The debate about ideal glycemic control in intensive care continues. In the last years, large intervention studies including the NICE-SUGAR trial unexpectedly failed to replicate the initial and very promising finding of the first Leuven study showing a significant survival benefit for TGC in surgical ICU patients. Substantial differences between these studies were present in study design, execution of glucose control and treated population. Likely one of the main reasons why clinical trials and meta-analyses showed negative results for TGC was the high incidence of hypoglycemia induced by intensive insulin therapy. Probably it was impossible to prevent this because no reliable tool facilitating GC was available. The used (paper-based) protocol was often not extensively tested before implementation, and adherence to it was not reported or registered, a fact that may be especially problematic in a multi-center approach where different sites often display substantial differences in standards of critical care including important aspects in nutrition and glycemic control. Despite all efforts, a large gap is still evident between what is achievable and what is desirable in glucose control – (near) normoglycemia besides minimal variability.
SGC with the implemented eMPC algorithm is a well-validated, CE-certified nurse-driven tool to assist glucose control in critical care. As demonstrated in our two centers, it was possible to achieve excellent adherence and glycemic control with low variability. The target range of 4.4 to 8.3 mmol/l could be achieved in 83% (Graz) and 93% (Zurich) of the time. Despite excellent efficacy, usability and performance parameters were assessed very differently by the nursing staff at the two study sites. In Zurich, glucose control was outstanding and significantly better than in Graz, yet the involved ICU staff turned out to be unsatisfied. We hypothesize that this difference was caused by 1) the intensive hands-on simulated training performed only in Graz and 2) the long lasting routine use of BBraun pumps at the same site. Although the training was a time-consuming and laborious process, it nevertheless seems to be necessary when implementing such a new tool in order to assure long-term functionality and operator satisfaction within the ICU team. The fact that infusion pumps from another manufacturer are in use in Zurich meant that ICU staff needed to become familiar with even more new equipment and had to handle two different types of pumps at the same time. This is particularly evident in the response to question 8 regarding potential routine use which was viewed favorably by 70% of nurses in Graz, but only 11% in Zurich.
Workload was perceived to be increased by the majority of the users which is explainable by the relatively frequent sampling and high documentation effort compared to current routine care at both sites.
With regard to safety, one patient experienced severe hypoglycemia (2.5% of patients or 0.03% of glucose readings) which seems an acceptable rate in this population. This incident was primarily attributable to a long (4-hour) measurement interval that was suggested by the device. However, although the responsible nurse felt uneasy and measured a dropping blood glucose level of 4.5 mmol/l, it was not entered in the SGC and insulin infusion continued with the same rate. With more extensive training of the involved staff, this hypoglycemic episode may have been prevented as the eMPC almost certainly would have advised to stop insulin infusion in view of a rapidly falling blood glucose concentration. The continued use of insulin at a glucose level below 5.6 mmol/l has recently been identified as one of the most common causes for hypoglycemia in an analysis of > 55,000 glucose readings in 1,657 patients. This incident also demonstrates that the possibility of user overruling may be crucial for optimal SGC use and the device's suggestions should only be followed in view of the clinical context.
Obviously, our study is limited by its small sample size and open design. However, the primary aim was to in detail evaluate the complete SGC system in a medical population of critically ill patients at two different sites. As demonstrated by the individual profiles of the participants, glucose control could be established and maintained over the whole study period in both centers. Another limitation is the use of bedside glucometers as they were originally developed for glucose measurement in another setting and are not accurate enough for glycemic control in critically ill patients, especially when anemia is present. Morever, the mean sampling interval of > 2 hours is a potential limitation as in comparison with continuous glucose monitoring, hypo- and hyperglycemic excursions of blood glucose may have been missed, as also described in the literature. Certainly, our observations need to be confirmed in larger populations with a focus on presence or absence of diabetes and also in a setting with CGM.
Under clinical trial conditions, subcutaneous continuous glucose monitoring (CGM) has been found to be sufficiently accurate when compared to arterial sampling. In a small pilot study, Kopecky and colleagues showed that combining the eMPC algorithm with CGM is a feasible method for glycemic control in cardiac surgery patients. Okabayashi et al. could demonstrate a reduced risk for surgical site infections and excellent perioperative tight glycemic control without hypoglycemic episodes using a closed loop glycemic control system with CGM in patients requiring hepato-biliary-pancreatic surgery.
These studies and ours are important steps on the way to the artificial pancreas at the bedside of the ICU patient - a real closed loop system that would ideally combine reliable continuous glucose sampling with a semi-automated method to improve glycemic control and reduce workload. Major efforts are also being undertaken to internationally standardize important aspects of glycemic control, as evident in the 2013 Consensus recommendations jointly written by the leading experts in the field. These developments will certainly greatly contribute in the implementation of safe, efficient and not overly elaborate glucose control in critical care.