Intensive glucose control in critically ill patients -- keeping glucose below 120 with a continuous insulin drip -- was all the rage for a few years in the early 2000s after it was shown to improve survival in surgery patients, and then seemed to do the same in non-surgical, critically ill MICU patients who were hospitalized for more than a few days. However, as astute observers noted, the fact that the overall result of the trial showed "no benefit" of intensive glucose therapy suggested that someone in the trial was getting hurt. Subsequent studies did not confirm the benefits of intensive glucose control.
That didn't slow the rising star of Greet van den Berghe, lead author of the pro-intensive-glucose-control articles, who could be seen at conferences around the world arguing passionately for the benefits of intensive sugar control in most patients, medical and surgical, with some extremely persuasive PowerPoint slides.
In the years after publication of van den Berghe's seminal papers, hospitals (including those where I worked) rushed to create protocols, buy pumps, train nurses, and print posters to meet this new metric, and (it was thought) to save patients' lives.
Just a few short years later, though, the NICE-SUGAR trial brought the crash after the sugar high: among 6,100 ICU patients, those randomized to intensive glucose control had an increased risk of death, compared to keeping sugars under 180 mg/dL. However, the authors couldn't establish a solid connection between hypoglycemia and increased mortality.
Now, they have. In the September 20, 2012 New England Journal of Medicine, the NICE-SUGAR investigators provide an epitaph to the short life of intensive glucose control for critically ill patients. In this post-hoc analysis of the NICE-SUGAR data, they show that intensive glucose control caused hypoglycemia, and that hypoglycemia was strongly associated with mortality:
- Patients with severe hypoglycemia (glucose < 40 mg/dL) were twice as likely to die compared to patients without hypoglycemia (adjusted hazard ratio 2.1). Those in the intensive control group were ten times as likely to have severe hypoglycemia (n~200 vs ~20).
- Those with moderate hypoglycemia (glucose 40-70) had a hazard ratio for death of 1.4, and 82% of these patients were in the intensive-control group.
- Those patients with moderate hypoglycemia had a 5% absolute greater risk of death (28% vs 23%) than those with no hypoglycemia.
- There was a dose-response relationship between hypoglycemia and mortality -- those with more than one day of low blood sugars were more likely to die than those with one day of hypoglycemia.
Hypoglycemia was also associated with longer ICU stays, and can be confounded by severity of illness, and so, yadda yadda, correlation does not mean causation.
Clinical Takeaway: Hypoglycemia seems to kill critically ill people, and tight glucose control under ideal circumstances (i.e., in clinical trials) is a likely iatrogenic cause of death. Keep blood glucose below 180 in critically ill patients, with an insulin drip if necessary, to best avoid complications of both hyperglycemia and hypoglycemia. When evaluating any potentially harmful therapy as the new standard of care, beware of zealots with cool accents and stylish glasses.
NICE-SUGAR investigators. Hypoglycemia and Risk of Death in Critically Ill Patients. NEJM 2012;367:1108-1118.
Discuss this article in the SDN Critical Care Forum.