Reducing Type 2 Diabetes Diagnostic Delays Using Decision Support

Overview

This study will focus on the cohort of 20,000 established patients cared for by 31 attending physicians in the outpatient, adult primary care practices at UT Southwestern (two general internal medicine one family medicine and one geriatric practice). The investigators will develop and implement an automated Diabetes Detection Tool (DDT) that does data mining on electronic medical record (EMR) lab data to systematically identify all primary care patients with elevated random plasma glucose results (RPGs) who are at high risk of diabetes and thus in need of further testing. In a cluster-randomized trial, primary care providers will be randomized to either the intervention/DDT arm or usual care. Providers in the intervention arm will receive visit-based, EMR-enabled case identification and real-time decision support. Outcomes will be tracked at a patient level. All subjects will be followed for 12 months to assess rates of follow-up diabetes testing, time to testing, rates of subsequent diabetes diagnosis, and time to diagnosis. The investigators hypothesize that the visit-based provider decision support will be superior to usual care.

The growing epidemic of type 2 diabetes affects over 8.3% of the US population and presents a major challenge to healthcare systems and public health. An additional 7 million people have undiagnosed diabetes and over 79 million have pre-diabetes, which if unrecognized and untreated can progress to full-blown diabetes. Although screening and diagnostic tests are routinely available, health systems struggle to diagnose patients with diabetes in a timely manner. In fact, clinical diagnosis lags 8-12 years behind the onset of glucose dysregulation, resulting in diagnostic delays and the presence of diabetes complications at the time of diagnosis. Among patients engaged in clinical care without a known diagnosis of diabetes, nearly all patients have random plasma glucose (RPG) data available which potentially provides valuable, early warning safety signals regarding the need for further diabetes testing. However, elevated glucose values are commonly unrecognized and over 60% of abnormal values are not followed-up with diabetes testing in a timely fashion. Opportunities exist to leverage existing data within electronic medical records (EMR) to identify patients in need of further diabetes testing and develop systems-based solutions to reduce: 1) failures in following-up abnormal glucose tests, 2) delays in diagnosing diabetes, and 3) frequency of missed diagnoses of diabetes. This proposal will leverage the Epic EMR at the University of Texas Southwestern Medical Center (UTSW) to improve the detection and follow-up testing rates of abnormal glucose values in real-world practice. The investigators will conduct a cluster randomized, pragmatic trial comparing the effectiveness of a clinical decision support strategy versus usual care to reduce failures in timely follow-up of abnormal RPGs. The investigators will focus on the cohort of 20,000 established patients cared for by 31 attending physicians in three outpatient, adult primary care practices at UTSW (two general internal medicine one family medicine and one geriatric practice). Primary care providers (PCPs) will be randomized to either the clinical decision support intervention or usual care. Providers in the clinical decision support/intervention arm will receive clinical decision support that identifies abnormal random glucose values and prompts providers to conduct diabetes screening. Outcomes will be tracked at the patient level and all subjects will be followed for 12 months to assess rates of follow-up diabetes testing, time to testing, rates of subsequent diabetes diagnosis, and time to diagnosis. Data on study eligibility, patient clinical risk factors and sociodemographics, provider and visit characteristics, and outcomes will be ascertained using the comprehensive Epic EMR. The investigators hypothesize that the visit-based provider decision support will be superior to usual care.