Objective: Common diabetes risk screening tools like the American Diabetes Association Diabetes Risk Test (ADADRT) requires survey administration. We studied if diabetes risk can be assessed from routinely available data in the electronic health record (EHR).
Methods: ADADRT was administered to 169 participants without preexisting diagnosis of prediabetes or diabetes, following which the EHR was reviewed for information on ADADRT questions. Standard validation measures were calculated using HbA1c as the gold standard.
Results: About 27% of the sample had prediabetes or diabetes. Of the 7 ADADRT variables, physical activity was the least available in the EHR. We observed that modifying ADADRT by dropping physical activity and changing the risk threshold from ≥5 to ≥4 preserves its diagnostic accuracy. The modified ADADRT (cutoff at ≥4) was then approximated using EHR data, yielding the following validation measures: proportion of high-risk persons 56.8%, sensitivity 77.8%, specificity 50.8%, positive predictive value 36.5%, negative predictive value 86.3%, Youden Index 28.6, and area under the receiver-operating characteristic curve 0.64.
Conclusion: Diabetes risk score can be estimated using routine EHR data. This could mean a quicker eligibility determination for prediabetes testing and referral to the National Diabetes Prevention Program.
C.C. Chima: None. N. Anikpezie: None. L.S. Pongetti: None. B.C. Wade: None. T. Powell: None. B. Beech: None.
