High-risk case identification for use in comprehensive complex care management.

Area of Science:

• Health Services Research
• Predictive Modeling in Healthcare
• Chronic Disease Management

Background:

• Effective patient selection is vital for optimizing multimorbid care management.
• Identifying patients at high risk for future costs who can benefit from proactive care is a key challenge.
• Existing predictive models require refinement with clinical considerations.

Purpose of the Study:

• To develop and validate a patient selection process for multimorbid care management.
• To balance the accurate identification of high-cost patients with clinical benefit from proactive interventions.
• To incorporate physician input for refining patient selection criteria.

Main Methods:

• Surveyed six physicians on clinical considerations for high-risk patient identification.
• Extracted data from Clalit Health Services' comprehensive database (2010-2011).
• Utilized the Adjusted Clinical Groups (ACG) predictive model for risk scores and assessed model performance (c-statistic, PPV) before and after applying physician-derived exclusion criteria.

Main Results:

• The ACG model demonstrated acceptable discriminatory power (c-statistic 0.80 pre-exclusion, 0.75 post-exclusion).
• After applying exclusion criteria (e.g., active cancer, age ≥95), the positive predictive value (PPV) for the top 6% highest risk patients was 40%.
• Selected high-risk patients exhibited significantly higher age, number of chronic conditions, and utilization compared to the general patient population.

Conclusions:

• A validated predictive modeling tool, even with clinical exclusions, effectively identifies multimorbid patients for proactive care management.
• Physician input is valuable for refining exclusion criteria to ensure clinical benefit.
• This approach supports targeted resource allocation in care management programs.