Comparing cardiovascular screening strategies in a health-examination population: a real-world markov model study

Cardiovascular disease (CVD) is the leading cause of death in China, yet few studies have used real-world data to systematically assess the health benefits and cost-effectiveness of different CVD screening strategies. To compare three CVD screening strategies in wide use in China, namely the China-PAR score (Strategy 1), atherosclerotic cardiovascular disease (ASCVD) risk assessment (Strategy 2), and ASCVD risk assessment combined with vascular assessment (Strategy 3), against no screening in a health-examination population, and to examine how age and sex shape screening performance. Data came from the electronic health records and the provincial cause-of-death surveillance database of the health management center of a tertiary hospital, covering 2015 to 2024. We included 14,124 people aged 40 to 80 years who had completed the examination, questionnaire, and vascular testing required for risk prediction. Rather than fitting a separate disease model for each strategy, we fitted a single covariate-adjusted three-state continuous-time Markov natural-history model (no CVD; CVD event, a composite absorbing endpoint; and non-CVD death) shared by all strategies, and represented each screening strategy as an intervention-allocation filter that determined which individuals received lifestyle intervention and, where indicated, statin plus antihypertensive therapy. Using per-person Markov cohort simulation over 10 years with 5% annual discounting, we compared each strategy against no screening. Costs comprised screening, intervention, and CVD-event costs. Outcomes were the incremental quality-adjusted life year (QALY), the number of CVD events prevented, the number needed to screen (NNS), the incremental cost-effectiveness ratio (ICER), and net monetary benefit; a cost-effectiveness acceptability frontier (CEAF) was constructed against a willingness-to-pay (WTP) threshold of ¥90,000/QALY. Probabilistic (1,000 Monte Carlo iterations), one-way, and scenario sensitivity analyses were performed, together with subgroup analyses by age group and by sex. Against no screening, the incremental QALYs per person were 0.063, 0.080, and 0.113 for Strategies 1 to 3, and the ICERs were ¥4,816, ¥9,314, and ¥12,089 per QALY, respectively, all far below the WTP threshold. Strategy 3 prevented the most CVD events (636.73 in the cohort) and had the lowest NNS to gain one QALY (8.8). In the fully incremental analysis, Strategy 2 was excluded by extended dominance, and Strategy 3 remained cost-effective against Strategy 1 (¥21,292/QALY). Age strongly shaped benefit: the incremental QALY of Strategy 3 in people aged 55 to 80 (0.224) was about 4.4 times that in the 40-to-54-year group (0.051). China-PAR (Strategy 1) modestly outperformed ASCVD risk assessment (Strategy 2) in women (incremental QALY 0.037 versus 0.035) and in adults aged 55 to 80 (0.160 versus 0.152). At the WTP threshold, Strategy 3 had about a 99% probability of being the optimal option, and every sensitivity analysis kept all strategies well below ¥90,000/QALY. All three CVD screening strategies were highly cost-effective relative to no screening. Strategy 3 was the preferred option overall and especially in adults aged 55 to 80, while China-PAR may be favored in women. Because the three strategies share a single natural-history model and differ only in intervention allocation, the differences in cost-effectiveness reflect the accuracy of risk classification rather than different disease trajectories.