Published in the October 2004 issue of Today’s Hospitalist
About two years ago, Beth Israel Deaconess Medical Center in Boston undertook a project to tightly control the glucose levels of its cardiac surgery patients. A growing body of research had shown that keeping glucose levels in the 80-110 mg/dL range could reduce infection rates and mortality, and nurses, mid-level practitioners and physicians all thought it was time to act.
Staff were so focused on tracking patients’ hourly glucose levels to measure how many patients were meeting the new goal, however, that it was several months before anyone recognized the obvious: The cardiac surgery service had not seen a single case of sternal organ space infection (mediastinitis) in months.
Within two months of giving IV insulin to patients with glucose levels of 125 mg/dL or greater, cases of mediastinitis in the unit had disappeared. Today, nearly 18 months later, the trend continues.
The milestone shows just how successful the hospital’s glycemic control program has been. When the cardiac surgery service began its glycemic control project in 2002, glucose levels were in control “defined as less than 130 mg/dL “about 42 percent of the time in the 48 hours after surgery. Today, by comparison, those levels are on target 68 percent of the time.
Beth Israel Deaconess Medical Center’s experience shows that tightly controlling glucose levels can greatly reduce infections in the hospital setting. It also demonstrates how hard “and how closely “everyone in the unit had to work to change not just practice, but the culture.
Here’s a look at how the hospital was able to control glucose levels, and how that process might translate into other hospitals and other patient care units.
The protocol
The glucose control project started when a group of physicians, nurses, NPs and PAs decided that the literature on glucose control was so compelling that they should institute a protocol in the cardiac surgery unit.
The hospital developed two sets of guidelines: an intraoperative protocol for anesthesiologists and a postoperative protocol for the cardiac surgery service.
Justine Carr, MD, a hematologist who is a director of health care quality at the medical center, says that one key to the protocol’s success was its flexibility. While the protocol was based on the evidence and practices outlined in the literature, the implementation has always been viewed as a work in progress. As a result, the protocol has undergone iterative changes directed by the patient care team.
When the project first began, for example, the protocol directed nurses to initiate an insulin infusion for diabetic patients with glucose levels greater than 150 mg/dL. Nondiabetic patients were initially treated with subcutaneous insulin injections and switched to IV only if two subcutaneous doses failed to control the blood sugar.
Kathy Clark Hussain, RN, nurse manager of the cardiac surgery service, says that nurses quickly realized that the algorithm was too complex. In the hectic pace of her unit, she says, it took extra time to track the response to subcutaneous insulin, and to follow the criteria to switch to an IV infusion.
Staff also learned that subcutaneous insulin often took too long to work. “These are patients who just came out of the OR,” Ms. Hussain says. “Absorption of subcutaneous insulin can vary with their blood pressure and perfusion.”
As a result, the team simplified the protocol and directed nurses to start an insulin infusion on every patient with elevated glucose on arrival in the cardiac surgery service. “The literature says to treat glucose, not just diabetes,” Dr. Carr says. “The benefits are independent of whether the patient has diabetes.”
The protocol was streamlined to target cardiac surgery patients with a glucose higher than 150 mg/dL. The goal was to keep glucose levels at 130 mg/dL or lower. “Going straight to IV made the process much simpler,” Dr. Carr says.
Close monitoring
Dr. Carr explains that the hospital was able to adjust the protocol so regularly because it closely tracked how often patients met the goal of 130 mg/dL. “We developed and automated a metric that summarized whether the patient was in or out of control for each hour during the first 48 hours after surgery,” she says.
A monthly report recorded the unit’s average hours in control, which was reviewed regularly by physicians, nurses and midlevel practitioners. By closely monitoring monthly trends and reviewing details on patients who weren’t well-controlled, Dr. Carr explains, the unit was able to modify the protocol.
Monthly performance reports were also shared with unit’s staff, something Dr. Carr says helped reinforce the work of the entire team.
To make sure that patients were meeting glucose goals, nurses checked patients’ blood glucose levels every hour. While Dr. Carr acknowledges that such frequent checking requires significant effort, she says it’s the only way to really control glucose levels in patients who have just undergone surgery.
“Adrenaline and cortisol are high in the immediate post-op- hours, with a commensurate insulin resistance and hyperglycemia,” she says. “After the first 24 to 48 hours, a lot of that settles down, and the patient’s insulin needs may fall. Attention to this change is important in avoiding hypoglycemia.”
Hourly measures also mean that if a patient’s glucose levels begin to move out of range, nurses can act quickly to correct it. “Even if you get off track,” Ms. Hussain says, “if you measure it an hour later, you’re right on track again. We found that when the protocol wasn’t working, it was often because one of the hourly measures had been missed, and it was two or three hours later and we were playing catch-up.”
Nurses, midlevel practitioners and physicians learned some other valuable lessons about keeping patients’ glucose levels in check. When glucose levels were high, the protocol called only for an adjustment to the infusion rate. Michelle Fey, NP, however, says staff quickly realized that by changing the infusion rate without adding an IV bolus, they could delay control by hours.
While the protocol addressed hyperglycemia, it also helped nurses prevent hypoglycemia. Ms. Fey acknowledges that because patients who have just undergone major surgery are often not awake, the clinical signs of hypoglycemia can be difficult to detect.
“It’s an advantage of hourly monitoring,” she says. “When we start getting into the low threshold, we can either stop insulin or in some cases give a small amount of D50. The protocol tells you what to do when glucose is high, and it tells you what to do when it’s low.”
Ms. Hussain acknowledges that insulin dosing for inpatients still requires “a bit of art,” but she says that the protocol helps guide the nurses. “If a patient is on a trajectory where they’re dropping quickly,” she says, “we’ve built adjustments into the protocol. If there’s a dramatic change, we go more slowly.”
Starting slowly
Another key to the initiative’s success? The hospital started to control glucose levels slowly. When staff first began tracking glucose levels, for example, they targeted only patients whose glucose was greater than 150 mg/dL.
While research had urged physicians to treat glucose levels between 80 mg/dL and 110 mg/dL, Dr. Carr says it was important to start slowly.
“It took us time to get comfortable with the idea that we wouldn’t be inducing hypoglycemia,” she explains. “Staff have to reach a certain level of comfort, and nurses need to adjust to the change in workflow. I would advise people to start with a threshold they’re comfortable with and lower the target range as they gain experience.”
In addition, diabetes specialists from the Joslin Clinic participated in the early development of the protocol. Dr. Carr says their experience helped make sure patients received a controlled level of glucose in their other intravenous infusions. Joslin staff also helped customize care for the small subset of patients with type I diabetes.
While the hospital initially established a threshold for insulin treatment of 150 mg/dL, the clinician team decided to lower it to 125 mg/dL six months later. About a year after that, it was lowered to 110 mg/dL. (Although the treatment threshold was gradually lowered over a year and a half, the goal has always been to keep patients’ glucose levels at 130 mg/dL or less.)
While the hospital has aggressively lowered the threshold at which it gives patients insulin, Dr. Carr says the jury is still out on whether a trigger of 110 mg/dL will control glucose levels better than a threshold of 125 mg/dL.
Within two months of lowering the threshold to 125 mg/dL, the cardiac surgery service had no cases of mediastinitis. In the 24 months before, the unit typically saw 1.2 cases of mediastinitis for every 100 operations.
“For more than a year,” Dr. Carr says, “mediastinitis has not reappeared. When we looked at how long it had been since our last organ space infection, we suddenly began to realize that tight glucose control had had a direct impact on our infection rate.”
But because the mediastinitis rate has already reached zero, she notes, it will be difficult to gauge how lowering the treatment threshold will affect patient care. “All of our success came when the level was at 125,” Dr. Carr explains, “so whether having it 110 is better than 125 is still a question.”
Exporting the model to other settings
Beth Israel Deaconess Medical Center’s glycemic control protocol has been successful in its cardiac surgery service, but can it be applied to other settings like general ICUs or even med/surg floors?
While the hospital is evaluating the idea for both settings, Dr. Carr notes that in many ways, the cardiac surgery population was the group on whom the initiative is most likely to be successful.
“It’s a uniform group of patients who have all had major heart surgery,” she explains. “Their stress hormones are uniformly elevated, they all haven’t eaten since midnight the night before, and they remain NPO on IVs for a day or two.”
She also notes that medical ICUs interested in replicating the hospital’s success face several challenges.
One consideration is whether the trigger is solely the glucose level or whether the clinical condition plays a role as well. A second consideration is the mode of insulin administration (IV or subcutaneous), whether the patient is eating, and how well extremities are perfused. Outside the ICU, glucose control is often restricted to subcutaneous administration, because IV infusion is more resource intensive.
Dr. Carr offers the following advice to any facility considering implementing a glucose control program: Keep it simple and small. Start by getting staff comfortable with a limited protocol, measure outcomes, confirm effectiveness and then “and only then “expand.
Equally critical is changing the culture, which means re-educating staff about the importance of glucose control in all patients, not just those with diabetes. Metrics are important and help with buy-in, because they show how adherence to the protocol helps patients meet the glucose target.
Finally, she says, tracking the infection rate alongside the glucose control metric visually demonstrates the connection and shows staff the benefits of their labor.
Edward Doyle is Editor of Today’s Hospitalist.
