Inpatient Diabetes Management With Automated Insulin Delivery Systems Compared to Multiple Daily Injections With a Basal-bolus Regimen - a Randomized Controlled Trial

Aim

The investigators aim to investigate if automated insulin delivery systems (AID) improve in-hospital glycemic and clinical outcomes in patients with type 2 diabetes compared to standard-of-care with a pen-basal-bolus insulin regimen manually titrated by general staff at Herlev-Gentofte Hospital and a clinical decision support system (GlucoTab) titrating the basal-bolus regimen automatically daily at Graz University Hospital.

Population

Hospitalized patients with type 2 diabetes in non-intensive care units (non-ICU) at medical wards at Copenhagen University Hospitals of Herley-Gentofte (affiliated with Steno Diabetes Center Copenhagen) and Medical University Hospital of Graz (N = 92).

Design

This is an investigator-initiated, two-armed, two-site, prospective, randomized, open-label, blinded endpoint (PROBE) trial.

Objectives

The objective is to determine the glycemic and clinical effects of inpatient AID systems in non-ICU patients with type 2 diabetes. Participants will be randomized in a usual-of-care and an AID arm. Diabetes management will be performed by usual care in the control arm based on a basal-bolus insulin regimen and point-of-care (POC) glucose testing. A continuous glucose monitoring (CGM) system (Abbott FreeStyle Libre 3) will be used in all groups for outcome analysis and comparison between the groups. The CGM will be blinded for the control arm, to not interfere with the usual of care because of the higher amount of glucose data. The AID-arm will be managed by an AID system with real-time CGM data transmitted to nursing stations.

Outcomes

Primary outcome: The primary outcome is the difference in CGM-recorded time in range (TIR) (70-180 mg/dl (3.9-10.0 mmol/l)) between the POC- and the CGM-arm according to the 2023 in-hospital CGM consensus during the entire hospital stay.

Secondary outcomes: Outcomes are reported according to the 2023 in-hospital CGM consensus and specified in the protocol during the entire hospital stay, including three levels of time above range (TAR) 180-250mg/dl (10.0-13.9 mmol/l), >250mg/dl (>13.9 mmol/l), and >180mg/dl (>10.0 mmol/l); three levels of time below range (TBR) 54-70mg/dl (3.0-3.9 mmol/l), <54mg/dl (<3.0 mmol/l), and <70mg/dl (<3.9 mmol/l); events of hypoglycemia in three levels, 54-68mg/dl (3.0-3.8 mmol/l), <54mg/dl (< 3.0 mmol/l), and <70mg/dl (<3.9 mmol/l), where the glucose values between the two hypoglycemic events must all be >70mg/dl (>3.9 mmol/l) for at least 15 consecutive minutes(1), including prolonged hypoglycemic events (> 120 minutes), recurrent hypoglycemic events (events preceded by another hypoglycemic event), and recurrent hypoglycemic days (percentage of days with at least one hypoglycemic event on separate days that is preceded by another in-hospital day with hypoglycemia(1)); mean glucose level; standard deviation (SD) of the CGM glucose distribution; coefficient of variation (CV); and insulin doses during hospitalization.

Clinical outcomes: The investigator assess the length of hospital stay as calculated from time of admission until discharge; in-hospital mortality; admissions to intensive care unit; any in-hospital-related complications occurring at least one day after randomization and until discharge, as documented and defined by the treating physician in the electronic health record (e.g., acute kidney injurie, sepsis, etc.)

Method

For the usual-of-care-arm, glucose assessment is done by standard POC glucose testing and insulin is manually titrated by general staff at Herlev-Gentofte Hospital and the glucose assessment is done by standard POC glucose testing and insulin is manually titrated by the GlucoTab system titrating the basal-bolus regimen automatically daily at Graz University Hospital. For the AID-arm, CGM data informs in real time the mylife Ypsopump for automated insulin delivery.

Device

The investigational device is the AID system, containing of the mylife YpsoPump and the FreeStyle Libre 3 sensor.