Modified metabolic syndrome criteria for identification of patients at risk of developing diabetes and coronary heart diseases: longitudinal assessment via electronic health records.

BACKGROUND: Metabolic syndrome has been shown to predict type 2 diabetes mellitus and cardiovascular events in well-studied cohorts, but lack of appropriate measures in real-life populations has limited its use in clinical settings. We developed and tested an algorithm to identify patients at risk for future diabetes or coronary heart disease (CHD) events using electronic health records (EHRs) at the Centre Hospitalier Universitaire de Sherbrooke (CHUS).
METHODS: Patients older than 18 years who had at least 1 visit (outpatient or inpatient) at the CHUS in 2002 or 2003 were included. We excluded patients with diabetes or CHD at baseline. Patients with at least 3 relevant measurements were classified as no metabolic syndrome (zero criteria met), at-risk for metabolic syndrome (1-2 criteria met), or having metabolic syndrome (≥ 3 criteria met). Incidence of diabetes and CHD were assessed through 2008.
RESULTS: Data from 31,823 patients were included at baseline: 2997 (9.4%) were classified as having metabolic syndrome, while 18,686 (59%) were classified as at risk for metabolic syndrome. During the 5-year follow-up, having metabolic syndrome was associated with a 20.0% risk of developing diabetes (age- and sex-adjusted odds ratio = 5.12 [95% confidence interval, 4.57-5.74]; P < 0.0001) and a 14.7% CHD event incidence (age- and sex-adjusted odds ratio = 1.83 [95% confidence interval, 1.62-2.07]; P < 0.0001).
CONCLUSIONS: An algorithm based on clinically available EHRs could identify patients at high cardiometabolic risk of future diabetes and CHD in the population receiving care at the CHUS.