Intensive insulin therapy did not reduce hospital mortality in critically ill patients.

The implementation of intensive insulin therapy in intensive care units at one institution was not related to a decrease in hospital mortality, according to study results.

Researchers at the University of Washington and Harborview Medical Center in Seattle compared outcomes in 10,456 patients admitted in the seven intensive care units at Harborview Medical Center after applying intensive insulin therapy. The study was broken down into three time periods.

In period 1 from March 2001 to February 2002, the intensive care units administered no protocol. In period 2 from March 2002 to June 2003, they administered an insulin therapy protocol with target glucose 80 to 130 mg/dL. Finally, during period 3 from July 2003 to February 2005, the intensive care units administered a protocol with target glucose 80 to 110 mg/dL.

Over time, insulin administration increased (period 1: 9%, period 2: 25%, period 3: 42%). Those patients in period 3 had higher adjusted hospital mortality than those in period 1 (OR, 1.15; 95% CI, 0.98-1.35), according to researchers. Hypoglycemia incidence increased about fourfold from periods 1 to 3.

The researchers concluded that future research is necessary before implementing an intensive insulin therapy in the critically ill. – by Stacey L. Adams

Critical Care. 2008;doi:10.1186/cc6807.

The authors show interesting evidence of increasing adoption of an intensive insulin policy in intensive care unit patients comparing 2002, 2003 and 2005, with average daily blood glucose decreasing from 146 to 142 to 133, with modest increase in hypoglycemia, from 1% to 1.6% to 2.2%. ICU mortality was 9%, 10.8%, and 9.8% during the three periods, suggesting no definite benefit. This is compatible with results of many database studies, in which the fact that no specific intervention is being assessed and that it is simply an observation of a cohort leads to there being no clear outcome difference. One must simply conclude that there is evidence that intensive treatment is being more widely adopted. The authors go on to make a statistically questionable step of then looking at subgroups of intensive care three or fewer days vs. more than three days, and subgroups by types of intensive care units, medical, surgical and trauma, with one of the comparisons. For trauma units, suggesting increased intensive care unit mortality although not increased total in-hospital mortality. The reason this is questionable is that: 1) the primary outcomes were negative, so in theory one should not do subgroup analysis; and 2) there are many possible subgroup analyses, of which only one is positive; and 3) many factors could lead to changes over time in outcome, which cannot be controlled in a cohort observation as reported.

– Zachary T. Bloomgarden, MD

Endocrine Today Editorial Board member