In-Hospital Glycemic Dysregulation Associated With Worse Outcomes in COVID-19

ABSTRACT

Introduction: Evidence establishes that COVID-19 patients with DM2 are at increased risk for severe disease and worse outcomes. Peer reviewed data is sparse comparing glycemic control and clinical outcomes among COVID-19 patients with vs. without DM2, and thus we sought to address this gap. Methods: We selected patients at least 18 years old who expired or were discharged between March 16, 2020 through September 15, 2020. Principal analysis compared glycemic patterns among patients with DM2 vs. non-DM2. Median, coefficient of variation (CV), maximum and minimum glucose parameters were computed to characterize longitudinal glycemic patterns. Logistic regression modeling identified significant (p<.05) associations between composite outcome vs. glycemic parameters and putative risks for progression to severe COVID-19. Receiver operating characteristic (ROC) curve identified cut points for glycemic parameters. Cox regression models were employed to control for significant confounders. Continuous data summarized as median was compared using Kruskal-Wallis test. Discrete data were compared with Pearson’s chi-square test. Two-tailed p<.05 was significant. Results: Among 494 patients, 157 (32%) had DM2 with no intergroup differences in age (68 [56–79]), sex (52% male, 48% female), or race (68% Caucasian, 19% Other, 13% African American). Insulin was administered to DM2 (93%) and non-DM2 (54%) patients (p<.0001). Comorbidities were more prevalent in DM2, including cardiovascular (68% vs. 54%, p=.003), renal (72% vs. 52%, p<.0001) and obesity (51% vs. 38%, p<.0001). Markers including D-dimer (0.98 [0.61–1.95] mg/L), lactate dehydrogenase (308 [230–392] U/L), ferritin (436 [174–856] ng/mL), and triglycerides (172 [109–239] mg/dL), were not different in DM2 vs. non-DM2 (p>.05). CRP was greater in patients with (8.6 [3.6–14.6]) vs. without (6.1 [2.0–12.6]) DM2 (p=.005). Baseline glucose in DM2 (163 [121–253] mg/dL) vs. non-DM2 (107 [96–124] mg/dL) was significantly greater, with former an independent predictor of composite outcome (p=.0005). Cox modeling of other glucose parameters in DM2 vs. non-DM2 demonstrated various impact regarding risk for composite outcome including median (155 [128–209], p=.46) vs. (103 [94–118], p=.09);coefficient of variation (28 [19–38], p=.08) vs. (15 [9–20], p=.002);maximum (252 [187–362], p=.0005) vs. (129 [110–156], p=.002);and minimum (99 [79–128], p=.95) vs. (89 [81–98], p=.02). The unified baseline glucose cut point for composite outcome risk controlled for significant covariates was 138 gm/dL (p<.0001), which included respectively 20% and 10% of patients with and without DM2. Conclusion: Glycemic dysregulation in COVID-19 patients is independently associated with ICU admission and/or hospital mortality. Presence of DM2 amplifies glycemic dysregulation, but risk stratification appears warranted in all COVID-19 patients.