by Hebrew University of Jerusalem

Credit: Pixabay/CC0 Public Domain

Risk adjustment models play a critical role in guiding clinical care, adjusting case mix in research and supporting health services planning and financing, including payment adjustments based on patient complexity. However, these models, such as the Elixhauser Comorbidity Measure, often focus solely on comorbid illnesses and may not fully capture a patient's medical complexity. This is especially pertinent in Israel, where hospitals are reimbursed a fixed fee per day, irrespective of patient complexity.

Researchers at The Hebrew University have enhanced and externally validated the Elixhauser Comorbidity Model to improve its accuracy. The lead author was Dr. Gidon Liebner, a resident in orthopedics at Hadassah hospital and a graduate of the Hebrew University School of Medicine. The senior authors are Drs. Shuli Brammli-Greenberg and Adam Rose, professors at the Hebrew University School of Public Health.

The study, published in BMC Health Services Research, incorporated additional clinical and demographic data into the Elixhauser model. By incorporating additional clinical and demographic data, the study has improved the model's accuracy in predicting key outcomes such as length of stay, in-hospital mortality, readmission within 30 days, and escalated care, meaning the intensive care unit or similar sorts of care.

This improved model not only aids health care professionals in making more informed clinical decisions but also enables more efficient resource allocation within health care facilities. The refined model enhances the overall quality of patient care, potentially leading to cost savings and contributing to advancements in health care research.

Leveraging Israel's unique centralized health data repository, the study conducted a retrospective observational cohort analysis of 55,946 admissions to the internal medicine service of the Shaare Zedek Medical Center (Jerusalem). It found that the enhanced model outperformed the standard Elixhauser Model. By including variables such as laboratory test results, vital signs, and demographic information, the new model achieved significant improvements in prediction accuracy over the basic Elixhauser model, which has itself been in use for decades for this purpose.

The study showed that the upgraded model was better at predicting certain health outcomes. For example, when it came to estimating how long someone would stay in the hospital, the improved model increased its predictiveness from an R² of 10.1% to 28.1%.

Additionally, in predicting whether someone might pass away during their hospital stay, the model's c-statistic, a measure of model predictiveness, went up from 71.1% to 87.9%. These improvements show that the enhanced model is better at figuring out and predicting how long someone will be in the hospital and the risk of in-hospital mortality compared to the standard model.

"Our enhanced model fills a crucial gap in the original Elixhauser Model by providing a more comprehensive assessment of patient complexity," said Prof. Rose. "This model has broad applicability to other health care 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."

More information: Gideon Leibner et al, Incorporating clinical and demographic data into the Elixhauser Comorbidity Model: deriving and validating an enhanced model in a tertiary hospital's internal medicine department, BMC Health Services Research (2024). DOI: 10.1186/s12913-024-11663-z

Provided by Hebrew University of Jerusalem