SEPTEMBER 30, 2015
These Algorithms Reduce Readmissions
Using glycemic management software that integrates with its EHR system, a Virginia hospital system has achieved a 79% improvement over the national average for hyperglycemia rates.
Doctors and nurses following the standard of care in hospitals aren’t just paying more attention to analytics. These days, the algorithm can be the cornerstone of superior care.
Consider the tricky process of controlling the glycemic levels of patients with diabetes.
Working out the algorithm on paper that will enable a patient to avoid either hyperglycemia or hypoglycemia, although a time-honored practice, is no match for the speed, efficiency, and accuracy of well-crafted algorithms.
At eight-hospital, Virginia-based Sentara Healthcare, this translates into a 79% improvement over the national average for hyperglycemia rates. These hospitals have achieved an overall hypoglycemia rate of 0.83 percent, which hospital officials says is a remarkable blended rate for IV and subcutaneous insulin therapy.
Through improved glycemic management initiatives, Sentara expects to be able to dramatically reduce readmissions and length of stay, as well as to improve patient safety.
The algorithms, provided in software from Glytec, take inputs such as the patient’s weight and type of diabetes, and calculate a target glucose range for the patient as specified by the physician, says Miller Trimble, director of IT at Sentara.
‘Prone to error’ on paper
“We had all those algorithms laid out on paper, and it was complicated, difficult to follow, inefficient, and certainly prone to error,” he says.
Complicating the task was understanding how other conditions the patient had dictate the glucose range, and the glucose trend – increasing or decreasing. “A decreasing level for a patient can cause brain damage, require a higher level of care, and can even cause death,” says Sentara CIO Bert Reese.
Of particular note to those who fret over the workflow complications that technology can introduce, Glytec’s Glucommander software has successfully been incorporated into the workflow of Epic electronic health record software.
In a prior incarnation of the technology at Sentara, clinicians either had to use an inferior home-grown algorithm incorporated into Sentara’s EHR, or had to exit the EHR, launch Glucommander from a separate desktop icon, and key in the patient’s information, Miller says.
“We’ve very tightly integrated the Glucommander software with our electronic medical record application, and it’s very seamless for the nurse, while they have the patient’s record open in our electronic medical record, to quickly hit an icon and have that same patient’s record pulled up in the Glucommander system, get a dose calculated with an input or two, and then know at that point what they need to give, and they can administer it right there back in the EMR,” Miller says.
Epic integration
Formal training on Glucommander alleviated nurses’ fears that the technology would not be used properly, and in four Sentara hospitals running EHR software other than Epic, discussions are underway to implement Glucommander, Miller says.
The Sentara Glucommander story is one of a growing number of stories I am hearing that puts the lie to conventional wisdom that Epic doesn’t work well with third-party software.
“We have a lot of these types of extensions that we’re talking about here, embedded, where we exit Epic and go back into it, and it’s been relatively straightforward for us,” Reese says.
Miller adds that Sentara did have to perform a small amount of custom programming to integrate Glucommander with Epic, but “we used the available Epic tools and it wasn’t difficult,” he says.
“We were starting at a rate of around 25% hyperglycemia, defined as any glucose determination greater than 180, and over the years, we implemented both IV and then subcutaneous Glucommander,” says Paul Chidester MD, Sentara vice president of medical affairs. “We’ve been able to take it down to a rate of around 19%, without a change in hypoglycemia. So we’ve had some really good results.”
It’s also worth noting that unlike garden-variety analytics software, the FDA regards Glucommander’s algorithms as a medical device, and so Glytec applied for and received FDA approval.
Modifying the order sets
Incorporating the Glytec algorithms also required tweaking some newly developed order sets at Sentara, Chidester says.
“We had basically one IV insulin order set for the whole system, with a few variations, whether the patient had a cardiac surgery, or whether they were in OB, where you have even tighter insulin control,” he says. “We basically swapped out that order set for an order set that has Glucommander in its place, and there was a need to make some minor modifications to the order set. Glucommander [is] basically the only way that folks now can treat patients with IV insulin.”
“For the subcutaneous module,” Chidester continues, “physicians still have the choice to use either our preexisting weight-based order sets in our EMR, or use Glucommander. We didn’t want to completely eliminate the preexisting order set, because number one, the subcutaneous module in Glucommander is not FDA approved for nondiabetics, and there are some patients that come in with stress hyperglycemia as a result of their surgery or illness, and they need to be managed on subcutaneous insulin, and again, there’s not FDA approval to use Glucommander for that.
“And some of the endocrinologists and other folks still have desire to manage it on their own, Chidester says.” We’re a teaching institution, and some of the medical school faculty really want the residents to learn how to dose insulin on their own, which is perfectly appropriate, because not every place has this tool.”
Adds Chidester: Sentara is in the midst of collecting and analyzing data on the outcomes from using these algorithms, so it will be able to quantify reduced length of stay, readmissions, surgical site infections and mortality.
In their Consensus Statement on Inpatient Glycemic Control, the American Association of Clinical Endocrinologists (AACE) and the American Diabetes Association (ADA) agree, “The complexity of inpatient glycemic management necessitates a systems approach that facilitates safe practices and reduces the risk for errors.”
I am certain we will see an ever-increasing number of such joint statements about the advent of more such technology-driven systems approaches across a wider and wider range of specialties in medicine.
