Background: Short-term, transient hyperglycemia is associated with adverse outcomes in acutely ill populations. Because parenteral nutrition (PN) is dextrose based, we hypothesized that exposure to PN would be associated with hyperglycemia and that greater levels of dextrose infusion would be associated with higher glucose concentrations. Our objective was to examine the temporality, incidence, and dose response from dextrose load upon hyperglycemia using several serum glucose cut points in PN vs non-PN HSCT recipients.
Methods: The medical records of adults admitted for initial autologous or allogeneic hematopoietic stem cell transplant (HSCT) at 2 university-affiliated hospitals between September 1999 and December 2003 were used in this retrospective cohort. To minimize the impact of disease acuity on serum glucose, patients with diabetes mellitus, steroid administration, patients with recently treated infections, or patients who died during therapy were eliminated from the study. Serum glucose values were recorded once per day from the first morning venous blood draw (2 AM-6 AM) to achieve uniformity among patients, to avoid measurements occurring more frequently among hyperglycemic patients, and to minimize the influence of oral intake. Hyperglycemia was examined using several serum glucose cut points (110, 125, 150, 175, and 200 mg/dL). Wilcoxon rank-sum tests were used to detect differences in hyperglycemic events between PN and non-PN subjects, and mixed-effects regression models were used to detect the association between PN exposure and hyperglycemia. To address the temporality and incidence of hyperglycemia between PN vs non-PN participants, before and "after" time frames were created. Preinfusion (before) and actual infusion (after) times were used for these intervals for PN patients; however, the average hospital days before (before) or during (after) PN infusion were used for comparison in non-PN recipients (ie, autologous non-PN before = hospital days 1-10, after = hospital days 11-21).
Results: Of the 208 patients who qualified for inclusion 49% (n = 101/208) received PN, which provided on average 26 kcal per kg, 1.3 g of protein per kg, and 2.7 mg/kg/min of dextrose (range 1.3-3.9 mg/kg/min). The proportion of hyperglycemic days before was not different between groups; however, it was significantly greater after in PN vs non-PN patients, regardless of serum glucose cut point. A dose response between dextrose administered (mg/kg/min) and serum glucose concentrations was not seen. When longitudinally presented, the temporal relationship between serum glucose and PN initiation was reflected approximately on hospital day 9. Using regression models that account for repeated measures, the odds of having hyperglycemia (yes/no; glucose >110 mg/dL) after PN exposure were nearly 4 times (odds ratio 3.9; 95% confidence interval, 2.7-5.5) that of non-PN exposed, after controlling for donor type, race, age, and conditioning chemotherapy. PN was the only variable to significantly interact with time (p < .0001), signifying not only the change in odds over time but also as powerful evidence that PN was the causative agent of hyperglycemic events.
Conclusions: The broad use of PN at levels within current clinical guidelines in HSCT adults was associated with profound hyperglycemia; however, greater dextrose dose, within the narrow levels administered in this cohort, was not associated with higher glucose concentrations.