BACKGROUND: In clinical practice, factors known to be associated with cardiovascular disease (CVD) like albuminuria, education level, or coronary calcium score are not directly incorporated in cardiovascular risk prediction models. The aim of the current study was to quantify the added value of potential risk modifying characteristics when added to the SCORE2 algorithm for individuals without diabetes mellitus (DM) or prior CVD.
METHODS AND RESULTS: Individuals without previous CVD or DM were included from the ARIC, MESA, EPIC-NL and HNR studies (n=46,285) in whom 2,177 CVD events and 2,062 non-cardiovascular deaths were observed over exactly 10.0 years of follow-up. The effect of each possible risk modifying characteristic was derived using Fine and Gray models that included an offset term for the SCORE2 linear predictor. The risk modifying characteristics were applied to individual predictions using the “naïve approach”, which modifies predicted risks based on the population prevalence and the SHR of the relevant predictor. Subdistribution hazard ratios are presented in the table. External validation was performed in the CPRD cohort (UK, n=518,015, 12,675 CVD events). In the external validation, adjustment of SCORE2 predicted risks with both single and with all available risk modifiers did not negatively affect calibration (see figure) and led to a modest increase in discrimination (C-index 0.742 [95% CI 0.737–0.746] versus unimproved SCORE2 risk C-index 0.737 [95% CI 0.732–0.741]). The net reclassification index or adding all these predictors was +0.032 (95% CI 0.025; 0.028) for future events and −0.008 (95% CI −0.009; −0.007) for future non-events. The coronary calcium score was found to the single strongest added predictor.
INTERPRETATION: The current analysis presents a method on how to integrate possible risk modifying characteristics that are not included in existing CVD risk models for the prediction of CVD event risk in apparently healthy people. This flexible methodology improves the accuracy of predicted risks and increases applicability of prediction models for individuals with additional risk known modifiers.
