Cardiovascular risk assessment in the 2021 European Guidelines on Cardiovascular Disease Prevention – A Population-Based Validation of SCORE2 and SCORE-OP

Yochai Schonmann

Keywords: Cardiovascular risk, SCORE2, SCORE-OP, population-based, validation study

Background:
The 2021 ESC guidelines on cardiovascular disease prevention recommend using updated versions of the Systematic COronary Risk Evaluation prediction model (SCORE2 and SCORE-OP) for risk stratification to guide therapeutic interventions in apparently healthy people (primary prevention).

Research questions:
To assess the accuracy and clinical performance of the 2021 SCORE2 and SCORE-OP in a contemporary population-based cohort.

Method:
A historical cohort using routinely collected data from Israel’s largest health provider, Clalit health services (CHS). We retrieved data on all 1,008,209 CHS members eligible for primary prevention and aged 40 years or over on 1/1/2012. We estimated participants' predicted ten-year cardiovascular risk, and followed them for ten years, or until they were hospitalized for any fatal or non-fatal atherosclerotic cardiovascular disease event (ASCVD).

Results:
We were able to calculate the SCORE and include 893,691 (88.7%) of all primary-prevention-eligible CHS members. The mean age was 57.1±11.8 years; 15.05% (134,493) were aged ≥70; 58.78% (525,318) were women, and 16.26% (145,358) were Arabs. The mean calculated ASCVD risk was 5.54 (SD±5.17). During a mean follow-up of 9.18 (SD±2.15) years, 44,959 (5.03%) people had experienced an ASCVD. The overall ratio of predicted-to-observed events (P/O) was 1.10 and the ROC area was 0.717 (95%CI 0.715-0.720). P/O was 0.97 for men, 1.25 for women, 0.70 for Arabs). 9.85% (88,097) of study participants were deemed by the 2021 guidelines as at a very-high-risk, and were more likely to experience an ASCVD (HR=4.24; 95%CI 4.15-4.33; P<0.0001); sensitivity 25.8% (95%CI 25.4-26.2); specificity 91.0% (95%CI 90.9-91.0); PPV 13.3% (95%CI 13.0-13.4); NPV 95.9% (95%CI 95.8-95.9).

Conclusions:
The 2021 versions of the SCORE overestimated ten-year ASCVD risk among women, and underestimated risk among Israeli Arabs, but provided overall good estimates for the general Israeli population. Guideline-based class I intervention cutoffs were specific, but not sensitive for predicting future events.

Points for discussion:
Are risk prediction calculators practical and useful for routine use in busy primary care?

What could be acceptable cutoffs for treatment in primary cardiovascular prevention?

How should individualized risk prediction calculation be incorporated into patients' and physicians' decision making