What question did this study set out to answer?

This study aims to evaluate the agreement of statin eligibility across different cardiovascular risk prediction models in a population with metabolic syndrome.

May 29, 2026Open Access

The Role of Cardiovascular Risk Prediction Model Selection in Primary Prevention: An Observational Study of Statin Eligibility Agreement Across Nine Scores in a Lithuanian Primary-Prevention Cohort

Key Points

This study aims to evaluate the agreement of statin eligibility across different cardiovascular risk prediction models in a population with metabolic syndrome.
Cross-sectional analysis of 11,174 adults aged 40–65 years with metabolic syndrome
Compared statin eligibility using nine cardiovascular risk prediction models
Assessed pairwise agreement using multiple statistical methods including Cohen’s κ and McNemar testing.
Eligibility rates varied significantly across models, from 67.39% with SCORE2 to 3.03% with AusCVDRisk.
Overall pairwise agreement was modest, with a median Cohen’s κ of 0.38 and stronger agreement for non-eligibility.
Consensus eligibility declined sharply with increased model stringency, dropping from 73.5% at k = 1 to 1.87% at k = 9.

Abstract

Background and Objectives: Cardiovascular risk prediction models (RPMs) are widely used to guide statin initiation in primary prevention, yet the extent to which different models produce concordant treatment decisions in the same population remains insufficiently characterized. We compared statin eligibility across nine commonly used RPMs: SCORE2, PREVENT, PCE, ASSIGN, FRS-hCHD, AusCVDRisk, MESA, QRISK3, and RRS. Materials and Methods: We performed a cross-sectional analysis of 11,174 adults aged 40–65 years with metabolic syndrome enrolled in the Lithuanian High Cardiovascular Risk primary prevention program (LitHiR) and evaluated them at a single tertiary center during 2006–2023. Statin eligibility was determined for each RPM using guideline-mapped treatment thresholds. Pairwise agreement was assessed using Cohen’s κ, Gwet’s AC1, Positive and Negative Percent Agreement (PPA/NPA), the Jaccard index, and McNemar testing. Analyses were repeated by sex. Consensus eligibility was defined as treatment recommended by at least k of nine models. Results: Eligibility varied more than twenty-fold, from 67.39% (7530/11,174) with SCORE2 to 3.03% (339/11,174) with AusCVDRisk; intermediate estimates included PREVENT at 44.83%, QRISK3 at 39.00%, and PCE at 37.97%. Overall pairwise agreement was modest: κ ranged from 0.03 (SCORE2 vs. AusCVDRisk) to 0.67 (QRISK3 vs. ASSIGN), with a median κ of 0.38 (IQR: 0.19–0.51). Median AC1 was 0.58 (IQR 0.37–0.68). Agreement was stronger for non-eligibility than for eligibility (median NPA: 0.82 vs. median PPA: 0.53). Consensus eligibility declined from 73.5% at k = 1 to 45.1% at k = 3, 30.0% at k = 5, and 1.87% at k = 9, with the greatest sex divergence at intermediate stringency. Conclusions: In this real-world cohort with elevated cardiometabolic risk, statin eligibility was highly dependent on RPM choice and showed only modest inter-model concordance. Increasing consensus stringency rapidly reduced eligibility, indicating that RPM selection and embedded thresholds substantially influence statin treatment decisions in primary prevention.

Read Full Paperexternally

AI से पूछें

Bookmark

View Full Paper