What the New England Journal can give it can take back.
In 2001 the Journal published what can be called the first Van den Berghe study in which it was demonstrated that in critically ill surgical patients intensive insulin therapy had a beneficial effect. It was a large trial and the results seemed impressive and from the data the authors presented they recommended that blood glucose be targeted to be less than 110 in critically ill patient in the surgical intensive care unit. We love to extrapolate- maybe it would be just as good for medical ICU patients. So another trial was launched.
The second Van den Berghe paper appeared in 2006 and investigated blood glucose control in medical ICU patients with different and somewhat puzzling results. Different in that the over all morality benefit to the tight control group did not occur and puzzling in that a subset ( those with ICU stays greater than 3 days) showed benefit while the complementary subgroup ( those
with stays less than 3 days) showed detriment. This lead some observers to recommend a middle ground namely treat to a a blood glucose of less than 150 for first few days and then go for more intense control ( 110) after that.
Maybe this is reasonable but I am always suspicious of subgroup analysis as some strange findings can appear and the more subgroups you look at the greater the likelihood of a false positive result misleading you. ( My favorite subgroup analysis story can be found here in the discussion of how the benefits found in the landmark thrombolysis trial (ISIS-2) trial did not apply to those patients unlucky enough to have been born under the sign of Gemini or Libra). We are not told about the astrological signs of the participants in the Van den Berghe study.
It has been pointed out that the benefits effects of intensive insulin therapy on postoperative survival was mainly seen in post-op surgical patients who were given rather large glucose loads on admission to the surgical ICU and it makes sense they needed more insulin to mitigate the potential harm of overzealous sugar loading.
In still another NEJM article on intensive insulin therapy in the critically ill published in the Jan 10,2007 issue we are told that the use of "intensive insulin therapy place critically ill patients with sepsis at increased risk for serious adverse effects related to hypoglycemia " and that there was no difference in mortality at 28 days. The authors of this article conclude that when one takes into account both of the Van den Berghe articles and their publication that
" intensive insulin therapy has no measurable,consistent benefit in critically ill patients in a medical ICU, regardless of whether the patients have severe sepsis..."
In what could be characterized as a effectiveness study using historical controls ( i.e. not a randomized trial) treating to a blood glucose of less than 140 it was demonstrated that that regimen seemed doable and was associated to a number of measurable benefits. Comparing results then and now is tricky business as many variables change over time and how well they can be controlled for is really never known.
DR RW gives a nice review of the glucose issue here.
The 2004 Surviving sepsis Guidelines suggest a blood glucose target of less than 150
using an infusion of glucose and insulin. The 140 to 150 range seems a reasonable and reasonably safe approach-pushing to the arbitrary 110 may well cause hypoglycemia episodes and do more harm than good.