Where's the target?

The NICE-SUGAR investigators’ finding that intensive control actually increased mortality among their more than 6,000 ICU patients will lead to a dramatic and rapid change in practice for hospitalists.

It may not actually take a spoonful of sugar to make the medicine go down, but critical care experts are looking a little more positively at blood sugar in the wake of recent research.

The appropriately named NICE-SUGAR trial, published in the March 26 New England Journal of Medicine, upended conventional wisdom, the evidence base and expert guidance about intensive glucose control in critically ill patients.

Since 2001, the model for intensive care unit (ICU) glucose control had been a study by Greet Van den Berghe, MD, and colleagues, which found that reducing patients' glucose to normal levels (80 to 100 mg/dL) reduced morbidity and mortality.


In the wake of that study, professional organizations issued guidelines calling for tight control and hospitals around the world changed their protocols to work toward normalization of glucose in ICU patients.

Now, that process is moving in reverse, thanks to the NICE-SUGAR investigators' finding that intensive control actually increased mortality in ICU patients.

For many hospitalists and others involved in ICU care, the results likely will lead to changes in glucose management practices.

Time for a change

“Intensive glucose control has been widely recommended based predominately on the results of Van den Berghe's study,” said NICE-SUGAR author Simon Finfer, MD. “I think our results suggest there's no clear protective effect of glucose control for critically ill patients. Therefore, it would be appropriate for the guidelines to be revised as a matter of urgency. Clinicians working in intensive care should examine their practices.”

The American Association of Clinical Endocrinologists and American Diabetes Association have already followed his advice, revising their recommendations on glucose management for critically ill patients. Their consensus statement, issued May 8th, calls for insulin therapy to be initiated if blood sugar goes over 180 mg/dL and then maintained to keep blood sugar at a level between 140 and 180. Previously, the groups had advocated keeping ICU patients' glucose below 110.

Not surprisingly, the new consensus statement correlates pretty closely with the findings from NICE-SUGAR. Patients in the conventional arm of the trial were given insulin if their glucose exceeded 180 mg/dL, and administration was discontinued when levels dropped below 144 mg/dL. Patients in the intensive control group were treated to a blood glucose target of between 81 and 108 mg/dL.

The patients in the 140 to 180 mg/dL range had lower rates of long-term mortality (24.9% vs. 27.5%) and severe hypoglycemia (0.5% vs. 6.8%) than those in intensive control group. Still, experts in the field aren't ready to write the new advice in stone just yet.

“It may be reasonable to aim for a more moderate level, maybe 144 to 180, but I don't think we really have evidence showing that that's what the optimal target for glucose control should be,” said Renda Soylemez Wiener, MD. A meta-analysis conducted by Dr. Wiener raised questions about intensive control last summer when it was published in the Journal of the American Medical Association. Her study found no difference in mortality whether glucose was kept below 110 or 150.

It's indicative of the uncertainty on the subject that even the NICE-SUGAR study authors are cautious in interpreting the impact of their results. Author Dean R. Chittock, MD, advocates the 140 to 180 range based on the trial data, but he doesn't rule out other solutions.

“It's difficult to go outside the bounds of the study to say that other [blood sugar targets] wouldn't be just fine,” he said.

Is hypoglycemia to blame?

One reason for the experts' caution is that no one knows why the NICE-SUGAR trial found an increase in mortality with intensive control, an association that no prior research had encountered. “We only completed the trial at the end of November, so the detailed analysis of the very large database is ongoing. The mechanism is unclear at the moment,” said Dr. Finfer.

Hypoglycemia, the only other difference between the groups to be discovered so far, is a prime suspect. “The majority of the deaths and the excess of deaths were from cardiovascular causes. There is evidence from diabetic literature and animal studies that hypoglycemia can cause autonomic dysfunction and autonomic failure,” said Dr. Finfer. “We need to examine whether not only severe hypoglycemia but also lesser degrees of hypoglycemia have adverse effects on either autonomic function or the function of the cardiovascular system as a whole.”

But some researchers are reluctant to place the blame on hypoglycemia. “Whether the hypoglycemia is something that's contributing or whether just really sick patients are more likely to have hypoglycemia and more likely to die, I don't think we know,” said Dr. Wiener.

Adding to the mystery are data showing that mortality rates were not different while the patients were in the ICU. The two study arms had separated at a month out, but the difference in deaths only became significant at 90 days.

“I don't think that the effects of hypoglycemia are really powerful enough to have that effect,” said Mark D. Siegel, ACP Member, who co-wrote an editorial published with NICE-SUGAR. “Maybe in some populations having a relatively higher glucose during a critical illness could be one of the body's protective mechanisms. There may be other biological explanations for it as well. At this point, I don't think anybody really knows mechanistically what the reason for the higher mortality is.”

The experts do have a lot of explanations for how the NICE-SUGAR study could have come to such different results from the Van den Berghe trial, however. Although the new data may appear to contradict the old, much of the difference can explained by variations in study design, authors of both studies said.

Different trials, different results

“We basically did other things,” explained Dr. Van den Berghe. Her trial, published in the Nov. 8, 2001 NEJM, was conducted in a single Belgian center and compared intensive (80 to 110 mg/dL) control with what was, at the time, conventional treatment—infusing insulin only when glucose hit 215 mg/dL and then maintaining it between 180 and 210.

The NICE-SUGAR trial, on the other hand, was conducted among more than 6,000 patients in 42 ICUs in Australia, New Zealand and Canada. Patients in the intervention group were treated to about the same target as in the Van den Berghe study. But the conventional treatment group had a target of 180 mg/dL or less.

The studies have a number of other differences, but most of the variation stemmed from these two factors—the study setting and the level of intervention, the experts said. “With the control group we didn't really do anything, because it could be that the hyperglycemia is an adaptation to the illness, beneficial even,” said Dr. Van den Berghe.

Obviously, she found that blood sugar over 180 wasn't beneficial. But while most observers took the lesson from the study that intensive control was the solution, even Dr. Van den Berghe's own follow-up analysis (which included both her surgical trial and a subsequent medical ICU study) indicated some evidence for a moderate course.

“It shows the largest benefit really comes from avoiding excessive hyperglycemia to an intermediate level. Then a little bit more can be obtained in our hands if you go below 110,” she said.

The hands of the researchers in the two trials were also engaged in some pretty different activities, noted Dr. Finfer. “One of the striking things is the nutritional management of the patient. In both her surgical ICU trial and her medical ICU trial, well over 70% of the calories given to patients were given as intravenous glucose or parenteral nutrition. That's very different from the practice in most of Australia, New Zealand, Canada, and, I think, most western intensive care units in the world,” he said.

Much attention has also been given to the differences in the studies' patient populations. Dr. Van den Berghe's first study, which showed the greatest benefit from intensive control, was conducted among surgical ICU patients.

The NICE-SUGAR study found no difference in benefit for medical or surgical patients, but a meta-analysis that included data from NICE-SUGAR's and Van den Berghe's trials did find a mortality benefit from intensive control in the surgical ICU subgroup of patients.

“People continue to raise the issue of whether or not there could be advantages to tight glucose control in the surgical ICU. I think the subgroup analysis from the NICE-SUGAR suggests there probably isn't but some people might like to see a large randomized trial in the surgical ICU,” said Dr. Wiener.

Subgroup analyses of NICE-SUGAR are currently under way, but the researchers don't expect any definite answers to come out of the work. “There was no evidence beyond a doubt that any particular subgroup benefited,” said Dr. Chittock. “All of these sub-studies will be hypothesis-generating to lead to new investigations.”

Dr. Van den Berghe has her own hypothesis as to why surgical patients may have a different response to intensive control. “Patients in the surgical ICU normally have not been ill for a long time. When they get to the ICU, we know the hyperglycemia must have been there only hours.” It's possible that medical ICU patients have already either suffered the damage of high glucose or adapted to it, she speculated.

The new evidence has only expanded the range of theories about effects of glucose in the ICU. “It might be that there are different periods during the ICU stay where different degrees of glucose control might actually turn out to be useful,” suggested Dr. Siegel.

Other experts wondered whether the increased mortality in NICE-SUGAR is necessarily caused by intensive control, or whether it could be that sicker patients were less able to stick to the intervention protocol.

Issues of equipment, nursing

One concept about which the experts were certain, and in agreement, is the need for better methods of measuring intensive care patients' blood sugar. Dr. Van den Berghe believes some of the success of her study lay in the equipment used to test glucose levels.

At her hospital, the same model of glucometer was used on every patient every time, and the one included in her study had an extra advantage, she said. “One of the possible explanations is that the blood glucometer we used also has potassium each time on the readout. If you give insulin in bolus injections, you shift potassium into the cell and if you don't measure potassium, that could cause arrhythmia which causes death.”

In the Van den Berghe study, nurses checked potassium along with each blood sugar reading, and supplemented it when the result was low. Dr. Van den Berghe also attributed much of her success with intensive control to the skill of nursing staff.

“In our center, we recruited something between 68% and 95% of all patients who were admitted, which means that our nurses were very experienced in doing it because it was always on the agenda,” she said. Nurses were also authorized to make intuitive decisions about insulin dosing, as opposed to following a set algorithm.

The burden on nurses has typically been an argument against intensive control and a reason to celebrate the NICE-SUGAR results. “That's the good news, because many hospitals were challenged with the previous glycemic targets. It's really difficult to get to those low numbers without increasing the risk of hypoglycemia, as was shown in multiple studies,” said Etie S. Moghissi, MD, co-chair of the AACE/ADA task force that issued the consensus statement.

Dr. Finfer agreed. “Maintaining blood glucose in the very tight range of 80 to 110 in unstable critically ill patients is very difficult and it uses a lot of nursing time. It's estimated up to two hours of nursing time per patient per day. Even then, the majority of people have had difficulty keeping the blood sugar in that range for more than 50% of the time,” he said.

Not at her hospital, according to Dr. Van den Berghe. There are two patients per nurse, and intensive control is “dead easy,” she said. “Everybody's welcome to come and see how they do it.”

But, given that other ICUs seem to have trouble replicating her results, even Dr. Van den Berghe sees a need for flexibility in recommendations.

“If I would have to rewrite the guidelines, I would say if the setting were comparable to the NICE-SUGAR study, be careful and take 140. If you have a very experienced team, let's leave it where it is,” she said. “I would make that sort of balanced statement—depends on where you work, in what setting you work.”

A rush to the bandwagon

That sort of balanced statement may have been what was missing in the initial flurry of guidelines and protocols favoring intensive control, the experts said. “I do believe our trial has a message for those who write guidelines, in that people should be more critical when looking at evidence, particularly if it comes from one trial in one hospital,” said Dr. Finfer.

A number of factors encouraged physicians and organizations to jump on the intensive control bandwagon after the Van den Berghe trial came out, Dr. Chittock explained. “The critical care literature at that time was really full of negative trials. It was one of the first positive trials we'd had in a long time. It's a relatively cheap intervention to put in place. Perhaps because of that it was easy to adopt.”

The lesson of intensive glucose control is one that can be applied in other areas. “We have a number of studies now—the hydrocortisone treatment of sepsis is a good example. These are studies that seem to change practice almost prematurely before they can be validated. The critical care community really needs to take notice that these enthusiastic responses to unicenter studies need to be attenuated until they can be validated,” said Silvio E. Inzucchi, MD, who co-wrote the NICE-SUGAR editorial.

Experts also want the critical care community to temper their enthusiasm in response to the new evidence against intensive control. “Hospitals that have not jumped on the bandwagon may feel vindicated and won't do anything about hyperglycemia. Some hospitals may take their successful protocols and abandon them. That's one of our concerns,” said Dr. Inzucchi.

Such an overcorrection is just what the AACE and ADA task force members worried about in their discussions, Dr. Moghissi said. “That should not be the conclusion from this study. We know that hyperglycemia is associated with poor outcomes, therefore we need to pay attention to hyperglycemia.”

“It's the balance of finding a reasonable, achievable and safe target—those are the key words,” she added. Three goals that no one, even on this hotly debated topic, could dispute.