Hyperglycemia and insulin use in patients with COVID-19 and severe hypoxemia allocated to 12 mg vs. 6 mg of dexamethasone: a secondary analysis of the COVID STEROID 2 randomized trial

Abstract

Background: While dexamethasone has been shown to improve survival in COVID-19, its dose-response relationship with plasma glucose (PG) levels and insulin requirements is poorly understood. This study investigated the impact of 12 mg (higher dose) versus 6 mg (standard dose) of dexamethasone on hyper- or hypoglycemic events and the use of insulin.

Methods: A secondary analysis of a subpopulation of the COVID STEROID 2 trial. Glycemic outcomes were assessed by time-to-event analysis of a hyperglycemic (two PG measurements ≥ 11.1 mmol/L), severe hyperglycemic (PG > 20 mmol/L), hypoglycemic (< 3.8 mmol/L) event or use of insulin, adjusted for age, diabetes status, hospital site, and mechanical ventilation. PG levels were compared before and after treatment allocation with linear mixed models to estimate changes in average PG levels over time.

Results: Of 321 participants, 170 were allocated to the higher dose and 151 to the standard dose of dexamethasone. Time to a hyperglycemic event did not differ between groups, whereas severe hyperglycemic events were more frequent in the higher dose group (36%) than in the standard dose group (31%) with an adjusted subdistributional hazard ratio of 1.76 (95% CI [1.22-2.54], p = 0.003). Insulin use and hypoglycemic events did not differ between groups. The higher vs. standard dose group had an average PG increase of 0.5 mmol/L (95% CI [- 0.2 to 1.4], p = 0.149).

Conclusion: Higher vs. standard doses of dexamethasone were associated with a higher incidence of severe hyperglycemia in patients with COVID-19 and severe hypoxemia, but the average increase in PG was similar between groups.

Keywords: Adverse events; COVID-19; COVID-steroid 2 trial; Dexamethasone treatment; Higher and standard dose; Hyperglycemia; Hypoglycemia; Insulin; Secondary analysis.

Fig. 1

a Cumulative incidence functions for a hyperglycemic event (two PG measurements ≥ 11.1 mmol/L) or a severe hyperglycemic event (≥ 20 mmol/L) after allocation to 12 mg vs. 6 mg of dexamethasone in patients with COVID-19 and severe hypoxemia. b Cumulative incidence functions for daily use of ≥ 20 IE or ≥ 40 IE rapid-acting insulin after allocation to 12 mg vs. 6 mg of dexamethasone in patients with COVID-19 and severe hypoxemia

Fig. 2

Cumulative incidence functions for a hyperglycemic event (A) (two PG measurements ≥ 11.1 mmol/L) or a severe hyperglycemic event (B) (≥ 20 mmol/L) within participants allocated to 12 mg or 6 mg of dexamethasone treatment stratified by diabetes status. The diabetes groups were defined as: Normoglycemia: HbA1c < 39 mmol/mol, prediabetes: HbA1c 39–47 mmol/mol, diabetes: HbA1c ≥ 48 mmol/mol, a known diagnosis of diabetes or on glucose-lowering treatment prior to admission

Fig. 3

Cumulative incidence functions for daily insulin use of ≥ 20 IE (A) or ≥ 40 IE (B) within dexamethasone treatment in participants treated with 12 mg or 6 mg of dexamethasone stratified by diabetes status. The diabetes groups were defined as: Normoglycemia: HbA1c < 39 mmol/mol, prediabetes: HbA1c 39–47 mmol/mol, diabetes: HbA1c ≥ 48 mmol/mol, a known diagnosis of diabetes or on glucose-lowering treatment prior to admission. For visibility, confidence intervals are not shown

Fig. 4

Glycemic variability from the pre-dexamethasone period to randomization to 12 mg or 6 mg of dexamethasone. The figure shows the predicted average plasma glucose values three days before and 6 days from randomization (day 0). The predictions were obtained by two linear mixed models adjusted for age, diabetes status, mechanical ventilation, baseline average plasma glucose, dexamethasone treatment prior to allocation, and hospital site with inverse probability of censoring weight for discharge. Confidence intervals were calculated by bootstrap resampling method