Dysglycaemia and the interaction of chronic and acute glycaemia with mortality post cardiac surgery

Abstract

Hyperglycaemia, glycaemic variability and hypoglycaemia, are associated with adverse outcomes in the critically ill. In the cardiac surgical population, the relationship between these metrics and outcomes is poorly defined. We aimed to characterise the relationships between acute dysglycaemia, chronic hyperglycaemia and outcomes in a cardiac surgical population. Using the Medical Information Mart for Intensive Care IV v1.0 (MIMIC IV-v1.0) database, we compiled a dataset of cardiac surgical patients from 2008 to 2019. Multivariable analysis was performed to assess the independent effect of glucose metrics on mortality while controlling for confounders. Prior hyperglycaemia was assessed by measurement of glycated haemoglobin (HbA_1c) pre-operatively. Of the 9132 patients included in the analysis, 27% had known diabetes and the prevalence of unrecognised diabetes was 11%. The mean, cumulative dose of hyperglycaemia and coefficient of variation of blood glucose level all increased significantly (P<0.001) with greater pre-operative HbA_1c. Acute hyperglycaemia was strongly associated with mortality (OR 5.88, 95% CI, 3.03 to 10.6), although this effect was diminished by exposure to chronic hyperglycaemia (HbA_1c ≥6.5%). Hypoglycaemia was strongly associated with mortality irrespective of premorbid glycaemic control. The findings indicate that chronic pre-operative hyperglycaemia attenuates the association between acute hyperglycaemia, glycaemic variability and mortality in the cardiac surgical population, but does not modify the mortality risk associated with hypoglycaemia.

Keywords: Blood glucose; Cardiothoracic surgery; Critical care; Glycaemic variability; Stress Hyperglycaemia.

Fig. 1

Marginal model predictions (with 95% confidence bands) for in-hospital death for pre-defined chronic glycaemia sub-groups: well controlled (HbA_1c <6.5%), adequately controlled (HbA_1c 6.5% - 7.9%), and poorly controlled (HbA_1c ≥ 8.0%), for (A) Cumulative dose hyperglycaemia (B) Number of Hyperglycaemic events (C) Number of Hypoglycaemic events (D) Glycaemic variability (coefficient of variation).

Fig. 1

Marginal model predictions (with 95% confidence bands) for in-hospital death for pre-defined chronic glycaemia sub-groups: well controlled (HbA_1c <6.5%), adequately controlled (HbA_1c 6.5% - 7.9%), and poorly controlled (HbA_1c ≥ 8.0%), for (A) Cumulative dose hyperglycaemia (B) Number of Hyperglycaemic events (C) Number of Hypoglycaemic events (D) Glycaemic variability (coefficient of variation).