Premium
Inclusion of Dietary Evaluation in Cardiovascular Disease Risk Prediction Models Increases Accuracy and Reduces Bias of the Estimations
Author(s) -
Panagiotakos Demosthenes B.,
Pitsavos Christos,
Stefanadis Christodoulos
Publication year - 2009
Publication title -
risk analysis
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.972
H-Index - 130
eISSN - 1539-6924
pISSN - 0272-4332
DOI - 10.1111/j.1539-6924.2008.01140.x
Subject(s) - confounding , medicine , anthropometry , body mass index , disease , mediterranean diet , risk assessment , coronary heart disease , demography , computer security , sociology , computer science
In the past few years, the prediction of CVD risk has received special attention; however, some investigators assert that risk models have so far not been very successful. Thus, we examined whether the inclusion of dietary evaluation in a risk prediction model that already contained the classical CVD risk factors increases the accuracy and reduces the bias in estimating future CVD events. The database of the ATTICA study (which included information from 1,514 men and 1,528 women) was used. At baseline, the HellenicSCORE values (based on age, gender, smoking, systolic blood pressure, and total cholesterol) were calculated, while overall assessment of dietary habits was based on the Mediterranean diet score (MDS) that evaluates adherence to this traditional diet. In 2006, a five‐year follow‐up was performed in 2,101 participants and development of CVD (coronary heart disease, acute coronary syndromes, stroke, or other CVD) was defined according to WHO‐ICD‐10 criteria. The MDS and the HellenicSCORE were significant predictors of CVD events, even after adjusting for various potential confounders ( p < 0.05). However, estimating bias (i.e., misclassification of cases) of the model that included HellenicSCORE and other potential confounders was 8.7%. The MDS was associated with the estimating bias of the outcome ( p < 0.001) and explained 5.5% of this bias. Other baseline factors associated with bias were increased body mass index, low education status, and increased energy intake/BMR ratio. The inclusion of dietary evaluation, as well as other Sociodemographic and anthropometric characteristics, increases the accuracy and reduces estimating bias of CVD risk prediction models.