Burnout’s Prevalence Estimations

To the Editor: Elmariah et al longitudinally assessed burnout among internal medicine residents after the “2011 duty hour changes.” The authors used an abbreviated 5-item version of the Maslach Burnout Inventory to assess burnout, relying on a 5-point scale (from 1 = Disagree strongly to 5 = Agree strongly). In order to estimate burnout’s prevalence, the authors opted for a cutoff score of 3, corresponding to a neutral response on the scale (ie, neither agree nor disagree with having experienced the mentioned symptom). Based on these categorization criteria, Elmariah et al found that about 65% of the participants suffered from burnout. The authors described the amount of burnout experienced on the night float rotation as “particularly troubling.” The mean burnout score in this condition was 3.84, a score that corresponds to an intermediate position between neither agree nor disagree and slightly agree with having experienced burnout symptoms. The authors concluded that “resident burnout remains a significant problem even with recent duty hour modifications.” We have 3 concerns regarding the conclusions drawn. The first concern regards the categorization criteria employed to identify cases of burnout. The authors’ decision to rely on a cutoff score of 3 is (a) devoid of any theoretically or clinically grounded rational and (b) surprising in view of the published response alternatives of the Maslach Burnout Inventory. Indeed, allowing participants who neither agreed nor disagreed with having burnout symptoms to be considered cases of burnout is problematic because such participants do not report burnout symptoms. Moreover, pending clear diagnostic criteria for burnout, it has been suggested that conservative cutoff scores be used when interested in studying cases of burnout. Such cutoff scores correspond to high frequency (or intensity) of burnout symptoms and, therefore, closely adhere to the available descriptions of the (clinical) state of burnout as a severe and debilitating condition. The authors’ modus operandi contrasts with such recommendations. All in all, the authors’ conclusion that 65% of the residents suffered from burnout appears to be unsubstantiated. Our second concern is with regard to the authors’ claim that the amount of burnout experienced on the night float rotation was particularly troubling. Judging from Table 2, the levels of burnout symptoms reported across conditions were actually low: Mean burnout scores were all below 4 (and often below 3). Recalling that a score of 4 means slightly agree with having experienced burnout symptoms, it can be hypothesized that most residents were in fact weakly affected by such symptoms. This observation further questions the authors’ conclusions. Finally, it would have been helpful to know what the symptom picture was like before duty hours were changed in 2011. Without this information, one cannot assess the extent to which the implemented changes have been useful. As noted elsewhere, more rigor is needed in burnout research if effective decisions are to be made in terms of occupational health policies. Because they most often lack a theoretical or clinical underpinning, current practices in burnout research may be detrimental to the identification and treatment of people who actually developed stress-related syndromes in relation to their work.