Discussion Board

I am attaching a paper I wrote recently on role of heuristics in diagnosis for review and comments.

I point out the notion of heuristics as a source of diagnostic errors is based entirely on the studies of Tversky and Kahneman who found that subjects make faulty probability judgments from given information by neglecting important probability rules due to influence of heuristics.

But when we analyze the process of diagnosis in practice, we find physicians do not make probability judgments (diagnose) from presentations. Instead, they suspect a disease from a presentation, formulate it as a hypothesis and diagnose it only if the hypothesis is verified to be correct by tests.

But this process of hypothesis generation and verification which occurs in diagnosis in practice is not a part of Tversky and Kahneman's version of heuristic reasoning. Therefore, I suggest, their version of role of heuristics is not applicable to diagnosis.

I point out that a heuristic plays an essential role in generation of a hypothesis without which diagnosis is not possible.

We may need to alter our view about the role of heuristics in diagnosis.

Thanks.

Bimal

Bimal Jain MD

Northshore Medical Center

Salem MA 01970. 

The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Mass General Brigham Compliance HelpLine at http://www.massgeneralbrigham.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.

A heuristic, which is a mental shortcut, plays a very different role in diagnosis than it does in daily life. In daily life, a heuristic leads to a judgment being made from initially available information such as a birdfeeder being empty when there are no birds in backyard. This judgment may or may not be correct. In diagnosis, on the other hand, a judgment (diagnosis) is almost never made from a presentation (initially available information). For example, acute MI is not diagnosed from chest pain in a patient. Instead, this presentation leads to acute MI being suspected due to the heuristic of resemblance and formulated as a  hypothesis which is evaluated by a test. The test results in the hypothesis being correct or not; then  acute MI is ruled in or ruled  out.

Thus while a heuristic may lead to an error in judgment in daily life, it does not lead to an error in diagnosis due to the hypothesis generation and verification method. It is unfortunate that the error prone results of Tversky and Kahneman's studies on heuristics in daily life have been applied to diagnosis to conclude they are a source of diagnostic errors. In reality, a heuristic is employed in diagnosis like it is employed in science which is to generate a hypothesis that is always evaluated by an experiment or test. Thus it plays an essential role in diagnosis in generating a hypothesis without which diagnosis is not possible.

In some pediatric conditions, heuristics serve as a reliable way to diagnose patients without formal hypothesis testing or obtain diagnostic studies. Two examples come readily to mind.

 

1. 2 AM, November, a crying toddler is carried past the physician workstation exhibiting a barky cough, stridor at rest and suprasternal retractions and his triage vitals show a fever, tachypnea, normal sats and only mildly elevated HR. The heuristic representativeness restraint (if looks, quacks and flies like a duck it's a duck) allows the pediatric ED physician to place orders for racemic epi neb and dexamethasone with this initial information and without having to take a history, or do more of an exam than is described above. The likelihood that the peds ED doc is right is very high because 1) few other conditions present in this way 2) immunization rates for Haemophilus influenza b are so high that the epiglottitis has virtually disappeared from the map. Indeed, I have seen children just like this present to non-peds EDs and have unnecessary neck radiographs obtained and then misinterpreted by non-peds radiologists as epiglottitis prompting unnecessary antibiotics and helicopter critical care transport to a peds hospital. The hypothesis generation and "rule out" strategy was more dangerous to the patient than reliance on heuristics. Of course, one should ask about FB aspiration and immunization status because a heuristic diagnosis without checking off a few other boxes could get one into trouble.
2. A 5 y/o boy presents with concerns of rash only on his legs, mild abdominal , ankle and wrist pain and mild swelling of his legs. His VS are all normal for age. The clinician "search satisfies" when the exam reveals palpable purpura and petechiae restricted to the BLE and she diagnoses Henoch Schoenlein Purpura. At minimum a UA should be obtained to identify signs of nephritis and, if found, perhaps a renal function panel to identify signs of AKI. Yet, no hypothesis or differential is likely to be of use to the seasoned clinician in this case and, indeed, doing an exhaustive search for other causes of purpura (say ITP, meningococcemia) to test the hypothesis very well places the patient at risk of complications of the hypothesis testing (e.g. false + blood culture leading to admission and unnecessary antibiotics).

 

I find that, similar to comments made by Geoff Norman and colleagues (McLaughlin, Eva & Norman. Reexamining our bias against heuristics. Adv in Health Sci Educ (2014) 19:457–464), "under certain circumstances using heuristics may lead to better decisions than formal analysis." Yet, I have also fallen prey to diagnosis momentum and confirmation bias as I mentioned in another recent post about missing a case of Guillain-Barre. There is a place in medicine for both intuitive diagnosis and more analytic reasoning. Both are prone to error under certain conditions. To say one is better than the other misrepresents the nuance of the diagnostic process.

 

 

---

CONFIDENTIALITY NOTICE:  This e-mail, including any attachments, is for the sole use of Childrens Hospital Colorado and the intended recipient(s). It may contain confidential and privileged information or may otherwise be protected by law. Any unauthorized review, use, disclosure, or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply e-mail and destroy all copies of the original message and any attachment thereto.

Hi everyone,

I just read the attached paper - it was just published today online ahead of print, in the journal Radiology.  I thought it would be of high interest to members of this list-serve and to the SIDM community.  You will undoubtedly recognize their Figure #1!

In this brand-new paper, hot of the press today, Dr. Steven Waite, a radiologist at SUNY-Downstate in Brooklyn and colleagues review and comment on the problem of racial disparities in radiology image utilization.  Of course, radiology imaging is only one piece of the diagnostic process, but it is gaining in utilization and importance.

In this group we have previously discussed the imperative of appropriate imaging utilization in the context of over-utilization, which is a recognized problem leding to significant risk of misdiagnosis/over-diagnosis for patients (among other risks).  This paper highlights the opposite problem, namely the underutilization of imaging, and explores the question as to what expent disparities remain in imaging utilization which might track with ethnicity/race and not only with S.E.S.  I think it is another important part of the conversation.

All the best,

Mike

Thanks, Neal - excellent summary and resources.

Mike