Finding future high-cost cases: comparing prior cost versus diagnosis-based methods

Area of Science:

• Health Services Research
• Medical Informatics
• Healthcare Management

Background:

• Identifying high-cost patient subgroups is essential for effective healthcare resource allocation and intervention.
• Traditional methods relying solely on past costs may not fully capture future healthcare needs.

Purpose of the Study:

• To evaluate the efficacy of patient-level cost and diagnosis data in identifying a small, high-cost population subgroup.
• To determine if these identified subgroups can benefit from medical management interventions to mitigate future costs.

Main Methods:

• Utilized the MEDSTAT Market Scan Research Database (1997-1998) containing inpatient and ambulatory healthcare encounter records.
• Employed prior cost and a diagnostic cost group (DCG) risk model to identify the top 0.5% of individuals with high predicted costs for the subsequent year.
• Compared the characteristics, costs, and prevalence of diseases targeted for management between the identified top groups and the general population.

Main Results:

• The overlap between prior cost and DCG-identified top groups was only 38%.
• Both identified groups comprised individuals with high future costs and significant rates of chronic conditions like diabetes, heart failure, lung disease, and depression.
• The DCG group identified individuals who were slightly more expensive and more likely to have targetable chronic conditions compared to the prior cost group.

Conclusions:

• Diagnosis-based risk models demonstrate comparable or superior power to prior cost data in predicting high future healthcare expenditures.
• Integrating both cost and diagnostic data offers enhanced predictive power and operational utility for identifying patients for targeted medical management.
• Diagnostic information is key to identifying specific medical problems amenable to management, leading to improved patient outcomes and reduced healthcare costs.