Improving Diabetes Risk Assessment and Screening in Minority-Predominant Community Health Center Patients

Aim 1: Compare the performance of alternate screening criteria in identifying clinically detected dysglycemia using national longitudinal EHR data from minority-predominant community health center patients. Hypothesis 1A: By failing to identify minority patients who develop dysglycemia while young (<40 years old) and lean [body mass index (BMI) <25kg/m2], the USPSTF criteria will have less favorable performance in predicting incident dysglycemia than the ADA criteria based on the receiver operating characteristic curve. Hypothesis 1B: A novel, pragmatic risk model that includes all potentially relevant EHR variables will have the most favorable performance in predicting incident dysglycemia, followed by models based on ADA and USPSTF criteria, respectively. Aim 2: From a health system perspective, estimate and compare the cost-effectiveness of screening minority-predominant community health center patients using the following approaches, based on: 1) variables from the novel EHR risk model; 2) USPSTF criteria (age and BMI); and 3) ADA criteria (at least 2 T2DM risk factors). Hypothesis 2: Screening community health center patients with novel EHR-based criteria and the ADA criteria will be more cost-effective than adopting the USPSTF criteria.