For glucoses, tight theory a tough practice

November 2006

Feature Story

Anne Paxton

What’s the difference between a protocol on paper and a protocol in practice? The cardiothoracic intensive care unit at Maine Medical Center, Portland, knows the answer from firsthand experience with implementing a program of tight glycemic control.

“We’ve probably been hammering away on tight glycemic control for about three years, but a couple of years ago we really started to take a closer look at compliance by the nursing staff to see whether they were actually following the protocol we thought was so wonderful,” says Cheryll St. Onge, RN, BSN, MS, the unit’s nursing director.

“The bottom line was, they weren’t.” More and more hospitals are adopting TGC—the active use of insulin to keep patients’ blood glucose at or near normal levels—as an effective strategy for improving patient outcomes. As they do, the data from the field are confirming what the literature has concluded: Tight glycemic control works. But it’s no cakewalk to implement it in a hospital.

At Maine Medical Center, in fact, the paper was the problem. After looking into the matter, St. Onge found that the original protocol, about five pages long and written in narrative form, was bogging down the process. “People just didn’t have or take the time to go through the fine details of what they should be doing,” she says.

Convincing the nursing staff that taking the time is worth it is important, St. Onge says. “Some of the pushback institutionally across the country is that it is very labor-intensive. When trying to convince people this is the right thing to do, it’s important to help them understand that TGC is not only a way to achieve improved glycometabolism, but it also improves immunocompetence and offers other direct beneficial effects of insulin for which the blood glucose is a barometer of effectiveness.”

In time, she says, outcome metrics such as deep sternal wound infections and length of stay will show improvement. After the TGC protocol was in place for a year at Maine Medical, the deep sternal wound infection rate decreased dramatically. And St. Onge noticed a distinct shift in attitude when infections did occur.

“People didn’t just say ‘Oh well, that’s one of the risks of surgery.’ They were very concerned and wanted to know what was different about this case and why there was an infection. Was the glucose controlled? What was the A1c when the patient came into the hospital? Did we start the insulin drip post-op when we were having a catheter put in or was it running in the OR, or did we slip up and they hit 200 a couple of times?”

A chronic issue, St. Onge says, is the nursing staff’s concern about hypoglycemia. “Nurses are taught to fear hypoglycemia, and always have on their minds not allowing the blood sugar to go too low. So we really tried to educate the staff that it is as important or as risky, if not more risky, to have the blood sugar too high, with all the risk factors it potentiates when it gets above a certain level.”

To stem these fears, the medical center took a gradual approach to adopting TGC. “We started this protocol with a high mark of 150 mg/dL, which to us today seems too high. But after people were comfortable with that level, then they were able to drop it to 140 and then down to 120. I recommend to anyone implementing this type of program to take that approach. If we had started too aggressively, there would have been more fear and anxiety about hypoglycemia.”

In fact, the number of patients that fall into the hypoglycemic category of glucose less than 50 is extremely low. “And since we have clear guidelines on how to treat hypoglycemia, we’ve had no negative outcomes related to hypoglycemia in over 1,000 patients.”

With leadership from endocrinology, the hospital is now considering a TGC implementation throughout the hospital, St. Onge says. “The endocrinologist involved in this is supportive of our protocol for use in cardiac surgery. “The protocol is aggressive, and cardiac surgery patients have a lot of stress from surgery and consequently high levels of insulin resistance. A protocol for a medical ICU might not be as aggressive.”

Can the same practices be applied to all ICU patients? A Maine state patient quality care initiative called eICU, which conducts live monitoring of ICU patients remotely, has shown there could be problems with such a move, St. Onge notes.

“The eICU program has been monitoring all of the glucose levels, so we have a very clear picture of the average glucose level for all our patients. One may have a sense that things are going well, but the data shows that a mixed ICU population will not do as well with TGC as cardiac surgery patients.”

Ultimately, it was a fairly low-tech innovation that helped boost Maine Medical Center’s compliance with the TGC protocol, she said. “We were fortunate that there were some graduate students from Dartmouth Medical School who needed a project to work on, so we asked if they could make our protocol more user-friendly.”

“They actually took the five-page narrative and condensed it into a one-page nomogram. It’s a very colorful grid with one axis being for glucose from the prior hour, the other axis for the current glucose level, and we use the intersection to determine what to do with that drip. So it really is a very easy, quick reference.”

As a tool, the nomogram “made the difference for the nurses regarding workflow and compliance,” St. Onge says, in addition to shifting the concern from hypoglycemia to hyperglycemia.

Compliance was the challenge, too, at Temple University Hospital in Philadelphia, says Ercele Reyes, MSN, nurse manager of Temple’s surgical trauma and intensive care unit. Until 2004, the unit had no protocol; it depended on physician orders and a sliding scale for giving insulin based on certain blood sugars.

“But that sliding scale was too high, because there were a lot of evidence-based studies showing if it is kept between 80 to 110, you can decrease ventilator-associated pneumonia and even sepsis, and help heal wounds faster,” Reyes says.

The unit joined the nationwide Volunteer Hospital Association, which has a program to prevent infections using evidence-based practices such as TGC. Starting in April 2004, when the unit adopted a version of the Yale Protocol, and set its initial target range at 85–139 mg/dL, it conducted the usual in-service training and initiated a mentor program to help initiate the protocol in each case where a patient’s blood glucose exceeded the standard. But after a month, compliance with the protocol had dropped from 91 percent to 74 percent.

“Some nurses were not following the protocol because they didn’t want to stick the patient every hour, and because of other challenges,” one of which was the limited number of LifeScan OneTouch Flexx systems. “Their scarcity was interrupting the workflow,” Reyes says.

“Initially we only had one or two, so I increased the number of machines to four. That really increased compliance.” In fact, after baseline ventilator bundle compliance rose from 70.2 to 91.9 percent, the Volunteer Hospital Association recognized Temple as one of the top five hospitals on improved compliance.

Even as the TGC target range was dropped to a fairly drastic 80 to 110 mg/dL, the unit was able to show striking improvement. Its ventilator-associated pneumonia, or VAP, rate, for example, declined to 6.2 per 1,000 ventilator days—a 20.8 percent drop since 2001. There was no significant change in mortality, Reyes notes, because the trauma unit usually cares for the sickest patients with multiple health problems.

“We found that our compliance was very good, our VAP declined, and we decreased the length of stay for patients in the ICU. We were able to save almost $207,000 within one year,” Reyes says.

She is now interfacing the hospital information system with glucose testing at the bedside. “The challenge I have is that when physicians look at my computer, the laboratory results don’t include the patients’ most recent blood sugar.” Reyes is conducting a study to show the cost-effectiveness of wireless point-of-care testing. “You can automatically see trending of blood sugar on the computer instead of assessing critical care flow sheets. It also improves patient care.”

The TGC program at Christiana Health Care System, Wilmington, Del., was launched three years ago, and in February 2006 it was elevated to a hospital wide program. But it hasn’t been easy, says D. Kristin Smith, RN, MSN, CDE, diabetes clinical nurse specialist who works in Christiana’s Department of Patient Care Services, Education.

“We now have clinical management guidelines for adult patients with diabetes that set blood sugar targets at less than 110 mg/dL in critical care, and 80–110 pre-meal, and less than 180 mg/dL post-meal for non-critical care areas.” These guidelines are based on the position statement of the American College of Endocrinologists.

“But getting hospitalized patients to those targets is a challenge for several reasons. They’re here a short time, they’re sick, which is making their blood sugar higher, and even if you adjust their insulin every day, you may not get to your target.”

Starting last January, before the hospital wide program was scheduled to launch, “We had a huge six-week initiative to educate the nurses on the revised Diabetes Medication Administration Record [DMAR] and the new Diabetes Management preprinted order set.” They set up numerous hour-long in-services using a PowerPoint presentation on the benefits of TGC and the concepts of basal and bolus insulin to achieve TGC as well as a worksheet using the new order set and DMAR.

She estimates about 1,200 nurses and unit clerks attended the in-services, and those who missed the sessions were trained later one-on-one on the units. “There was also a blast e-mail sent to all the physicians to enlighten them about the changes.”

Nevertheless, when the program was evaluated after six months, “we found our preprinted orders were being used only 50 percent of the time on admissions, and we had some issues with physicians using the forms incorrectly. The physicians in particular, it appears, didn’t get the message.

“They’re aware of the changes now, because they’ve been inundated by nurses trying to get them to use the order set. But they still opt not to use it,” Smith says. Though she is seeing long-acting basal insulin ordered more frequently, “lots of physicians still want to put patients on sliding scale only, either AC and HS or Q6 hours, without adding basal insulin.”

One of the problems of tracking progress is the sheer quantity of data, says Teresa Hawkins, BS, C(ASCP), Christiana point-of-care testing coordinator. “My current data only goes back to June 2005. We wanted to go back further, but we have so many test results we have to clear out of the system to keep functioning day to day that we couldn’t really do a retrospective study including all of last year.”

Since little clinical and financial data were collected before the protocol was implemented, the implementation team admits it will be difficult to determine the protocol’s effectiveness.

Anecdotally, however, Smith says, “We’re seeing more people under 200, and there used to be lots of patients over 200, and the general feeling is we are doing a better job of keeping patients’ glucose under control.”

The laboratory has software that comes with the LifeScan meters, she notes, “and we just this year started to take advantage of the fact we can get data.” For example, the laboratory started tracking the levels of hypoglycemia at different times of day.

“We are really promoting the use of basal insulin at bedtime instead of rapid acting insulin, which is more likely to cause nighttime hypoglycemia,” Smith says. “Physicians are in the habit of ordering the same scale for meals and bedtime, which causes a lot of hypoglycemia in the middle of the night.” As an alternative, their mild nighttime scale coverage starts at 250. “Again, we are promoting basal insulin at HS, not rapid acting,” she says.

The system’s Diabetes Management Group is in charge of implementing the program. “This is a large institution, and it takes a long time to get things done because there are so many committees you have to deal with. So I’m finding it’s just a very cumbersome experience.”

The physicians are the biggest hurdle to almost any change, she says. “It’s easy to educate nurses because we have an educational system in place and it works well. There really isn’t a good system to reach physicians.” Her goal for the next year is to make a lot of physician contact, either in groups or individually, to discuss the benefits of TGC, the shortcomings of the sliding-scale-only approach, and the tools they have to offer them.

The team is preparing to study some specific surgical diagnoses to see if length of stay and complications in those patients who have diabetes are increased or worsened based on blood sugar control. So it’s too early to say how well the hospital wide program is working. “I would say in a couple of years we might have a much better handle on whether this is making a difference,” Smith says.

Getting a better fix on the data is essential to making tight glycemic control work, says Charles C. Reed, BSN, RN, CNRN, patient care coordinator for the surgical trauma intensive care unit at University Hospital, San Antonio.

As part of a collaborative effort across the University of Texas affiliated hospital systems, the 25-bed unit started an unofficial word-of-mouth glycemic control program in 2004. At that point, “physicians knew there was the Greet Van den Bergh study (N Engl J Med. 2001; 345: 1359– 1367) that was driving the boat on tight glycemic control, and we started working on protocols.” By 2005 the hospital had implemented a physician order set and developed a protocol for the surgical trauma unit.

“We didn’t do any sort of staging,” Reed says. “We did an educational piece for the staff, we sent out articles supporting TGC, and basically the physicians signed off the order sets, the nursing staff entered it into the computer system, and off it would go.”

The move accelerated a startling decrease in mean glucose levels, which fell from 156 in 2003 to 140 in 2004, and to 131 in 2005. But after impressive progress, the percentage of patients kept at levels within the target range stopped improving.

“We were averaging 35 percent to 40 percent in the 80 to 110 range, and we couldn’t seem to break that barrier. All through January 2005 we were about 27 percent in the target range, then in July 2005 we were at 35 or 36 percent, and we were hitting close to 39 percent through August. So we had a nice increase, but then it sort of stalled, and we couldn’t get over the hump.”

Once they started using the Rals-Plus tight glycemic control module, made by Medical Automation Systems, they were able to turn that around. “Prior to that, our point-of-care office would send us information once a month, we’d process it, look at the data, present information to our staff, and our clinical management team would review it.”

With the implementation of the Rals-Plus module, they were able to look at the data right away. “The Roche Diagnostics Accu-Chek gets downloaded, and we can run a report from any computer in the unit, then see which of three surgical trauma modules are not showing results in the target range, then focus on working with those nurses and physicians.”

“Part of our clinical evaluation is demonstrating that evidence-based practice is being used, but we really had no way of showing how well it was actually doing. The Rals module helped take us to the next level.”

Now, when Reed comes in to work, he can run a report to see what’s been happening in the last eight hours and which side of the unit needs the most attention. As a result, the unit is averaging 50 percent of patients between 80 and 110. “We’d like to get it closer to 60 percent. But right now we’re also averaging almost 85 percent between 80 and 140, which is really excellent.”

Tight glycemic control has caused the number of Accu-Cheks to rise sharply. “With the Rals program sent every week to show them where they’re at, we’re averaging between 6,000 and 8,000 Accu-Cheks a month, which is pretty astronomical; just before that we were averaging 3,000 to 5,000. So, yes, there is the increased cost. If you have someone on insulin drip, and our average length of stay is five days, you figure it’s about $100 more per patient.”

But that cost is more than recouped, he believes. University Hospital plans to present the hard figures at a December conference of the Southern Surgical Society, but Reed says the unit’s outcomes correlate with the TGC studies. “Our mortality rate was consistent for four years until our TGC kicked in, and then we did have a decrease in it,” he says. “When you have decreased length of stay, lower antibiotic use, and lower morbidity and mortality, how can you say it’s not worth it?”

Like most other programs that have adopted TGC, University Hospital has had to contend with the fear of hypoglycemia, but there has been no increase in hypoglycemic episodes since the protocol was implemented; they remain at less than one percent. Reed says the best control for both hyper- and hypoglycemia is the insulin drip.

“Up to now, the practice has been to do an Accu-Chek when the patient comes in, put them on a sliding scale, do another Accu-Chek, and if the glucose is not controlled we’ll increase the sliding scale, and if that doesn’t work then we move to insulin drip.”

Now all patients who are admitted get an Accu-Chek right away, and anybody over 110 will be put on the drip. “You can imagine how labor-intensive that is,” Reed admits. “But right now we’re averaging a mean glucose in the one-teens, and our patients greater than 180 have been less than four percent.” That makes University Hospital—out of the nine hospitals participating in the TGC program—the best controlled of all, Reed reports.

Despite the successes they are reporting, the TGC programs at Texas, Maine, Temple, and Christiana are still the exception. “Everybody’s talking about tight glycemic control, but not many people have actually done it,” says Christiana’s Teresa Hawkins. “The biggest obstacles are getting physicians to accept it and getting nurses to do it, because really the bulk of the work falls on nurses who are doing POC glucoses. Once nurses are educated about it, they are very enthusiastic, but physicians still have to direct the process.”

For hospitals considering the optimal approach, Maine Medical Center’s St. Onge believes slow and steady adoption of tight glycemic control will work best in the long run. “I would recommend starting in incremental steps. Don’t expect to conquer the mountain on the first attempt, take it one step at time, and don’t give up—because you need to keep in focus that it’s the right thing to do for patients.”

At Maine Medical Center, the five-page narrative was condensed into a one-page nomogram. Download the complete nomogram.

Related Links

• Insulin Protocol (PDF, 419 K)