Validation of the cardiovascular risk model NORRISK 2 in South Asians and people with diabetes

Abstract

Objectives To evaluate the predictive ability of the previously published NORRISK 2 cardiovascular risk model in Norwegian-born and immigrants born in South Asia living in Norway, and to add information about diabetes and ethnicity in an updated model for South Asians and diabetics (NORRISK 2-SADia). Design. We included participants (30–74 years) born in Norway (n = 13,885) or South Asia (n = 1942) from health surveys conducted in Oslo 2000–2003. Cardiovascular disease (CVD) risk factor information including self-reported diabetes was linked with information on subsequent acute myocardial infarction (AMI) and acute cerebral stroke in hospital and mortality registry data throughout 2014 from the nationwide CVDNOR project. We developed an updated model using Cox regression with diabetes and South Asian ethnicity as additional predictors. We assessed model performance by Harrell’s C and calibration plots. Results. The NORRISK 2 model underestimated the risk in South Asians in all quintiles of predicted risk. The mean predicted 13-year risk by the NORRISK 2 model was 3.9% (95% CI 3.7–4.2) versus observed 7.3% (95% CI 5.9–9.1) in South Asian men and 1.1% (95% CI 1.0–1.2) versus 2.7% (95% CI 1.7–4.2) observed risk in South Asian women. The mean predictions from the NORRISK 2-SADia model were 7.2% (95% CI 6.7–7.6) in South Asian men and 2.7% (95% CI 2.4–3.0) in South Asian women. Conclusions. The NORRISK 2-SADia model improved predictions of CVD substantially in South Asians, whose risks were underestimated by the NORRISK 2 model. The NORRISK 2-SADia model may facilitate more intense preventive measures in this high-risk population.