Prediction Abilities of SCORE2 Risk Algorithms for Incident Dementia and All-Cause Mortality: Results From the UK Biobank Cohort Study

Abstract

Background: Whether the updated Systematic COronary Risk Evaluation (SCORE2) risk algorithm is suitable for the prediction of incident dementia and all-cause mortality and whether its discrimination abilities for these outcomes are higher than those of the SCORE and Cardiovascular Risk Factors, Aging, and Dementia (CAIDE) risk algorithms are unclear.

Methods: The present study included 429 033 participants (mean age: 57.1 ± 8.1 years; male: 46.2%; White: 94.1%) free of dementia from the UK Biobank at baseline, with a median follow-up of 12.8 years. Cox regression models were adopted to investigate the longitudinal relationships of SCORE2 risk categories with outcomes, and receiver operating characteristic curve analyses were used to compare the discrimination abilities of the 3 algorithms.

Results: During 5 376 778 person-years of follow-up, 6 477 all-cause dementia, 2 726 Alzheimer's disease (AD), 1 439 vascular disease (VD), and 31 981 all-cause deaths were identified. We found that higher SCORE2 risk was associated with higher risks of all-cause dementia, AD, VD, and all-cause mortality. The C-indices of SCORE2 risk for discriminating incident all-cause dementia, AD, VD, and all-cause death were 0.750 (95% confidence interval [CI]: 0.745 to 0.755), 0.750 (95% CI: 0.743 to 0.757), 0.800 (95% CI: 0.791 to 0.809), and 0.721 (95% CI: 0.718 to 0.724), respectively, which were significantly improved in comparison to those of the SCORE and CAIDE risk algorithms.

Conclusion: The SCORE2 risk algorithm is applicable in predicting incident all-cause dementia, AD, VD, and all-cause mortality in European populations, and its discrimination abilities for dementia and death are significantly higher than those of the SCORE and CAIDE risk algorithms. Further validations in other populations are warranted.

Keywords: All-cause mortality; Dementia; SCORE2; UK Biobank.