PRO103 REASSESSMENT OF A REAL-WORLD POPULATION OF ADULTS WITH TYPE 2 DIABETES (T2D): WHEN LOW LEVELS OF C-PEPTIDE OR PRESENCE OF DIABETES AUTOANTIBODIES MIGHT INDICATE POSSIBLE TYPE 1 DIABETES (T1D)

To identify a cohort of patients from a US EMR database that might meet a definition for T1D based on C-peptide or diabetes autoantibodies laboratory findings despite having been classified by their medical provider as T2D patients. From the IQVIA US Ambulatory EMR database of 75 million patients (2006 forward), adults (18 years of age) were selected if they had any problem record between 1/1/2006-10/31/2019 for T2D (ICD-9/10-CM), and excluding those with evidence of T1D. From available labs, C-peptide <0.80ng/mL was assessed and autoantibody status determined for glutamic acid decarboxylase (GAD), IA-2, islet cell (ICA) and insulin antibody tests. An abnormal indicator for each antibody result (as documented in EMR) was used to determine antibody status. Patients were indexed on date of first qualifying test from patient selection step. Patient demographics and clinical profiles were summarized, including insulin and oral antidiabetic prescriptions and HbA1c levels. 7,410 patients with C-peptide or diabetes autoantibody findings were identified from over 5 million T2D patients: mean database observation period 5.0±3.7 years, with 46.3% women, mean age at index 55.8±14.5 years, and 59.6% Caucasian. Diabetes prescriptions within ±12 months of index were identified for 6,816 patients (92.0%), of which 93.5% were prescribed insulin (either with or without oral) and 6.5% were prescribed oral medications. Among insulin patients, mean (SD) for the max-HbA1c within ±12 months was 9.5%±2.2% pre-index and 8.9%±2.0% post-index. The non-insulin group had max-HbA1c of 9.1%±2.3% pre-index but had improved post-index results of 7.9%±2.0%. A cohort of 7,410 patients with laboratory test profile consistent with T1D was identified from EMR data despite having T2D as the provider-specified diagnosis. These patients were primarily prescribed insulin medications, but a small subset was prescribed oral medications and had no evidence of insulin orders in their history, and another small subset were not prescribed any diabetes medications.