Bayesian latent class modelling to examine the diagnostic accuracy of the first hetero-assessment instrument for occupational burnout

Occupational burnout has no standardized diagnostic or screening criteria. Following a dozen of patient-reported outcome measures (PROMs) for occupational burnout, Belgian researchers developed the first hetero-assessment instrument (HAI) designed for health professionals’ use. The HAI’s sensitivity and specificity was previously assessed with reference to the OLdenburg Burnout Inventory (OLBI) using frequentist statistics in Belgium (100 participants) and Switzerland (42 participants). This study aimed at assessing the HAI’s diagnostic performance using Bayesian latent class modelling (BLCM). We applied Hui-Walter framework for two tests and two populations and ran models with minimally informative priors, with and without conditional dependency between HAI and OLBI results. We further performed sensitivity analysis by replacing one of the minimally informative priors by the distribution beta (2,1) at each time for all priors. We also performed the analysis using literature-based informative priors for OLBI. Using the BLCM without conditional dependency, the sensitivity and specificity of the HAI was 0.91 (0.77-1.00) and 0.82 (0.59-1.00), respectively. The sensitivity analysis did not yield any significant changes in these results. In all models, the sensitivity was never below 0.82 and the specificity was never below 0.78. The HAI’s sensitivity and specificity determined in this study are better compared to the previous studies conducted using frequentist statistics. These finding suggests that the use of BLCM is preferred in the absence of the diagnostic gold standard and precludes underestimating the diagnostic accuracy of the tested instrument.