Lawyers, Engineers, and Ménière’s: The Representativeness Bias

Updated for pithier title . . .

Composite case presentation: An elderly woman presents for two spells of vertigo. Both had abrupt onset and were quite severe, with associated diaphoresis and nausea. There were no other neurological symptoms such as diplopia. She can’t recall whether there were aural symptoms such as hearing loss or tinnitus. There was no specific provocation, such as positional change. She felt unsteady, but doesn’t recall lateralized ataxia. She simply sat down right away and the symptoms passed within 10 minutes. The medical history includes multiple vascular risk factors. She’s had progressive hearing loss over years and also chronic tinnitus. Neurological exam suggests conductive hearing loss A.U. but is otherwise normal.

Localization: This is a classically difficult problem to localize. The symptom severity and especially the prominent nausea suggest a peripheral vestibulopathy, as does the absence of other symptoms or signs referable to the brainstem or cerebellum. However, isolated vertigo can definitely occur as a manifestation of CNS disease. Another potential clue is the hearing loss and tinnitus; these suggest labrynthopathy, which may or may not be related to the presenting problem.

Etiologic diagnosis: We’ve been trying to emphasize to the students and new residents the importance of two things in arriving at an etiology (and shortly, I’ll argue for a third): time course and risk factors. Regarding CNS pathologies, the abrupt onset and short duration of these spells is suggestive of a vascular process, as is her vascular risk factor profile. Migrainous vertigo is a consideration, but there is no migraine history. Regarding peripheral processes, BPPV is a consideration, but those spells are usually brief (seconds or a minute–not 10 minutes). Vestibular neuritis is longer-lasting (days). Another consideration is Ménière’s disease, of which the paroxysmal time course and the patient’s tinnitus and hearing loss are suggestive. However, the patient’s spells lasted ~ 10 minutes whereas in Ménière’s they typically last some hours. Also, the hallmark of Ménière’s is fluctuating hearing loss and tinnitus, which wasn’t described.

The representativeness bias: Now we finally get to the topic for today. In addition to those above, there’s another reason to be skeptical of the Ménière’s diagnosis, which is its base rate (or prior probability). According to PubMed Health, the incidence of Ménière’s disease is 50-100K cases per year in the U.S. Not a trivial number, but far less than the 800K strokes and additional 240K TIAs that occur each year. That is, stroke/TIA is ten to twenty times more common than Ménière’s disease. The subtle but important point here is that the “representativeness” of the case–how closely the pattern matches the stereotype for a disorder–is independent of its base rate. We don’t think of base rates when engaging in the the mental shortcut (heuristic) of pattern-matching. However, the base rate has a powerful effect on the probability of the diagnosis.

The classic work on this subject was done by Tversky and Kahneman, who termed it the representativeness bias (this is a great paper-I highly recommend it):

Subjects were shown brief personality descriptions of several individuals, allegedly sampled at random from a group of 100 professionals-engineers and lawyers. The subjects were asked to assess, for each description, the probability that it belonged to an engineer rather than to a lawyer. In one experimental condition, subjects were told that the group from which the descriptions had been drawn consisted of 70 engineers and 30 lawyers. In another condition, subjects were told that the group consisted of 30 engineers and 70 lawyers. The odds that any particular description belongs to an engineer rather than to a lawyer should be higher in the first condition, where there is a majority of engineers, than in the second condition, where there is a majority of lawyers . . .

. . . In a sharp violation of Bayes’ rule, the subjects in the two conditions produced essentially the same probability judgments. Apparently, subjects evaluated the likelihood that a particular description belonged to an engineer rather than to a lawyer by the degree to which this description was representative of the two stereotypes, with little or no regard for the prior probabilities of the categories.

Factoring the prior probabilities of cerebrovascular vs. Ménière’s disease into our clinical judgment, it becomes clear that TIA manifesting with isolated vertigo (and the coincidence of the also very common presbycusis and tinnitus) is much more likely than Ménière’s attack. It’s as if, instead of the 70:30 split that Tversky and Kahneman used above, we were sampling from a population that had 94 stroke patients and 6 Ménière’s patients, presenting a case history that had some features of both, and asking which is the more likely diagnosis.

Now, this doesn’t mean that we can never diagnose uncommon conditions–just that base rates must be considered along with the time course and risk factor profile lest we fall into the representativeness trap. I propose that we make an effort, especially with our students, to consider this dimension of the formulation explicitly.

About Justin A. Sattin

I'm a vascular neurologist and residency program director. I created this blog in order to share some thoughts with my resident and other colleagues, and to foster my own learning as well.
This entry was posted in Medical Knowledge and tagged , . Bookmark the permalink.

One Response to Lawyers, Engineers, and Ménière’s: The Representativeness Bias

  1. Pingback: Risk vs. Uncertainty |

Comments are closed.