Translating Subphenotypes of Newly Diagnosed Type 2 Diabetes from Cohort Studies to Electronic Health Records in the United States

Read the full article See related articles

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

Novel subphenotypes of type 2 diabetes mellitus (T2DM) are associated with differences in response to treatment and risk of complications. The most widely replicated approach identified four subphenotypes (severe insulin-deficient diabetes [SIDD], severe insulin-resistant diabetes [SIRD], mild obesity-related diabetes [MOD], and mild age-related diabetes [MARD]). However, the widespread clinical application of this model is hindered by the limited availability of fasting insulin and glucose measurements in routine clinical settings. To address this, we pooled data of adults (≥18 years) with newly diagnosed T2DM from six cohort studies (n = 3,377) to perform de novo clustering and developed classification algorithms for each of the four subphenotypes using nine variables routinely collected in electronic health records (EHRs). After operationalizing the classification algorithms on the Epic Cosmos Research Platform, we identified that among the 727,076 newly diagnosed diabetes cases, 21.6% were classified as SIDD, 23.8% as MOD, and 40.9% as MARD. Individuals classified as SIDD were more likely to receive insulin and incretin mimetics treatment and had higher risks for microvascular complications (retinopathy, neuropathy, nephropathy). Our findings underscore the heterogeneity in newly diagnosed T2DM and validated T2DM subphenotypes in routine EHR systems. This offers possibilities for the subsequent development of treatment strategies tailored to subphenotypes.

Article activity feed