A risk stratification tool for hospitalisation in Australia using primary care data

Area of Science:

• Health Informatics
• Predictive Analytics
• Primary Care Research

Background:

• General practice (GP) data can identify patients for targeted care, potentially reducing hospitalisation rates, especially for those with chronic conditions.
• Validated risk models for hospitalisation prediction using Australian GP data are lacking.
• Existing models often rely on prior hospitalisation history, which is not consistently available in GP systems.

Purpose of the Study:

• To develop and validate a predictive risk model using Australian GP data to predict one-year hospitalisation risk.
• To create a tool for GPs to identify patients needing coordinated healthcare, without using prior hospitalisation data.
• To support the Health Care Homes initiative by facilitating targeted care for patients with chronic and complex conditions.

Main Methods:

• Developed and validated a predictive risk model using a retrospective primary care cohort linked to hospitalisation records in Victoria, Australia.
• Utilised predictors including demographics, prescription history, pathology results, and disease diagnoses.
• Excluded prior hospitalisation information as a predictor variable.

Main Results:

• The developed predictive risk model demonstrated good performance in predicting hospitalisation within one year.
• The model successfully predicts hospitalisation risk without relying on historical hospitalisation data.
• The model has been implemented in primary care practices within the Health Care Homes initiative.

Conclusions:

• A novel predictive risk model using readily available GP data can effectively identify patients at risk of hospitalisation in Australia.
• The model provides a valuable tool for GPs to target coordinated care, particularly for patients with chronic conditions.
• Implementation within the Health Care Homes initiative demonstrates the practical utility of the model in improving patient management and potentially reducing hospital resource utilisation.