Aims: To describe the factors predicting c‐peptide within 60 days of diagnosis, in a multi‐ethnic, incident Type 1 diabetes cohort.

Methods: ADDRESS‐2 recruits patients with clinician‐assigned Type 1 diabetes within six months of diagnosis. Clinical, demographic and routine laboratory data are collected. Islet autoantibodies: glutamic acid decarboxylase (GADA), insulinoma‐associated protein 2 (IA‐2A) and zinc transporter‐8 (ZnT8A) are measured in those opting to give a blood sample (52%). We analysed data collected between 01 September 2011 and 30 June 2017.

Results: Of the 4,606 participants recruited, 341 (7.4%) had a random c‐peptide measured in clinic within the first 60 days of diagnosis (80% within the first week of diagnosis). The median c‐peptide was 0.23nmol/l (IQR 0.14–0.38nmol/l). C‐peptide was more likely to be lower if participants: were children (0.19 vs 0.26nmol/l, p = 0.01); presented with diabetic ketoacidosis (DKA) (0.19 vs 0.25nmol/l, p < 0.001); and autoantibody positive (0.23 vs 0.32nmol/l, p = 0.004). There were no significant differences in c‐peptide with gender, ethnicity (White vs non‐White), body mass index (BMI) or time from diagnosis to date of c‐peptide measurement. On multiple linear regression of all significant variables (n = 179), autoantibody positivity (coeff. –0.014, p = 0.005) and presenting with DKA (coeff. −0.13, p = 0.002) were strong independent predictors of lower c‐peptide. When excluding antibody status from the regression model (n = 333), presenting with DKA (coeff. −0.16, p = 0.02) remained the only significant independent predictor of lower c‐peptide.

Conclusion: Patients with Type 1 diabetes have lower c‐peptide close to diagnosis if they present with DKA and they are autoantibody positive, irrespective of age.

Acknowledgement: On behalf of the ADDRESS‐2 Management Committee, Patient Advocate Group and Investigators