Improvement of Blood Glucose Control on the Bone Marrow Transplant (BMT) Unit: A Retrospective Review of Our Quality Improvement Pilot Program

Background:

Multiple studies of improved glycemic control in critically ill patients have yielded contradictory results. Few studies on inpatient hyperglycemia exist in the BMT population. We undertook a quality improvement project to improve blood glucose (BG) control with a goal of increasing the proportion of time that patients on our BMT service spent within the range of 70-200 mg/dl.

Methods:

With the Division of Endocrinology, an algorithm for the initiation and modification of finger sticks and anti-hyperglycemic medications was created and implemented on the Tufts Medical Center BMT service for admissions between 4/1-6/30/13 that were predicted to be > 48 hours in duration (intervention). Using the Remote Automated Laboratory System (RALS), the percent of time in each BG range (<70, 71-110, 111-140, 141-199, >200) was calculated for the entire floor in the three months prior to implementation(baseline) and during the three months of the pilot program. As the oncology service is included in this calculation and was not part of our intervention, admissions were analyzed for comparison. With IRB approval, retrospective data of admissions >48 hours was collected to evaluate BG, length of stay, and infectious complications.

Results:

The baseline cohort included 64 BMT admissions, while there were 70 BMT admissions in the intervention cohort and 102 oncology admissions not part of the intervention. 14% of patients in each of the three admission groups had a history of diabetes. 30% of all patients on BMT were discharged on steroids, compared to 10% on oncology. On admissions when finger stick evaluation of BG was initiated (36% in the BMT intervention cohort, 25% in the BMT baseline corhort, (P = 0.25), more patients received short acting insulin as per the algorithm (21% vs 6%, P = 0.016), but there was no difference in the number transitioned to long acting insulin.  In the intervention cohort, the proportion of time spent in the BG range of 71-199 increased, with less time spent with a BG < 70 or > 200 (Figure 1, P < 0.0001). Fewer BMT patients were hyperglycemic within 48 hours of a documented infection in the intervention group compared to the baseline cohort, but the overall rate of infection among the three groups was low. Within each cohort on BMT, 6 admissions had a discharge BG>200, and 3 were discharged on new anti-hyperglycemic medications.

Conclusions:

We were able to demonstrate the feasibility of implementing a program to control and track blood glucose. Not only were we able to limit hypoglycemic episodes, there was a lower rate during the intervention compared to baseline. The results of this retrospective study will allow the design of larger trials to determine whether BG control has an impact on length of stay, infectious complications, and mortality.

Figure 1: