Researchers: Adam Rose¹, Shuli Brammli-Greenberg¹, David Katz²

Background: Risk adjustment models, which characterize patients by complexity, can support differential payment to hospitals by complexity. Such models are not currently used in Israel; instead, there is a uniform payment per patient-day.

Objectives: 1) To demonstrate that a widely-used risk adjustment model, the Elixhauser index, can be successfully applied to an Israeli hospital; 2) To enhance the model with additional clinical information, and to measure the improved prediction; 3) To apply the enhanced model to support differential payment and clinical risk stratification.

Method: We fit the base Elixhauser model using admissions to the internal medicine departments of Shaarei Zedek between 2016-2019. We enhanced the model with additional clinical variables and measured the improved model performance. We simulated hospitals treating patients of varying complexity and examined how complexity-adjusted payment would compare to the current state. We compared the ability of the enhanced model to predict risk for patients admitted with pneumonia, compared with the widely-used CURB-65 model.

Findings: We fit the Elixhauser model to data from Shaarei Zedek. The enhanced model improved outcome prediction substantially. Complexity-adjusted payment would make a major difference in how Israeli hospitals are paid. Our enhanced model predicted outcomes including in-hospital mortality much better than CURB-65.

Conclusions: Risk adjustment models are possible in the Israeli context and can be made to work with data from an Israeli hospital. Added clinical data, which are widely available, greatly improves the model’s predictiveness.

Recommendations: To our knowledge, Israel is the only developed country that does not use risk adjustment to support differential payment by patient complexity. This may starve hospitals that treat the most complex patients of necessary resources. Our study shows that there is no reason why this approach, used by the rest of the developed world, cannot work here as well.

Research end date: 03/2024