Advantages of Continuous-Valued Risk Scores for Predicting Long-Term Costs: The Framingham Coronary Heart Disease 10-Year Risk Score

Abstract

Background: The few studies that have examined the relationship between midlife cardiovascular disease risk and longer-term costs have differentiated risk using a small number of risk categories. In this paper, we illustrate the advantages of a continuous-valued score to examine the relationship between risk and longer-term costs: the Framingham 10-year coronary heart disease risk score.

Methods: Our study cohort consisted of 1333 Second Generation Framingham Heart Study participants enrolled in fee-for-service Medicare for at least 8 quarters and who had a risk score assessment between age 40 and 50 years. We used generalized linear models to examine the relationships between quarterly Medicare costs and risk scores.

Results: Using risk categories defined by the Framingham score, the cost differences between a low and high risk group were 40% to over 200% greater than differences in comparable studies using a small number of risk categories. A continuous-valued score facilitates comparison of the cost consequences of impacting risk score changes. For example, an intervention that is able to reduce a person's score change between midlife and later-life from the 75th percentile to the 25th percentile would result in almost a 20% reduction in longer-term costs. In contrast, an intervention that is able to reduce a person's midlife score from the 75th percentile to the 25th percentile would result in a 38% reduction in costs.

Conclusions: A continuous-valued risk score has advantages compared to defining risk based on a small number of risk categories.

Keywords: Framingham risk score; Medicare costs; cardiovascular disease; risk scores.

Conflict of interest statement

CONFLICTS OF INTEREST Rhoda Au receives research funding from Pfizer, through Boston University, to conduct a callback cognitive assessment for the Framingham Gen 2/OmniGen 1 cohorts.