Diagnostic bias in clinical decision making: An example of L-tryptophan and the diagnosis of eosinophilia-myalgia syndrome

Objective. Eosinophilia-myalgia syndrome (EMS) has been defined as the clinical presentation of eosinophilia, severe myalgia, and the exclusion of other infectious/malignant illnesses. Since the case definition does not require exposure to L-tryptophan (LT), diagnostic bias would occur if a physician's decision to diagnose EMS were influenced by knowledge of LT use. Methods. A random sample of 813 physicians practising in the United States and Canada was obtained. Physicians were asked to provide diagnoses for 6 case vignettes having diverse resemblances to EMS. Six weeks later, participants were asked to provide diagnoses for a complementary series of cases described in identical text except for different data regarding LT use. Results. Physicians who responded (N = 227, 28%) were more likely to diagnose EMS when LT exposure was present compared to the same case without LT use. In the most striking difference, EMS was diagnosed by 48% of physicians when the case was described in a man using LT, but by only 8% of physicians for the same case without LT use. The McNemar bias ratios, which compare responses provided by physicians completing both series, ranged from 0.65 to 1.0. Conclusion. These data indicate that the diagnosis of EMS may be biased by knowledge of LT. By showing the presence of diagnostic bias in clinical decision making, we suggest an important methodological problem that may arise in both clinical and research settings.