Underestimation of Cardiovascular Risk by the SCORE2 Model in Primary Care: A Call for Recalibration

The SCORE2 model is a derived risk prediction model that estimates the CVE risk. Originally developed with population-based cohort data, this model is also intended to be used in routine primary care to calculate the risk of first-onset CVE and guide treatment decisions.

In this study, we followed adult patients in the Netherlands who visited their general practitioner (GP) and underwent a CVE risk assessment at their GP office. The aim was to relate the results of the SCORE2 model to the observed risk of a first CVE during follow-up.

A retrospective cohort study was conducted of 205,548 patients, analyzing factors including age, sex, smoking status, diabetes, total and HDL cholesterol, and mean systolic blood pressure. The performance of the SCORE2 model was assessed using observed-to-expected (O/E) risk ratios, calibration plots, and C-statistics. Analyses were stratified by sex, age groups (< 50 and ≥ 50 years), and low and moderate risk country categories.

Results indicated a mean observed 10-year risk of 10.1%, notably higher than the model-predicted risk of 6.2%. The O/E ratio in females and men was 1.54 and 1.68 respectively.The mean observed 10-year CVE was in the age groups < 50 years and ≥ 50 years 6.9% and 11%, respectively. (O/E ratio: 1.78 vs. 1.62).Consequently, approximately 35% of patients potentially missed preventive treatments due to SCORE2's underestimation.

These findings suggest the SCORE2 model may underestimate CVE risk in primary care, highlighting the need for calibration in this setting.