Real-time Continuous Glucose Monitoring in the Intensive Care Unit: The Fifth Vital Sign

Inpatient hyperglycemia is prevalent in 32–38% of hospitalized patients and is associated with more complications, longer length of stay, and higher mortality rates (1). Perioperative hyperglycemia is noted in 20–40% of postsurgical patients, with even greater prevalence of 80% in patients after cardiac surgery (2). Most notably, ∼12–30% of patients do not have a known history of diabetes (2). Without close monitoring, hyperglycemia could easily be missed in patients without diabetes, with devastating consequences. Hyperglycemia is closely associated with poor surgical outcomes, including increased length of stay, delayed wound healing, infections, myocardial infarction, stroke, and death in patients with and without diabetes (3,4). Similarly, hypoglycemia and high glucose variability also adversely affect surgical outcomes (5).

In hospitalized critically ill and noncritically ill patients, point-of-care glucose monitoring remains the gold standard. Before the coronavirus disease 2019 (COVID-19) pandemic, there were few studies on using real-time continuous glucose monitoring (rtCGM) in hospitalized patients, but the pandemic escalated the need for minimizing patient contact and compelled us to explore rtCGM to replace finger-stick blood glucose measurements to protect health care workers and save personal protective equipment while safely managing glycemic excursions in patients. In March 2020, the U.S. Food and Drug Administration issued an unprecedented allowance for using rtCGM in the inpatient setting (6). Consequently, multiple centers have published their experience with rtCGM in intensive care unit (ICU) and non-ICU hospitalized patients (7–9).

Prior studies have helped identify target glucose of 140–180 mg/dL in critically ill patients who use intravenous insulin infusion, which is often a tedious and challenging process, as it requires finger-stick blood glucose checks every 1–2 h (10,11). In this issue of Diabetes Care, Voglová Hagref et al. (12) aim to assess the efficacy and feasibility of rtCGM in critically ill patients undergoing major abdominal surgery and requiring intravenous insulin infusion. To achieve this goal, they compared paired glucose values from rtCGM with glucose values of arterial blood measured on an acid-base analyzer (ABL) as the reference value. They also compared rtCGM with a arterial blood measured by a point-of-care glucose meter, more commonly used at bedside. Upon comparing the glucose values from the rtCGM to the ABL values, they noted a reassuring mean absolute relative difference of 9.4%, while 98.9% of values fell in zones A and B of the surveillance error grid, suggesting attainment of accuracy levels in these critically ill patients comparable with the manufacturer’s reporting in the healthier ambulatory population. The accuracy of bedside blood glucose monitoring was superior, however, with a mean absolute relative difference of 5.8% and 98.8% of values in zones A and B of the surveillance error grid.

Additionally, feasibility was further demonstrated by placing the glucose sensor at a previously unreported site, the infraclavicular region, for two reasons: 1) the abdomen was a less practical site in patients undergoing major abdominal surgery, and 2) the upper arm sites were more likely to be dislodged during routine inpatient activities. In fact, the authors point out that the infraclavicular region was more easily accessible by the nursing staff. Of the 65 participants enrolled, data from 61 participants were included in the analysis, with the exclusion of 4 patients with limited data because of transmitter failure or hematoma at the sensor insertion site, an additional demonstration of the technical and operational challenges of rtCGM in this complicated setting.

Overall, this is a remarkable study of rtCGM in a high-risk, intensive care patient population that adds valuable evidence to the growing literature for the use of rtCGM in the hospital. The most notable contribution of this study to the literature is the investigators’ selection of patients with high acuity and complex health care needs. Currently, there are limited data on this patient population, because these individuals are commonly excluded from studies because of the many potential sources for inaccuracies, including mechanical, physiological, and chemical influences, resulting in significant fluctuations in glucose. However, it is the patients with challenging glycemic control for whom the delivery of safe and effective insulin therapy has the greatest potential for benefits. Traditionally, this glycemic control requires frequent, often hourly, blood glucose monitoring, which is laborious and, when delayed, can cause harm with hypoglycemia as well as hyperglycemia. This study supports use of rtCGM for minimizing the risk of adverse effects of insulin therapy as well as the burden of glycemic monitoring for the staff. Hospital use of rtCGM is a highly anticipated practice, with universal demand but has been historically limited by concerns of inaccuracy due to medication interferences, physiological derangements, and conditions with highly fluctuating glucose values in patients who are critically ill. These concerns are all well addressed by the methodology and analysis used in this study. Finally, by using an alternative site for sensor application, the study added further feasibility for patients in the ICU.

However, one major limitation of the study design hinders its universal applicability. The study used blood glucose sampling from arterial lines and reference values from arterial blood gas analyzers for the frequent calibrations. Neither of these is the standard practice for many hospitals ICUs, particularly in the U.S. Therefore, it remains unclear if the same accuracy can be replicated using venous or capillary blood samples. Nevertheless, it should be noted that the current standard of care of finger-stick capillary blood glucose measurements can create inaccuracies as well, influenced by technique, temperature, perfusion, medications, and other variables.

The authors also acknowledge that while the accuracy of rtCGM in the critically ill is comparable to that reported by the manufacturer in more stable ambulatory settings, it did not meet ISO criteria for blood glucose meters for inpatient use. However, the increased quantity of glucose data with trends and alarms may be an added benefit that can negate harmful glycemic excursions until further criteria are established specifically for acceptable accuracy limits for inpatient rtCGM. At the very least, this continuous, automated technology is no worse than the existing practice and offers additional benefits to the more tedious point-of-care blood glucose monitoring.

Armed with positive results for feasibility, accuracy, and ease of use in the hospital setting, we now are embarking on a new era of rtCGM glucose values as a fifth vital sign (Fig. 1), with potentially new targets for glycemic control using CGM glucometrics. Larger studies are needed to highlight the accuracy and feasibility of rtCGM across a varied inpatient population and to evaluate its efficacy on clinical outcomes, length of stay, and readmission. It would be reasonable to repeat prior intensive insulin therapy trials, such as Normoglycemia in Intensive Care Evaluation–Survival Using Glucose Algorithm Regulation (NICE-SUGAR), using rtCGM to elucidate the ideal glucose targets without the risk of hypoglycemia. Furthermore, demonstrating the cost-effectiveness of rtCGM compared with current blood glucose monitoring will persuade hospital administrators to adopt the technology more readily. Regardless, it is indisputable, that there is now a growing momentum for regulatory approval and implementation of rtCGM in high-risk hospitalized patients.