Validity and reliability of an abbreviated Copenhagen Burnout Inventory in Canadian emergency physicians and residents

Emergency physicians have the highest rates of burnout among all medical specialties. There is a need for accurate and reliable burnout assessment tools to monitor changes and assess the effects of interventions. However, existing tools are typically long and/or costly. We sought to validate an abbreviated Copenhagen Burnout Inventory among emergency physicians and trainees in Canada.

We conducted a planned secondary analysis of a national, cross-sectional survey of emergency physicians and trainees in Canada. Exploratory factor analysis was performed followed by confirmatory factor analysis. Kaiser's eigenvalues rule, a scree plot, and Horn's parallel analysis guided the number of factors to extract. Structural validity fit indices and internal consistency were compared to pre-specified cutoffs. Criterion validity was assessed compared to the full Copenhagen Burnout Inventory (burnout defined as mean ≥ 50/100).

One hundred eighty-two responses were randomly split into separate cohorts for exploratory factor analysis and confirmatory factor analysis. Data were confirmed to be statistically suitable for factor analysis. Using exploratory factor analysis, a ten-item, two-factor abbreviated Copenhagen Burnout Inventory was reached after removing items based on over correlation (≥ 0.80), cross-loading (≥ 75%), and low factor loading (< 0.60). In confirmatory testing, the abbreviated inventory had a good Comparative Fit Index (0.91) though did not meet cutoffs for the remaining fit indices. Internal consistency was 0.92 (95%CI 0.90-0.95). Using a cutoff of 33/50, sensitivity was 0.99, specificity was 0.82, and area under the ROC curve was 0.86.

With further validation, an abbreviated ten-item Copenhagen Burnout Inventory has potential to serve as a short, freely available burnout assessment tool among Canadian emergency physicians and trainees. This abbreviated inventory has evidence to support its internal consistency and criterion validity, albeit with inconsistent structural validity. Future validation with larger samples is required, with special attention paid to content validity, test-retest reliability, and correlation with important outcomes.