SCORE2 tables do not assign the same risk category as the underlying model in almost 20% of Polish primary care patients

Abstract Purpose To compare the estimates provided by the SCORE2/SCORE2-OP charts with the estimates obtained by using untransformed SCORE2 and SCORE2-OP equations presented in their respective original papers. Methods A sample of 13,724 Polish primary care patients was studied. Patients were recruited from 16 major Polish administrative regions proportionally to the number of inhabitants. Each eligible patient was assigned SCORE2/SCORE2-OP risk estimate and category via SCORE2 chart and the equations for high risk countries provided in the original SCORE2/SCORE2-OP papers. Results After excluding patients with risk parameters exceeding permitted values and with comorbidities putting them in high risk category regardless of other factors, a total 6,621 patients were included in the analysis. A total of 2,164 (32.7%) patients had the equation-based estimate at least 20% lower than SCORE2 chart estimate, while a 20% higher equation-based estimate was found in only 382 (5.6%) patients; the assigned risks differed significantly (p0.0001, Wilcoxon rank test). It was also found that SCORE2 charts assigned 1,277 (19.3%) patients to a higher risk category and only 122 (1.8%) of the patients to a lower category (p0.001 for overall differences in risk category assignment, McNemar-Bowker test). When the differences in average SCORE2/SCORE2-OP estimates provided by SCORE2 charts and equations were plotted against the parameters used in the SCORE2 charts, a clear saw-tooth pattern emerged for SBP, age, and non-HDL (Picture 1; negative values correspond to lower estimate being provided by equations), which reflects the fact that SCORE2 charts display a clear tendency to overestimate risk at the lower boundary of each SBP, age and non-HDL interval. Additionally, a clear bias toward risk overestimation at higher HDL concentrations was observed due to the fact that SCORE2 charts simplify HDL and total cholesterol into non-HDL (likelihood of misclassification by the SCORE2 charts vs. HDL is presented in Picture 2). Conclusions SCORE2/SCORE2-OP charts tend to overestimate cardiovascular risk relative to the core SCORE2/SCORE2-OP risk models. The use of calculators that directly apply the model might help to more accurately determine the indications for blood pressure- and lipid-lowering therapies.Figure 1 Figure 2