New Model for Evaluating Hospital Decision-Making Developed by Hebrew University Researchers
JERUSALEM, December 13, 2024 - A new model that helps hospitals evaluate patient treatment, developed by Hebrew University of Jerusalem researchers, outperforms the widely used Elixhauser Comorbidity model used to predict key outcomes.
The new, expanded model, described in BMC Health Services Research, incorporates additional clinical, diagnostic, and demographic data into the existing Elixhauser model, a software program that identifies 38 pre-existing conditions based on secondary diagnoses (i.e., comorbidities). These comorbidities, included on hospital administrative data lists, were originally selected because they impact resource allocation (e.g., length of stay or charges) and affect healthcare outcomes, such as in-hospital mortality and readmissions.
"Our enhanced hospital decision-making model fills a crucial gap in the original Elixhauser Model by providing a more comprehensive assessment of patient complexity," said Prof. Adam J. Rose of the Hebrew University School of Public Health. "This new model has broad applicability to other healthcare settings, both within and beyond internal medicine, and could support decisions regarding admission and care settings, home hospitalization suitability, and payment adjustments based on patient complexity."
The researchers showed that including variables such as laboratory test results, vital signs, and demographic information improved the model's accuracy for predicting key outcomes such as length of stay, in-hospital mortality, readmission within 30 days, and the need for escalated care, such as an intensive care unit.
Widely used decision-making models, such as the Elixhauser Comorbidity Measure, often focus solely on comorbid conditions, which may not fully capture a patient’s medical complexity. These risk adjustment models play a critical role in guiding clinical care, adjusting case mix for research, and supporting health services planning and financing, which includes making payment adjustments based on patient complexity.
Researchers leveraged Israel's unique centralized health data repository to conduct a retrospective, observational cohort analysis of 55,946 admissions to the internal medicine service of the Shaare Zedek Medical Center in Jerusalem. The study showed the upgraded, expanded model was better at predicting certain health outcomes. For example:
- The augmented model was better able to estimate how long a patient would stay in the hospital, increasing R2 predictability from 10.1% to 28.1%.
- It significantly improved the ability to determine in-hospital mortality(whether a patient might die in the hospital), with the c-statistic, a measure of model predictiveness, rising from 71.1% to 87.9%.
This more comprehensive decision-making model is especially pertinent in Israel, where hospitals are reimbursed a fixed fee per day, irrespective of patient complexity.
The research paper, titled “Incorporating Clinical and Demographic Data into the Elixhauser Comorbidity Model: Validation and Impact on Outcome Predictions in a Tertiary Hospital's Internal Medicine Department” is now available in BMC Health Services Research, and can be accessed at https://rdcu.be/d2umv. DOI 10.1186/s12913-024-11663-z
Researchers:
Gideon Leibner1, David E. Katz1,2, Yaakov Esayag2, Nechama Kaufman3,4, Shuli Brammli-Greenberg1, Adam J. Rose1
Institutions:
- Faculty of Medicine, Hebrew University of Jerusalem, Jerusalem Israel
- Department of Internal Medicine, Shaare Zedek Medical Center, Jerusalem Israel
- Department of Quality and Patient Safety, Shaare Zedek Medical Center, Jerusalem Israel
- Department of Emergency Medicine, Shaare Zedek Medical Center, Jerusalem Israel
