Aim: To evaluate factors predictive of insulin discontinuation in subjects with ketosis-prone Type 2 diabetes.
Methods: One hundred and six subjects with ketosis-prone Type 2 diabetes were recruited during the index episode of diabetic ketoacidosis (DKA). All subjects were followed in a special clinic for at least 6 months. If the subject's glycaemic control reached specified glycaemic goals, exogenous insulin was gradually decreased until discontinuation. Baseline and follow-up characteristics were compared between the off-insulin and the on-insulin groups.
Results: At the end of the follow-up period (915+/-375 days) insulin was discontinued in 47% subjects. Subjects in the off-insulin group were significantly older at the time of diagnosis of diabetes. In the off-insulin group the majority of subjects were newly diagnosed with diabetes. After 6 months of follow-up, subjects in the off-insulin group had significantly lower mean HbA(1c), higher mean C-peptide-to-glucose ratio and had more clinic visits per year. In the proportional hazard analysis, new-onset diabetes [hazard ratio (HR) 1.54; 95% confidence interval (CI) 1.02-2.45], and a higher C-peptide-to-glucose ratio at 6 months of follow-up (HR 1.77; 95% CI 1.22-2.63) were significant predictors of insulin discontinuation.
Conclusions: In subjects with ketosis-prone Type 2 diabetes, the best predictors of insulin discontinuation are having new-onset diabetes, and higher beta-cell functional reserve (as measured by the C-peptide-to-glucose ratio).