Screening for Diabetes Risk Using Integrated Dental and Medical Electronic Health Record Data

Abstract

Undiagnosed diabetes and prediabetes present a serious public health challenge. We previously reported that data available in the dental setting can serve as a tool for early dysglycemia identification in a primarily Hispanic, urban population. In the present study, we sought to determine how the identification approach can be recalibrated to detect diabetes or prediabetes in a White, rural cohort and whether an integrated dental-medical electronic health record (iEHR) offers further value to the process. We analyzed iEHR data from the Marshfield Clinic, a health system providing care in rural Wisconsin, for dental patients who were ≥21 y of age, reported that they had never been told they had diabetes, had an initial periodontal examination of at least 2 quadrants, and had a glycemic assessment within 3 mo of that examination. We then assessed the performance of multiple predictive models for prediabetes/diabetes. The study outcome, glycemic status, was gleaned from the medical module of the iEHR based on American Diabetes Association blood test cutoffs. The sample size was 4,560 individuals. Multivariate logistic regression revealed that the best performance was achieved by a model that took advantage of the iEHR. Predictors included age, sex, race, ethnicity, number of missing teeth, percentage of teeth with at least 1 pocket ≥5 mm from the dental EHR, and overweight/obesity, hypertension, hyperlipidemia, and smoking status from the medical EHR. The model achieved an area under the receiver operating characteristic curve of 0.71 (95% confidence interval, 0.69-0.72), yielding a sensitivity of 0.70 and a specificity of 0.62. Across a range of populations, informed by certain patient characteristics, dental care team members can play a role in helping to identify dental patients with undiagnosed diabetes or prediabetes. The accuracy of the prediction increases when dental findings are combined with information from the medical EHR. Knowledge Transfer Statement: Prediabetes and diabetes often go undiagnosed for many years. Early identification and care can lead to improved glycemic outcomes and prevent wide-ranging morbidity, including adverse oral health consequences, in affected individuals. Information available in the dental office can be used by clinicians to identify those who remain undiagnosed or are at risk; the accuracy of this prediction increases when combined with information from the medical electronic health record.

Keywords: dentists; hyperglycemia; periodontitis; prediabetic state; prevention & control; risk.

Figure.

Receiver operating characteristic curves of candidate predictive models. Base model (dashed line curve): number of missing teeth + % of teeth with at least 1 deep pocket. Integrated model (solid line curve): number of missing teeth + % of teeth with at least 1 deep pocket + age + sex + race + ethnicity + hypertension + overweight/obesity + hyperlipidemia + smoking status.