A new look at an old well-being construct: evaluating the psychometric properties of 9, 5, and 1-item versions of emotional exhaustion metrics

Objective: To compare the relative strengths (psychometric and convergent validity) of four emotional exhaustion (EE) measures: 9- and 5-item scales and two 1-item metrics.Patients and methods: This was a national cross-sectional survey study of 1409 US physicians in 2013. Psychometric properties were compared using Cronbach's alpha, Confirmatory Factor Analysis (CFA), Exploratory Factor Analysis (EFA), and Spearman's Correlations. Convergent validity with subjective happiness (SHS), depression (CES-D10), work-life integration (WLI), and intention to leave current position (ITL) was assessed using Spearman's Correlations and Fisher's R-to-Z.Results: The 5-item EE scale correlated highly with the 9-item scale (Spearman's rho = 0.828), demonstrated excellent internal reliability (alpha = 0.87), and relative to the 9-item, exhibited superior CFA model fit (RMSEA = 0.082, CFI = 0.986, TLI = 0.972). The 5-item EE scale correlated as highly as the 9-item scale with SHS, CES-D10, and WLI, and significantly stronger than the 9-item scale to ITL. Both 1-item EE metrics had significantly weaker correlation with SHS, CES-D10, WLI, and ITL (Fisher's R-to-Z; p < 0.05) than the 5- and 9-item EE scales.Conclusion: The 5-item EE scale was repeatedly found equivalent or superior to the 9-item version across analyses, particularly with respect to the CFA results. As there is no cost to using the briefer 5-item EE scale, the burden on respondents is smaller, and widespread access to administering and interpreting an excellent wellbeing metric is enhanced at a critical time in global wellbeing research. The single item EE metrics exhibited lower convergent validity than the 5- and 9-item scales, but are acceptable for detecting a signal of EE when using a validated EE scale is not feasible. Replication of psychometrics and open-access benchmarking results for use of the 5-tem EE scale further enhance access and utility of this metric.