Improved prediction of medical expenditures and health care utilization using an updated chronic disease score and claims data

J Clin Epidemiol. 2013 Oct;66(10):1118-27. doi: 10.1016/j.jclinepi.2013.04.011. Epub 2013 Jul 8.

Abstract

Objective: To predict future medical expenditures, health care utilization, and mortality in Switzerland using an updated chronic disease score (CDS), a chronic morbidity measure based on pharmacy data.

Study design and setting: We performed a cohort study using medical claims data from insured persons enrolled in 2009 and 2010. Patient's characteristics, chronic conditions, and health care costs from baseline were used to calculate each patient's disease score. Two-part regression models were fit to predict health care expenditures, utilization, and mortality during the following year using the score's baseline values. We calculated the proportion of explained variation for each regression model to assess their performance.

Results: The CDS model, controlled for sociodemographics and health insurance plan, showed a significant improvement in explained variance of health care costs, outpatient costs, and outpatient visits in 2010. Future outpatient visits were predicted best with an R(2) of 29.2% (age group: 18-65 years) and 22.9% (>65 years), and models predicted future mortality with a c-statistic of 0.8.

Conclusion: The CDS showed reasonable predictive validity of future health care utilization and medical expenditure based on pharmacy dispensing data, which may support health care decision makers in the planning delivery systems and resource allocation.

Keywords: Chronic diseases; Claims data; Health care costs; Mortality; Predictive models; Utilization.

MeSH terms

  • Adolescent
  • Adult
  • Aged
  • Aged, 80 and over
  • Chronic Disease* / drug therapy
  • Chronic Disease* / economics
  • Chronic Disease* / mortality
  • Cohort Studies
  • Delivery of Health Care / statistics & numerical data*
  • Female
  • Health Expenditures / statistics & numerical data*
  • Humans
  • Insurance Claim Review
  • Insurance, Pharmaceutical Services
  • Male
  • Middle Aged
  • Models, Statistical*
  • Reproducibility of Results
  • Switzerland / epidemiology
  • Young Adult