Skip to main content

23-12-2010 | Mental health | Article

Clinicians show heuristic bias in bipolar disorder diagnosis


Free abstract

MedWire News: Only 40% of clinicians make a correct diagnosis of bipolar disorder when presented with a definitive case and are subject to heuristic bias, a survey shows.

Clinicians were more likely to make a correct diagnosis when more symptoms were present - meaning patients who just meet the threshold criteria are at risk for undertreatment, report Larissa Wolkenstein (University of Tübingen, Germany) and colleagues.

"Given the high prevalence and the high suicidal risk of bipolar disorder, it seems essential that clinicians are well trained to diagnose bipolar disorder correctly," they comment in the Journal of Affective Disorders.

Some studies have shown heuristic bias in bipolar disorder diagnosis, for example giving disproportionate weight to certain "prototypic symptoms" such as reduced need for sleep.

To investigate further, the researchers presented a case vignette that fulfilled the criteria for bipolar disorder to 204 psychotherapists.

The basic vignette was a patient who presented with depression and evidence of three out of the seven possible hypomanic symptoms on the DSM-IV. This was modified to include an additional fourth hypomanic symptom of reduced need for sleep or distractibility - thus giving three variant vignettes.

In addition, half of all vignettes included a potential casual explanation for hypomania - meeting a new partner (which does not preclude a diagnosis on the DSM-IV).

Overall, bipolar disorder was correctly diagnosed in 41.0% of the cases; in 59.0% of the cases another diagnosis was made, mainly unipolar depression (50.3%). Seven (3.8%) of those therapists, however, made a note indicating that they suspected a diagnosis of bipolar disorder.

Analysis showed that case vignettes with four hypomanic symptoms were more often diagnosed correctly (for reduced sleep, 47.3% and distractibility, 57%) than were those with only the basic three symptoms (20%), giving a significant odds ratio for misdiagnosis of 5.5, when only the basic three symptoms were present.

Regression analysis revealed no significant difference between the additional symptoms in terms of correct diagnosis, suggesting the clinicians were working on an additive model, which is at odds with the polythetic model of the DSM.

There was a borderline significant trend for a causal influence on diagnosis, such that vignettes where the case reported recently meeting a new partner were slightly less likely to be correctly diagnosed as having bipolar disorder.

Finally, the researchers found that clinicians who misdiagnosed were less likely to recommend appropriate medication.

"Given that therapeutic strategies depend on the assigned diagnostic label, which might not only be inefficient but even harmful when making a wrong diagnostic decision, it becomes clear that a standardized diagnostic proceeding is in great demand," Wolkenstein et al comment.

MedWire ( is an independent clinical news service provided by Current Medicine Group, a trading division of Springer Healthcare Limited. © Springer Healthcare Ltd; 2010

By Andrew Czyzewski

Related topics