Validation of continuous clinical indices of cardiometabolic risk in a cohort of Australian adults
1 Spatial Epidemiology and Evaluation Research Group, School of Population Health and Sansom Institute for Health Research, University of South Australia, Adelaide, South Australia, Australia
2 Research Centre of the Douglas Mental Health University Institute, Verdun, Québec, Canada
3 Discipline of Medicine, The University of Adelaide, Adelaide, South Australia, Australia
4 Department of Medicine, The University of Melbourne, St. Vincent’s Hospital, Melbourne, VIC, Australia
BMC Cardiovascular Disorders 2014, 14:27 doi:10.1186/1471-2261-14-27Published: 27 February 2014
Indicators of cardiometabolic risk typically include non-clinical factors (e.g., smoking). While the incorporation of non-clinical factors can improve absolute risk prediction, it is impossible to study the contribution of non-clinical factors when they are both predictors and part of the outcome measure. Metabolic syndrome, incorporating only clinical measures, seems a solution yet provides no information on risk severity. The aims of this study were: 1) to construct two continuous clinical indices of cardiometabolic risk (cCICRs), and assess their accuracy in predicting 10-year incident cardiovascular disease and/or type 2 diabetes; and 2) to compare the predictive accuracies of these cCICRs with existing risk indicators that incorporate non-clinical factors (Framingham Risk Scores).
Data from a population-based biomedical cohort (n = 4056) were used to construct two cCICRs from waist circumference, mean arteriole pressure, fasting glucose, triglycerides and high density lipoprotein: 1) the mean of standardised risk factors (cCICR-Z); and 2) the weighted mean of the two first principal components from principal component analysis (cCICR-PCA). The predictive accuracies of the two cCICRs and the Framingham Risk Scores were assessed and compared using ROC curves.
Both cCICRs demonstrated moderate accuracy (AUCs 0.72 – 0.76) in predicting incident cardiovascular disease and/or type 2 diabetes, among men and women. There were no significant differences between the predictive accuracies of the cCICRs and the Framingham Risk Scores.
cCICRs may be useful in research investigating associations between non-clinical factors and health by providing suitable alternatives to current risk indicators which include non-clinical factors.