Blog

Todd Fraser wrote 05-18-2010 04:54:09 pm
I agree Matt. And it appears the literature supports that. Where is the right mark to aim at, low enough to prevent infectious (and consequent) complications, while high enough to have margin for error.

My personal view is to maintain it somewhere in single figures.

There were a couple of interesting reviews on this in the September CICM magazine. Worth a read.

alexander browne wrote 08-18-2011 02:20:37 pm
I believe that the whole glucose debate centres on the premise of normalising physiology, something that has been disappointing with attempted normalisation of other parameters. Release of glucose is related to a stress response, with a greater degree of illness creating a greater stress response, clearly there are "issues with the tissues" in terms of their ability to utilise the glucose, hence the rises we see.

Clearly too much glucose is bad, but it just reflects the underlying badness within the patient, and perhaps it is something that we need to keep in check, but not fully normalise. I agree somewhat with keeping BGLs in single figures, but my parameters are 5-12. However, I work in a unit that uses a glucose feed algorithm that works extremely well, a kind of set and forget thing.

I think the NICE sugar method didn't work, rather than tight glucose is not beneficial, and perhaps a better system that prevents extremes in glucose levels would be more appropriate.

Matt Mayer wrote 06-18-2009 02:43:51 am
It seems to make physiological sense that a low BSL rather than a high one poses a more acute medical risk to the patient, as the risks of a BSL <4 seem
greater than a those of a value >9. I think this landmark study will prevent a lot of potentially avoidable middle of the night hypo's.

Anthony Tzannes wrote 02-05-2011 04:24:24 pm
Hi Neil,
I recently read your insightful review of the study and its place in evidence behind Mx of blood glucose levels in critically ill patients.

A few comments from my reading of the study:

1. A concern with the protocol of the study: namely that if hypoglycaemia was found on testing BSL, then it would not be treated until formal blood result back confirming the result - resulting in potentially long periods of hypoglycaemia (I may have misinterpreted this, as it seems illogical unless greater concern was made for high BSL than low as part of study design)

2. Tight control group had an average BSL of 6.8 and thus either had periods of very high BSL or (more likely) had majority of time with BSL outside target range.

My interpretation of the the second point is that the study failed to achieve its target an thus it cannot be stated that more relaxed BSL control is better for critically ill patients than tight control. What can be stated is that with current technology, attempting to achieve tight control in the critically ill patients results in overall worse outcomes (taking into account point 1 above)

A subgroup analysis of those patients that did achieve tight control without any hypoglycaemic events would be interesting (or may just muddy the waters further!)

Regards, Anthony

Neil Orford wrote 02-05-2011 04:28:15 pm
Hi Anthony,

1. we didn't enter the study so I am not sure how it worked in reality. On the whole issue of hypopgylcaemia, I am uncertain of its significance. For
instance brief periods of biochemical hypoglycemia may be inconsequential, as in ambulatory diabetics it takes a long duration of severe hypoglycaemia to lead to measurable cognitive deficits. Also the insulin tolerance test for adrenal insufficiency deliberately leads to severe hypoglycemia.

The German Sepnet study stopped there 2x2 IIT and starch study due to high incidence of hypo's, as a serious adverse effect, with no emonstratable harm. I think at the same time they felt there was no evidence of benefit ( ie reduced mortality, although underpowered at interim). It seemed a bit unfair to compare death to biochemical hypoglycaemia.

I don't want to dismiss hypoglycemia, I am just uncertain of the significance of brief episodes of biochemical hypoglycemia.

2. I understand what you are saying, although I am not sure on a few points. The 2 groups had significantly different time weighted BSL( 6.4 vs 8.0 ), and significantly different insulin doses. So they where different. I think you are correct in interpreting this as meaning they had a lot of time out of range, I am not sure if it is the majority. So presumably both groups oscillated around the average, with more time above than below ( presuming that algorithms are built to keep you from having hypo's). I don't think we know how much time was spent abov

Anthony Tzannes wrote 02-05-2011 04:31:53 pm
Hi Neil,

I agree with your comments re: hypoglycaemia being of uncertain significance, the studies definitely weren't designed or powered to look specifically at the effects of hypoglycaemia. I think you have summed up my feelings as well when you stated that the Sepnet study considered hypoglycaemia as a serious adverse effect with no demonstrable harm, people seem to recover well from episodes but we want to avoid them if at all possible.

As for the issue with average BSL (I had recalled this as 6.8, not 6.4) , If the average BSL was outside the target range; when the purpose of the study was to determine if keeping patients within the target range changes outcomes; then the study cannot answer the question it was designed to ask, what is has answered, is that the attempt to achieve that range was associated with increased mortality. I agree that the 2 groups were different in terms of amount of insulin received.

One issue here is one that makes all studies in ICU difficult, namely the difficulty in getting good data on the outcomes that are potentially of greater
import than mortality. Whether a person lives or dies is easy and to determine and very hard to argue with as an end point, however the functional outcome of survivors (ie. able to live independently vs needing care) is in some ways a more important endpoint and much harder to both measure (time and effort compared to death: yes or no) and statistically deal with all the confounding factors associated with post