Why the study?
Guideline-recommended CVD prediction models differ in populations, predictors, and outcome definitions, leaving the implications for risk estimation unclear and direct comparisons limited.
Do the PREVENT, SCORE2, and QRISK3 CVD risk prediction models show comparable performance and transportability in the UK Biobank population?
Population
502,157 UK Biobank participants
Comparison
PREVENT vs SCORE2 vs QRISK3 risk prediction models
Design
Independent external validation and head-to-head comparison study
Key result
PREVENT, SCORE2, and QRISK3 CVD risk prediction models showed similar performance, with AUCs ranging from 0.7092-0.7468 in females and 0.6813-0.6946 in males, despite suboptimal calibration.
Authors
Loading...
Models perform similarly despite heterogeneity but overestimate risk in older adults; leaves open need for prospective standardized validation.
Observational (n=502,157)
Yes
Do the PREVENT, SCORE2, and QRISK3 CVD risk prediction models show comparable performance and transportability in the UK Biobank population?
Effect estimate: AUC 0.7092 to 0.7468 for females and 0.6813 to 0.6946 for males
Despite heterogeneity in their development, PREVENT, SCORE2, and QRISK3 showed similar discriminative performance in the UK Biobank, though all systematically overestimated risk in older individuals.
Kastrati et al. (2026) conducted an observational in Cardiovascular disease (n=502,157). PREVENT, SCORE2, and QRISK3 models was evaluated on Overall predictive performance (discrimination and calibration) (AUC 0.7092 to 0.7468 for females and 0.6813 to 0.6946 for males). PREVENT, SCORE2, and QRISK3 CVD risk prediction models showed similar performance, with AUCs ranging from 0.7092-0.7468 in females and 0.6813-0.6946 in males, despite suboptimal calibration.