SMART2 recurrent CVD event risk for stratification of cumulative healthcare costs
J Ebbers, J A N Dorresteijn, F L J Visseren, S H J HagemanAbstract
Background
Risk prediction models are widely recommended and used to guide and optimize preventive treatment of atherosclerotic cardiovascular diseases (ASCVD). While these models quantify the probability of recurrent events, their relation to subsequent healthcare costs remains unclear. Understanding whether higher predicted risk translates into greater healthcare expenditure is essential to evaluate the economic implications of risk-guided prevention. Demonstrating a clear association between predicted risk and costs would support the use of risk stratification not only to estimate health benefit, but also to inform resource allocation and reimbursement decisions in secondary prevention.
Purpose
To quantify the relationship between predicted 10-year recurrent cardiovascular event risk using the SMART2 model and cumulative healthcare costs in individuals with established ASCVD.
Methods
EHR-data were extracted from patients with established ASCVD who had an internal medicine outpatient visit a Dutch university hospital from January 1st 2015 until July 1st 2024. Established ASCVD was defined by previously registered ICD or reimbursement codes indicating a history of symptomatic coronary artery disease, cerebrovascular disease, peripheral artery disease and/or abdominal aortic aneurysm. The SMART2 risk score was calculated for each patient. Longitudinal healthcare expenditure until December 31st 2023 was obtained through linkage with Statistics Netherlands. Healthcare costs included all insured medical costs, including hospital, primary care, medication, home care and long-term care costs. Yearly cumulative costs were analysed across quintiles of risk. Sensitivity analysis with quintiles of age was performed to discern whether potential differences between risk groups was not caused just by age differences. Data were accessed and processed in accordance with Dutch, European, and institutional regulations.
Results
A total of 18,894 patients were included with a median follow-up time of 5.0 years (IQR; 2.7-7.7). The mean total 8-year medical care costs were €77,608, mainly driven by hospital costs (€45,463; 58.6%). The mean total healthcare costs were 54% higher in the highest quintile (€98,790, 95%CI €90,434 to €107,145) compared with the lowest quintile (€64,324, 95% CI €58,899 to €69,748). Risk stratification consistently distinguished between lower and higher mean costs across all cost categories, including long-term care (Figure 1). Discrimination between low and high costs was significantly better with risk stratification than with age-based stratification (Figure 2a-b).
Conclusion
In patients with established ASCVD, Higher SMART2-predicted risk corresponded to progressively higher healthcare costs. These findings suggest that risk-guided prevention may not only optimize health outcomes but also support more efficient allocation of healthcare resources in secondary prevention.