Tuesday, January 31, 2006

Blood sugar control in the critically ill patient-the plot thickens

The Feb. 2, 2006 of the NEJM features a randomized trial regarding "tight" blood sugar control in intensive care patients.Physicians caring for the critically ill would have been happier with a more definitive answer but clinical reality continues to be messy and typically resists our efforts to understand it.

Belgian physicians randomized 1200 medical ICU patients into an intensive insulin control group (target blood sugar 80-110) and a conventional treatment group (give insulin if blood sugar greater than 215).

Overall, there was no difference in mortality. When data were analyzed based on length of stay in ICU various answers were forthcoming. Using a 3 day dividing line it seemed that those in ICU beyond 3 days enjoyed a mortality benefit ( 52.5 % vrs 43%). Those who stayed less than 3 days actually showed an increase in mortality. Using five days, the longer stay group enjoyed several benefits in terms of a variety of morbidity indicators but those less than 5 days showed no difference. The authors report less kidney damage in the treated group and earlier weaning from respirators.

The results here are complicated and we are inundated with data and with so much data and with medical ICU patients being so heterogeneous in presentation and in the other treatments received simple answers elude the researchers .Controlling one variable in an experiment with so many other variables influencing the outcomes may well lead to a bottom line more ambiguous than certain. Various sub-group analysis which may give hints as to what is going on may also give false positives due the multiple comparison phenomenon and false negative due to low power statistical power with the small numbers in each group. Which of the multiple outcomes analyzed should be determinative for the clinical physician caring for ICU patients? The trial that seemed to be the tipping point for more aggressive glucose control was done in a surgical ICU units with less seriously ill patients whose medical conditions were less multifaceted and complex.

Dr.Atul Malhotra, in his editorial tries to give the reader some suggestions. He looks at the bright side choosing to emphasize the reported improvement in some aspects of morbidity rather than ambiguity in mortality outcome. He suggests target blood sugar of less than 150 for the first 3 days and then use the 80-100 target used in this trial recognizing the possibility of harm being done to the patients in the short stay in ICC. Certaintly, hypoglycemia is not a good thing and there were more episodes in the treatment group. There is another trial in the wings (the NICE-SUGAR trial).Ignoring the overtly cute name of this trial, hopefully we can look forward to more clear cut answers when that is published.

Dr. Malhotra's suggestions seem reasonable based on the analysis of the Belgian data;it makes sense to be less vigorous in the first few days with insulin therapy and then tightening more on glucose control afterward. However the 3 day value may be more an artifact of the post-hoc analysis of the data than a magic number that may or may not hold up when further trials become available.

No comments: