Algorithm Improves Blood Sugar Control in Hospitalized Patients

UCSF physicians develop unique integrated health record calculator that automatically determines insulin dosing, decreasing the rate of hyperglycemia and hypoglycemia, while reducing the need for physicians to write new insulin orders by a factor of 12.

By Melinda Krigel

Hospitalized patients with complex dietary restrictions often develop hyperglycemia, or high blood sugar. This occurs in roughly one-quarter to one-half of these patients, leading to serious complications, particularly in those with pre-existing diabetes. Controlling blood sugar in the hospital setting is challenging for a variety of reasons including inconsistent caloric intake, changes in kidney and liver function, surgery, infections and limitations in labor-intensive glucose monitoring and insulin administration.

To meet these challenges, UCSF Health endocrinologist Robert J. Rushakoff, MD, MS, developed a self-adjusting subcutaneous insulin algorithm (SQIA) and embedded it in the medication administration record (MAR) of the electronic medical record. The SQIA can automatically calculate the next insulin dose needed and has been used with patients in UCSF Health’s three San Francisco hospitals. The results of his study showed that, compared to conventional insulin dosing, the SQIA was associated with lower rates of severe hyperglycemia and hypoglycemia (low blood sugar) and a decreased length of hospital stay.

Rushakoff reports on the glucose control results from the study at the American Diabetes Association's annual Scientific Sessions June 22 in Orlando, Florida.

In the first three years of full implementation (September 2020 to September 2023), the SQIA was used with thousands of hospitalized patients who were on feeding restrictions in three categories: nothing by mouth (NPO), continuous tube feeds (TF) and intravenous food through a vein (TPN).

When a physician ordered rapid-acting insulin to be administered to a patient in one of these nutritional categories, the physician was given the choice to use the SQIA or proceed with conventional insulin dosing orders. Physicians who used the SQIA entered only an initial starting insulin dose, which the algorithm automatically adjusted for subsequent doses. Those who used conventional insulin dosing, by contrast, were required to manually enter new dosing orders when needed.

Over time, clinicians built a better algorithm 

At insulin dosing times, a nurse entered the patient’s current glucose level in the MAR, and the SQIA used the prior insulin doses, prior glucose levels and current glucose levels to automatically calculate the new insulin dose. Based on constant monitoring and feedback from Rushakoff’s team of nurses, pharmacists, physicians and programmers, adjustments were made to the algorithm and the calculator interface to improve titration of the appropriate insulin dose for the patient.

Impressively, the researchers found the SQIA reduced the number of insulin orders physicians wrote by more than 12 times compared to conventional insulin dosing.

“When compared to conventional insulin orders, use of the SQIA decreased the rate of hyperglycemia, further reduced our already low hospital-wide rate of hypoglycemia and improved physician efficiency with physicians rarely needing to write or adjust new orders in the SQIA order set,” said senior author Rushakoff, UCSF professor of Medicine and medical director for inpatient diabetes at UCSF Health.

The SQIA resulted in higher doses of insulin administered in NPO and TPN diets with reduced rates of severe hyperglycemia and without an increase in hypoglycemia, suggesting conventional, physician-driven orders may be undertreating patients. In addition, rates of severe hyperglycemia decreased progressively over the course of the study, suggesting that the SQIA’s continued implementation has provided increasing benefits for patients over time.

“Our findings suggest that typical insulin inertia seen in adjusting insulin doses in many institutions would be overcome by an automated algorithm like the SQIA that reduces physician workload,” said Rushakoff. 

The success of the SQIA led it to being the primary method of ordering insulin for hospitalized patients across UCSF hospitals and is selected for approximately 80% of eligible hospitalized patients by physicians.

This work builds on previous innovations in inpatient diabetes management at UCSF. In 2013, the virtual glucose management service (vGMS) was conceived and implemented. Each morning, the vGMS generates an automated report of all inpatients with uncontrolled blood glucose. A diabetes specialist reviews this report remotely, along with the insulin-glucose chart, and enters insulin dosing recommendations into each patient’s EMR. These recommendations are available for clinician review by 6:30 a.m. daily. Since implementing the vGMS, UCSF has seen a 50% decrease in the number of inpatients on the daily high-glucose report and rates of hypoglycemia have been consistently low. Publication about the success of the vGMS has led to local versions of the vGMS being implemented at medical centers around the world.

Other UCSF authors: Gwendolyn Lee, Michael Kohn, Esther Rov-Ikpah, Paras B. Mehta, Craig San Luis, Craig Johnson, Suneil Koliwad and Cynthia Fenton.

Funding: This work was supported by research-directed discretionary funds provided by the Division of Endocrinology and Metabolism at UCSF Health

About UCSF Health: UCSF Health is recognized worldwide for its innovative patient care, reflecting the latest medical knowledge, advanced technologies and pioneering research. It includes the flagship UCSF Medical Center, which is a top-ranked specialty hospital, as well as UCSF Benioff Children’s Hospitals, with campuses in San Francisco and Oakland, Langley Porter Psychiatric Hospital and Clinics, UCSF Benioff Children’s Physicians and the UCSF Faculty Practice. These hospitals serve as the academic medical center of the University of California, San Francisco, which is world-renowned for its graduate-level health sciences education and biomedical research. UCSF Health has affiliations with hospitals and health organizations throughout the Bay Area. Visit Follow UCSF Health on Facebook or on Twitter.