Validity Assessment of Low-risk SCORE Function and SCORE Function Calibrated to the Spanish Population in the FRESCO Cohorts

José Miguel Baena-Díez, Isaac Subirana, Rafael Ramos, Agustín Gómez de la Cámara, Roberto Elosua, Joan Vila, Alejandro Marín-Ibáñez, María Jesús Guembe, Fernando Rigo, María José Tormo-Díaz, Conchi Moreno-Iribas, Joan Josep Cabré, Antonio Segura, José Lapetra, Miquel Quesada, María José Medrano, Paulino González-Diego, Guillem Frontera, Diana Gavrila, Eva Ardanaz, Josep Basora, José María García, Manel García-Lareo, José Antonio Gutiérrez-Fuentes, Eduardo Mayoral, Joan Sala, Irene R. Dégano, Albert Francès, Conxa Castell, María Grau, Jaume Marrugat

To assess the validity of the original low-risk SCORE function without and with high-density lipoprotein cholesterol and SCORE calibrated to the Spanish population. Pooled analysis with individual data from 12 Spanish population-based cohort studies. We included 30 919 individuals aged 40 to 64 years with no history of cardiovascular disease at baseline, who were followed up for 10 years for the causes of death included in the SCORE project. The validity of the risk functions was analyzed with the area under the ROC curve (discrimination) and the Hosmer-Lemeshow test (calibration), respectively. Follow-up comprised 286 105 persons/y. Ten-year cardiovascular mortality was 0.6%. The ratio between estimated/observed cases ranged from 9.1, 6.5, and 9.1 in men and 3.3, 1.3, and 1.9 in women with original low-risk SCORE risk function without and with high-density lipoprotein cholesterol and calibrated SCORE, respectively; differences were statistically significant with the Hosmer-Lemeshow test between predicted and observed mortality with SCORE (P < .001 in both sexes and with all functions). The area under the ROC curve with the original SCORE was 0.68 in men and 0.69 in women. All versions of the SCORE functions available in Spain significantly overestimate the cardiovascular mortality observed in the Spanish population. Despite the acceptable discrimination capacity, prediction of the number of fatal cardiovascular events (calibration) was significantly inaccurate.