Performance of the 2015 US Preventive Services Task Force Screening Criteria for Prediabetes and Undiagnosed Diabetes

Matthew J. O’Brien, Kai McKeever Bullard, Yan Zhang, Edward W. Gregg, Mercedes R. Carnethon, Namratha R. Kandula, Ronald T. Ackermann

In 2015, The US Preventive Services Task Force (USPSTF) recommended screening for prediabetes and undiagnosed diabetes (collectively called dysglycemia) among adults aged 40-70 years with overweight or obesity. The recommendation suggests that clinicians consider screening earlier in people who have other diabetes risk factors. To compare the performance of limited and expanded screening criteria recommended by the USPSTF for detecting dysglycemia among US adults. Cross-sectional analysis of survey and laboratory data collected from nationally representative samples of the civilian, noninstitutionalized US adult population. A total of 3643 adults without diagnosed diabetes who underwent measurement of hemoglobin A1c (A1c), fasting plasma glucose (FPG), and 2-h plasma glucose (2-h PG). Screening eligibility according to the limited criteria was based on age 40 to 70 years old and overweight/obesity. Screening eligibility according to the expanded criteria was determined by meeting the limited criteria or having ≥ 1 of the following risk factors: family history of diabetes, history of gestational diabetes or polycystic ovarian syndrome, and non-white race/ethnicity. Dysglycemia was defined by A1c ≥ 5.7%, FPG ≥ 100 mg/dL, and/or 2-h PG ≥ 140 mg/dL. Among the US adult population without diagnosed diabetes, 49.7% had dysglycemia. Screening based on the limited criteria demonstrated a sensitivity of 47.3% (95% CI, 44.7-50.0%) and specificity of 71.4% (95% CI, 67.3-75.2%). The expanded criteria yielded higher sensitivity [76.8% (95% CI, 73.5-79.8%)] and lower specificity [33.8% (95% CI, 30.1-37.7%)]. Point estimates for the sensitivity of the limited criteria were lower in all minority groups and significantly different for Asians compared to non-Hispanic whites [29.9% (95% CI, 23.4-37.2%) vs. 49.8% (95% CI, 45.9-53.7%); P < .001]. Diabetes screening that follows the limited USPSTF criteria will identify approximately half of US adults with dysglycemia. Screening other high-risk subgroups defined in the USPSTF recommendation would improve detection of dysglycemia and may reduce associated racial/ethnic disparities.