The strong association between hyperglycemia and increased risk of mortality among critically ill patients is well known. But an increasing body of research is showing that high blood glucose levels are most problematic for patients who had no or well-controlled diabetes prior to admission.
Studies have found, for example, that glycemic variability can be a clinical predictor of poor outcome and that preadmission glucose control can have a significant impact on how diabetic patients fare in the ICU.
James Krinsley, MD, director of critical care at Stamford Hospital in Connecticut and clinical professor of medicine at Columbia University College of Physicians and Surgeons in New York, said that research conducted over the past few years has underscored the need for “personalized” treatment protocols that stratify hyperglycemic patients by their diabetic status. Treatment of inpatient hyperglycemia should not be based on a “one-size-fits-all” solution, he added.
“We may be heading toward an era where a patient comes in and there might be multiple targets that the ICU will handle. Is that feasible using current technology? Probably not, but the era of continuous or nearly continuous glucose monitoring is approaching,” he said.
Stress hyperglycemia ratio
A new formula for quantifying the association between hyperglycemia and bad outcomes for patients without a preexisting diagnosis of diabetes was introduced in the December 2015 Journal of Clinical Endocrinology and Metabolism.
The investigators developed a method to quantify relative, or stress, hyperglycemia among inpatients, using HbA1c and admission glucose level. The stress hyperglycemia ratio was defined as admission blood glucose level divided by estimated average glucose level derived from HbA1c.
The ratio was found to be independently associated with critical illness (while glucose level itself was not), making it “a better biomarker of critical illness than absolute hyperglycemia in patients across the glycemic spectrum,” the authors wrote. Helpfully, the ratio identifies patients at increased risk of critical illness even at glucose levels below 180 mg/dL.
The research “reinforces that you should not discount hyperglycemia in the patient without a diagnosis of diabetes,” said Gregory Maynard, MD, MSc, ACP Member, clinical professor and chief quality officer at the University of California Davis Medical Center in Sacramento.
Whether they use the specific formula or not, it's important for physicians to remember that hyperglycemia in a patient with a normal HbA1c or slightly elevated HbA1c is a stronger marker of stress and potential complications than it is in a patient with diabetes, he said. “You should take that very seriously, and you should take that at least as seriously as you do in a patient with diabetes because the correlation between poor outcomes and high glucose is stronger in a patient without diabetes. Don't ignore it.”
Even among patients with diabetes, relative hyperglycemia can be important. “Preadmission glucose control can have a very significant impact on how diabetics fare in the ICU,” Dr. Krinsley said, adding that recent literature suggests that for diabetics with low HbA1c levels (below 7%) on admission, the relationship between hyperglycemia during ICU stay and mortality is similar to that of patients without diabetes. If they develop stress hyperglycemia in the hospital, they are at greater risk of mortality, he reported.
Change in glycemic levels during a hospital stay is another important predictor of a poor outcome. A patient's glucose level might seem acceptable at an average of 140 mg/dL, but it could actually be fluctuating between 200 mg/dL and 110 mg/dL over the course of the day. “There is accumulating evidence that people who have that variability do worse,” Dr. Maynard said.
An international multicenter investigation with nearly 45,000 patients, conducted by Dr. Krinsley and colleagues and published in Critical Care in March 2013, found that increased glycemic variability was independently associated with increased risk of mortality among critically ill patients without diabetes but not among patients with diabetes. A large single-center observational cohort study, also published in Critical Care in March 2013, found similar results regarding increasing glycemic variability.
This research shows the necessity of frequent blood glucose measurement among critically ill patients. “We miss a lot of the glycemic excursions when we only do periodic testing,” Dr. Maynard said.
A minimum measurement interval of every hour is likely to be necessary for optimal glucose control, said Dr. Krinsley, noting that the time in the targeted blood glucose range may become an important measure. He wrote in the April 2015 issue of Critical Care that for patients without diabetes, staying in the targeted blood glucose range of 70 mg/dL to 140 mg/dL at least 80% of the time was independently associated with survival.
This evidence is increasing interest in devices that perform continuous or nearly continuous glucose monitoring, the experts noted.
Although existing guidelines on inpatient glucose control differ on a number of factors (see sidebar on page 17), they have yet to address the issues of relative and variable hyperglycemia.
Guidelines need to be rewritten to “recognize the fact that all patients are not the same. Why should we assume that all patients will react the same to derangements of glucose in the ICU?” said Dr. Krinsley.
Susan S. Braithwaite, MD, FACP, clinical professor of medicine at the University of Illinois College of Medicine at Chicago, anticipates that there will be “modification of guidelines in the future, such that stress-induced exacerbation will be recognized among persons having diabetes as well as persons having no diabetes, and perhaps will have greater import than diabetes itself with respect to prognosis.”
The International Symposium on Intensive Care and Emergency Medicine, which is meeting in Brussels in March, is expected to hold discussions on appropriate targets for different patient groups, Dr. Krinsley said.
Action in the interim
Hospitalists don't have to wait for guidelines to respond to the risks of stress hyperglycemia and glycemic variability. For starters, they can routinely look at measures of premorbid glycemic control when patients are admitted, Dr. Krinsley noted.
At Stamford Hospital's ICU, the targets for blood glucose depend on the patient's diabetic status. For nondiabetics and for diabetics with an HbA1c level below 7%, the target range is 80 mg/dL to 140 mg/dL. For diabetics with an HbA1c level above 7%, the target range is 110 mg/dL to 160 mg/dL, which may eventually be adjusted higher pending further analysis and research, Dr. Krinsley said.
Patients with poorly controlled diabetes on admission to the hospital seem to fare better with a higher glucose target, although how high it should be is unknown, he noted.
Dr. Maynard said more research is needed to determine whether nondiabetic patients with stress hyperglycemia should be treated to a lower glucose target than patients with diabetes. He currently recommends a target of 100 mg/dL to 180 mg/dL or 140 mg/dL to 180 mg/dL. Values greater than 180 mg/dL should be avoided, he said.
“Basically we follow the Endocrine Society guidelines,” Dr. Maynard said. “We are trying to get people to be uncomfortable and to do something when glucose is greater than 180 and to seek that range between 100 and 180 mg/dL. In some patients who are easily controlled or who may be at higher risk from complications, such as surgical patients, a lower goal may be appropriate, but I am not ready to put people in that category just because they have a normal A1c.”
Concerns about stress hyperglycemia should be balanced against the known risks of hypoglycemia. “You should also not have a lower-end target lower than 140 unless you can demonstrate that you can do that safely without excessive hypoglycemia. No matter what your target is, if you can't reach it safely, you shouldn't go there,” said Dr. Maynard.
It's still uncertain how much reducing glycemic variability and relative hyperglycemia will lead to better outcomes, as opposed to these markers simply identifying patients at higher risk of poor outcome, noted Dr. Maynard. That question, and others, require more research, but in the meantime, “the minimal approach should be to strive to meet existing guidelines,” he concluded.