Emergency department physician-level and hospital-level variation in admission rates

Ann Emerg Med. 2013 Jun;61(6):638-43. doi: 10.1016/j.annemergmed.2013.01.016. Epub 2013 Feb 15.


Study objective: We explore the variation in physician- and hospital-level admission rates in a group of emergency physicians in a single health system.

Methods: This was a cross-sectional study that used retrospective data during various periods (2005 to 2010) to determine the variation in admission rates among emergency physicians from 3 emergency departments (EDs) within the same health system. Patients who left without being seen or left against medical advice, patients treated in fast-track departments, patients with primary psychiatric complaints, and those younger than 18 years were excluded, as were physicians with fewer than 500 ED encounters during the study period. Emergency physician-level and hospital-level admission rates were estimated with hierarchic logistic regression, which adjusted for patient age, sex, race, chief complaint, arrival mode, and arrival day and time.

Results: A total of 389,120 ED visits were included in the analysis, and patients were treated by 89 attending emergency physicians. After adjusting for patient and clinical characteristics, the hospital-level admission rate varied from 27% to 41%. At the physician level, admission rates varied from 21% to 49%.

Conclusion: There was 2.3-fold variation in emergency physician adjusted admission rates and 1.7-fold variation at the hospital level. In the new era of cost containment, wide variation in this common, costly decision requires further exploration.

MeSH terms

  • Adolescent
  • Adult
  • Aged
  • Cross-Sectional Studies
  • Emergency Service, Hospital / statistics & numerical data*
  • Female
  • Hospitals, Teaching / statistics & numerical data
  • Humans
  • Logistic Models
  • Male
  • Middle Aged
  • Patient Admission / statistics & numerical data*
  • Physicians / statistics & numerical data*
  • Practice Patterns, Physicians' / statistics & numerical data
  • Retrospective Studies
  • Young Adult