Abstract
Background/Aims Multivariate cardiovascular disease (CVD) risk calculators, such as the Framingham risk equations can be used to identify populations most likely to benefit from treatments to decrease risk. We wanted to determine what proportion of adults in a large integrated health plan had sufficient automated laboratory data to calculate risk and when insufficient laboratory data was available whether BMI could be used.
Methods The setting was 26 primary care clinics in Washington (Group Health). We used automated databases to identify patients’ age 30–74 with at least 2 years continuous enrollment prior to April 1, 2010 and no CVD. The goal was to define two different Framingham risk scores using only automated data; one based on lipids (laboratory-based), and one based on BMI (clinic-based). We describe the proportion of patients for whom we were able to calculate risk, the data missing when we could not, and the concordance between these scores when both measures could be computed.
Results We identified 122,270 eligible patients. Of these, 59.7% (n=73,023) had sufficient data to calculate the lab-based risk score and 88.1% (102,795) clinic-based risk score. Neither score could be calculated for 14.5% (n=17,732). The most common reason for not being able to calculate was missing data on cholesterol. Using the laboratory-based score only, we found 12.9% of the population were at high risk (risk >20%), 24.5% moderate risk (10–20%), and 62.6% low risk (<10%). For those with both risk scores (n=71,280), the lab-based risk score was lower than the clinic-based score for 84.3% of patients (60,060/71,280). The lab-based score was 3.1% lower on average, but the two risk scores were within ±5% for 77.0% of patients (54,874/71,280). The risk scores differed by more than 10% for only 8.7% of patients (n=6236), and in most cases (6098 of 6236), the clinic-based score was higher.
Conclusion Electronic data can be used to classify CVD risk for most adults age 30–74. Risk scores based on BMI tend to estimate risk as higher than scores based on laboratory data. However, the risk scores do not differ by more than 5% for most patients.
