Continuous GlucoseMonitoring inthe Intensive Care Unit During theCOVID-19 Pandemichttps://doi.org/10.2337/dc20-2219

OBJECTIVE

Real-time continuous glucose monitoring (rtCGM) in critically ill hospitalizedpatients holds promise; however, real-world data are needed.

RESEARCH DESIGN AND METHODS

WeplacedDexcomG6CGMon intensive careunit (ICU)patients atMontefioreMedicalCenter with confirmed coronavirus disease 2019 (COVID-19) infection and glycemicvariability. We analyzed inpatient CGM accuracy using point-of-care (POC) glucose-CGMmatched pairs and included patients for analysis regardless of clinical status.

RESULTS

We included 11 patients with CGM: 8 on continuous insulin infusion (CII), 8 onvasopressors, 8 intubated, 4 on high-dose glucocorticoids, 6 on renal replacementtherapy, and 2 with anasarca. Accuracy was 12.58% for mean and 6.3% for medianabsolute relative difference. CGM reduced POC testing by∼60% for patients on CII.

CONCLUSIONS

In this real-world preliminary analysis of rtCGM during critical illness, we dem-onstrate early feasibility, considerable accuracy, and meaningful reduction in thefrequency of POC glucose testing.

The coronavirus disease 2019 (COVID-19) pandemic has highlighted the urgent needtodevelop innovative caresolutions tomanagehyperglycemia inhospitalizedpatientswhile preserving resources. CGM use in the critically ill patient in the intensive careunit (ICU) offers unique benefits by allowing real-time tracking of glucose levels todetect often unrecognized severe hypo- or hyperglycemic events, while decreasingthe high burden of frequent point-of-care (POC) testing during clinically tenuousstates and continuous insulin infusion (CII) (1).Recent studies of real-time (rt)CGM use in non-ICU patients have demonstrated

promising results in inpatient settings (2–4). However, these studies excludedcritically ill patients. During the pandemic, use of CGM in the ICU could havetransformative results inenablingoptimizedglycemic control, reducingexposure timeof staff, and decreasing resource use. Nevertheless, scant data exist on the use ofmodern rtCGM devices in the ICU.We implemented CGM in the ICU atMontefioreMedical Center during the COVID-

19 pandemic and report our results on the performance and accuracy in a real-worldsetting.Wepurposely studied critically ill patients,who received therapies historicallyexcluded from inpatient CGM studies, to provide vital early data on CGM use in theICU.

1Fleischer Institute for Diabetes andMetabolism,New York Regional Center for Diabetes Trans-lational Research, Albert Einstein College ofMedicine, Bronx, NY2Montefiore Medical Center, Bronx, NY3Division of Endocrinology, Department of Med-icine, Emory University School of Medicine, At-lanta, GA

Corresponding author: Shivani Agarwal, shivani.agarwal@einsteinmed.org

Received 4 September 2020 and accepted 5December 2020

This article is part of a special article collectionavailable at https://care.diabetesjournals.org/collection/diabetes-and-COVID19.

© 2020 by the American Diabetes Association.Readersmayuse this article as longas thework isproperly cited, the use is educational and not forprofit, and the work is not altered. More infor-mation is availableathttps://www.diabetesjournals.org/content/license.

Shivani Agarwal,1,2 Justin Mathew,2

Georgia M. Davis,3 Alethea Shephardson,2

Ann Levine,2 Rita Louard,1,2

Agustina Urrutia,3 Citlalli Perez-Guzman,3

Guillermo E. Umpierrez,3 Limin Peng,3 and

Francisco J. Pasquel3

Diabetes Care 1

NOVEL

COMMUNICATIO

NSIN

DIABETES

Diabetes Care Publish Ahead of Print, published online December 23, 2020

RESEARCH DESIGN AND METHODS

Following the U.S. Food and Drug Ad-ministration statement of nonobjectionto inpatient CGM use during the pan-demic (1,5), hospital leadership allowedclinical use of CGM in the ICU at Mon-tefiore Medical Center. This retrospec-tive analysis of clinical datawas approvedby the Albert Einstein College of Medi-cine Institutional Review Board (Bronx,NY).

Patient Eligibility and SelectionPatientswere eligible if they hadmultipleblood glucose readings in the hypergly-cemic (.180 mg/dL) or hypoglycemicrange (,70 mg/dL) and were hospital-ized in the ICU during the COVID-19pandemic from 1 April to 30 May2020 at Montefiore Medical Center. Pa-tients remained eligible if they werereceiving mechanical ventilation, vaso-pressors, CII, or renal replacement ther-apy (RRT). Exclusion criteria involvedreceiving high-dose ascorbic acid or acet-aminophen .4 g/day. Selection of pa-tients was based on requests for CGMplacement by hospital staff and reviewofinclusion/exclusion criteria by an internalinpatient CGM committee (authors S.A.,R.L., A.S., and A.L.).

CGM Placement and DocumentationAfter patient selection, a trained diabe-tes nurse practitioner or nurse specialistplaced a G6 sensor and transmitter ontothe patient’s abdomen or upper arm incases of prone positioning. CGM re-ceivers were placed outside of the pa-tient’s door for visualization of real-timeglucose data.Recorded blood glucose values were

monitored in real-time by the inpatientCGM committee for validation of sensorand POC glucose readings according toFood and Drug Administration guide-lines (620% of each other [for bloodglucose $100 mg/dL] or 620 mg/dL ofeach other [for POC blood glucose#100 mg/dL]) (6,7). Alarm triggerswere set to ,100 and $250 mg/dLand 3 mg/dL/min rate of glucose riseor drop. After validation, CGM valueswere used to monitor glycemic trends,adjust insulin dosing, and alarm forsevere hypo- or hyperglycemic events,which triggered confirmatory POC test-ing. At minimum, once-daily POC mea-surement was required to ensuremaintained CGM validation.

Statistical AnalysisDescriptive data are presented as mean6SD for continuous variables (age, HbA1c,time in range) and count (%) for categoricalvariables (race/ethnicity, diabetes type,comorbidities). We collected CGM andPOC glucose-matched pairs (n 5 493) tocalculate the mean and median averagerelativedifference (MARD)betweenmeas-urements (median ARD %). We also cal-culated the proportion of CGM valueswithin615, 20, and 30% of POC referencevalues for glucose levels.100 mg/dL and615, 20 or 30 mg/dL for POC glucoselevels #100 mg/dL (%15/15, %20/20,%30/30) (8,9). Clarke error grid anal-ysis was used to assess clinical reliabil-ity of matched POC-CGM glucose pairs(9).

RESULTS

Overall (n 5 11), mean age was 56 6 17years, 6 were male, 8 belonged to un-derrepresented racial-ethnic groups (His-panic or non-Hispanic Black). Six patientshad preexisting type 2 diabetes, three hadtype 1 diabetes, and two had no knowndiabetes. Mean HbA1c was 8.7 6 2.5%.

At the time of CGM use, nine patientswere on CII, eight received vasopressors,eightwere intubated, four received high-dose glucocorticoids, six were on RRT,and two had anasarca.

Median number of days on CGM was9.0 (3.0–10.0). Mean time in range was46.1 6 15.8%, above range was 53.3 615.8%, and below range was 0.66 0.8%.Clarke error grid analysis and MARD arereported in Fig. 1. In sum, 493 pairedPOC-CGM measurements were used foranalysis. Regarding accuracy, mean ARDand median ARD were 12.58% and 6.3%,respectively. We found 77.7% of valuesfell within zone A (CGM within 20% ofPOC), 20.5%zoneB (.20%difference, noincorrect treatment), 0.2% zone C (hy-perglycemia or hypoglycemia leading toinappropriate treatment), 1.2% zone D(undetected hypoglycemia or hypergly-cemia needing treatment), and 0.4%zone E (hypoglycemia mistaken for hy-perglycemia, and vice versa) (Fig. 1).

To calculate reduction in POC testingwhileusingCGM,wecalculated themeannumber of hours/potential POC tests whileon CII versus actual POC tests (n 5 8 pa-tients). Mean number of hours/potentialPOC tests on CII was 72.125 h/;72 tests,compared with mean number of actualPOC tests on CGM of 28, representing an

estimated 60% reduction in POC testsduring rtCGM use.

CONCLUSIONS

We present our firsthand real-world ex-perience of rtCGM in the ICU during theCOVID-19 pandemic in New York City.Our findings show that use of rtCGM inthe ICU setting is feasible, acceptable,and reliable as an adjunctive modality toPOC glucose measurements. CGM useresulted inameaningful reduction inPOCtesting frequency for patients requiringCII. Overall, our study underscores thegreat potential of CGM to aid in glucosetesting for critically ill patients. Whileother studies have systematically ex-cluded the sickest patients, we includedpatients with multiple critical illnesses andtherapies, which is a particular strength ofour study.

Overall, 98% of all sensor readings hadclinically acceptable correlation (zonesA 1 B). Currently, there are few reportsof remotemonitoringusing theG6devicein the ICU. Compared with findings fromrecent small studies using G6 in medical-surgical patients on hospital floor units(2,10,11), ourMARD (12.58%) fallswithinthe range of these reports (9.4–12.8%)(2,10,11). Our accuracy results are evenmore encouraging given our eligibilitycriteria including critically ill patientswith potentially interfering treatments,compared with prior studies. Neverthe-less, concerns remain regarding completereplacement of POC glucose testing withCGM (1,12–14), given that our data revealseveral instances of potential failures todetect hypo- orhyperglycemia (zoneD) ormistreatment (zone E). Thus, while prom-ising, cautionmust be exercised in utilizingCGM as the sole modality for glucosetesting.

Limitations of our study include itssmall sample size and observational na-ture. While inclusion of critically ill pa-tients is a strength and our accuracyresults are in line with previous studiesexcluding sick patients, further studiesare required to evaluate CGM safety andaccuracy in a larger critically ill popula-tion. Differentiation of results by poten-tially interfering treatments is needed.

In sum, our results provide proofof concept that rtCGM can be usedsafely and effectively during criticalillness, while offering reliable accuracyand decreasing POC glucose testingfrequency. Overall, CGM use in the

2 rtCGM in Critically Ill COVID-19 Patients Diabetes Care

ICU setting should be further exploredas a treatment aid for inpatient diabe-tes management of the critically illpatient.

Acknowledgments. The authors greatly appre-ciate the efforts of their fellow health careworkers and support staff at the MontefioreMedical Center for providingoutstandingpatientcare at considerable personal risk on the frontlines of this pandemic. The authors express theirsolidarity with those who are or have been illwith COVID-19 and their families.Funding. S.A. is supported in part by NationalInstitutes of Health (NIH) under award numbersK23DK115896 and P30DK111022. G.M.D. is sup-ported in part by NIH under award number1K23DK122199-01A1. G.E.U. is partly supportedby research grants from the NIH UL1 TR002378and 1P30DK111024-04 and has received unre-stricted research support from Sanofi, Novo Nor-disk, and Dexcom. F.J.P. is supported in part byNIH under award numbers 1K23GM128221-03,

P30DK11102404, and P30DK111024-05S and hasreceived research support from Dexcom andMerck.Dexcom provided device equipment support,

but hadno role in thedesignof this study, its datacollection, analysis, or findings.Duality of Interest. F.J.P. has received consul-ting fees from Boehringer Ingelheim. No otherpotential conflicts of interest relevant to thisarticle were reported.Author Contributions. S.A. and F.J.P. concep-tualized the study, researched and analyzed thedata, and wrote the manuscript. J.M. input data,researchedandanalyzed thedata, andwrote andreviewedthemanuscript.G.M.D. researchedandanalyzed the data and wrote, reviewed, andedited the manuscript. A.S., A.L., R.L., A.U.,and C.P.-G. input and analyzed the data andreviewed and edited the manuscript. G.E.U.reviewed and edited the manuscript. L.P. ana-lyzed the data and reviewed and edited themanuscript. S.A. is the guarantor of this workand, as such, had full access to all the data in thestudy and takes responsibility for the integrity ofthe data and the accuracy of the data analysis.

References1. Galindo RJ, Aleppo G, Klonoff DC, et al. Im-plementation of continuous glucose monitoringin the hospital: emergent considerations forremote glucose monitoring during the COVID-19 pandemic. J Diabetes Sci Technol 2020;14:822–8322. Reutrakul S, Genco M, Salinas H, et al. Fea-sibility of inpatient continuous glucose monitor-ing during the COVID-19 pandemic: earlyexperience. Diabetes Care 2020;43:e137–e1383. Shehav-Zaltzman G, Segal G, Konvalina N,Tirosh A. Remote glucose monitoring of hospi-talized, quarantined patients with diabetes andCOVID-19. Diabetes Care 2020;43:e75–e764. UshigomeE, YamazakiM,HamaguchiM, et al.Usefulness and safety of remote continuousglucose monitoring for a severe COVID-19 pa-tient with diabetes. Diabetes Technol Ther.15 July 2020 [Epub ahead of print]. DOI: 10.1089/dia.2020.02375. Pasquel FJ, Umpierrez GE. Individualizing in-patient diabetes management during the coro-navirus disease 2019 pandemic. J Diabetes SciTechnol 2020;14:705–7076. Garg SK, Akturk HK. A new era in continuousglucose monitoring: Food and Drug Administra-tion creates a new category of factory-calibratednonadjunctive, interoperable class II medicaldevices.DiabetesTechnol Ther2018;20:391–3947. Shah VN, Laffel LM, Wadwa RP, Garg SK.Performance of a factory-calibrated real-timecontinuous glucose monitoring system utilizingan automated sensor applicator. Diabetes Tech-nol Ther 2018;20:428–4338. Freckmann G, Pleus S, Grady M, Setford S,Levy B. Measures of accuracy for continuousglucose monitoring and blood glucose monitor-ing devices. J Diabetes Sci Technol 2019;13:575–5839. Clarke W, Kovatchev B. Statistical tools toanalyze continuous glucose monitor data. Di-abetes Technol Ther 2009;11(Suppl. 1):S45–S5410. Nair BG, Dellinger EP, Flum DR, Rooke GA,Hirsch IB. A pilot study of the feasibility andaccuracy of inpatient continuous glucose mon-itoring. Diabetes Care 2020;43:e168–e16911. Tripyla A, Herzig D, Joachim D, et al. Per-formance of a factory-calibrated, real-time con-tinuous glucose monitoring system duringelective abdominal surgery. Diabetes ObesMetab 2020;22:1678–168212. Davis GM, Galindo RJ, Migdal AL, UmpierrezGE. Diabetes technology in the inpatient settingfor management of hyperglycemia. EndocrinolMetab Clin North Am 2020;49:79–9313. Galindo RJ, Migdal AL, Davis GM, et al.Comparison of the FreeStyle Libre Pro Flashcontinuous glucose monitoring (CGM) systemand point-of-care capillary glucose testing inhospitalized patients with type 2 diabetes trea-ted with basal-bolus insulin regimen. DiabetesCare 2020;43:2730–273514. Ancona P, Eastwood GM, Lucchetta L, EkinciEI, Bellomo R, Martensson J. The performance offlash glucose monitoring in critically ill patientswith diabetes. Crit Care Resusc 2017;19:167–174

Figure 1—Clarke error grid analysis showing the relationship between CGM and POC glucosevalues in critically ill patients with COVID-19 (n 5 11).

care.diabetesjournals.org Agarwal and Associates 3