Development and validation of a new algorithm for improved cardiovascular risk prediction

Abstract

QRISK algorithms use data from millions of people to help clinicians identify individuals at high risk of cardiovascular disease (CVD). Here, we derive and externally validate a new algorithm, which we have named QR4, that incorporates novel risk factors to estimate 10-year CVD risk separately for men and women. Health data from 9.98 million and 6.79 million adults from the United Kingdom were used for derivation and validation of the algorithm, respectively. Cause-specific Cox models were used to develop models to predict CVD risk, and the performance of QR4 was compared with version 3 of QRISK, Systematic Coronary Risk Evaluation 2 (SCORE2) and atherosclerotic cardiovascular disease (ASCVD) risk scores. We identified seven novel risk factors in models for both men and women (brain cancer, lung cancer, Down syndrome, blood cancer, chronic obstructive pulmonary disease, oral cancer and learning disability) and two additional novel risk factors in women (pre-eclampsia and postnatal depression). On external validation, QR4 had a higher C statistic than QRISK3 in both women (0.835 (95% confidence interval (CI), 0.833-0.837) and 0.831 (95% CI, 0.829-0.832) for QR4 and QRISK3, respectively) and men (0.814 (95% CI, 0.812-0.816) and 0.812 (95% CI, 0.810-0.814) for QR4 and QRISK3, respectively). QR4 was also more accurate than the ASCVD and SCORE2 risk scores in both men and women. The QR4 risk score identifies new risk groups and provides superior CVD risk prediction in the United Kingdom compared with other international scoring systems for CVD risk.

Fig. 1. Final model-adjusted hazard ratios for CVD.

Adjusted hazard ratios in 5,155,595 women and 4,820,711 men, presented at the mean age of 39 years for variables with age interactions. The hazard ratios were adjusted for fractional polynomial terms for age and BMI (see Supplementary Fig. 1, which shows the relevant fractional polynomial terms). SBP is per 20-unit increase. Adj HR, adjusted hazard ratio; FH of CHD, family history of coronary heart disease.

Fig. 2. Effect of the new risk factors on prediction of 10-year CVD absolute risk.

Ten-year CVD risk predictions for men and women over different ages. Predictions for an individual with each of the new risk factors are compared to those of a similar individual of the same age but without the new risk factors (reference individual). In this analysis, the reference individual is a White nonsmoker and has no adverse health conditions, an SBP of 125 mm Hg, a cholesterol/HDL ratio of 4.0 and a BMI of 25 kg m⁻².

Fig. 3. Decision curves for QR4, QRISK3 and Model A.

Decision curves showing net benefit in men and women aged 18–84 years in England and the devolved administrations. Decision curves for QR4, QRISK3 and Model A are compared to those for ‘Treat All’ (intervention in all individuals irrespective of risk threshold) and ‘Treat None’ (intervention in no individuals).

Fig. 4. Calibration of QRISK3 and QR4.

Centile calibration plots of the observed and predicted risks for QR4 and QRISK3 in men and women aged 18–84 years in the English validation cohort. The red crosses show the observed risk versus the 10-year risk of CVD at each level of mean predicted risk. The blue line shows a perfect calibration scenario in which the mean predicted risk is equal to the observed risk.

Extended Data Fig. 1. CVD (primary outcome) and non-CVD death rates per 1000 by calendar year and month over the full study period in the English derivation cohort in those aged 18–84 years.

The two red spikes on the monthly graph show the non-CVD deaths occurring during the first and second COVID-19 pandemic waves in April 2020 and Jan 2021. Data presented in left panel are mean rates with 95% confidence intervals.

Extended Data Fig. 2. Adjusted hazard ratios for CVD risk for fractional polynomial terms for age and BMI and age interactions in the derivation cohort.

Fractional polynomial terms for adjusted hazard ratios for CVD risk for age, BMI and age interactions in the derivation cohort.

Extended Data Fig. 3. Decision curves for QR4, ASCVD and SCORE2 in people aged 40+ in the England validation cohort using the primary CVD outcome definition.

The greatest net benefit is observed for QR4 followed by ASCVD followed by SCORE2.

Extended Data Fig. 4. Predicted and observed 10-year CVD risks for ASCVD and SCORE2 in the English validation cohort in people aged 40+ using the primary CVD outcome.

Predicted and observed 10-year CVD risks for ASCVD and SCORE2 in the cohorts from England in people aged 40+ using the primary CVD outcome definition.