Randomized clinical trials work best for discrete outcomes in well defined groups of patients. In studies involving more complicated, heterogeneous groups of patients with multiple variables, results tend to be less clear cut as biases, confounding factors, the play of chance and the problems of squeezing subset analysis to make some sense of it all conspire to often make the conclusions less than satisfying or useful for real life clinical medicine.
A case in point is the recent prematurely stopped trial of tight glycemic control in ICU patients reported here. No advantage in terms of 28 day mortality rate or duration of stay in the ICU was seen in the tighter control group although other trials have suggested that mortality is lowered with stricter glucose control.Two studies from Belgium had been hailed by some as landmark studies ( at least the 2001 surgical ICU study was) and largely on the basis of the 2001 study both the ADA and the AACE issued recommendations regarding tighter control of blood sugar, the former organization targeting glucose values less than 180 and as close as possible to 110. A new standard of care emerged. Here is a thoughtful analysis of the 2001 NEJM paper and a partial review of other data tending to dampen enthusiasm for tight sugar control.
The 2006 medical ICU Belgium study, which I have commented on before, gave intensivists ( a group of physicians who treat very sick patients and used to be called internists) pause for concern in their enthusiasm to "treat to goal" (a recent mantra that makes me wonder what about the old saw about treat the patient not the numbers) as the data had to be tweaked to show a mortality advantage to the tight control group.(One analysis seemed to indicate increased mortality for the first 3 days and then a mortality advantage for the tighter group). There did seem to be less need for dialysis and there was earl er weaning from ventilators in the tight control arm of the study.
The recently reported study (European Glucontrol trial) found no mortality difference for the 536 patients treated to a blood glucose of 80-110 as compared to a group of 546 who glucose was maintained in the 140-180 range.One of the investigators, Dr. Presier from Belgium, is quoted as saying that there is no reason to recommend the use of 110 mg as a target and that 140 or 150 would be wise.
I think we have learned that elevated blood sugars should not be ignored in the critically ill patient but that we can go the way of "too much of a good thing" in attempting to impose a blood glucose of 80-110 particularly in light of recent evidence of lack of clear cut benefit and the very real increased risk of hypoglycemia about which I am just old-fashioned enough to believe it is not a good thing. In the Glucontrol trial of those with a serious hypoglycemic the mortality was 18% versus 11.6 % in those without such a reaction. For those of us still optimistic enough to believe a bigger better randomized trial will show the way, we can wait for the results of the Nice-Sugar trial.