PHILADELPHIA – Glycemic variability appeared to be independently associated with increased length of stay and mortality in noncritically ill hospitalized patients, in a large retrospective study presented at the annual scientific sessions of the American Diabetes Association.
"Glycemic variability" refers to oscillations of blood glucose levels around the mean. Two people can have identical hemoglobin A1c values, yet one may have far greater variability than the other and therefore be considered to have poorer control because of frequent bouts of hyper- and hypoglycemia, said Dr. Carlos E. Mendez, director of the diabetes management program at the Albany (N.Y.) Stratton VA Medical Center.
Previous studies done in critical care settings have demonstrated increased mortality in patients with high glycemic variability, independent of hypo- and hyperglycemia (Crit. Care Med. 2008;36:3008-13). Both in vitro and in vivo data suggest that "rather than hyperglycemia, glycemic variability has a more profound effect and greater production of reactive oxygen species and what we think is more reactive oxidative stress," Dr. Mendez said.
To examine this phenomenon in noncritically ill patients, he and his associates retrospectively reviewed glucose values from a total of 960 patients who were admitted to medical (81%) or surgical (19%) wards during 2008-2010 and for whom a minimum of two point-of-care blood glucose values per day had been ordered. A GSD (glucose standard deviation) value was calculated as a surrogate for glycemic variability.
The patients were typical of a VA population: They had a mean age of 69.8 years, and 95% were male. Their mean blood glucose was 181.3 mg/dL, with a mean GSD of 57.4 mg/dL. Average length of stay was 5.7 days. A mean of 4.8 glucose readings per day was performed for the group as a whole. Nearly a quarter (24%) had at least one hypoglycemic episode; about two-thirds were receiving insulin. The 90-day mortality was 11%. The majority (85%) had a diagnosis of diabetes.
Length of stay increased significantly with increasing GSD, from 3.3 days for the 166 patients with the least variability (0-30 mg/dL), to 6.5 days for the 245 patients with GSD of 61-90 mg/dL, to 7.4 days for the 39 who had glycemic variability greater than 120 mg/dL. Those increases translated to a mean 6% increase in length of stay for every 10-mg/dL increase in the GSD. The relationship remained significant for patients with and without diabetes, for medical and surgical patients, and for those who had and had not experienced hypoglycemia, Dr. Mendez said.
Mortality at 90 days also increased significantly with greater GSD, from 9% in those with the lowest variability, up to 28% for the patients with the highest variability. Here, there was a 9% increased risk of death for every 10 mg/dL of increased GSD. The mortality findings were significant for the patients with diabetes and for the medical ward patients, but didn’t reach significance for nondiabetic and surgical patients, probably because of low numbers, he said.
Prospective studies using continuous glucose monitoring would confirm these results and better elucidate factors that may influence glycemic variability in noncritically ill inpatients, he concluded.
Dr. Mendez disclosed no conflict of interest.