[ACG model can predict large consumers of health care. Health care resources can be used more wisely, individuals at risk can receive better care].

We describe a method, which uses already existent administrative data to identify individuals with a high risk of a large need of healthcare in the coming year. The model is based on the ACG (Adjusted Clinical Groups) system to identify the high-risk patients. We have set up a model where we combine the ACG system stratification analysis tool RUB (Resource Utilization Band) and Probability High Total Cost >0.5. We tested the method with historical data, using 2 endpoints, either >19 physical visits anywhere in the healthcare system in the coming 12 months or more than 2 hospital admissions in the coming 12 months. In the region of Västra Götaland with 1.6 million inhabitants, 5.6% of the population had >19 physical visits during a 12 month period and 1.2% more than 2 hospital admissions. Our model identified approximately 24,000 individuals of whom 25.7% had >19 physical visits and 11.6% had more than 2 hospital admissions in the coming 12 months. We now plan a small test in ten primary care centers to evaluate if the model should be introduced in the entire Västra Götaland region.