Measuring population health risks using inpatient diagnoses and outpatient pharmacy data

Area of Science:

• Health Services Research
• Health Economics
• Predictive Modeling

Background:

• Accurate prediction of future healthcare expenditures is crucial for resource allocation and financial planning.
• Existing models often rely on limited data sources, potentially impacting their predictive accuracy.

Purpose of the Study:

• To evaluate the effectiveness of models using inpatient encounter and outpatient pharmacy claims data for predicting future healthcare costs.
• To compare the predictive performance of models utilizing pharmacy data, inpatient diagnosis data, and a combination of both.

Main Methods:

• Utilized the MEDSTAT Market Scan Research Database (1997-1998) for the privately insured, under-65 population.
• Developed pharmacy and disease profiles from pharmacy claims and inpatient encounter data, respectively.
• Assessed the predictive power of individual and combined data sources for subsequent-year healthcare expenditures.

Main Results:

• The inpatient diagnosis model explained 8.4% of future cost variation in low-hospitalization populations.
• Pharmacy and inpatient diagnosis models showed comparable overall performance, with distinct strengths in identifying low-cost and high-cost individuals.
• A combined model significantly outperformed individual models, achieving an R2 of 11.8%.

Conclusions:

• Both pharmacy and inpatient diagnosis data are valuable for stratifying individuals by expected healthcare costs.
• Integrating comprehensive pharmacy and inpatient diagnosis data presents a promising strategy for enhancing the prediction of future healthcare expenditures.