You’re likely familiar with the adage, “When you hear hoofbeats, think of horses, not zebras”—meaning, look for the suspected diagnosis, not the exotic or surprising one (the zebra). But failure to acknowledge the zebra can lead to unintended bias.
Diagnosis and treatment are central to clinical care—physicians draw from years of education, training, and practice to deduce an assessment and create a treatment plan. But what happens if there is a flaw in this process? Biases touch on countless aspects of cognition, including those involved in clinical decision-making.
Confirmation bias can be detrimental to patient health by sending the clinician down a rabbit hole of mistaken treatment. In turn, a mistaken diagnosis can alter the trajectory of patient care and impact decisions made by other clinicians who may not question the validity of the initial diagnosis.
Here’s a closer look at confirmation bias, a type of bias that can taint initial diagnosis, and what to do about it.
What is confirmation bias?
According to the author of a letter published in the Journal of the American Academy of Dermatology (JAAD), clinical reasoning occurs in two parts. First, an initial diagnostic hypothesis is formed. Second, this hypothesis is tested and a process of elimination occurs, resulting in confirmation. Confirmation bias refers to a propensity to give greater weight to the initial hypothesis and then advance that hypothesis without adequate testing.
In an article published in the AMA Journal of Ethics, authors formally characterize this two-part pattern of clinical reasoning as the dual-process theory, with the first part being intuitive and the second part being analytical. Intuitive reasoning relies on heuristics, or mental shortcuts, whereas analytical reasoning is deductive in nature.
An example of intuitive reasoning would be the automatic assumption during the winter season that a patient presenting with fever, fatigue, and joint pain has influenza. Analytical thinking would open the door to other less obvious options, such as meningococcal meningitis (ie, the zebra).
Confirmation bias is associated with brain maturation, and it is more prevalent in adults vs children or teens. In clinical practice, common examples of confirmation bias include disregarding lab results that don’t support the favored diagnosis, and failure to acknowledge the possibility of other diagnoses.
Cognitive bias is commonly associated with another form of bias called anchoring bias. The latter refers to bias based on initial data. Cognitive bias comes in the following forms:
Availability bias: the propensity to think that the diagnoses that come to mind first are correct
Response bias: assuming responses from patients are inaccurate
Omission bias: the predilection to favor omission over commission
Sunk-cost bias: resolution to stay with a concept because of all that has been invested so far
Premature closure: heavily favoring the initial diagnosis
Overconfidence: feeling a level of confidence that outweighs the data
Framing effect: semantics favoring a diagnosis
Risk factors for confirmation bias
Physicians face numerous stressors that prove to be perfect breeding grounds for confirmation bias, according to the authors of the aforementioned AMA Journal of Ethics article.
”Consideration should be given to the difficulty physicians face in employing analytical thinking exclusively,” they wrote. “Beyond constraints of time, information, and resources, many physicians are also likely to be sleep-deprived, work in an environment full of distractions, and be required to respond quickly while managing heavy cognitive loads. These are working conditions in which analytical thinking strategies are difficult to apply, especially given that they require the cooperation of brain structures that suffer greatly from sleep deprivation.”
Overcoming confirmation bias
Humans are naturally inclined to confirmation bias. Fortunately, there are ways to surmount confirmation bias and open the doors to a proper diagnosis.
One method involves making a conscious effort to disprove the favored diagnosis by balanced testing in lieu of confirmatory testing.
Ironically, reviewing the patient chart and past medical history may actually introduce confirmation bias. In addition, assigning too much weight to disease prevalence and patient characteristics may also lead to confirmation bias. Although these steps are integral in establishing a correct diagnosis, their potential to mislead must also be considered.
Other strategies to decrease confirmation bias are reflection and cognitive forcing.
With reflection, the physician seeks out and is open to alternative diagnoses, and is willing to engage in effortful reasoning when coming to a conclusion. On the other hand, cognitive forcing involves conscious evaluation of alternative diagnoses in addition to that which comes naturally.
Confirmation bias is a natural human reaction. Physicians should fight the temptation to treat the first diagnosis that comes to mind. Reflection on other options is integral to forming a robust treatment plan. Although horses are more common than zebras, zebras are out there.