Management of blood glucose in critically Ill patients with pre-existing type 2 diabetes

Abstract

This thesis comprises four distinct but complementary chapters on blood glucose management during critical illness, with a focus on management of patients with a pre-existing type 2 diabetes. The work submitted includes a narrative review, survey of clinical practice, study protocol and statistical analysis plan, and the conduct and report from a bi-national, multi-centre, parallel-group, randomised clinical trial. Dysglycaemia, or disordered glucose metabolism, is almost ubiquitous with severe critical illness, with marked increases in endogenous glucose production and counter-regulatory responses. The magnitude of dysregulation is associated with severity of illness on presentation to hospital and subsequent mortality. Despite longstanding knowledge about these associations, the threshold at which hyperglycaemia causes harm remains unknown. There is a further complicating factor to understanding the relationship between hyperglycaemia and harm, is that over a quarter of patients admitted to intensive care units (ICUs) have type 2 diabetes, and a pre-existing disordered glucose metabolism. Evidence to inform the management of blood glucose in critically ill patients is predominately from studies conducted in sample populations that have only a small proportion of patients with pre-existing diabetes. This creates clinical uncertainty in discrete populations, such as those with type 2 diabetes (Chapter 1.2). A survey of clinicians was performed to understand current practice and to determine whether they required further evidence to better care for patients (Chapter 1.3). Observational and exploratory studies have reported that patients with a higher HbA1c on ICU admission had a lower mortality rate if they had a modest elevation of blood glucose concentrations to >10 mmol/L during their ICU admission. Based on such data it is plausible that in patients with type 2 diabetes mild hyperglycaemia during critical illness is protective. Because hypoglycaemia has consistently been shown to be harmful to all patients, and the threshold blood glucose for harm may even be slightly greater in patients with pre-existing diabetes, ‘personalising’ or allowing for different blood glucose ranges based on an individual patient’s pre-existing glucose metabolism has the potential to improve care. Exploratory studies indicate that such a personalised approach did not detect a signal for harm, although the studies included relatively small sample populations and study methodology risked bias, such that there was inadequate evidence to inform practice. For these reasons, further evaluation using a rigorously designed randomised clinical trial was warranted (Chapter 2.2). Treatment of hyperglycaemia in critically ill patients typically utilises intravenous insulin, which poses a risk of causing hypoglycaemia if appropriate commencement and titration parameters are not selected. A pragmatic approach to achieving a more personalised target in patients with type 2 diabetes is to commence insulin at a greater blood glucose concentration (e.g., ≥14 mmol/L) and compare this to what occurs in usual practice (e.g., ≥10 mmol/L), which has been informed by trials conducted in populations predominately comprising patients without pre-existing diabetes. Given the strong relationships between harm and hypoglycaemia, hypoglycaemia is an appropriate outcome by which to measure the impact of an elevated commencement point for intravenous insulin administration (Chapter 3.2). The multicentre, open label, randomised clinical trial was conducted in critically ill patients with pre-existing type 2 diabetes and established that day-28 incidence hypoglycemia (<4.0mmol/L) was significantly reduced by the intervention. While this study was not powered for patient-centred outcomes, there was no benefit in any of the clinical outcomes measured. The implications for clinical practice from the trial conducted is that while optimal management of hyperglycaemia in critically ill patients with type 2 diabetes remains uncertain, current or usual practice should continue. Future trials may benefit from determination of a patients pre-existing glycaemia though HbA1c testing, as long as this can be conducted in a time efficient manner, and offers the potential to further personalise physiological targets. Technological advancements such as dynamic protocols, continuous blood glucose monitors and close-loop systems may offer the opportunity to achieve greater time within target ranges, and reduce the amount of time required to manage blood glucose. This has the potential to reduce the net cost while achieving optimal blood glucose management. Novel therapeutics, such as glucagon-like peptide-1 (GPL-1), may have innate properties that mitigate some of the limitations of intravenous insulin administration. These interventions have the possibility to improve patient outcomes and improve the care provided in intensive care units (Chapter 4.2). In summary, this program of work has contributed new and important information in the fields of glycaemic management, acute glycaemic targets in critically ill patients with type 2 diabetes, and the implications of a personalised approach to blood glucose management.