unconscious bias on the diagnostic thinking
processes of doctors adds to the complexities of the entire doctor-patient experience. The best doctors acknowledge their vulnerabilities and try hard to retrain themselves or monitor themselves and their thinking processes as they move through any given diagnostic challenge.

The last type of cognitive error I want to talk about is what’s often called in the cognitive literature diagnostic momentum. This is a kind of medical groupthink in which once a diagnostic label is attached to a patient, it tends to become “stickier and stickier.” Doctors are taught in medical school that they should not simply accept a diagnosis given to a patient but should reevaluate the data for themselves before accepting or sometimes rejecting this diagnosis. That we should, as former president Ronald Reagan often exhorted (in a very different setting), “Trust but verify.” Rather than accept a previous diagnosis, doctors are supposed to start fresh by thinking things through for themselves. This, of course, is much easier said than done.

If a doctor is tired or in a hurry, she is far less likely to take the time to review all the test results and other evidence that went into the diagnosis. And even if she does expend the effort to do that, it’s difficult to not fall into the same well-defined disease pattern—potentially mistaken or not—that those who have seen the patient already have defined. But that kind of extra effort can sometimes pay off dramatically.

The Doctor of Last Resort

Graciela Moity spoke in a slow, husky vibrato. She sounded weary, discouraged. “I can remember clear as day when it all began,” she said. “It was just over a year ago. I woke up and felt like my legs were on fire.”

She was talking to Dr. David Podell—the most recent of a train of doctors who had evaluated the woman since that day she awoke in such pain. The three previous doctors couldn’t figure out what was going on. Their best guess was scleroderma, a disease caused by the overproduction of one
of the connective tissues, collagen. The patient’s symptoms weren’t a great fit, but sometimes the disease could manifest itself in unusual ways. She was referred to Podell for confirmation of the diagnosis and treatment of this unusual autoimmune disorder.