Published in the April 2007 issue of Today’s Hospitalist
Patrick Williams, MD, the medical director of Cogent Healthcare’s hospitalist service at Jewish Hospital in Louisville, Ky., is proud that his hospital is now rolling out a new glycemic control protocol in the ICU. The protocol, which is the hospital’s first attempt to expand control efforts beyond cardiac surgery patients, is expected to reduce the ICU’s rates of infections and complications.
“Anybody in the ICU with a blood glucose over 130 automatically gets on the protocol,” Dr. Williams explains. “We’ll try to keep them between 80 and 110.”
But when it comes to plans for tighter control beyond the ICU, his tone shifts to one of frustration. That’s because the much broader goal of better glycemic control for everyone in the hospital, not just cardiac surgery and ICU patients, is still very far off.
Jewish is far from alone in not moving to adopt a glycemic control protocol beyond sliding scale insulin on the wards. While it’s been more than five years since a landmark study found that tightly controlling blood sugars in a surgical ICU almost halved mortality rates, U.S. hospitals have moved slowly to change control efforts in the ICU. Even fewer have tackled that issue on the floors.
Now, however, hospitals are turning to their hospitalists to make glycemic control happen, and hospitalists nationwide are trying to bring standardization to the diverse population of non-critical inpatients with hyperglycemia. Physicians who have crafted glycemic control protocols say such initiatives are not only possible, but have big payoffs in terms of tighter control.
But, they caution, hospitalists should give themselves plenty of time to design protocols and order sets, and plenty of patience to weather what is likely to be a lot of trial and error.
A question of evidence
Several factors are behind hospitals’ reluctance to tackle glucose control beyond the ICU. While there is strong evidence supporting control in critically ill patients, data paint a less clear picture for medical patients. Gregory Maynard, MD, one of the hospitalist leaders heading up national standardization efforts, concedes that to date, no randomized evidence proves that hitting a tighter target for non-critical medicine patients leads to fewer complications or shortened length of stay. But despite a lack of hard data, he says it’s not a big leap to conclude that tight glucose control makes sense.
Part of the problem is what Dr. Maynard calls “misinterpretation” of a study published in the Feb. 2, 2006, New England Journal of Medicine. (That trial was led by Greet Van den Berghe, MD, who also headed up the landmark study on surgical ICU patients.) The 2006 study found that tighter glycemic control was associated with higher mortality rates among patients who stayed fewer than three days in a medical ICU.
While the study’s results may seem less than encouraging, Dr. Maynard says there are reasons why. “The Van den Berghe II protocol had a lot more hypoglycemia than other centers have with the same target,” points out Dr. Maynard, who is chief of the hospital medicine division at University of California, San Diego (UCSD) Medical Center. “I think it’s a matter of tweaking their protocol.”
Basal/bolus vs. sliding scale
As for tighter control for non-critical patients, “there is an overwhelming amount of pathophysiologic and cohort-type studies, as well as epidemiological evidence, that hyperglycemia is associated with bad outcomes in a variety of inpatient settings, not just the ICU,” says Dr. Maynard, who will discuss improving glycemic control at the Fall 2007 Hospitalist CME meeting in September. (Go online for more information.)
His own preliminary results from implementing subcutaneous (sub-Q) basal/bolus insulin order sets on UCSD’s medicine wards show that tighter glucose reduced rates of both hypoglycemia and extreme hypoglycemia by about 30%.
Guillermo Umpierrez, MD, a diabetologist at Atlanta’s Emory University, agrees that the body of evidence in support of sub-Q basal/bolus regimens is growing. Yet he notes that a study in the May/June 2006 Journal of Hospital Medicine found that 65% of inpatients with diabetes or hyperglycemia at the Mayo Clinic in Scottsdale, Ariz., received sliding scale insulin.
“Only a minority of patients with blood glucose greater than 200 had their therapy changed,” Dr. Umpierrez says. “Sliding scale just perpetuates the state of hyperglycemia.”
While debate over the evidence continues, most hospitalists have very practical concerns about quality improvement efforts to tightly control blood sugars. They fear that trying to standardize the insulin requirements of everyone on the wards, who all have different levels of insulin resistance and nutritional needs, will be a logistical nightmare.
“I’m looking for a doable protocol that is simple and won’t be all-consuming for staff,” says Jewish Hospital’s Dr. Williams. “But there are other considerations, such as nursing workload, to being proactive.”
Hospitalists who have implemented or are designing protocols admit that the process is neither easy nor foolproof. Lakshmi Halasyamani, MD, a hospitalist at St. Joseph Mercy Medical Center in Ann Arbor, Mich., is working with a glycemic control team to implement a basal insulin protocol for non-critical patients. “There’s no one size fits all,” says Dr. Halasyamani, “and no off-the-shelf way to do this.”
Right now, the team is incorporating input on protocols from several groups of clinicians. “We’re reviewing the educational needs of those users,” she adds, “to make sure the protocol won’t be used in a vacuum, but as part of a patient’s comprehensive care.”
Make it a team effort
That underscores the importance of starting any glucose control initiative by forming a multidisciplinary team, which should include not only physicians, nurses and pharmacists, but nutritionists, tray delivery representatives and other staff who play a central role in patient care. Without that input, physicians say, you won’t have a good sense of staff workflow, which will make implementing a protocol much harder.
Cheryl O’Malley, MD, a hospitalist who championed a sub-Q insulin protocol in 2003 at Banner Good Samaritan Medical Center in Phoenix, Ariz., says that getting an overview of nursing workflow was critical to the initiative’s success.
“Before we put in the protocol, nurses from the night shift would check morning blood sugars around 5:30 or 6 a.m.,” she says. “But the trays wouldn’t arrive until 7, so that had to be changed.” When the protocol was put in place, checking morning sugars became the job of the day nurses.
The same goes for setting glucose targets. While studies and guidelines may say that a certain range is ideal, it’s critical to pick a target that’s flexible enough to be adapted to the working conditions at your hospital.
While the American Diabetes Association advises a random glucose of less than 180 and a preprandial glucose between 90 and 130, Dr. Maynard says physicians should worry more about how their peers will react than the exact science. “What is important is to have one target for the entire unit and to take the same standardized approach for all the patients in that unit,” he says.
A common “and effective “strategy, he says, is to set a target on the high side, then work your way down as your institution and nursing staff feel more comfortable.
“The main barrier to better control is the fear of hypoglycemia,” Dr. Maynard points out. “You have to reassure yourself and your staff that you can reach the higher target safely.”
Simplify standardization efforts
One of the keys to a successful control strategy is making it easy for clinicians to comply. That means streamlining your protocol to eliminate as much variability as possible.
To start, aim for a protocol that works for 80% of your patients. “Don’t confuse the issue by listing all the reasons why a physician may want to vary that suggestion for a minority of patients,” says Dr. Maynard.
The sub-Q order set at UCSD, for example, lists only insulin-dosing parameters that work for patients with either type 1 or type 2 diabetes, rather than separate parameters.
And to promote standardization, the basal insulin order set at Banner Good Samaritan is fashioned as a series of checkboxes. Several of the boxes come checked, including one to discontinue insulin sliding scale if previously ordered. The checked boxes become standing orders, unless physicians manually cross them out.
“Make orders as automatic as possible,” says Dr. O’Malley. “Instead of having physicians opt in for orders, you want to have them opt out.”
Simplification also applies to another thorny area in glycemic control: what dosing strategies to use for patients with different nutritional status. When designing a protocol, you need to select one out of several appropriate options for each patient population “NPO, TPN, on enteral feeds or on solid food “and make that your default choice.
According to Dr. Maynard, a good place to start when considering dosing options is the technical review published by the American Diabetes Association in the February 2004 issue of Diabetes Care.
“We selected dosing options that we thought we could standardize most reliably,” he explains. For instance, the UCSD basal/bolus order set calls for using regular instead of rapid-acting analogue insulin for a correction scale in NPO patients, and as the nutritional and correctional insulin for patients with tube feeds. “Nurses can monitor the regular insulin less frequently,” he points out.
He also recommends using a peakless, long-acting basal insulin, such as glargine or detemir.
“With these insulins, the nurses can give the dose of basal insulin whether the patient is NPO or not,” Dr. Maynard says. If NPH is used, on the other hand, “you usually have to decrease the dose by one-third to one-half when the patient goes NPO.”
And to get more buy-in from clinicians, Dr. Maynard urges physicians spearheading a protocol to prune their inpatient formularies, doing away with insulins that are not their preferred choice. “A lot of people are just eliminating NPH or 70/30-type insulins because those aren’t the preferred choice for any setting,” he says. “That’s another way to make orders more uniform.”
Start small, expect changes
While the goal may be tighter control for everyone, experts suggest starting small. Consider rolling out a protocol on one ward, for example, or among only a few physicians.
UCSD, for example, piloted its sub-Q insulin management algorithm on three hospitalist services. That gave residents and attendings a chance to iron out kinks before incorporating the algorithm into the order sets for its computerized physician order entry (CPOE) system.
Another reason to start small: You’ll need to tweak the protocol. At Banner Good Samaritan, Dr. O’Malley explains, the original basal insulin protocol has been revised several times to promote wider adoption.
One group that regularly opted out of using it was the endocrinologists; the original order set didn’t include orders they typically write. To motivate them to adopt the order set, the latest iteration includes two new options: one to use carbohydrate counting for patients with type 1 diabetes, and another to determine a patient’s supplemental insulin dose based on their correction factor.
You also need to expect outliers and to spend time educating nurses and physicians. That’s a challenge, Dr. O’Malley admits, given the turnover among hospitalists and nurses. Because Banner Good Samaritan does not yet have CPOE, she says that the glucose control committee relies on chart reviews to see who is using the protocol and if physicians are following the recommended doses.
Ultimately, technology should help glycemic control efforts. As more hospitals adopt CPOE systems, embedding basal/bolus order sets should get easier.
At Jewish Hospital, Dr. Williams currently sits on the committee that is picking the hospital’s next generation of information management system, one that will have CPOE capacity. The ability to more closely tailor insulin dosing to individual patients holds out the possibility, he says, of doing away with sliding scale insulin.
At Banner Good Samaritan, where a protocol has been in place for four years, Dr. O’Malley says she’s also waiting for the advent of CPOE. That’s because the biggest problem with physicians adopting the order set is physical availability.
“Even with all our checkboxes, doctors really have to opt into the order set because they have to get the form and fill it out,” she says. “Sometimes they’ll just write their own orders on the regular order sheet, which is missing all the standing orders “for NPO and other types of patients “that are built into the order set.”
Dr. Maynard is convinced that automation will provide the eventual solution, with closed-loop algorithms offering automated glucose monitoring and dose assessment. There are several automated tools already on the market “including Glucomander Plus “he says, “but a lot of people are creating their own,” finding that off-the-shelf ones are either too expensive or don’t work for their institutions.
“It’s a way to automate the calculations and spit out for the nurse what the adjustment should be,” he says. Another advantage? The tools are portable and can be easily integrated into nurse workstations.
For now, hospitalists who are either using or planning on using protocols are figuring out how to track data to help them get other physicians on board.
At St. Joseph Mercy, says Dr. Halasyamani, “we’re developing a measurement system to ensure that implemented interventions can be measured and modified.”
Phyllis Maguire is Executive Editor of Today’s Hospitalist.
“Wedded to the protocol”
Ask Woodruff J. English, MD, about the insulin protocol he’s spent the last three years piloting in a medicine nursing unit, and he offers this caveat: “I honestly have no idea if it’s going to work.”
The protocol is the famous Portland Protocol, pioneered at Providence St. Vincent Medical Center in Portland, Ore., where Dr. English spent eight years as lead hospitalist before turning his full-time efforts there earlier this year to quality improvement and palliative care.
But the Portland Protocol, which relies on insulin infusion rather than subcutaneous insulin, was designed for cardiac surgery patients who have a “clearly choreographed clinical pathway in terms of nutritional intake and metabolic stress,” Dr. English points out. The medical patient, on the other hand, does not.
“When you try to run a heterogeneous population through a standardized protocol, the protocol has much less precision,” he says. “It’s still unclear whether we will be able to use the same protocol for all medical patients.”
The protocol significantly ups the ante for nurses who, if a patient is unstable, may have to check blood sugars every 30 minutes. That’s a huge issue for a ward nurse who’s responsible for up to five patients, Dr. English says, rather than just one patient in an ICU.
So if the protocol is so labor-intensive, why not switch to one with subcutaneous insulin? Dr. English admits that he and the pilot team have considered switching. But he points to the fact that they’d want to have insulin infusion as another tool (besides subcutaneous insulin) to manage glucose in post-op patients and unstable medical patients outside of the very full critical care units. They continue to test the protocol in a general ward, he says, to standardize nursing practice.
“As a hospitalist, I look at the hospital as having to function optimally for all patients,” Dr. English says. “If we are using multiple glucose control tools, the nurses get confused and make mistakes.”
The pilot team at Providence St. Vincent isn’t entirely convinced that medical patients will suddenly have better morbidity and mortality because their sugars are lower, he points out. “But we may improve morbidity and mortality as a result of making fewer mistakes, because we have a clearer process. And we are not having any significant hypoglycemia issues with the protocol so far.”
In the meantime, the pilot gives clinicians a wide target range to practice. While the team eventually hopes to get targets in the 90 to 140 range, “we’re tolerating sugars in the upper 100s,” says Dr. English, “just to get our hands on how the protocol is going to function.”
He stresses that he wouldn’t recommend an infusion protocol on the wards to anyone. “But we are wedded to this particular protocol,” he says, “and that makes our circumstance unique.”
Sign up now for glycemic control newsletter
In May 2007, Today’s Hospitalist will launch a new Inpatient Glycemic Control newsletter, with topics ranging from dosing tips to what works in implementing and tracking glucose control programs.
Go to the Today’s Hospitalist Web site, click on “E-mail alerts” and submit your e-mail address. Look for your e-mail copy of Inpatient Glycemic Control in May.
New glycemic control workbook
Gregory Maynard, MD, headed up the Society of Hospital Medicine’s glycemic control task force, which posted its “Improving Glycemic Control” workbook online earlier this year.
The workbook includes a detailed blueprint for how to target tighter glucose control in different settings in the hospital and craft protocols and order sets. Several protocol examples are included.