Cardiovascular disease risk prediction for people with type 2 diabetes in a population-based cohort and in electronic health record data

Abstract

Objective Electronic health records (EHRs) have become a common data source for clinical risk prediction, offering large sample sizes and frequently sampled metrics. There may be notable differences between hospital-based EHR and traditional cohort samples: EHR data often are not population-representative random samples, even for particular diseases, as they tend to be sicker with higher healthcare utilization, while cohort studies often sample healthier subjects who typically are more likely to participate. We investigate heterogeneities between EHR- and cohort-based inferences including incidence rates, risk factor identifications/quantifications, and absolute risks. Materials and methods This is a retrospective cohort study of older patients with type 2 diabetes using EHR from New York University Langone Health ambulatory care (NYULH-EHR, years 2009–2017) and from the Health and Retirement Survey (HRS, 1995–2014) to study subsequent cardiovascular disease (CVD) risks. We used the same eligibility criteria, outcome definitions, and demographic covariates/biomarkers in both datasets. We compared subsequent CVD incidence rates, hazard ratios (HRs) of risk factors, and discrimination/calibration performances of CVD risk scores. Results The estimated subsequent total CVD incidence rate was 37.5 and 90.6 per 1000 person-years since T2DM onset in HRS and NYULH-EHR respectively. HR estimates were comparable between the datasets for most demographic covariates/biomarkers. Common CVD risk scores underestimated observed total CVD risks in NYULH-EHR. Discussion and conclusion EHR-estimated HRs of demographic and major clinical risk factors for CVD were mostly consistent with the estimates from a national cohort, despite high incidences and absolute risks of total CVD outcome in the EHR samples.

Precision Prognostics for Cardiovascular Disease in Type 2 Diabetes: A Systematic Review and Meta-analysis

“…For example, the FDS study achieved high c-statistics (>0.80) when validated in an Australian cohort, but lower ones (0.58-0.69) when tested in European countries. In line with previous studies 5,6,389 , discrimination for the UKPDS and FRS was generally poor on external validation. Most prediction models focused on baseline characteristics and did not account for time-varying factors that may modify CVD risk (e.g., statin, SGLT-2i, GLP-1 RA).…”

“…However, predicting CVD risk in those with T2D remains a challenge, and existing risk algorithms, such as the UK Prospective Diabetes Study (UKPDS) Risk Engine and Framingham Risk Score (FRS), have shown only modest predictive value in external validation studies. [4][5][6] Thus, it is essential to develop readily available and cost-effective measures that can accurately identify individuals with a higher absolute risk of developing CVD beyond the risk estimated from established risk factors. Precision medicine provides a promising approach to optimize risk prediction by integrating multidimensional data (i.e., genetic, clinical, sociodemographic), accounting for individual differences.…”

Precision Prognostics for Cardiovascular Disease in Type 2 Diabetes: A Systematic Review and Meta-analysis

“…For example, the FDS study achieved high c-statistics (>0.80) when validated in an Australian cohort, but lower ones (0.58-0.69) when tested in European countries. In line with previous studies 6,7,387 , discrimination for the UKPDS and FRS was generally poor on external validation. Most prediction models focused on baseline characteristics and did not account for time-varying factors that may modify CVD risk (e.g., statin, SGLT-2i, GLP-1 RA).…”

“…More than 500 million individuals worldwide are affected by this chronic disease, resulting in significant human and economic costs 3,4 . However, predicting CVD risk in T2D remains a challenge, and existing risk algorithms, such as the UK Prospective Diabetes Study (UKPDS) Risk Engine and Framingham Risk Score (FRS), have shown only modest predictive value in external validation studies [5][6][7] .…”