Is a meta‐analytic approach to burnout's prevalence timely?

This letter to the editor concerns the following article: Cañadas‐De la Fuente GA, Gómez‐Urquiza JL, Ortega‐Campos EM, et al. Prevalence of burnout syndrome in oncology nursing: a meta‐analytic study. Psychooncology in press. doi:10.1002/pon.4632 Cañadas‐De la Fuente et al performed ameta‐analysis of the prevalence of emotional exhaustion, depersonalization, and diminished personal accomplishment—the three definitional components of the burnout syndrome—among nursing professionals working in oncology units. They found that 30% of these workers presented with emotional exhaustion; 15%, with depersonalization; and 35%, with diminished personal accomplishment. The authors concluded that burnout is a problem of considerable importance among nursing professionals employed in oncology units. Although we agree that oncology nursing can be a stressful occupation, we have concerns regarding the relevance and interpretability of the findings of Cañadas‐De la Fuente et al. The estimates of burnout's prevalence that the authors included in their meta‐analysis have been established with (different versions of) the Maslach Burnout Inventory, using the tercile‐based cutoff scores presented in themanual of the questionnaire. As pointed out in several recent publications, these categorization criteria are devoid of any clinical or theoretical underpinning and are not diagnostic criteria. Therefore, relying on such categorization criteria for estimating the prevalence of burnout is unwarranted. In order to be valid, meta‐estimates need to be based on primary estimates that are themselves valid. Unfortunately, this condition cannot be met in research on burnout's prevalence. We note that some investigators have begun the groundwork required for defining burnout in a clinically meaningful way. Lundgren‐Nilsson et al, for instance, attempted to establish a cutoff score for identifying burnout based on an 18‐item version of the Shirom‐Melamed Burnout Questionnaire—an alternative measure of burnout symptoms. In order to accomplish their goal, these authors relied on a sample of patients seeking medical care at a specialized outpatient stress clinic and fulfilling the ICD‐10 criteria for “other reactions to severe stress” (F.43.8A; Swedish amended version). The work of Lundgren‐Nilsson et al was only preliminary. Moreover, it did not reflect a consensual view of burnout. Indeed, the definition of burnout endorsed by these authors differs from the definition of burnout associated with the Maslach Burnout Inventory. However, the work of Lundgren‐Nilsson et al at least provides investigators with a cutoff score having explicit external reference points rather than reflecting a clinically groundless sample split. All in all, the meta‐analysis of Cañadas‐De la Fuente et al is inconclusive because of the very state of burnout research. Because there is