Published in the September 2004 issue of Today’s Hospitalist
In the not-so-distant past, treating sepsis could be a frustrating experience. Few treatments were shown by evidence to be effective in treating the devastating disease, and most of the treatment options were largely supportive.
What a difference a few years can make. A flurry of groundbreaking studies has shed light on some highly effective ways to treat sepsis, in some cases dramatically reducing mortality rates.
Although the news is good news for patients and physicians alike, much of the evidence runs counter to decades of conventional wisdom about how to treat sepsis patients. A study may show that lower tidal volumes drastically reduce mortality rates, but your instincts “and years of experience “tell you to do something entirely different.
In an article in the August 2004 issue of Anesthesia & Analgesia that highlights five major developments in sepsis treatment, Michael A. Gropper, MD, points out that these treatments can also produce very real complications, from bleeding to hypoglycemia. In severe cases, these problems can be nearly as severe as sepsis itself.
That’s why Dr. Gropper, the study’s lead author and director of the ICU at the University of California, San Francisco, says that many of the new treatments require physicians to embrace a new mindset. While your knee-jerk reaction may be to avoid using a treatment like activated protein C because of the risk of bleeding, you need to balance those concerns with the sometimes-stunning reductions in mortality these treatments offer.
Today’s Hospitalist talked to Dr. Gropper about developments in sepsis care that are most pertinent to hospitalists “and how they are forcing physicians to take a new view of treating critically ill patients.
Of the treatments discussed in the article, hospitalists are probably most familiar with the recommendations on glucose control. What are the pros and cons of tight control for glucose levels?
We grossly underestimated the importance of glucose control in the past. A few years ago, if you looked at average ICU patients, they would have glucose levels in the 180 to 220 range, and we thought that was normal, that it was a stress response. Some physicians actually thought it was good, that it provided more energy for the cells.
Before the landmark study by van den Berghe, a few studies had examined the role of glucose control in cardiac surgery patients, but most of the subjects were diabetics. The van den Berghe study was so important because it looked at not only diabetics, but any inpatient who was hyperglycemic.
Everybody was surprised about how strict you have to be with glucose control, because 80 to 110 mg/dL is very tight control. This represented a huge change in mindset.
Everyone is afraid of hypoglycemia as a side effect of insulin. If you look at the harm that can come from a relatively low glucose level and compare it to the harm that can be produced by a higher glucose, however, it’s much less of an issue. But we had to change the mindset of physicians to get them to view a glucose of 180 or 200 mg/dL as harmful.
The study was surprising because it represented such a big reduction in mortality, more than 30 percent. I don’t think we were surprised that it was better, I think we were surprised at just how much better it was.
Has there been any new research on the role of glucose control?
Another study since the van den Berghe article looked at things a little differently. From a practical standpoint, it looks like keeping glucose below 140 to 150 mg/dL is nearly as good as keeping it at 110 mg/dL.
The latter study also shed some light on another issue. When the van den Berghe study came out, everyone wanted to know what increased survival. Was it the act of giving insulin to patients, or was it controlling glucose levels? The second study showed that the benefit clearly came from controlling the glucose, because patients who needed more insulin to control their glucose still had higher mortality rates.
In the article, you mention that UCSF implemented a glucose protocol, and that it was difficult. What made the protocol so challenging?
It required a lot of effort on the part of nursing staff. The protocol we used was so restrictive that a nurse with two patients following it could spend up to two hours a day doing nothing but adjusting insulin, testing glucose, and performing other related tasks. Every time you adjust the insulin infusion, you have to recheck it.
You point to research showing that low tidal volumes help improve mortality in patients suffering from acute respiratory distress syndrome. How can physicians make sense of this finding, which flies in the face of their experience?
When patients have ARDS, they are obviously hypoxemic. If you increase the tidal volume on these patients, their oxygenization usually improves, and that’s why we typically used higher tidal volumes in those patients. You look at their CT scans, and they have atelectasis, so it makes sense to inflate the lung.
In the ARDSnet trial, patients were randomized to either 6 mL/kg or 12 mL/kg. While patients with the higher tidal volumes had significantly better oxygenization in the first two or three days, it was a classic false outcome. While these patients had better oxygenization, they also had a greater risk of death.
Again, this represents a change in mindset from trying to expand the lungs from the state where they’ve collapsed to using smaller tidal volumes. Another Brazilian trial by Amato had shown that lower tidal volumes made a big difference, but we were skeptical based on a number of factors. But that trial actually showed a 25 percent reduction in mortality, and the strategy doesn’t cost anything. You’re just adjusting the dial on the ventilator.
Are other factors besides tidal volume at work?
The strategy does involve more than just reducing tidal volume. For example, we used protocols to wean patients off mechanical ventilators, which was a component of the ARDSnet trial. Because that was a component in the control group and the intervention group, though, it’s pretty clear that the low tidal volume made the difference.
Along those lines, a recent study in the New England Journal of Medicine looked at the effect of using higher or lower levels of positive end-expiratory pressure, and it showed no difference. That seems to provide further evidence that lower tidal volumes are the key.
How have physicians embraced these results in practice?
I still occasionally see patients transferred from other facilities who are coming in with very high tidal volumes. When you actually put low tidal volume ventilation in practice, patients can appear uncomfortable. They might require a little more sedation, although there are no data to support that, so that makes some people a little uncomfortable.
It gets back to the fact that physicians are nervous because these patients are very sick, and they turn up the tidal volume and the oxygen improves. It’s hard to understand how that’s a bad thing, but it is. If you look at the patients on the ARDS protocol, they don’t usually die from lung failure. They die from the failure of other organs, especially renal failure.
The bottom line is that you’re much better off having respiratory failure than renal failure. The mortality rate is higher with renal failure.
The article talks about activated protein C, which has shown to decrease mortality risk, but raises concerns about bleeding. What do hospitalists need to know about this treatment?
People are very worried about bleeding, but you have to choose your patients. If the patient had neurosurgery yesterday, that person is obviously an extremely high risk for bleeding. But if the patient has a soft tissue infection or a broken hip, I’m willing to take that risk, because you can control that bleeding and keep an eye on it.
If you look at the mortality risk of a patient who is in shock and think about relative risk benefit, bleeding usually represents a lower risk than dying from sepsis.
The cost of this therapy is another factor, but a long length of stay in an ICU is also very expensive.
Physicians were also skeptical because the drug was tested in only a single trial. It’s rare that the FDA approves drugs based on a single trial, but it was a large, very well done trial, and the treatment effect was large. The treatment was so effective that the trial was halted midstream.
Why is this article important now? The research is a few years old.
With sepsis patients, it’s important that we use a multipronged approach. I’m trying to get people to think of this whole group of therapies when it comes to treating sepsis patients. These treatments are all new and have appeared just in the last few years.
What’s encouraging is that if you look at risk-adjusted mortality in the control groups of these studies, it’s already going down. I think that’s because of the cumulative effect of all these treatments.