How Doctors Think When They Make the Wrong Diagnosis
Neuroscience evidence elucidates some of the workings behind inaccurate medical diagnosis
Making the right diagnosis is the heart and soul of medicine. Making the wrong diagnosis can have disastrous consequences.
In medical school, we were told to approach diagnosis by starting with the “differential diagnosis,” a consideration of all the possibilities. Then, we would use data from the patient’s history, physical examination, laboratory tests, and imaging results to narrow down those possibilities and arrive at the definitive diagnosis, the one upon which we would base our treatment recommendations. A great deal has been written about this process in an attempt to make it more efficient and accurate. Among other important topics has been the elucidation of how doctors actually think when confronted with a patient who has a complaint, symptom, or sign.
Let’s take an imaginary but illustrative case. A 45-year-old patient comes to the doctor and reports that she has been been coughing for the past week. She complains of headache, chest pain when she coughs, sore throat, and general malaise, and notes that she has had oral temperatures as high as 101 F during this period.
Based on this information alone, the doctor can already make a reasonable assumption that this woman is suffering from an upper respiratory infection caused by a virus. That would be the most common cause of these symptoms. Yet other diagnoses are possible, including strep throat, bacterial sinusitis, pneumonia, lung cancer, and lymphoma.
How does the doctor go about making a diagnosis? A fascinating paper published in May, 2017 in the journal Scientific Reports by collaborating scientists from Brazil and the UK used functional magnetic resonance imaging (fMRI) to get a picture of how a physician’s brain works when considering diagnostic information.
Data were analyzed from 31 primary care doctors who were given several tasks while undergoing functional brain imaging. Functional MRI takes a snapshot of brain activity and the shorter the interval between these snapshots, the greater the chance to capture the rapid fluctuations in brain activity that occur as we think about things. In this study, images were recorded every 400 milliseconds, an unusually rapid sequence that permitted seeing the brain’s activity as the doctors mull over information and formulate their ideas.
When the doctors were asked to name animals and objects, a network was activated in the brain called the frontoparietal attention network (FPAN), sometimes also called the salience network. The FPAN is a group of brain structures extending from the back of the brain (parietal lobe) to the front (frontal lobe) that is activated whenever we pay attention to a stimulus. It helps us focus our attention and ignore extraneous stimuli.
Giving the doctors diagnostic information and asking them to formulate a diagnosis activated the same brain network as naming animals and objects did. So did asking them to make a treatment decision.
Of most interest, the study showed that giving the doctors diagnostic information that has less information so that the diagnosis is more uncertain was accompanied by greater activation of the FPAN than when they were given information that is highly informative. In other words, as the doctor becomes more certain of the diagnosis, his attention to the problem decreases.
In our example, the constellation of fever, cough, chest pain, headache, sore throat, and general malaise in a 45-year old woman has relatively high uncertainty as to the actual diagnosis. As the doctor learns that the patient is a non-smoker and hears the sounds called “rales” with her stethoscope at the base of the patient’s right lung, diagnostic certainty increases and the doctor’s FPAN activity decreases. She is already considering a diagnosis of right lower lobe pneumonia and considering which antibiotic to prescribe.
Finally, the doctor gets the results of a chest X-ray, showing an a cloudy haze called an infiltrate in the right lower lobe and both the diagnosis and treatment are certain. When this happened in the study, the researchers observed “an unexpected and remarkable switch” in brain activity: activity switched from the FPAN that had been working so far to a set of regions on the right side of the brain that are used to monitor sounds as we hear them. They propose that this switch is the neural correlate “of becoming aware of one’s own responses.”
The study’s findings raise several concerns. As the authors speculate, making a diagnosis prematurely likely shuts off the FPAN attention network, so that no further possibilities are considered. Premature diagnosis is an important component of medical errors. If our doctor heard “fever, cough, chest pain” and immediately thought “cold” she might experience a decrease in FPAN activity so that further information would be ignored and the correct diagnosis of pneumonia missed. The point at which activity switches to the auditory centers of the brain may signal the end of openness to consider new data and reconsider the original diagnosis.
This study shows what happens in the brain as a doctor shifts through evidence and reaches a decision. It gives us insight into how the human brain makes it difficult for us to change our minds. It would have been interesting if the authors had tested this latter concept out directly by examining brain activation patterns when doctors had already shown the neural shift and giving them new information that challenges the original diagnosis.
For example, what would happen if the doctor in our example were told that a more senior radiologist re-read the X-ray and noted a small mass in the right lower lobe. This suggests a tumor. Would our physician reopen the diagnostic process and would her FPAN attention network reactivate? Or would she find it difficult to integrate this new information and reconsider the pneumonia diagnosis because her brain had already reached closure?
What does this kind of study actually add anything to what we already know? For one thing, seeing the way the physical brain responds to a diagnostic challenge helps us understand the rapidity with which doctors make diagnostic decisions and the roots of premature diagnosis. Using brain imaging, future studies can test out methods to improve physicians’ diagnostic skills with objective markers to show whether their thinking process is really being altered.
Despite our many technological advances, making the right diagnosis still depends on a doctor’s cognitive skills. Human error looms as a threat to every patient’s outcome and safety. We need more studies like this one that help us understand the mental processes physicians use to make diagnoses, how these can go awry, and what we can do make them more accurate.