Effect of Insulin Delivery Strategies on Glycemic Control and Nursing Exposure in Critically Ill Patients With COVID-19: An Observational Cohort Study

Poor glycemic control is associated with increased risk of mortality, infection, and hospital length of stay in critically ill patients (1). The severe acute respiratory syndrome coronavirus 2—the virus that causes coronavirus disease 2019 (COVID-19)—causes pancreatic β-cell injury, resulting in insulin resistance and hyperglycemia. The hyperglycemic state glycosylates the ACE2 receptor expressed on tissues, promoting viral entry (2). The ensuing insulin resistance is associated with a greater severity of infection (3). There is no formal guidance directing the use of different glycemic management strategies in this patient population (4). The objective of this study was to evaluate glycemic management of patients with COVID-19 using continuous infusion insulin (CII) and subcutaneous insulin (SCI).

This was an observational, single-center cohort study at an academic medical center (August 2020–March 2021). The study included critically ill adult patients with a positive COVID-19 test who were treated with insulin for hyperglycemia for at least 12 h. Selection of glycemic control strategy was at the discretion of the treating team. Management of hyperglycemia was protocolized and conducted by the bedside nurse. Initiation and transition off CII were at the physician’s discretion. Blood glucose (BG) values were assessed at 4-h intervals. The impact of each glycemic management strategy on nursing workflow and personal protective equipment (PPE) use was inferred from the number of interventions for the first 24 h of insulin therapy. The primary outcome was the proportion of BG values within target range for 72 h after the start of insulin therapy. Secondary objectives included the average insulin requirements per day, length of stay, and BG excursions. Odds ratios and Student t test were used to analyze outcomes.

A total of 121 patients were screened, and 98 patients met criteria for inclusion in the study. There was a statistically significant difference in the proportion of BG values within target range of 60–180 mg/dL (Table 1). The CII group was 2.72 times more likely to be within target BG range (95% CI 2.1–3.5, P < 0.001). Mean ± SD BG during the first 72 h of glycemic management was lower for the CII group (158.2 ± 61.5 mg/dL) than the SCI group (181.7 ± 65.9 mg/dL, P < 0.001). There was also a significant difference in glycemic variability (SD of BG values) between the CII and SCI groups (65.3 ± 34.3 mg/dL vs. 43.9 ± 24.1 mg/dL, respectively; P = 0.002).

Bedside nurses entered the patient room mean ± SD 21.6 ± 3.2 times when taking care of a CII patient compared with an average of 6.20 ± 1.8 times when taking care of a SCI patient over the first 24 h (P < 0.001). The CII group had longer hospital (32.9 ± 27.4 vs. 21.0 ± 18.0 days) and intensive care unit (ICU) (18.7 ± 10.2 vs. 11.2 ± 9.6 days) length of stay than the SCI group.

CII was superior for glycemic control in terms of proportion of BG values within target range, average BG, and glycemic variability. Patients managed with CII were nearly three times more likely to achieve target BG goals. The only parameter for which the SCI strategy proved favorable was attributable nursing care intensity and resource demand. The degree of dysglycemia likely signaled profound inflammation and metabolic dysregulation. The magnitude of insulin resistance was consistent with a sustained decoupling of dose response and glucose metabolism. Common medications for supportive care in the ICU setting—such as corticosteroids and vasopressors—exacerbated the underlying pathophysiology and amplify insulin resistance (2,5). There was a striking disparity in insulin requirements between the CII and SCI groups, i.e., ∼100 units of insulin in the first 24 h. This difference cannot be easily explained by severity of illness, uncontrolled diabetes, dextrose support, or concurrent medications. The lack of randomization and baseline differences observed limit conclusions that may be drawn. However, baseline BG appeared to be a reasonable branch point for determining which patients needed CII.

Nurses entered the room on average three times more frequently over a 24-h period to titrate CII compared with SCI. The rate of nursing exposures may have decreased beyond the first 24 h as dysglycemia stabilized in patients on CII. Nursing exposure and use of PPE to provide bedside patient care remain critical concerns. The capacity to support CII likely depends on institution-specific nursing and PPE resources and demand. Further, continuous glucose monitoring may mitigate nurse exposure and PPE use.

Patients in the CII group experienced better glycemic control; monitoring and delivery of glycemic therapy for these patients resulted in substantially greater nursing exposures.