Taking glucose management online

New algorithms and virtual systems can help hospitals to optimize inpatient glycemic control.

Numerous studies have shown an association between poor glycemic control and negative outcomes in hospitalized patients, but recommended glucose targets remain elusive for many hospitals. Recently, however, some institutions have made inroads by leveraging the power of computerized systems to optimize insulin dosing.

Photo by Thinkstock
Photo by Thinkstock.

Methods range from commercially developed electronic glucose management systems (GMS) that use algorithms to guide insulin delivery—such as Glucommander, EndoTool Glucose Management System, and GlucoStabilizer—to customized programs that use electronic health records (EHRs) to incorporate specialist expertise.

“Many physicians still aren't familiar with computerized glucose management,” said Robert Tanenberg, MD, MACP, professor of medicine at East Carolina University and medical director of the inpatient diabetes program at Vidant Medical Center in Greenville, N.C., which currently uses EndoTool. “But based on the data that's come out recently showing hyperglycemia rates approaching zero, these systems deserve more widespread use.”

A variety of systems

Paper-based protocols have historically involved progressively increasing IV insulin doses to treat hyperglycemia, but without careful monitoring and dose adjustments, that can lead to hypoglycemia, said Dr. Tanenberg.

Additional challenges include that hospitalists aren't necessarily focused on glycemic control in patients who are in the hospital for other reasons, said ACP Member Greg Maynard, MD, chief quality officer at University of California, Davis, Medical Center.

“Educating providers and creating order sets and documentation around glucose management is a lot of work,” he said. “Computerized tools help reinforce appropriate dosing and adjustment and encourage providers to do things in a more standardized way.”

In his previous position at University of California, San Diego, Dr. Maynard used an automated infusion protocol developed in house that considers the rate of descent or change in glucose levels as well as the patient's last glucose value. Whenever a clinician uploads a new value, the protocol automatically generates a recommendation to increase, lower, or maintain the current drip rate.

“That program was very useful in the ICU to get safe control and set up the transitional dose to subcutaneous insulin,” he said.

Newer GMS tools are typically integrated with EHRs and take account of multiple variables to prevent both hyper- and hypoglycemia.

At Vidant Medical Center, which is a 900-bed tertiary teaching hospital, Dr. Tanenberg and colleagues found that implementing EndoTool's GMS led to significant improvements in blood glucose control, according to a seven-year retrospective study published in the March 2017 Endocrine Practice.

In the study, patients with uncontrolled diabetes, stress hyperglycemia, and/or postoperative blood glucose levels above 140 mg/dL achieved target levels within 1.5 to 2.3 hours (4.5 to 5 hours for cardiovascular patients), with minimal hypoglycemia. Hypoglycemia incidence decreased from 1.04% in 2009 to 0.05% in 2015, a period during which patient volume more than doubled.

The tool's algorithm is based on blood glucose readings and insulin doses entered into the EHR by clinicians, along with patients' individual age, weight, diabetes type, response to carbohydrate intake, steroid use, serum creatinine, and glomerular filtration rate. As new information is added, the system continuously recalculates optimal timing for glucose checks and recommended insulin dosing. The IV version of the software also factors in stacking of IV insulin to help prevent over- or under-dosing, said Dr. Tanenberg.

“Since we've been using an electronic GMS, we've achieved better overall control and almost zero hypoglycemia—in less time,” he said. “It's greatly reduced the number of midnight phone calls to physicians about blood sugar levels and virtually eliminated the need for endocrinologist consults.”

Virtual clinicians

Photo by Thinkstock
Photo by Thinkstock.

On the other hand, new electronic systems can also more closely involve specialists in day-to-day care. At the University of California, San Francisco (UCSF), a virtual GMS, developed in-house and integrated with the EHR, allows one diabetes specialist to make daily insulin recommendations remotely for patients at three different hospitals.

The process takes less than an hour compared with a full day for diabetes teams to visit each patient individually, said Robert Rushakoff, MD, FACP, professor of medicine and medical director for inpatient diabetes at UCSF. It's changed the culture around diabetes management, he noted.

“Knowing that a note might show up in the chart forces providers to pay attention,” he said. “Since the system was launched, we've seen a change toward more aggressive glycemic control.”

UCSF also significantly reduced hyper- and hypoglycemia, said Dr. Rushakoff. Compared with the year before the electronic system was launched, its use led to a 39% decrease in hyperglycemic patients and a 36% decrease in hypoglycemic patients at the three UCSF hospitals, according to results published in the May 2, 2017, Annals of Internal Medicine. During the same time period, the number of severe hypoglycemic events fell from 40 to 15.

The process involves generating an automated daily report of patients with uncontrolled blood glucose, which is then reviewed remotely by a diabetes specialist with access to patients' insulin/glucose charts in the EHR. The charts contain enough information for the specialist to make recommendations without contacting clinicians or interviewing patients, said Dr. Rushakoff. If an intervention is needed, the specialist puts a glucose management note in the patient's chart.

Over time, physicians have learned to associate the notes with failure to control diabetes, which has helped overcome clinical inertia and encouraged more aggressive management, he noted. Physicians are now much more likely to proactively write appropriate insulin orders or modify current orders.

“Our teams have learned to open up the note, implement it, and learn from it,” he said. “It's become almost a badge of dishonor to get a note.”

In an era of bundled payments and value-based reimbursement, virtual management can be a cost-effective strategy, he said.

“When you have complex patients with ever-changing nutritional needs and enteral feedings, you need to be more hands-on, but having an endocrinologist on site for daily rounds requires extensive resources,” said Dr. Rushakoff. “With this system, we can effectively manage patients at multiple hospitals from a distance.”

Another of the many barriers to effective management of hyperglycemia is the complexity of coordinating insulin administration with meal delivery, said Joseph Aloi, MD, FACP, professor of internal medicine and chief of endocrinology at Wake Forest School of Medicine in Winston-Salem, N.C.

In his previous position, at Eastern Virginia Medical School, he had success with the Glucommander electronic GMS, which receives a patient's initial insulin dose, then automatically calculates all daily insulin titrations and displays them as on-screen prompts. It takes nutritional status into account in making recommendations for full, partial, or held doses.

“It's very important to coordinate insulin with meals, especially if a patient will not be eating for some reason, such as an upcoming procedure or surgery,” said Dr. Aloi. “This system can recognize an NPO order and put all insulin or medical orders on hold.”

In a nine-hospital study led by Dr. Aloi, 62% of patients using the system achieved recommended blood glucose targets compared with 47% before implementation. The results were published in the January 2016 Journal of Diabetes Science and Technology.

Using an electronic GMS can also improve workflow and save staff time by eliminating the need for clinicians to submit daily orders, noted Dr. Aloi.

“The system keeps everything on a dashboard so that nurses do not have to refer to paper sheets while entering data into the EHR,” he said. “It tells them when a patient's readings are changing rapidly, requiring more frequent glucose checks.”


Not all hospitals' efforts have met with such success. Another study of non-ICU patients looked at the impact of a computerized insulin order set and titration algorithm at a large academic medical center.

The standardized protocol recommended initial starting doses for basal and bolus insulin for all adult patients with hyperglycemia. It included preselected orders for diet, glucose monitoring, correction and mealtime insulin, treatment of hypoglycemia, and diabetes education.

Implementation was associated with decreases in hypoglycemia, but rates of hyperglycemia increased, according to results in the November 2016 Hospital Practice. While the algorithm helped physicians intensify doses based on glucose patterns, it still required them to calculate exact doses and enter them into the computerized physician order entry system.

“Despite the distribution of a pocket card of the algorithm as an additional resource at the time of the educational intervention, adoption was slow,” the authors reported. They concluded that integration with the hospital's EHR is key to improving clinical practice.

Improvements in glycemic control from such interventions are dependent on the implementation methods, as shown by two different studies at the same hospital that examined implementation of standardized insulin order sets—both paper and electronic.

In the first study, published in the September 2012 American Journal of Medicine, introduction of a standardized insulin order set reduced hyperglycemia by 39% in cardiovascular surgery patients compared to a control group. Clinicians had access to educational workshops and materials as well as an overnight helpline, resulting in 100% usage.

In the second study, published in the March 2017 American Journal of Medicine, the same authors evaluated the impact of a computerized system on medical patients and found no improvement due to low institutional uptake.

“When we rolled out the electronic order set hospitalwide, there was not as much training or education,” said Catherine Yu, MD, assistant professor of medicine at the University of Toronto and a staff endocrinologist at St. Michael's Hospital in Toronto, Canada, who was the lead author on both studies. However, a subanalysis of the second study showed that glycemic control did improve when the order sets were used appropriately, suggesting that effective implementation is key.

“Just getting physicians to enter order sets isn't enough,” she said. “You have to teach people to use the system properly in order to affect outcomes.”

Engaging all stakeholders early on is critical, she added. Teams should know not only how to use the new system but also the rationale behind it, because understanding promotes behavior change. In addition, user groups should test new processes and offer feedback on potential refinements and improvements.

Lack of an effective implementation strategy can undermine the effectiveness of an electronic GMS, said Dr. Maynard, who has advised Glytec on its Glucommander system.

“If you just drop in the protocol without training, providers have a tendency to work around it, and it ends up not being very effective,” he said. “Once up and running, these systems can be relatively easy to adopt, but it takes planning and clinical leadership up front.”