The Hoofbeats of Zebras

That clinical shorthand for prioritizing a diagnostic work-up is called an anchoring bias when physicians focus on an initial piece of information “when formulating a diagnosis without sufficiently adjusting to later information.” A common and concerning presentation by patients in the Emergency Department is a complaint of shortness of breath – for which there are several possible causes and diagnoses.

High on the list is congestive heart failure (CHF), for which an increase in B-type natriuretic peptide (BNP) is an important test, and pulmonary embolism (PE), for which CT imaging is a more appropriate test. The researchers looked at which path physicians in the VA Emergency Departments took when they knew that the patient had a history of CHF. In the setting of a patient with a history of CHF, did the doctor hear the hoofbeats of CHF or those of PE?

Using a national database of VA patients presenting to the Emergency Department between 2011 and 2018, they identified patients aged 30 or older who presented with “shortness of breath.” They excluded patients on anticoagulants as this might suggest a higher likelihood of a PE, as anticoagulants are a standard part of PE care. They considered 108,019 patient visits, with a mean age 71.9, and overwhelmingly male (97.5%). To generalize,

• 4.1% of visits mentioned CHF. Among those patients, the duration of a diagnosis was CHF was more extended, and they had more inpatient admissions for CHF that year. They were less likely to have a recent diagnosis of cancer, less likely to have a prior deep vein thrombosis or PE, current elevation of their heart rate, or diminished oxygen saturation – all findings that went against a diagnosis of PE
• 13.2% of patients were tested for PE, 71.4% tested for BNP
• 0.23% had a diagnosis of PE made in the ED, and another 0.87% were diagnosed with PE in the next 30 days.

After adjusting for various clinical and demographic variables, a mention of CHF before evaluation resulted in an anchoring bias; those CHF hoofbeats resulted in 8.8% of patient visits involving PE testing. Without the mention of CHF, the rate of PE testing rose to 13.4%. An acute PE was diagnosed less frequently in those with a comment of CHF in 0.08% of visits, compared with 0.23% of visits where CHF was not mentioned. But before we believe this bias has resulted in disparate care,

“… we failed to find a difference in the rates of ultimately diagnosed acute PE between these visits compared with visits with no mention of CHF (1.2% vs. 1.1%).”

The researchers also point to quicker testing for PE when CHF was not mentioned by about 15 minutes. However, as any physician who has spent time in a VA or civilian Emergency Department will attest, time to testing involves much more than when the physician orders the test. [1]  

“…these findings suggest that the initial visit label of CHF, which may have anchored physicians away from PE, was associated with delayed work-up and diagnosis of PE.”

I certainly cannot argue that this finding was correct, but so what? The metric of care is not necessarily the speed of diagnosis; it is whether the patient has been treated and not harmed. There is no information on whether those with a diagnosis of PE made after that ED visit suffered medical harm. More importantly, if you return to the definition of anchoring bias presented by the authors,

“when formulating a diagnosis without sufficiently adjusting to later information.”

There is no evidence that physicians did not sufficiently adjust based on later information. Even in their discussion, the authors note that while the triage note mentioning CHF may have biased physicians, they cannot exclude that the patient was not also the influencer, framing their shortness of breath as a “recurrence” of their CHF.

The medical heuristic about hoofbeats is not based on statistical analysis or contemporary understanding of our cognitive biases; it is based on what Greek philosophers called phronesis, practical wisdom, and as this study has not disproved, results in another Greek term, praxis, thoughtful doing.

[1] Among the other factors are how busy the CT scanner or ultrasound departments are, whether laboratory work to allow for the use of dye has been ordered and reported, and available transportation to the imaging areas, among other factors.

Source: Evidence for Anchoring Bias During Physician Decision-Making JAMA Internal Medicine DOI: 10.1001/jamainternmed.2023.2366