MON-010 Prediabetes Detection Using Electronic Health Records Data: A Path to Population Health

Diabetes (DM) prevalence continues rising at alarming rates. The disease is debilitating resulting in significant morbidity, mortality, and psychosocial challenges for nations across the globe. Models of care largely aim at treating and rehabilitating, while emphasis on wellness promotion, prevention, and early detection requires greater attention. Data capture through electronic medical records (EMRs) has become more prevalent. Reliable mechanisms to assess people at risk for DM and to intervene early are needed. In this study we developed and validated an EMR detection tool to identify under recognized cases of prediabetes (preDM). A series of decision support algorithms were deployed to query EMR data in an academic medical center to identify adults who met industry acceptable clinical criteria of glycohemoglobin (A1c) for preDM, but were not diagnosed within a defined diagnostic code range. Additional data included demographics and medications for DM or preDM. For verification, the collected data was reviewed for accuracy. Within a 22 weeks period, the algorithm screened among 37,997 adult patients discharged from inpatient or observation units, same day admission and emergency department. The screening tool searched for subjects ≥ 18 years whose primary care was provided within the organization, with A1c levels between 5.7 to 6.4%, not receiving pharmacological treatment for preDM, and lacking a discrete diagnosis of preDM. A total of 255 subjects were identified with these criteria and 209 cases (82%) confirmed for accuracy. The imprecision resulted from having one of the following: diagnosis code of preDM in their record in another format (37), diagnosis code for DM (7), or A1c ≥ 6.5% (2). Our first iteration of a preDM detection algorithm recognizes an important condition that precedes DM among people at risk of being unrecognized. The algorithm can be applied to all care areas; ambulatory and inpatient settings. This tool provides a mechanism beyond practice performance to accountably identify preDM. It represents an effective strategy to use EMR data for disease identification, with potential demonstrable outcomes for meaningful use of the EMR and for population management. Future refinement of this tool will revise and expand on diagnostic documentation codes to increase its accuracy. Our actionable strategy in progress includes a clinical decision support program for awareness and recommendations to providers and patients. Shaw JE, Sicree RA, Zimmet PZ. Global estimates of the prevalence of diabetes for 2010 and 2030. Diabetes Res Clin Pract. 2010;87(1):4-14 Blumenthal D, Tavenner M. The "meaningful use" regulation for electronic health records. N Engl J Med 2010;363(6):501-4 Hersh W, Ehrenfeld, J.M. Clinical informatics. In: Skochelak SE, Hawkins, R.E., Lawson, L.E., Starr, S.R., Borkan, J.M., Gonzalo, J.D., editor AMA Education Consortium. Elsevier; 2017, p. 105-16