Endocrine Abstracts

Aim: To determine the association between recorded inpatient glycaemic variability and long-term mortality in patients with diabetes mellitus.

Methods: Inpatient capillary blood glucose (CBG) readings from eight acute hospitals were included. Data was analysed from first admission within the dataset with >4 measured CBGs. 28 353 admissions were identified (24 181 type 2 DM, 4,172 type 1 DM). Matching parameters were: age, diabetes duration, admission duration, median CBG and interquartile range of CBG values. Mortality analysis was performed from 90 days post discharge over five years, investigating 1) those with CBG IQR in the top half of all IQR measurements (matched for all except IQR), vs those in the lower half and 2) those with the lowest quartile median glucose (matched for all except median).

Results: 1) Glycaemic variability: 3099 matched pairs. Total mortality over the period of analysis: 31.2% (high IQR) vs 26.5% (low IQR) (P<0.001, HR 1.22 – Cox proportional hazard model of survival analysis). 2) Median glucose: 4177 matched pairs. Total mortality over the period of analysis: 32.4% (lowest quartile median glucose) vs 26.1% (top 3 quartiles median glucose) (P<0.001, HR 1.25 – Cox proportional hazard model of survival analysis).

Conclusion: Higher inpatient glycaemic variability is strongly associated with increased long-term mortality. With cohort IQR matching, lower median CBG is associated with higher long-term mortality. CBG variability may increase risk by increasing hypoglycaemia exposure. This will occur more frequently for any given variability level at lower median CBG. Variability per se has also been postulated to increase oxidative stress. This may act synergistically with hypoglycaemia increasing cardiovascular morbidity.

Glycaemic targets and treatment modalities for inpatients with diabetes should primarily aim to minimise glycaemic variability. Where greater CBG variability is unavoidable a less stringent CBG target should be considered.