(393) Glycemic Ratio, Diabetes Status, and Outcome of Patients Admitted to the ICU With Neurologic Illness

Introduction: Background
Poor glucose control is associated with adverse outcomes of patients hospitalized with neurologic disease for patients with and without diabetes. The Stress Hyperglycemia Ratio (SHR), calculated as the quotient of admission blood glucose (BG) and estimated preadmission BG (EPAG) ( [(28.7 × HbA1c%) – 46.7 mg/dL1]. High SHR has been associated with early neurological deterioration. Recently, a new metric, Glycemic Ratio, the quotient of mean intensive care unit (ICU) BG and EPAG, has been associated with mortality in a large single center cohort, with a U-shaped relationship and a nadir at GR 0.8-1.0. The purpose of this investigation is to assess the relationship of GR to mortality in critically ill patients with non-surgical neurologic disease.

Methods: This is a retrospective analysis of prospectively collected data, including all 634 patients admitted to the ICU at Stamford Hospital with a non-surgical neurological diagnosis between 2019 and 2023, a subset of the 4,148 patients who met the inclusion criteria: . 4+BG, HgbA1c obtained at admission. We investigated the relationships among diabetes (DM) status, mean ICU BG and GR. The primary outcome was mortality, defined as status at hospital discharge. We stratified mortality using bands of GR and created multivariable models to assess the independent association of DM and GR with mortality, using the APACHE IV prediction of mortality (APIV PM).

Results: The median age was 68 and 21.6% had DM. Their mean APIV PM was 23.8% and mortality was 10.1% (p=NS comparing patients with and without DM for all of these measures). Mean ICU BG was 151 vs. 115 mg/dL for patients with and without DM (p < 0.0001).

For the entire cohort, there was a U-shaped relationship between GR and mortality: GR < 0.8, 0.8- < 1.0, 1.0- < 1.2, 1.2- < 1.4, >1.4: 5.6%, 3.8%, 12.3%, 16.5%, 25.9%. Multivariable analysis demonstrated that GR (p=0.03) but not DM (p=0.96) or mean ICU BG (p=0.23) was independently associated with mortality.

Conclusions: Our investigation demonstrates that patients with GR 0.8-1.0, representing BG levels that are 80-100% of preadmission levels, is associated with lowest mortality in patients with non-surgical neurologic illness. These data support the need for trials assessing the role of personalized BG targets.