Cognitive Bias and Error in Diagnosis

Given the enormous amount of information that doctors must learn, retain, and process daily, it is no surprise that they are subject to a number of cognitive biases and errors (just as all people are) that are significant contributors to nondiagnosis and misdiagnosis. Having an understanding of these errors can help doctors to avoid or address them and allow patients to better manage their care.

Cognitive errors in medicine have garnered increasing attention as more and more studies have shown them to play an important role in diagnosis and treatment. Dr. Jerome Groopman, Chaired faculty member at Harvard Medical School, details many types of these errors in his book How Doctors Think, an insightful and accessible exploration for patients and doctors alike. Dr. Gurpreet Dhaliwal, Assistant Clinical Professor of Medicine at UCSF has also contributed important elements to our understanding of cognitive biases and errors. A few of these that particularly hinder effective diagnosis include:

Anchoring: The tendency to disproportionately weight (anchor in on) a subset of information to make a decision. Once a patient receives a preliminary diagnosis, it can often be hard for doctors to think of other potential diagnoses as they anchor on this hypothesis.

Confirmation bias: The tendency to seek out or interpret data in a way that confirms a hypothesis. If a doctor has a diagnosis in mind, he may be more likely to test for and recognize signs and symptoms that support this diagnosis and dismiss those that contradict it.

Premature closure: The tendency to consider a case to be solved once a satisfactory solution has been achieved without sufficiently considering alternative solutions. A doctor may stop considering other diagnoses once a diagnosis has been reached.

Inter and intra observer error: Variation in the interpretation of evidence across multiple views by an individual (intra) or between different individuals (inter). For example, the same radiologist looking at the same image may see different things on different occasions, while different radiologists looking at the same image will see different things.

How pervasive are these problems? Groopman’s research found that 80% of medical errors were cognitive. In a Harvard study, Leape et al report that “The proportion of adverse events due to negligence was highest for diagnostic mishaps (75 percent).” In a study at Veterans Affairs, Graber et al found that “System-related factors contributed to the diagnostic error in 65% of the cases and cognitive factors in 74%.” Raab et al, in a study supported by AHRQ, found that cancer diagnosis errors “ranged from approximately 2 to 10 percent of gynecologic cases and from approximately 5 to 12 percent of nongynecologic cases at various hospitals.” These findings highlight the pervasive nature of these challenges and the importance of doctors and patients facing complex and difficult diagnoses to work to mitigate them.