Clinician-level variation in admission practices for common conditions in Veteran Affairs emergency departments.

Abstract

Background: The Veterans Affairs (VA) Healthcare System seeks to improve value of care. Variations in hospital admission practices may be a target to improve the efficiency and quality of emergency care.

Method: This was a retrospective cross-sectional analysis of 113 VA emergency departments (EDs) from October 1, 2015 to June 8, 2023. ED visits were included if their primary diagnosis matched one of 535 ICD-10 codes within the previously developed admission intensity measure, encompassing 16 clinical conditions with known variation in admission practices and opportunities for outpatient management. The primary outcome was the standard deviation (SD) of clinician-level, risk-adjusted admission rates for each condition, calculated using multilevel, hierarchical models accounting for patient sociodemographic and clinical characteristics. We then summarized the SD of clinician admission rates, based on total admission volume at VA. We further assessed the association between clinician admission intensity and 72-h ED return visits resulting in hospitalization, a safety indicator, using Pearson correlation coefficients.

Results: During the study period, 18.5 million ED visits were recorded across 9350 clinicians, with 3,815,094 visits (20.6%) having primary diagnoses matching one of the 16 clinical conditions included in the admission intensity measure. Patients' mean (±SD) age was 63.3 (SD 15.4) years, and 89.9% were male. Among the 16 conditions, chest pain accounted for the highest number of admissions (229,507) and exhibited the greatest variation (SD ±14.7%) in clinician-level admission rates, while nephrolithiasis had the lowest variation (SD ±4.6%). There was no significant association between clinician admission rates and 72-h ED returns (Pearson correlation 0.018).

Conclusion: A previously developed admission intensity measure may help identify improvement opportunities in hospitalization practices across VA EDs. Chest pain was the most common condition among those in the measure, and subject to the most variability in admission practices among clinicians.