The impact of survey nonresponse on estimates of healthcare employee burnout

Healthc (Amst). 2020 Sep;8(3):100451. doi: 10.1016/j.hjdsi.2020.100451. Epub 2020 Aug 2.

Abstract

Accurate estimates of burnout prevalence are critical for workforce planning. We assessed survey nonresponse bias and its impact on burnout estimates by linking 27,226 primary care employees to administrative data, categorized by whether they responded to a 2016 workforce survey (19.2% response). We adjusted burnout prevalence by response propensity using mixed-effects logistic regression. Thirty-six percent of respondents screened positive for burnout. There were significant differences between respondents and non-respondents (e.g.,gender, tenure), but no difference between unadjusted (i.e., respondents only) and propensity-adjusted estimates of burnout among the workforce. This provides support that workforce surveys may yield valid burnout estimates despite low response.

Keywords: Burnout; Nonresponse bias; Primary care; Survey methods; Veteran.

MeSH terms

  • Adult
  • Burnout, Professional / epidemiology
  • Burnout, Professional / psychology*
  • Cross-Sectional Studies
  • Female
  • Humans
  • Male
  • Middle Aged
  • Prevalence
  • Surveys and Questionnaires / statistics & numerical data*
  • United States
  • United States Department of Veterans Affairs / organization & administration
  • United States Department of Veterans Affairs / statistics & numerical data
  • Workforce / statistics & numerical data