SHM Converge Scientific Abstract ePoster Gallery-Discharge Percentage as a Clinician-Level Length of Stay Metric: Associations With Increased Patient Volume

Pdf Summary

This retrospective cross-sectional study examined a new clinician-level length-of-stay metric called <strong>Discharge Percentage (DP)</strong>, which estimates the proportion of a hospitalist’s patients who are discharged versus followed up during a month. The goal was to assess how patient, hospital, and hospitalist factors relate to DP across four hospitals in one health system from 2020–2022.

The study included <strong>183 hospitalists</strong>, <strong>35,890 patients</strong>, and <strong>56,542 hospitalizations</strong>. Clinicians who appeared to be “admitters” were excluded, as were patients with zero length of stay or age 100 years. DP was modeled using hierarchical beta regression with clinician, hospital, and patient random effects, along with fixed effects for characteristics such as volume, experience, season, and discharge destination.

Results showed substantial variation in DP across hospitalists. Variance partitioning indicated that <strong>hospitalist factors explained 91.5%</strong> of the random-effects variance, while <strong>hospital-level factors explained 7.3%</strong> and <strong>patient factors only 1.2%</strong>. Higher monthly patient volume and more years in practice were associated with higher DP. Home discharge was also positively associated with DP, while advanced practice provider status and winter/spring seasons were associated with lower DP.

The findings suggest that DP is driven mainly by <strong>individual hospitalist practice patterns</strong>, not by patient mix or hospital setting. This supports DP as a potentially useful and more interpretable clinician-level metric of length of stay. However, the study was limited by its retrospective design, single health system setting, and lack of data on practice patterns. Future work should validate DP across systems, test whether it predicts future length-of-stay outcomes, and explore integration into hospitalist feedback and performance frameworks.

Asset Subtitle

Alexander Roseman

Meta Tag

Author List Alexander Roseman, Ebrahim Barkoudah, Peter Lindenauer, Shankar Viswanathan

Category Research

Concept Discharge Percentage

Concept Hierarchical Beta Regression

Concept Hospitalist

Concept Hospitalist Practice Patterns

Concept Length of Stay Metric

Distinguished Non-Finalist

Presenter Organization University of Massachusetts Chan School of Medicine - Baystate

Presenting Author Alexander Roseman

Track Value in Hospital Medicine

Keywords

Discharge Percentage

hospitalist

length of stay

hierarchical beta regression

cross-sectional study

variance partitioning

patient volume

hospital factors

discharge destination

practice patterns

Discharge Percentage

Hierarchical Beta Regression

Hospitalist

Hospitalist Practice Patterns

Length of Stay Metric