A Randomised Study to Assess the Efficacy and Safety of Automated Closed-loop Glucose Control in Insulin Treated Type 2 Diabetes (Phase 1), Inpatient Hyperglycaemia Requiring Subcutaneous Insulin Therapy (Phase 2 and Phase 3) and to Evaluate Use of Closed-loop Applying Faster Insulin Aspart Versus Standard Insulin Aspart (Phase 4)

Hyperglycaemia in hospitalized patients is becoming a common clinical problem due to the increasing prevalence of diabetes mellitus . Hyperglycaemia in this cohort can also occur in patients with previously undiagnosed diabetes, or during acute illness in those with previously normal glucose tolerance. As a result, the prevalence of acute or stress hyperglycaemia in hospitalised patients has been widely reported. A growing body of evidence currently suggest that the degree of hyperglycaemia upon admission and the duration of hyperglycaemia during their illness are associated with adverse outcomes.In-patient hyperglycaemia is now widely recognised as a poor prognostic marker in terms of morbidity and mortality, increased length of stay and cost to the healthcare system.

The current management of in-patient hyperglycaemia in non-critical care is still far from ideal, and vary widely between different centres. The discordance between clinical evidence and practice is due to a number of factors which could potentially undermine patient care and safety. Of these, hypoglycaemia remains one the biggest barriers to managing in-patient hyperglycaemia. There is therefore a need to develop and validate a more effective and safer system to manage in-patient hyperglycaemia.

A closed-loop insulin infusion system has previously been tested and reported to be feasible and safe in intensive care patients. Its utilisation in non-critical patients in the general medical and surgical wards currently remains unproven. Its use in this cohort however could potentially be of significant practical and clinical value, especially in a busy ward environment. The Model Predictive Control (MPC) algorithm developed by our group at the University of Cambridge utilises fundamental glucoregulatory processes and predicts future glucose excursion resulting from projected insulin infusion rates. The algorithm can also account for the patient's meal intake and the duration of action of the short acting insulin used. This has the distinct advantage over the "reactive" approach of sliding scale insulin protocols, which treats hyperglycaemia after it has already occurred.

The MPC algorithm has been studied in intensive care and cardiac surgery patients, and results from these studies to date have been encouraging. It is shown to be associated with a significantly higher percentage of time within the blood glucose target range, without increasing the risk of severe hypoglycaemia. The expectant role of a closed-loop system using the MPC algorithm in non-critical care patients would therefore be to provide clinicians with an effective and safe method to manage hyperglycaemia in hospital.

In early 2017, faster-acting insulin aspart (Fiasp, Novo Nordisk, Copenhagen, Denmark) received marketing authorisation from the European Commission. Due to the more favourable pharmacokinetic profile, Fiasp has the potential to further improve safety and efficacy of fully automated closed-loop glucose control.