Adding 19 easily available risk markers to the SCORE algorithm resulted in only small improvements in CVD mortality risk stratification, such as hsCRP improving AUC by 0.006 (p=0.015).
Cohort (n=8,476)
No
Does adding 19 easily available risk markers to the SCORE algorithm improve risk stratification for CVD mortality in asymptomatic individuals without prior CVD or diabetes?
The addition of 19 easily available risk markers to the SCORE algorithm provides only minimal improvements in cardiovascular mortality risk stratification, supporting the continued use of the standard SCORE model.
Estimación del efecto: AUC +0.006 (hsCRP)
valor p: p=0.015
AIM: European society of cardiology (ESC) guidelines recommend that cardiovascular disease (CVD) risk stratification in asymptomatic individuals is based on the Systematic Coronary Risk Evaluation (SCORE) algorithm, which estimates individual 10-year risk of death from CVD. We assessed the potential improvement in CVD risk stratification of 19 easily available risk markers by adding them to the SCORE algorithm. METHODS AND RESULTS: We followed 8476 individuals without prior CVD or diabetes from the Copenhagen City Heart study. The 19 risk markers were: major and minor electrocardiographic (ECG) abnormalities, heart rate, family history (of ischaemic heart disease), body mass index (BMI), waist-hip ratio, walking duration and pace, leisure time physical activity, forced expiratory volume (FEV)1%pred, household income, education, vital exhaustion, high-density lipoprotein (HDL) cholesterol, triglycerides, apolipoprotein A1 (ApoA1), apolipoprotein B (ApoB), high-sensitive C-reactive protein (hsCRP) and fibrinogen. With the exception of family history, BMI, triglycerides and minor ECG changes, all risk markers remained significantly associated with CVD mortality after adjustment for SCORE variables. However, the addition of the remaining 15 risk markers resulted in only small changes in discrimination calculated by area under the curve (AUC) and integrated discrimination improvement (IDI) and no improvement in net reclassification improvement (NRI). HsCRP improved AUC by 0.006 (p = 0.015) and IDI by 0.012 (p = 0.002); FEV1%pred improved AUC by 0.006 (p = 0.032) and IDI by 0.006 (p = 0.029). In the intermediate risk group FEV1%pred, education, vital exhaustion and ApoA1 all improved NRI but FEV1%pred was the only risk marker to significantly improve both IDI, AUC and NRI. CONCLUSION: The SCORE algorithm predicted CVD mortality in a Danish cohort well. Despite strong association with CVD mortality, the individual addition of 19 easily available risk makers to the SCORE model resulted in small risk stratification improvements.
Graversen et al. (Tue,) conducted a cohort in Cardiovascular disease risk (n=8,476). Addition of 19 risk markers to SCORE algorithm vs. SCORE algorithm alone was evaluated on CVD mortality risk stratification improvement (AUC, IDI, NRI) (AUC +0.006 (hsCRP), p=0.015). Adding 19 easily available risk markers to the SCORE algorithm resulted in only small improvements in CVD mortality risk stratification, such as hsCRP improving AUC by 0.006 (p=0.015).