Glucose control and mortality in critically Ill patients, based on real world evidence

George Dafoulas; Ilias Kalamaras; Konstantinos Votis; Alexandra Bargiota

doi:10.1530/endoabs.81.OC1.2

Objective: Published results of reduced morbidity and mortality with tight Glycemic Control of critical patients could not be reproduced in large prospective trials. Glycemic goals according to current respective guidelines include a target blood glucose range 140-180 mg/dl, while lower blood glucose targets may be appropriate for some patients. This study aims to provide real world evidence to this field.

Methods: We performed a retrospective study using the Medical Information Mart for Intensive Care Units (ICU) IV open access, anonymized database (MIMIC-IV) based on 15619 ICU admissions between 2008 and 2019 at Beth Israel Deaconess Medical Center, USA. Logistic regression was performed, using age, sex, SOFA, OASIS and proportion of time in glucose bands per ICU stay as predictors, and death in ICU as the target. Glucose bands and time proportions were defined as in Finney et al.¹ (hypoglycemic: blood glucose level <80 mg/dl, stringent: 80-110 mg/dl, normal: 111-144 mg/dl, intermediate: 145-180 mg/dl, liberal: 181-200 mg/dl, hyperglycemic: ≥201 mg/dl), where proportions were time-weighted to cope with variable measurement frequency. The study protocol was approved by the respective Institutional Review Boards.

Results:

Table 1 Relationship between ICU mortality and proportion of time spent in each glucose band, controlling for age (<0.001)., gender, SOFA score (<0.001). and OASIS score (<0.001). Odds Ratios correspond to 1 unit increase
ICU type	Model	ICU mortality (OR)	CI (95%)	P-value
Cardio(n= 7758)	hypoglycemic	4.583	(1.12, 18.69)	0.034
	stringent	0.111	(0.06, 0.19)	<0.001
	normal	0.096	(0.06, 0.16)	<0.001
	intermediate	5.050	(2.69, 9.48)	<0.001
	liberal	254.094	(73.46, 878.95)	<0.001
	hyperglycemic	28.657	(17.18, 47.8)	<0.001
Medical (n= 3463)	hypoglycemic	2.431	(0.95, 6.19)	0.063
	stringent	0.508	(0.34, 0.75)	<0.001
	normal	0.937	(0.62, 1.41)	0.753
	intermediate	1.707	(1.01, 2.9)	0.048
	liberal	2.596	(0.92, 7.31)	0.071
	hyperglycemic	1.230	(0.82, 1.83)	0.311
Surgical (n= 4398)	hypoglycemic	5.366	(1.85, 15.54)	0.002
	stringent	0.364	(0.24, 0.56)	<0.001
	normal	0.578	(0.39, 0.87)	0.008
	intermediate	2.558	(1.59, 4.12)	<0.001
	liberal	4.276	(1.57, 11.62)	0.004
	hyperglycemic	2.318	(1.36, 3.96)	0.002

Table 1 Relationship between ICU mortality and proportion of time spent in each glucose band, controlling for age (<0.001)., gender, SOFA score (<0.001). and OASIS score (<0.001). Odds Ratios correspond to 1 unit increase
ICU type	Model	ICU mortality (OR)	CI (95%)	P-value
Cardio(n= 7758)	hypoglycemic	4.583	(1.12, 18.69)	0.034
	stringent	0.111	(0.06, 0.19)	<0.001
	normal	0.096	(0.06, 0.16)	<0.001
	intermediate	5.050	(2.69, 9.48)	<0.001
	liberal	254.094	(73.46, 878.95)	<0.001
	hyperglycemic	28.657	(17.18, 47.8)	<0.001
Medical (n= 3463)	hypoglycemic	2.431	(0.95, 6.19)	0.063
	stringent	0.508	(0.34, 0.75)	<0.001
	normal	0.937	(0.62, 1.41)	0.753
	intermediate	1.707	(1.01, 2.9)	0.048
	liberal	2.596	(0.92, 7.31)	0.071
	hyperglycemic	1.230	(0.82, 1.83)	0.311
Surgical (n= 4398)	hypoglycemic	5.366	(1.85, 15.54)	0.002
	stringent	0.364	(0.24, 0.56)	<0.001
	normal	0.578	(0.39, 0.87)	0.008
	intermediate	2.558	(1.59, 4.12)	<0.001
	liberal	4.276	(1.57, 11.62)	0.004
	hyperglycemic	2.318	(1.36, 3.96)	0.002

Conclusion: Overall, increased time in the hypoglycemic, intermediate, liberal and hyperglycemic bands is related to increased ICU mortality, while increased time in the stringent and normal bands is related to decreased mortality. However, in medical ICUs, these results are not statistically significant with the available data, e.g. the OR for the “normal” band being around 1.

References

1. Finney, S.J., T.W., 2003. Jama, 290(15), pp.2041-2047.

Endocrine Abstracts

OC1.2