Cardiovascular disease risk in people with diabetes in the era of universal diabetes screening

Abstract Background/Introduction Traditionally, most people with diabetes have been considered at high cardiovascular disease (CVD) risk and CVD preventive treatment has been recommended for almost all patients after diagnosis. However, guidelines are increasingly recommending a more stratified approach to preventive treatment informed by predicted CVD risk. In the early 2000’s, New Zealand initiated the world-first national diabetes screening programme, as a component of a national CVD risk assessment programme, and by 2016, approximately 90% of eligible New Zealand adults had been screened. We hypothesised that in the era of universal diabetes screening, many recently diagnosed patients will be at low CVD risk and that contemporary CVD risk prediction equations would be required to accurately assess their risk. Purpose We aimed to develop sex-specific 5-year CVD risk prediction models in people with diabetes in New Zealand using a contemporary natural-population based cohort, and to characterise the distribution of CVD risk. Methods De-identified individual-level linkage of multiple health administrative datasets was undertaken to establish a cohort of almost all adults with diabetes in the northern region of New Zealand (one-third of the national population), without CVD or heart failure, and aged 30–74 years on January 1, 2014, with follow-up linkage to hospitalisations and mortality until December 31, 2018. Sex-specific models were developed to estimate the 5-year risk of CVD hospitalisation or death, with 15 pre-defined predictors, including sociodemographic variables, diabetes-related and renal function measures. Results 64,645 participants (median age: 57 years [IQR: 49–65]) were included in the final cohort, of whom 4,992 had first CVD events during 301,272 person-years follow-up. Models calibrated well in both women and men (Figure 1). Overall, increasing age; high deprivation level; smoking-related hospitalisations or use of smoking cessation medications; history of atrial fibrillation; high levels of haemoglobin A1c, total cholesterol to HDL cholesterol ratio, and albumin-to-creatinine ratio; use of blood pressure-lowering medications and insulin were associated with elevated CVD risk. Median 5-year CVD risks were 4.9% (IQR: 2.9%–8.1%) in women and 7.7% (4.8%–12.0%) in men. There was a wide range of predicted 5-year CVD risk, with 51% of women and 27% of men having a predicted risk below 5%, and 6% of women and 16% of men having a risk of 15% or above (Figure 2). Conclusions This study demonstrated marked heterogeneity in predicted CVD risk in people with diabetes in an era of universal diabetes screening. As diabetes screening is increasingly undertaken as part of a routine CVD risk assessment, these findings will have important clinical implications for accurately targeting new but expensive CVD preventive medications to the highest risk patients.