Novel Inpatient Automated Self-Adjusting Subcutaneous Insulin Algorithm: 3-Year Experience. An Observational Study

Achieving inpatient glycemic control in patients who are nil per os (NPO), on enteral tube feedings (TF), or total parental nutrition (TPN) remains extremely challenging.

To determine if, for inpatients with hyperglycemia who are NPO, on TF, or on TPN, an automated self-adjusting subcutaneous rapid-acting insulin algorithm (SQIA) we developed and programmed into the electronic medical record (EMR) leads to improvements in glucose control compared to conventional insulin treatment (CI).

Retrospective cohort study using EMR data from September 3, 2020, to September 2, 2023, of all adult inpatients, comparing point-of-care (POC) glucose measurements between patients on SQIA vs CI and either NPO, or on TF, or on TPN. The analysis looked at the proportion of q4 hour POC glucose levels in the following ranges: hypoglycemia (<70 mg/dL), in range (71-180 mg/dL), moderate hyperglycemia (181-250 mg/dL), and severe hyperglycemia (>250 mg/dL).

There were 5031 intervals (associated with 4310 hospitalizations) in which the patient was NPO or on TF or TPN and on the SQIA (73.5%) or CI (26.5%). The proportion of glucose values in the hypoglycemic and severely hyperglycemic ranges were significantly lower in the SQIA group vs the CI group (hypoglycemia: 0.65% vs 1.10%; difference -0.45%; -0.62 to -0.28%; P < .001; hyperglycemia: 5.40% vs 6.65%; difference -1.25%; -2.03% to -0.46%; P = .002). With glucocorticoids, severe hyperglycemia rates were lower for patients on SQIA, particularly those receiving high-dose glucocorticoids (11.1% lower).

Patients had a lower proportion of both hypoglycemic and severely hyperglycemic measurements when their blood glucose levels were managed using the SQIA than when managed using conventional physician-driven insulin orders.