INTRODUCTION: The optimal performance of cardiovascular disease (CVD) risk models in various populations (such as the Framingham Heart Sheet or the ESC SCORE) is of major interest in risk prediction modeling nowadays. We evaluated whether the inclusion of socioeconomic status (SES) in the HellenicSCORE would increase the accuracy of prediction, irrespectively of dietary information and the classical CVD risk factors. METHODS: Data from 1514 men and 1528 women (age >18 years), who were free of known CVD on enrolment in 2001-02, were studied (the ATTICA study). Five years later a follow up was performed and the development of CVD was defined (WHO-ICD-10 criteria). As SES indicators, education status and mean annual income were recorded, and a special SES 3-class index was calculated (low, moderate and high). The MedDietScore, which incorporates the inherent characteristics of the Mediterranean diet, was used as a dietary assessment tool, while the HellenicSCORE, which reflects the level of CVD risk factors, was also calculated. Additive logistic regression models were used to test the additive effect of SES and dietary assessment on the predictive ability of the HellenicSCORE. RESULTS: SES assessment did not improve the predictive ability of the estimated risk model compared to the model that included the HellenicSCORE, physical activity status, waist-to-hip ratio, diabetes and family history of CVD. Additionally, SES did not improve the predictive ability of the estimated risk model even when dietary assessment was added to the above model. CONCLUSIONS: Socioeconomic status does not improve the predictive ability of a CVD risk model, even when dietary information is also taken into account.
INTRODUCTION: The optimal performance of cardiovascular disease (CVD) risk models in various populations (such as the Framingham Heart Sheet or the ESC SCORE) is of major interest in risk prediction modeling nowadays. We evaluated whether the inclusion of socioeconomic status (SES) in the HellenicSCORE would increase the accuracy of prediction, irrespectively of dietary information and the classical CVD risk factors. METHODS: Data from 1514 men and 1528 women (age >18 years), who were free of known CVD on enrolment in 2001-02, were studied (the ATTICA study). Five years later a follow up was performed and the development of CVD was defined (WHO-ICD-10 criteria). As SES indicators, education status and mean annual income were recorded, and a special SES 3-class index was calculated (low, moderate and high). The MedDietScore, which incorporates the inherent characteristics of the Mediterranean diet, was used as a dietary assessment tool, while the HellenicSCORE, which reflects the level of CVD risk factors, was also calculated. Additive logistic regression models were used to test the additive effect of SES and dietary assessment on the predictive ability of the HellenicSCORE. RESULTS: SES assessment did not improve the predictive ability of the estimated risk model compared to the model that included the HellenicSCORE, physical activity status, waist-to-hip ratio, diabetes and family history of CVD. Additionally, SES did not improve the predictive ability of the estimated risk model even when dietary assessment was added to the above model. CONCLUSIONS: Socioeconomic status does not improve the predictive ability of a CVD risk model, even when dietary information is also taken into account.
Authors: Alexandros C Dimopoulos; Mara Nikolaidou; Francisco Félix Caballero; Worrawat Engchuan; Albert Sanchez-Niubo; Holger Arndt; José Luis Ayuso-Mateos; Josep Maria Haro; Somnath Chatterji; Ekavi N Georgousopoulou; Christos Pitsavos; Demosthenes B Panagiotakos Journal: BMC Med Res Methodol Date: 2018-12-29 Impact factor: 4.615