Hypoglycemia in patients with type 2 diabetes newly initiated on basal insulin in the US in a community setting: impact on treatment discontinuation and hospitalization.

ABSTRACT

OBJECTIVE: To evaluate the impact of 6 month hypoglycemia on treatment discontinuation and hospitalization of patients initiating basal insulin for type 2 diabetes (T2D) in real-world practice. METHODS: This was a retrospective cohort study of patient-level data using electronic medical records (EMRs) in the Predictive Health Intelligence diabetes dataset. Data from adult patients with T2D initiating basal insulin glargine, insulin detemir, or Neutral Protamine Hagedorn insulin between January 2008 and March 2014 was analyzed. The date of first basal insulin prescription in an outpatient setting was the index date. A 12 month baseline prior to the index date was established; follow-up was 6-24 months from the index date. Patients were assigned to cohorts by experience of hypoglycemia (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] code or blood glucose test) in the first 6 months following the index date; with hypoglycemia and without hypoglycemia cohorts were compared for basal insulin treatment discontinuation and hospitalization. RESULTS: Overall, 49,062 patients were included; 5159 (10.5%) experienced hypoglycemia in the 6 months following basal insulin initiation. In the first 12 months, 68.1% of patients in the with hypoglycemia cohort discontinued basal insulin versus 53.9% in the without hypoglycemia cohort (p < .0001); more patients in the with hypoglycemia cohort had at least one hospitalization in the first year of follow-up (50.1% vs. 14.6%; p < .0001). CONCLUSION: Patients with hypoglycemia soon after initiating basal insulin are at greater risk of discontinuation of their basal insulin therapy and hospitalization versus those who did not have hypoglycemic events within the first 6 months of basal insulin initiation. A limitation of this study is that it was a retrospective analysis of EMR data and the study may not be representative of all US patients with T2D on basal insulin and it cannot be assumed that every hypoglycemic event was recorded.