Until recently, continuous glucose monitoring (CGM) systems were reserved for use in the outpatient setting or for investigational purposes in hospitalized patients. However, during the coronavirus disease 2019 pandemic, use of CGM in the inpatient setting has grown rapidly. This review outlines important details related to the accuracy, limitations, and implementation of, as well as necessary staff education for, inpatient CGM use and offers a glimpse into the future of CGM in the inpatient setting.

Inpatient glucose management is challenging given the complexities of care for hospitalized patients with diabetes, including changes in diet status, medications, glucose metabolism, and schedules. Glucose variability and hypoglycemia have been associated with worse outcomes in hospitalized patients (13). In addition, diabetes and hyperglycemia are associated with worse coronavirus disease 2019 (COVID-19) outcomes (4,5).

Until recently, blood glucose measurements using point-of-care (POC) and whole-blood laboratory testing were the only means of assessing glycemia in the hospital setting. In contrast, the advent of continuous glucose monitoring (CGM) revolutionized diabetes care in outpatient settings. CGM measures interstitial fluid glucose levels via a subcutaneous sensor that transmits data to an insulin pump, smartphone, or stand-alone receiver or reader, either in real-time when using a real-time CGM (rtCGM) system or when a user scans the sensor with the receiver when using an intermittently scanned CGM (isCGM) system (Table 1) (69). Before the COVID-19 pandemic, inpatient CGM was limited to use for research purposes, in which it demonstrated a reduction in inpatient hypoglycemia (10,11) and hyperglycemia (12). However, in April 2020, the U.S. Food and Drug Administration (FDA) expanded the availability and implementation of noninvasive patient monitoring devices during the pandemic by exercising enforcement discretion so that CGM could be used temporarily in the inpatient setting (13). CGM was well suited to address some challenges of the COVID-19 pandemic by reducing nursing exposure, preserving personal protective equipment (PPE), and providing additional glucose data including trends (14). However, there are many important considerations for its continued inpatient use.

Table 1

Comparison of CGM Systems

FeatureFreeStyle Libre 14-Day (6)FreeStyle Libre 2 (7)Dexcom G6 (8)Medtronic Guardian 3 (9)
Type of system isCGM isCGM rtCGM rtCGM 
Calibration required? No No No Yes (at least two daily) 
Glucose data display options Receiver or smartphone (requires scanning sensor to display glucose data) Receiver or smartphone (requires scanning sensor to display glucose data) Receiver or smartphone (continuous) Smartphone (continuous) 
Glucose data storage Measures glucose every minute; stores data every 15 minutes (requires scanning to upload last 8 hours of data to receiver or smartphone) Measures glucose every minute; stores data every 15 minutes (requires scanning to upload last 8 hours of data to receiver or smartphone) Stores data every 5 minutes to receiver or smartphone (if within range of device) Stores data every 5 minutes (if smartphone is within range of device) 
Range of transmission between sensor and receiver or smartphone Not applicable 20 feet (unobstructed) to receive alerts/alarms only 20 feet (unobstructed) 20 feet 
Sensor wear time, days 14 14 10 
Warm-up time, hours up to 2 
Glucose display range, mg/dL 40–400 40–400 40–400 40–400 
Remote monitoring options LibreView: Cloud-based software with real-time data for smartphone users; LibreLinkUp: real-time data-sharing with smart device LibreView: Cloud-based software with real-time data for smartphone users; LibreLinkUp: real-time data-sharing with smart device Dexcom Clarity: Cloud-based software with 45-minute delay of glucose data; Dexcom Share: real-time data-sharing with smart device CareLink: Cloud-based software with real-time data (uploaded every 24 hours from personal to professional website); CareLink Connect: real-time data-sharing with smartphone 
Real-time alarm for glucose thresholds? No Yes Yes Yes 
Predictive glucose alerts? No No Yes Yes 
Reported MARD for outpatient use, % 9.30 9.30 8.7–10.6 
Potentially interfering substances Vitamin C Vitamin C Hydroxyurea, acetaminophen >1 g every 6 hours Acetaminophen 
FeatureFreeStyle Libre 14-Day (6)FreeStyle Libre 2 (7)Dexcom G6 (8)Medtronic Guardian 3 (9)
Type of system isCGM isCGM rtCGM rtCGM 
Calibration required? No No No Yes (at least two daily) 
Glucose data display options Receiver or smartphone (requires scanning sensor to display glucose data) Receiver or smartphone (requires scanning sensor to display glucose data) Receiver or smartphone (continuous) Smartphone (continuous) 
Glucose data storage Measures glucose every minute; stores data every 15 minutes (requires scanning to upload last 8 hours of data to receiver or smartphone) Measures glucose every minute; stores data every 15 minutes (requires scanning to upload last 8 hours of data to receiver or smartphone) Stores data every 5 minutes to receiver or smartphone (if within range of device) Stores data every 5 minutes (if smartphone is within range of device) 
Range of transmission between sensor and receiver or smartphone Not applicable 20 feet (unobstructed) to receive alerts/alarms only 20 feet (unobstructed) 20 feet 
Sensor wear time, days 14 14 10 
Warm-up time, hours up to 2 
Glucose display range, mg/dL 40–400 40–400 40–400 40–400 
Remote monitoring options LibreView: Cloud-based software with real-time data for smartphone users; LibreLinkUp: real-time data-sharing with smart device LibreView: Cloud-based software with real-time data for smartphone users; LibreLinkUp: real-time data-sharing with smart device Dexcom Clarity: Cloud-based software with 45-minute delay of glucose data; Dexcom Share: real-time data-sharing with smart device CareLink: Cloud-based software with real-time data (uploaded every 24 hours from personal to professional website); CareLink Connect: real-time data-sharing with smartphone 
Real-time alarm for glucose thresholds? No Yes Yes Yes 
Predictive glucose alerts? No No Yes Yes 
Reported MARD for outpatient use, % 9.30 9.30 8.7–10.6 
Potentially interfering substances Vitamin C Vitamin C Hydroxyurea, acetaminophen >1 g every 6 hours Acetaminophen 

Inpatient CGM may be a promising alternative to periodic POC glucose measurement and offers advantages that include automatic measurement of glucose values displayed at intervals of 1–5 minutes, automatic transmission of glucose values to physically distant display devices, and programmable alerts and alarms to warn the inpatient care team of existing or impending dysglycemia. However, it is important to understand that CGM measures interstitial glucose rather than blood glucose, as is measured with capillary POC or laboratory glucose testing. Intestinal glucose can lag behind blood glucose by up to 40 minutes, with an average delay of ∼9 minutes (15). The delay is noted to be more prolonged and clinically significant in the setting of rapidly changing glucose levels such as after meals (15,16). Few studies of interstitial glucose have been conducted in settings such as hypotension, hypoperfusion, hypoxia, or other states common in hospitalized patients. Nevertheless, since April 2020, because of the need to adapt inpatient workflows in response to the COVID-19 pandemic, inpatient use of CGM has expanded significantly.

Patients admitted to the hospital may wish to continue using their own personal CGM system during their hospital stay. Published guidelines overwhelmingly support this option (1719) and call for policy development, patient and staff education, and appropriate electronic health record (EHR) documentation to foster its safe and successful use. In this review, we discuss current guidelines, measures of accuracy, steps for successful implementation, current data, and future research recommendations for use of CGM in the adult inpatient population.

Several consensus guidelines have been published (1719), including the Diabetes Technology Society’s (DTS’s) statement titled “Continuous Glucose Monitors and Automated Insulin Dosing Systems in the Hospital Consensus Guideline” (19). The DTS statement was published in September 2020, which was early in the period immediately following the FDA’s announcement of its intended enforcement discretion (13) and before inpatient CGM was adopted more widely and research articles on its expanding use were published. The DTS’s consensus guideline writing group emphasized the need for hospitals to develop policies, processes, and staff education programs and provided basic recommendations for initiating CGM in the hospital. The statement called on hospitals to:

  • Consider the use of inpatient CGM “to reduce the need for frequent nurse contact for POC glucose testing and the use of PPE for patients in isolation with highly contagious infectious diseases (e.g., COVID-19)”

  • Avoid initiating CGM “in patients with severe hypoglycemia or hyperglycemia (i.e., blood glucose levels <40 or >500 mg/dL) or during periods of rapid glucose fluctuations”

Key DTS recommendations are outlined below for hospitals with patients who want to continue using their own CGM system while hospitalized (19).

  • If available, consult with an inpatient diabetes team when continuing or initiating CGM in the inpatient setting.

  • Avoid using CGM for management decisions in the following settings:

    • Severe hypoglycemia or hyperglycemia (i.e., blood glucose levels <40 or >500 mg/dL)

    • Diabetic ketoacidosis (DKA) until glucose is in the CGM measurement range, and then for adjunctive use

    • Rapidly changing glucose levels and fluid/electrolyte shifts

    • Skin infections near the sensor site (or placing sensors in areas with significant edema)

    • Use of vasoactive agents or with poor tissue perfusion

  • Use a CGM checklist for elective procedures during preoperative visits to ensure proper documentation of devices and real-time data reporting.

  • Check the correlation between capillary or venous blood values and CGM values after procedures for noncritically ill patients and between venous/arterial blood glucose and CGM values for critically ill patients to ensure that the patient’s CGM system is functioning well.

  • Use glycemic trend arrows and rates of change to help prevent extreme glycemic excursions, and set alarm thresholds for inpatient glycemic targets to predict hypoglycemia and hyperglycemia.

  • Develop hospital policies to ensure documentation of CGM in the EHR with standard CGM data reports and workflows and that capillary glucose is tested to calibrate CGM systems that require calibration.

Additionally, guidelines recommend that pregnant women continue using CGM during hospitalization (19). It should be noted that, although CGM is not FDA-approved for use in pregnancy, clinical benefits of such use have been demonstrated, including the use of glycemic trend arrows to prevent impending hypoglycemia or hyperglycemia (20). The American Diabetes Association’s 2022 Standards of Medical Care in Diabetes (17) acknowledge the lack of FDA approval and thus do not offer recommendations for use in hospitalized pregnant women until further data on improved glycemic and hospital outcomes are available (17).

The accuracy of CGM is of the utmost importance. It is especially important for clinicians to understand and interpret measures of CGM given the current lack of regulation or standardization of CGM in the inpatient setting.

Several standards exist for determining the accuracy of traditional blood glucose meters, as summarized in Table 2 (2123). The FDA has published standards for the use of prescription POC glucose meters in U.S. hospitals (24). This guidance is similar to that of the International Organization for Standardization (ISO) standard ISO 15197:2013 (21), which is widely used outside of the United States. In addition, the Clinical and Laboratory Standards Institute has published consensus guidelines for hospital POC glucose monitoring (25). These guidelines are specific to glucose meters but do provide a frame of reference for current meter accuracy standards when interpreting CGM accuracy data. Included in Table 2 are the FDA’s recommendations for systems it calls “integrated CGM,” including advanced hybrid closed-loop insulin pump systems that use CGM data to calculate and deliver insulin dose adjustments in real time.

Table 2

Glucose Monitor Accuracy Standards

ISO 15197:2013 (21)FDA Blood Glucose Meter (22)FDA Integrated CGM (23)
Blood Glucose RangeCriterionBlood Glucose RangeCriterionBlood Glucose RangeCriterion**
Home use Home use <70 mg/dL ±15 mg/dL 85% 
Blood glucose ≥100 mg/dL ±15% 95% All blood glucose values* ±15% ±40 mg/dL 98% 
Blood glucose <100 mg/dL ±15 mg/dL 95% All blood glucose values* ±20% 70–180 mg/dL ±15% 70% 
Hospital use Hospital use ±40% 99% 
99% of all measurements must fall within regions A and B of the consensus error grid Blood glucose ≥75 mg/dL ±12% 95% >180 mg/dL ±15% 80% 
Blood glucose <75 mg/dL ±12 mg/dL 95% ±40% 99% 
  All blood glucose values ±20% 87% 
      <70 mg/dL No reference values >180 mg/dL 
      >180 mg/dL No reference values <70 mg/dL 
ISO 15197:2013 (21)FDA Blood Glucose Meter (22)FDA Integrated CGM (23)
Blood Glucose RangeCriterionBlood Glucose RangeCriterionBlood Glucose RangeCriterion**
Home use Home use <70 mg/dL ±15 mg/dL 85% 
Blood glucose ≥100 mg/dL ±15% 95% All blood glucose values* ±15% ±40 mg/dL 98% 
Blood glucose <100 mg/dL ±15 mg/dL 95% All blood glucose values* ±20% 70–180 mg/dL ±15% 70% 
Hospital use Hospital use ±40% 99% 
99% of all measurements must fall within regions A and B of the consensus error grid Blood glucose ≥75 mg/dL ±12% 95% >180 mg/dL ±15% 80% 
Blood glucose <75 mg/dL ±12 mg/dL 95% ±40% 99% 
  All blood glucose values ±20% 87% 
      <70 mg/dL No reference values >180 mg/dL 
      >180 mg/dL No reference values <70 mg/dL 
*

The range of blood glucose values for which the meter has been proven accurate and will provide readings (other than low, high, or error).

**

All FDA integrated CGM Class II measurements must achieve a one-sided confidence level of 95% to be considered statistically significant. The FDA integrated CGM Class II imposes additional requirements on reporting blood glucose rates of change such that false indications of positive and negative rates of change will occur with ≤1% of measurements.

One frequently reported analytic statistical metric for CGM is the mean absolute relative difference (MARD). Error grid analysis results are also frequently reported to describe CGM accuracy.

MARD Values

The MARD is the percentage difference in either direction between the reference method (e.g., Yellow Springs Instrument method) and the sensor value (26). Most current glucose sensors have MARDs ranging from 8.5 to 13.6% over the entire range of glucose levels for outpatient use (69,27) (Table 1).

MARD values can vary based on glucose level, rate of change, sensor wear time, and need for calibration, as well as differences between individuals (26,28,29). The MARD is also dependent on the accuracy of the comparator glucose measurement, which is often a blood glucose meter that carries its own variation in accuracy from its reference instrument (26,30), particularly POC blood glucose measurements in the setting of hypotension (31).

In general, the lower a system’s MARD value is, the greater its accuracy is. As discussed above, there are currently no guidelines or cut-off values for CGM MARD values (32) in the outpatient setting, much less in the inpatient setting. MARD values have also been demonstrated to vary within a day (33) and should not be used as the sole criterion to assess CGM system accuracy (26).

Error Grid Analysis

Error grids are useful to help define how clinically accurate a glucose monitoring device is compared with a gold standard glucose measurement. The Clarke error grid (CEG), Parkes consensus error grid, and surveillance error grid are commonly used error grids (3436). Error grid results are presented both visually and as a percentage of data points falling within grid zones (36). Readings falling into zone A on the CEG are considered clinically accurate; those falling into zone B are less accurate but are considered benign errors. Readings from both zones A and B are considered acceptable for making clinical decisions (Figure 1) (37).

FIGURE 1

Error grid analysis for evaluating the clinical accuracy of blood glucose measurements. Measurements from the blood glucose monitoring device being evaluated are plotted against those from a reference method. Values in zone A (clinically accurate) and zone B (benign estimate errors) are accurate or acceptable. Values in zone C (unnecessary corrections), zone D (dangerous failure to detect and treat), or zone E (erroneous treatment) are potentially dangerous and therefore not acceptable. Reprinted with permission from ref. 37.

FIGURE 1

Error grid analysis for evaluating the clinical accuracy of blood glucose measurements. Measurements from the blood glucose monitoring device being evaluated are plotted against those from a reference method. Values in zone A (clinically accurate) and zone B (benign estimate errors) are accurate or acceptable. Values in zone C (unnecessary corrections), zone D (dangerous failure to detect and treat), or zone E (erroneous treatment) are potentially dangerous and therefore not acceptable. Reprinted with permission from ref. 37.

Close modal

Planning for CGM Use

Key stakeholders should be involved before implementing CGM in the hospital setting. This effort should include the development of a policy, which should reflect input from hospital leadership, diabetes specialists, prescribers, nursing leadership, nurse educators, bedside nurses, pharmacists, laboratory specialists, information technologists, the risk management team, quality champions, and other stakeholders. The policy should encompass information regarding definitions, purpose, scope, exclusion and inclusion criteria, details of procedure, and documentation. Medical leadership and legal and other appropriate institutional departments and committees should review the policy before it is implemented.

Patient Selection and Exclusion

CGM has been used primarily in two types of inpatients: those receiving intravenous (IV) insulin infusions in the intensive care unit (ICU) and step-down units and those on subcutaneous (SQ) insulin protocols in non-ICU settings. In addition, CGM has been used in the surgical setting. During the COVID-19 pandemic, CGM has been used in patients both with and without COVID-19 infection. The literature on these inpatient CGM experiences is summarized in Table 3 (11,12,14,3858).

Table 3

Inpatient CGM Studies

ArticlePopulationStudy DesignCGMAimResults
Critical care studies 
Joshi et al. (38ICU (n = 183) Retrospective Dexcom G6 Feasibility and accuracy 
  • MARD: 14%

  • Pearson correlation coefficient R value:

    • 0.838 for patients without vasopressors and CRRT

    • 0.798 for vasopressor use

    • 0.76 for CRRT use

 
Chow et al. (39ICU, COVID-19, diabetes (n = 30) Retrospective Dexcom G6 Feasibility, utility, and accuracy 
  • Discordance between CGM and POC glucose (>20%) observed in 11 patients, but differences were not considered clinically significant

  • Mean sensor glucose decreased from 235.7 ± 42.1 mg/dL (13.1 ± 2.1 mmol/L) to 202.7 ± 37.6 mg/dL (11.1 ± 2.1 mmol/L) with rtCGM management

  • 14% reduction in mean sensor glucose during rtCGM management compared with pre-rtCGM management (P = 0.0003)

  • Reductions in mean sensor glucose observed in 23 of 30 patients

  • 63% of nurses reported use of rtCGM helped to improve clinical care for severe COVID-19 patients with diabetes

  • 49% of nurses indicated use of rtCGM reduced their use of PPE

 
Faulds et al. (40ICU, COVID-19
(n = 19) 
Retrospective Dexcom G6 Feasibility and accuracy 
  • MARD day 1: 13.9 ± 7.8% (median 11.9%, IQR 3.3–29.4%)

  • MARD day 2: 13.5 ± 8.1% (median 10.6%, IQR 9.0–15.0%)

  • For glucose <100 mg/dL, 62% (23 of 37) paired CGM values were within 20 mg/dL of the POC value

  • For glucose >100 mg/dL, 70% (447 of 639) paired CGM values were within 20% of POC value

  • 71% reduction in frequency of POC compared with the institution’s standard of care

 
Agarwal et al. (41ICU, COVID-19
(n = 11) 
Prospective Dexcom G6 Feasibility and accuracy 
  • Overall MARD: 12.58%

  • 60% reduction in POC use

 
Sadhu et al. (42ICU, COVID-19
(n = 11) 
Retrospective Medtronic Guardian Connect
(n = 6), Dexcom G6 (n = 5) 
Accuracy 
  • Medtronic Guardian Connect:

    • MARD: 13.1%

    • 100% zones A and B on CEG

    • Mean bias: 17.76 mg/dL

  • Dexcom G6:

    • MARD: 11.1%

    • 98% zones A and B on CEG

    • Mean bias: −1.9 mg/dL

 
Bichard et al. (43ICU, DKA (n = 10) Prospective FreeStyle Libre Feasibility 
  • Correlation between POC and CGM glucose: R = 0.84 (P <0.001)

  • Difference between POC- and CGM-driven insulin infusion rates not significantly different from 0, at −0.11 units/hour (95% CI 0.28–0.06 units/hour)

    • Differences were apparent at higher infusion rates

 
Davis et al. (44ICU, COVID-19, diabetes (n = 9) Prospective Dexcom G6 Feasibility 
  • 75.7% of sensor readings >100 mg/dL were within 20% of reference POC

  • 63% reduction in POC compared with standard protocols

  • Negative sensor bias noted during hypoperfusion (i.e., pulseless electrical activity, shock), cardiac arrest and defibrillator use, therapeutic hypothermia protocols, and pronation or position changes causing sensor compression (e.g., bathing)

 
Noncritical care studies 
Davis et al. (45Non-ICU, diabetes
(n = 218) 
Pooled analysis, multicenter Dexcom G6 Accuracy 
  • Overall MARD: 12.8%

  • Proportion of readings meeting:

    • ±15%/15 mg/dL: 68.7%

    • ±20%/20 mg/dL: 81.7%

    • ±30%/30 mg/dL: 93.8%

  • CEG analysis: 98.7% values in zones A and B

 
Fortmann et al. (12Non-ICU, type 2 diabetes (n = 110) RCT Dexcom G6 Feasibility CGM group compared with usual care:
  • Significantly lower mean glucose (MΔ: −18.5 mg/dL)

  • Significantly lower percentage of time in hyperglycemia >250 mg/dL (−11.41%)

  • Significantly higher time in range 70–250 mg/dL (+11.26%)

  • P <0.05 for all

 
Galindo et al. (14Non-ICU, diabetes
(n = 97) 
Prospective FreeStyle Libre Pro Accuracy 
  • Mean daily glucose was significantly higher by POC (188.9 ± 37.3 vs. 176.1 ± 46.9 mg/dL)

  • Estimated mean difference of 12.8 mg/dL (95% CI 8.3–17.2 mg/dL)

  • Proportions of patients with glucose readings detected by POC versus CGM (P <0.001):

    • <70 mg/dL (14 vs. 56%)

    • <54 mg/dL (4.1 vs. 36%)

  • MARD (glucose 70–250 mg/dL): 14.8% (range 11.4–16.7%)

  • MARD (glucose 51–69 mg/dL): 28.0%

  • Proportions of glucose readings within:

    • ±15%/15 mg/dL: 62%

    • ±20%/20 mg/dL: 76%

    • ±30%/30 mg/dL: 91%

  • Error grid analysis: 98.8% of glucose pairs within zones A and B

 
Wright et al. (46Non-ICU, diabetes
(n = 77) 
Prospective FreeStyle Libre 14-day
(n = 43), FreeStyle Libre 2
(n = 34) 
Accuracy Libre 2:
  • MARD: 17.7%

  • CEG analysis: 98.8% of all values within zone A or B

    Libre 14-day:

    • MARD: 21.4%

    • CEG analysis: 98.8% of all values within zone A or B

 
Singh et al. (11Non-ICU, diabetes
(n = 72) 
RCT, single center Dexcom G6 Hypoglycemia prevention When compared with the POC group, the rtCGM GTS group experienced:
  • Fewer hypoglycemic events (<70 mg/dL) per patient (0.67 [95% CI 0.34–1.30] vs. 1.69 [95% CI 1.11–2.58], P = 0.024)

  • Fewer clinically significant hypoglycemic events (<54 mg/dL) per patient (0.08 [95% CI 0.03–0.26] vs. 0.75 [95% CI 0.51–1.09], P = 0.003)

  • A lower percentage of time spent below range (<70 mg/dL: 0.40% [95% CI 0.18–0.92%] vs. 1.88% [95% CI 1.26–2.81%], P = 0.002) and <54 mg/dL: 0.05% [95% CI 0.01–0.43%] vs. 0.82% [0.47–1.43%], P = 0.017)

 
Murray-Bachmann et al. (47Noncritical care, diabetes
(n = 52) 
Prospective Freestyle Libre Pro Accuracy FreeStyle Libre Pro versus serum glucose:
  • MARD: 13.2%

  • R2 = 0.89

  • Proportion of values with >15% difference between POC and CGM: 36%

  • Proportion of pairs that would have yielded results requiring the same intervention (i.e., treatment for hypoglycemia or hyperglycemia): 89%

    FreeStyle Libre Pro versus POC:

  • MARD: 15.6%

  • R2 = 0.83

  • Proportion of pairs with difference >15% between POC and CGM: 42%

  • Proportion of pairs that would have yielded results requiring the same intervention (treatment of hypoglycemia or hyperglycemia): 85%

 
Gómez et al. (48Non-ICU, COVID-19, diabetes
(n = 38) 
Prospective Medtronic iPro2 Feasibility and accuracy 
  • MARD: 12.9%

  • Proportion of POC/CGM pairs with difference >15 mg/dL: 0.6%

  • Pearson correlation analysis: R = 0.79

  • CEG analysis: 91.9% of values within zone A or B

  • CGM detected a higher number of hypoglycemic episodes than POC (55 vs. 12, P <0.01)

 
Sweeney et al. (49Post-ICU, cardiac surgery
(n = 11) 
Prospective Dexcom G6 Feasibility and accuracy 
  • Overall MARD: 14.80%

  • MARD for patients with eGFR >20 mL/min/1.73 m2: 12.13 ± 7.67%

  • MARD for patients with eGFR <20 mL/min/1.73 m2: 21.27 ± 20.81%

  • CEG analysis: 97% of CGM values within zones A and B

  • Overall MARD for first 24 hours: 15.42 ± 14.44%

  • Overall MARD beyond 24 hours: 14.54 ± 13.21%

  • MARD with eGFR >20 mL/min/1.73 m2:

    • For first 24 hours: 12.80 ± 7.85%

    • Beyond 24 hours: 11.86 ± 7.64%

 
Baker et al. (50Non-ICU, COVID-19, diabetes
(n = 10) 
Observational Dexcom G6 Accuracy 
  • MARD: 10.3%

  • CEG analysis: 99.2% of points in zones A and B; SEG analysis: 89.1% of points in the lowest risk category

  • For 25% of the rtCGM values, discordances between rtCGM and POC values would likely have resulted in different insulin doses (by 1–3 units)

 
Reutrakul et al. (51Non-ICU, COVID-19
(n = 9) 
Prospective Dexcom G6 Feasibility and accuracy 
  • MARD: 9.77%

  • Mean bias: 2.45 mg/dL

  • Correlation coefficient: R = 0.927

  • CEG analysis: zone A: 84.8%; zones A or B: 100%

 
Mixed studies 
Price et al. (52Diabetes on basal-bolus insulin
(n = 20),
DKA on IV insulin infusion (n = 16) 
Prospective FreeStyle Libre Pro Accuracy 
  • Overall MARD: 22.3%

  • CGM consistently reported lower glucose values than POC

  • Mean difference: on basal-bolus insulin, 44.8 mg/dL; on IV insulin infusion, 19.7 mg/dL

  • Absolute difference in insulin dose: basal-bolus insulin 1.34 units, IV insulin infusion 0.74 units

 
Longo et al. (53COVID-19, ICU
(n = 10), non-ICU
(n = 18) 
Prospective Dexcom G6 Feasibility and accuracy 
  • MARD:

    • All patients (POC or laboratory serum glucose): 13.2%

      • ▪ Critical care: 12.1%

      • ▪ General floor: 14%

    • All patients POC versus CGM: 13.9%

    • All patients laboratory serum glucose versus CGM: 10.9%

  • Percentage of values within 15%/15 mg/dL:

    • Serum laboratory glucose versus CGM: 93%

    • POC versus CGM: 87.6%

  • CEG analysis: percentage of pairs in no-risk zone:

    • Serum laboratory versus CGM: 86.1%

    • POC versus CGM: 82.5%

 
Surgery/radiology studies 
Migdal et al. (54Non-ICU patients with diabetes undergoing radiological procedures (n = 49) Prospective Dexcom G6 Accuracy of CGM before and after radiological procedure (X-rays [n = 28], CT scan [n = 13], catheterization/angiography
[n = 8]) 
  • No significant difference in mean CGM blood glucose or %CV for all imaging procedures combined (blood glucose mean Δ −7.7 ± 26.0 mg/dL, P = 0.051; %CV 18.0% before versus 19.1% after, P = 0.65)

  • Overall MARD: 13.3% before and 12.7% after imaging

  • Overall proportion of glucose values before and after imaging within:

    • ±15%/15 mg/dL: 69 versus 68%

    • ±20%/20 mg/dL: 80 versus 82%

    • ±30%/30 mg/dL: 94 versus 93%

  • CEG analysis: 98.1% of glucose values before imaging and 99.7% after imaging in zones A or B

 
Sugiyama et al. (55Healthy volunteers (n = 15), neurosurgery patients (n = 15), cardiac surgery patients (n = 15) Prospective Medtronic 620 (Japan) Accuracy 
  • CEG analysis: proportion in zone A:

    • Healthy volunteers: 82.7%

    • Neurosurgery patients: 86.8%

    • Cardiac surgery patients: 65.3%

      • ▪ Post-operative day 1: 85%

      • ▪ Post-operative day 3: 86.3%

  • Mean biases:

    • Healthy volunteers: −2.1 mg/dL

    • Neurosurgery patients: −8.3 mg/dL

    • Cardiac surgery: −23.5 mg/dL

 
Tripyla et al. (56Patients with prediabetes or diabetes undergoing abdominal surgery of >2 hours (n = 20) Prospective Dexcom G6 Accuracy 
  • Perioperative period MARD: 12.7 ± 8.7%

  • 67.4% of sensor readings within ISO 15197:2013 limits

  • CEG analysis: readings in zones A or B: 99.2%, in zone A: 78.8%, in zone B: 20.4%

  • Median perioperative sensor availability: 98.6% (IQR 95.9–100.0%)

  • No clinically significant adverse events

 
Perez-Guzman et al. (57Operating room and cardiac ICU patients (n = 15) Prospective Dexcom G6 Accuracy 
  • MARD: 12.9%

  • CEG analysis: 98.6% of glucose values in zones A or B (83.2% in zone A)

  • Proportion of sensor glucose values within:

    • ±15%/15 mg/dL: 69%

    • ±20%/20 mg/dL: 82%

    • ±30%/30 mg/dL: 94%

 
Nair et al. (58Non-COVID
diabetes, surgery patients
(n = 10) 
Prospective Dexcom G6 Accuracy 
  • MARD: 9.4%

  • Correlation coefficient: 0.7

  • Mean bias: −0.37 mg/dL

  • SEG analysis: 89% of paired glucose values within the no-risk zone

 
ArticlePopulationStudy DesignCGMAimResults
Critical care studies 
Joshi et al. (38ICU (n = 183) Retrospective Dexcom G6 Feasibility and accuracy 
  • MARD: 14%

  • Pearson correlation coefficient R value:

    • 0.838 for patients without vasopressors and CRRT

    • 0.798 for vasopressor use

    • 0.76 for CRRT use

 
Chow et al. (39ICU, COVID-19, diabetes (n = 30) Retrospective Dexcom G6 Feasibility, utility, and accuracy 
  • Discordance between CGM and POC glucose (>20%) observed in 11 patients, but differences were not considered clinically significant

  • Mean sensor glucose decreased from 235.7 ± 42.1 mg/dL (13.1 ± 2.1 mmol/L) to 202.7 ± 37.6 mg/dL (11.1 ± 2.1 mmol/L) with rtCGM management

  • 14% reduction in mean sensor glucose during rtCGM management compared with pre-rtCGM management (P = 0.0003)

  • Reductions in mean sensor glucose observed in 23 of 30 patients

  • 63% of nurses reported use of rtCGM helped to improve clinical care for severe COVID-19 patients with diabetes

  • 49% of nurses indicated use of rtCGM reduced their use of PPE

 
Faulds et al. (40ICU, COVID-19
(n = 19) 
Retrospective Dexcom G6 Feasibility and accuracy 
  • MARD day 1: 13.9 ± 7.8% (median 11.9%, IQR 3.3–29.4%)

  • MARD day 2: 13.5 ± 8.1% (median 10.6%, IQR 9.0–15.0%)

  • For glucose <100 mg/dL, 62% (23 of 37) paired CGM values were within 20 mg/dL of the POC value

  • For glucose >100 mg/dL, 70% (447 of 639) paired CGM values were within 20% of POC value

  • 71% reduction in frequency of POC compared with the institution’s standard of care

 
Agarwal et al. (41ICU, COVID-19
(n = 11) 
Prospective Dexcom G6 Feasibility and accuracy 
  • Overall MARD: 12.58%

  • 60% reduction in POC use

 
Sadhu et al. (42ICU, COVID-19
(n = 11) 
Retrospective Medtronic Guardian Connect
(n = 6), Dexcom G6 (n = 5) 
Accuracy 
  • Medtronic Guardian Connect:

    • MARD: 13.1%

    • 100% zones A and B on CEG

    • Mean bias: 17.76 mg/dL

  • Dexcom G6:

    • MARD: 11.1%

    • 98% zones A and B on CEG

    • Mean bias: −1.9 mg/dL

 
Bichard et al. (43ICU, DKA (n = 10) Prospective FreeStyle Libre Feasibility 
  • Correlation between POC and CGM glucose: R = 0.84 (P <0.001)

  • Difference between POC- and CGM-driven insulin infusion rates not significantly different from 0, at −0.11 units/hour (95% CI 0.28–0.06 units/hour)

    • Differences were apparent at higher infusion rates

 
Davis et al. (44ICU, COVID-19, diabetes (n = 9) Prospective Dexcom G6 Feasibility 
  • 75.7% of sensor readings >100 mg/dL were within 20% of reference POC

  • 63% reduction in POC compared with standard protocols

  • Negative sensor bias noted during hypoperfusion (i.e., pulseless electrical activity, shock), cardiac arrest and defibrillator use, therapeutic hypothermia protocols, and pronation or position changes causing sensor compression (e.g., bathing)

 
Noncritical care studies 
Davis et al. (45Non-ICU, diabetes
(n = 218) 
Pooled analysis, multicenter Dexcom G6 Accuracy 
  • Overall MARD: 12.8%

  • Proportion of readings meeting:

    • ±15%/15 mg/dL: 68.7%

    • ±20%/20 mg/dL: 81.7%

    • ±30%/30 mg/dL: 93.8%

  • CEG analysis: 98.7% values in zones A and B

 
Fortmann et al. (12Non-ICU, type 2 diabetes (n = 110) RCT Dexcom G6 Feasibility CGM group compared with usual care:
  • Significantly lower mean glucose (MΔ: −18.5 mg/dL)

  • Significantly lower percentage of time in hyperglycemia >250 mg/dL (−11.41%)

  • Significantly higher time in range 70–250 mg/dL (+11.26%)

  • P <0.05 for all

 
Galindo et al. (14Non-ICU, diabetes
(n = 97) 
Prospective FreeStyle Libre Pro Accuracy 
  • Mean daily glucose was significantly higher by POC (188.9 ± 37.3 vs. 176.1 ± 46.9 mg/dL)

  • Estimated mean difference of 12.8 mg/dL (95% CI 8.3–17.2 mg/dL)

  • Proportions of patients with glucose readings detected by POC versus CGM (P <0.001):

    • <70 mg/dL (14 vs. 56%)

    • <54 mg/dL (4.1 vs. 36%)

  • MARD (glucose 70–250 mg/dL): 14.8% (range 11.4–16.7%)

  • MARD (glucose 51–69 mg/dL): 28.0%

  • Proportions of glucose readings within:

    • ±15%/15 mg/dL: 62%

    • ±20%/20 mg/dL: 76%

    • ±30%/30 mg/dL: 91%

  • Error grid analysis: 98.8% of glucose pairs within zones A and B

 
Wright et al. (46Non-ICU, diabetes
(n = 77) 
Prospective FreeStyle Libre 14-day
(n = 43), FreeStyle Libre 2
(n = 34) 
Accuracy Libre 2:
  • MARD: 17.7%

  • CEG analysis: 98.8% of all values within zone A or B

    Libre 14-day:

    • MARD: 21.4%

    • CEG analysis: 98.8% of all values within zone A or B

 
Singh et al. (11Non-ICU, diabetes
(n = 72) 
RCT, single center Dexcom G6 Hypoglycemia prevention When compared with the POC group, the rtCGM GTS group experienced:
  • Fewer hypoglycemic events (<70 mg/dL) per patient (0.67 [95% CI 0.34–1.30] vs. 1.69 [95% CI 1.11–2.58], P = 0.024)

  • Fewer clinically significant hypoglycemic events (<54 mg/dL) per patient (0.08 [95% CI 0.03–0.26] vs. 0.75 [95% CI 0.51–1.09], P = 0.003)

  • A lower percentage of time spent below range (<70 mg/dL: 0.40% [95% CI 0.18–0.92%] vs. 1.88% [95% CI 1.26–2.81%], P = 0.002) and <54 mg/dL: 0.05% [95% CI 0.01–0.43%] vs. 0.82% [0.47–1.43%], P = 0.017)

 
Murray-Bachmann et al. (47Noncritical care, diabetes
(n = 52) 
Prospective Freestyle Libre Pro Accuracy FreeStyle Libre Pro versus serum glucose:
  • MARD: 13.2%

  • R2 = 0.89

  • Proportion of values with >15% difference between POC and CGM: 36%

  • Proportion of pairs that would have yielded results requiring the same intervention (i.e., treatment for hypoglycemia or hyperglycemia): 89%

    FreeStyle Libre Pro versus POC:

  • MARD: 15.6%

  • R2 = 0.83

  • Proportion of pairs with difference >15% between POC and CGM: 42%

  • Proportion of pairs that would have yielded results requiring the same intervention (treatment of hypoglycemia or hyperglycemia): 85%

 
Gómez et al. (48Non-ICU, COVID-19, diabetes
(n = 38) 
Prospective Medtronic iPro2 Feasibility and accuracy 
  • MARD: 12.9%

  • Proportion of POC/CGM pairs with difference >15 mg/dL: 0.6%

  • Pearson correlation analysis: R = 0.79

  • CEG analysis: 91.9% of values within zone A or B

  • CGM detected a higher number of hypoglycemic episodes than POC (55 vs. 12, P <0.01)

 
Sweeney et al. (49Post-ICU, cardiac surgery
(n = 11) 
Prospective Dexcom G6 Feasibility and accuracy 
  • Overall MARD: 14.80%

  • MARD for patients with eGFR >20 mL/min/1.73 m2: 12.13 ± 7.67%

  • MARD for patients with eGFR <20 mL/min/1.73 m2: 21.27 ± 20.81%

  • CEG analysis: 97% of CGM values within zones A and B

  • Overall MARD for first 24 hours: 15.42 ± 14.44%

  • Overall MARD beyond 24 hours: 14.54 ± 13.21%

  • MARD with eGFR >20 mL/min/1.73 m2:

    • For first 24 hours: 12.80 ± 7.85%

    • Beyond 24 hours: 11.86 ± 7.64%

 
Baker et al. (50Non-ICU, COVID-19, diabetes
(n = 10) 
Observational Dexcom G6 Accuracy 
  • MARD: 10.3%

  • CEG analysis: 99.2% of points in zones A and B; SEG analysis: 89.1% of points in the lowest risk category

  • For 25% of the rtCGM values, discordances between rtCGM and POC values would likely have resulted in different insulin doses (by 1–3 units)

 
Reutrakul et al. (51Non-ICU, COVID-19
(n = 9) 
Prospective Dexcom G6 Feasibility and accuracy 
  • MARD: 9.77%

  • Mean bias: 2.45 mg/dL

  • Correlation coefficient: R = 0.927

  • CEG analysis: zone A: 84.8%; zones A or B: 100%

 
Mixed studies 
Price et al. (52Diabetes on basal-bolus insulin
(n = 20),
DKA on IV insulin infusion (n = 16) 
Prospective FreeStyle Libre Pro Accuracy 
  • Overall MARD: 22.3%

  • CGM consistently reported lower glucose values than POC

  • Mean difference: on basal-bolus insulin, 44.8 mg/dL; on IV insulin infusion, 19.7 mg/dL

  • Absolute difference in insulin dose: basal-bolus insulin 1.34 units, IV insulin infusion 0.74 units

 
Longo et al. (53COVID-19, ICU
(n = 10), non-ICU
(n = 18) 
Prospective Dexcom G6 Feasibility and accuracy 
  • MARD:

    • All patients (POC or laboratory serum glucose): 13.2%

      • ▪ Critical care: 12.1%

      • ▪ General floor: 14%

    • All patients POC versus CGM: 13.9%

    • All patients laboratory serum glucose versus CGM: 10.9%

  • Percentage of values within 15%/15 mg/dL:

    • Serum laboratory glucose versus CGM: 93%

    • POC versus CGM: 87.6%

  • CEG analysis: percentage of pairs in no-risk zone:

    • Serum laboratory versus CGM: 86.1%

    • POC versus CGM: 82.5%

 
Surgery/radiology studies 
Migdal et al. (54Non-ICU patients with diabetes undergoing radiological procedures (n = 49) Prospective Dexcom G6 Accuracy of CGM before and after radiological procedure (X-rays [n = 28], CT scan [n = 13], catheterization/angiography
[n = 8]) 
  • No significant difference in mean CGM blood glucose or %CV for all imaging procedures combined (blood glucose mean Δ −7.7 ± 26.0 mg/dL, P = 0.051; %CV 18.0% before versus 19.1% after, P = 0.65)

  • Overall MARD: 13.3% before and 12.7% after imaging

  • Overall proportion of glucose values before and after imaging within:

    • ±15%/15 mg/dL: 69 versus 68%

    • ±20%/20 mg/dL: 80 versus 82%

    • ±30%/30 mg/dL: 94 versus 93%

  • CEG analysis: 98.1% of glucose values before imaging and 99.7% after imaging in zones A or B

 
Sugiyama et al. (55Healthy volunteers (n = 15), neurosurgery patients (n = 15), cardiac surgery patients (n = 15) Prospective Medtronic 620 (Japan) Accuracy 
  • CEG analysis: proportion in zone A:

    • Healthy volunteers: 82.7%

    • Neurosurgery patients: 86.8%

    • Cardiac surgery patients: 65.3%

      • ▪ Post-operative day 1: 85%

      • ▪ Post-operative day 3: 86.3%

  • Mean biases:

    • Healthy volunteers: −2.1 mg/dL

    • Neurosurgery patients: −8.3 mg/dL

    • Cardiac surgery: −23.5 mg/dL

 
Tripyla et al. (56Patients with prediabetes or diabetes undergoing abdominal surgery of >2 hours (n = 20) Prospective Dexcom G6 Accuracy 
  • Perioperative period MARD: 12.7 ± 8.7%

  • 67.4% of sensor readings within ISO 15197:2013 limits

  • CEG analysis: readings in zones A or B: 99.2%, in zone A: 78.8%, in zone B: 20.4%

  • Median perioperative sensor availability: 98.6% (IQR 95.9–100.0%)

  • No clinically significant adverse events

 
Perez-Guzman et al. (57Operating room and cardiac ICU patients (n = 15) Prospective Dexcom G6 Accuracy 
  • MARD: 12.9%

  • CEG analysis: 98.6% of glucose values in zones A or B (83.2% in zone A)

  • Proportion of sensor glucose values within:

    • ±15%/15 mg/dL: 69%

    • ±20%/20 mg/dL: 82%

    • ±30%/30 mg/dL: 94%

 
Nair et al. (58Non-COVID
diabetes, surgery patients
(n = 10) 
Prospective Dexcom G6 Accuracy 
  • MARD: 9.4%

  • Correlation coefficient: 0.7

  • Mean bias: −0.37 mg/dL

  • SEG analysis: 89% of paired glucose values within the no-risk zone

 

%CV, coefficient of variation; eGFR, estimated glomerular filtration rate; IQR, interquartile range; RCT, randomized controlled trial; SEG, surveillance error grid.

It is important for hospitals to review the most current published data and consensus guidelines to determine the most appropriate patients for CGM because the field is rapidly evolving. Current guidelines for patient selection and some exclusions to CGM use in the hospital are summarized above. With regard to exclusions, DTS guidelines recommend avoiding the use of inpatient CGM in patients with blood glucose levels <40 or >500 mg/dL and those with DKA, in the setting of rapidly changing glucose levels and fluid and electrolyte shifts, and in those with significant edema requiring vasoactive agents or with poor perfusion (15) until more research is available. Several small studies have reported on the feasibility of CGM use in the critical care setting during the COVID-19 pandemic (Table 3).

Interfering Substances

Various interfering substances can affect the accuracy of each CGM system (Table 1). It is important to be aware of these substances and refer to the user guides of each CGM system for more information. None of the current CGM systems, with the exception of an implantable CGM sensor (27), are labeled for use in the setting of X-rays, computed tomography (CT) scans, MRI, diathermy, or radiation therapy (19). Guidelines recommend removing CGM sensors before CT scans and MRI (19); however, some small studies did not show any effect of CT scan or even simulated MRI radiation on the accuracy of the Dexcom G6 (53,59) or the FreeStyle Libre Pro (60) CGM systems.

Education on Placement of CGM Sensors

Detailed education on CGM sensor insertion and setup procedures is essential for successful inpatient CGM implementation. This education can be provided through in-person training, videos, or tip sheets. Bedside nurses, or preferentially, if resources allow, specially trained CGM champions can be educated on CGM sensor insertion (10,14,48,53). Recommended sensor placement depends on the specifications of the specific CGM system being used. However, the Dexcom system, which is approved for use on the abdomen, was successfully worn with alternative arm placement (because of issues of prone positioning of COVID-19 patients) (44,53).

Protocols for CGM Use and POC Blood Glucose Correlation

For institutions already using CGM for glucose monitoring and insulin dosing, there is significant variability in protocols with regard to the frequency (if any) of confirmatory glucose testing (12,53). Both venous/arterial and POC capillary glucose monitoring have been used for correlation measurements (44,53). Laboratory glucose testing has demonstrated better correlation to CGM than POC capillary glucose testing (53). Although limited in volume and quality, data thus far have indicated the feasibility of CGM compared with POC testing in general ward settings (48) and in the ICU (14).

In most studies, acceptable correlation has been defined as a difference of <20% or <20 mg/dL between POC and CGM values if glucose is <100 mg/dL (14). At one institution, nursing education deemed the 20% calculation too difficult for nurses to calculate in the setting of the COVID-19 pandemic; therefore, an arbitrary difference of >35 mg/dL between POC and CGM was used for glucose levels >100 mg/dL (representing 20% variation from a glucose level of 175 mg/dL) (53).

Recommendations for when to use POC blood glucose testing to determine a patient’s current glucose level are similar to those reported in a 2021 review by Perez-Guzman et al. (57). These included instances in which:

  • Glucose values are <85 or >300 mg/dL

  • Hypoglycemia symptoms occur

  • Glucose values and/or glycemic trend arrows are not present on the monitor

  • A blood drop symbol appears on the monitor (7)

  • Hemodynamic instability occurs

  • A patient is in the immediate postoperative period (57)

Several different protocols for use of CGM in the settings of IV insulin infusion and SQ insulin dosing have been reported. The most frequently reported protocols are as follows (12,38,40,41,44,45):

  • Initial start-up (after the warm-up period is complete)

    • Patients must have two consecutive sensor-meter pairs ∼1 hour apart in which CGM and POC values differ by <20% or <20 mg/dL if glucose is <100 mg/dL, or

    • CGM data are not used for the first 24 hours and then must meet acceptable correlation criteria, as described above.

  • Maintenance

    • POC blood glucose and CGM glucose values must meet the above criteria every 2–6 hours for continued use in the setting of IV insulin infusion.

    • POC blood glucose and CGM glucose values must meet above criteria every 6–12 hours in the setting of SQ insulin dosing.

CGM System Setup

Hospitals using inpatient CGM must adapt the existing technology, which was designed for outpatient use, to the inpatient setting. Setup procedures depend on the type of CGM system used (rtCGM vs. isCGM), as well as the layout of the nursing unit and choice of devices to receive CGM data.

For rtCGM, some institutions have kept the receiver, smartphone, or reader outside of the patient’s room to prevent nursing exposure and have reported good connection (38,53). Many institutions choose to use smartphones and also to set up smartphones or tablets at the nurses station equipped with real-time “following” applications (e.g., the Dexcom Follow app) to provide real-time alerts at a distanced location (53). For isCGM, nurses or patients have been directed to scan with a stand-alone reader or smartphone at the bedside.

If smartphones are used as receivers, de-identified dummy accounts must be created to provide delayed remote monitoring from Cloud-based reporting software. If a stand-alone receiver is used, data stored in the Cloud-based data-reporting software must be manually downloaded for review.

EHR Integration

Integration of CGM data into EHR systems to allow for remote and immediate access to CGM data are integral to the development of future inpatient CGM programs. Currently, CGM data can reside on a Cloud-based website, which increases concern regarding the safety of these data and the need for protection from cybersecurity breaches (57). Because CGM systems sample glucose values every 1–5 minutes, the amount of storage space needed to save every data point is extraordinarily large. CGM manufacturers need to work with diabetes care teams in conjunction with EHR and middle-ware software vendors to integrate the most useful CGM data into existing EHR platforms.

In some EHR systems, CGM has been manually documented in the glucose flow sheet with a separate column for POC entry and correlation between the two devices. Separate EHR reports that include a patient’s demographics, diagnosis, length of stay, disposition, medications, average placement time for CGM, and data on vasopressors and continuous renal replacement therapy (CRRT) have also been created (38).

Espinoza et al. (61) proposed the development of a consortium of key stakeholders known as the Integration of Continuous Glucose Monitoring Data into the Electronic Health Record (iCoDE) Project, in collaboration with the DTS. The intention of this group would be to 1) develop standards for the exchange, classification, and mapping of CGM data and 2) develop policies to guide the integration of CGM data into EHR systems.

Using Trend Arrows

For outpatients using CGM, several tools have been developed to integrate trend arrows into insulin dosing algorithms (6264). Very few studies using trend arrows in the inpatient setting have been published. Scripps Hospital reported on a remote telemetry service that monitors CGM data, including trend arrows, and reports significant trends to nursing staff. This study demonstrated lower average glucose, improved time in range, and rare hypoglycemia (12).

CGM in Critical Care

Multiple studies have been conducted in ICU patients. Most of these were small and focused on feasibility and accuracy (Table 3). The MARDs reported in these studies ranged from 11.1 to 14%. The use of POC blood glucose measurements was reported to be reduced by 60–71% (44). Many of these studies demonstrated the feasibility of CGM use during the critical time of the COVID-19 pandemic, but more robust studies are clearly needed in this complex patient population.

Davis et al. (44) performed a proof-of-concept study using CGM and a computerized decision-support tool for IV insulin infusions in ICU patients. Negative sensor bias was noted during periods of hypoperfusion (i.e., with pulseless electrical activity or shock), therapeutic hypothermia, and position changes resulting in sensor compression, with a sharp decline in values and loss of signal during cardiac arrest and defibrillation (44).

Joshi et al. (38) reported on a retrospective study of 165 patients in the ICU, including patients on vasopressors and CRRT, with an overall MARD of 14% (POC blood glucose vs. CGM) and a Pearson correlation coefficient of R = 0.89 (in patients requiring CRRT/hemodialysis, R = 0.798, and for patients using vasopressors, R = 0.798).

There are more studies in non-ICU settings, many of which are case studies and pooled analyses. These studies reported wide variations in MARD ranging from 9.77 to 28% (Table 3). There was also significant variation in study design and patient selection criteria. Singh et al. (11) published data from a randomized controlled trial involving 72 patients using a glucose telemetry system (GTS) that integrated CGM with the EHR and demonstrated fewer hypoglycemic episodes per patient using GTS (GTS 0.67 [95% CI 0.34–1.3] vs. usual care 1.69 [95% CI 1.11–2.58], P = 0.024). Assessment of the quality of comparator glucose measures was also identified as important in a study by Longo et al. (53), with an overall MARD of 13.9% when using POC blood glucose measurement as a comparator versus 10.9% when CGM was measured against laboratory-measured glucose values.

There have been very few studies assessing CGM in the setting of surgical and radiological procedures. Although an earlier study did not show a good correlation during cardiac surgery (55), a recent study by Perez-Guzman et al. (57) involving 15 patients undergoing coronary artery bypass graft documented a MARD of 12.7%. There was some loss of sensor activity during surgery, but six sensors recovered completely after surgery and maintained accuracy despite vasopressor use.

CGM Satisfaction

CGM has been shown to save nurses time (17 vs. 36 minutes per day spent on glucose management, P <0.001) (65). In addition, in a survey of 112 medical staff using inpatient CGM, 92% reported that they felt CGM reduced their own exposure to COVID-19 (38). Dillman et al. (66) reported that 95% of nurses found CGM to be “useful,” specifically by anticipating hyperglycemia and hypoglycemia, and 64% reported CGM to be time-saving. Faulds et al. (67) also reported positive experiences from nursing staff, including the ability to continuously monitor glucose levels. However, they did find some aspects of the protocol, such as setting up smartphones to receive data, to be lengthy.

As Winston Churchill famously said, “Never let a good crisis go to waste.” The COVID-19 pandemic has allowed for the rapid expansion of inpatient CGM use. Although there is now a growing number of reports of successful integration of CGM in the inpatient setting with reasonable accuracy data, the need for further research focused on safety, accuracy, and outcomes is clear. Moving forward, it will be important to consider the limitations of CGM such as the lag time between interstitial fluid and blood glucose levels, which can vary, particularly in settings of rapidly changing glucose levels (16). Appropriate patient selection is key, and subgroup analyses of accuracy and outcomes data are also important, particularly in critically ill patients requiring vasopressors and CRRT and in states of hypoperfusion. As the DTS guidelines note, studies of the impact of lag time should be done, and standardized protocols for when confirmatory POC blood glucose testing is needed should be created (14,15).

Cost-benefit outcomes data analyses will be crucial for adoption of CGM technology by hospitals and payers. As discussed above, successful implementation of any inpatient CGM program requires extensive staff training in the placement, maintenance, and troubleshooting of CGM systems, as well as clear understanding and interpretation of CGM data.

Future protocols that integrate glycemic trend arrows into insulin dosing decisions may provide opportunity for improvement in inpatient glycemic management. In addition, education of nurses, nurse practitioners, physician associates, pharmacists, and physicians is crucial not only for the successful implementation of inpatient CGM, but also to foster its optimal use and actionable interpretation of the data it provides.

Many institutions have successfully integrated CGM into inpatient glucose management to save PPE, limit nursing exposure, and possibly improve glycemic outcomes during the COVID-19 pandemic. More expansive adoption of CGM may be a much-needed paradigm shift in inpatient glycemic management compared with occasional POC blood glucose checks. Further careful study of the accuracy and outcomes of CGM use in the hospital will be crucial to facilitate its widespread adoption and regulatory approval.

Acknowledgment

The authors thank Laurie Schwing of University of Pittsburgh Medical Assistance for her editorial assistance.

Duality of Interest

R.J. is a speaker for Medscape. No other potential conflicts of interest relevant to this article were reported.

Author Contributions

Both authors researched data and wrote and edited the manuscript. Both authors are the guarantors of this work and, as such, take responsibility for the integrity of the data presentation and accuracy of the review.

1.
Zhao
Q
,
Zhou
F
,
Zhang
Y
,
Zhou
X
,
Ying
C
.
Fasting plasma glucose variability levels and risk of adverse outcomes among patients with type 2 diabetes: a systematic review and meta-analysis
.
Diabetes Res Clin Pract
2019
;
148
:
23
31
2.
Garg
R
,
Hurwitz
S
,
Turchin
A
,
Trivedi
A
.
Hypoglycemia, with or without insulin therapy, is associated with increased mortality among hospitalized patients
.
Diabetes Care
2013
;
36
:
1107
1110
3.
Turchin
A
,
Matheny
ME
,
Shubina
M
,
Scanlon
JV
,
Greenwood
B
,
Pendergrass
ML
.
Hypoglycemia and clinical outcomes in patients with diabetes hospitalized in the general ward
.
Diabetes Care
2009
;
32
:
1153
1157
4.
Guo
W
,
Li
M
,
Dong
Y
, et al
.
Diabetes is a risk factor for the progression and prognosis of COVID-19
.
Diabetes Metab Res Rev
2020
;
36
:
e3319
5.
Zhu
L
,
She
ZG
,
Cheng
X
, et al
.
Association of blood glucose control and outcomes in patients with COVID-19 and pre-existing type 2 diabetes
.
Cell Metab
2020
;
31
:
1068
1077.e3
6.
Abbott Diabetes Care
.
FreeStyle Libre 14 Day Glucose Monitoring System User’s Manual
.
7.
Abbott Diabetes Care
.
FreeStyle Libre 2 flash glucose monitoring system provider information
.
8.
Dexcom
.
Dexcom G6 Continuous Glucose Monitoring System User Guide
.
10.
Spanakis
EK
,
Levitt
DL
,
Siddiqui
T
, et al
.
The effect of continuous glucose monitoring in preventing inpatient hypoglycemia in general wards: the glucose telemetry system
.
J Diabetes Sci Technol
2018
;
12
:
20
25
11.
Singh
LG
,
Satyarengga
M
,
Marcano
I
, et al
.
Reducing inpatient hypoglycemia in the general wards using real-time continuous glucose monitoring: the glucose telemetry system, a randomized clinical trial
.
Diabetes Care
2020
;
43
:
2736
2743
12.
Fortmann
AL
,
Spierling Bagsic
SR
,
Talavera
L
, et al
.
Glucose as the fifth vital sign: a randomized controlled trial of continuous glucose monitoring in a non-ICU hospital setting
.
Diabetes Care
2020
;
43
:
2873
2877
13.
U.S. Food Drug Administration
.
Enforcement policy for non-invasive remote monitoring devices used to support patient monitoring during the coronavirus disease 2019 (COVID-19) public health emergency (revised): guidance for industry and Food and Drug Administration staff
.
14.
Galindo
RJ
,
Aleppo
G
,
Klonoff
DC
, et al
.
Implementation of continuous glucose monitoring in the hospital: emergent considerations for remote glucose monitoring during the COVID-19 pandemic
.
J Diabetes Sci Technol
2020
;
14
:
822
832
15.
Basu
A
,
Dube
S
,
Slama
M
, et al
.
Time lag of glucose from intravascular to interstitial compartment in humans
.
Diabetes
2013
;
62
:
4083
4087
16.
Schmelzeisen-Redeker
G
,
Schoemaker
M
,
Kirchsteiger
H
,
Freckmann
G
,
Heinemann
L
,
Del Re
L
.
Time delay of CGM sensors: relevance, causes, and countermeasures
.
J Diabetes Sci Technol
2015
;
9
:
1006
1015
17.
American Diabetes Association Professional Practice Committee
.
16. Diabetes care in the hospital: Standards of Medical Care in Diabetes—2022
.
Diabetes Care
2022
;
45
(
Suppl. 1
):
S244
S253
18.
Wallia
A
,
Umpierrez
GE
,
Rushakoff
RJ
, et al.;
DTS Continuous Glucose Monitoring in the Hospital Panel
.
Consensus statement on inpatient use of continuous glucose monitoring
.
J Diabetes Sci Technol
2017
;
11
:
1036
1044
19.
Galindo
RJ
,
Umpierrez
GE
,
Rushakoff
RJ
, et al
.
Continuous glucose monitors and automated insulin dosing systems in the hospital: consensus guideline
.
J Diabetes Sci Technol
2020
;
14
:
1035
1064
20.
Feig
DS
,
Donovan
LE
,
Corcoy
R
, et al.;
CONCEPTT Collaborative Group
.
Continuous glucose monitoring in pregnant women with type 1 diabetes (CONCEPTT): a multicentre international randomised controlled trial
.
Lancet
2017
;
390
:
2347
2359
21.
International Organization for Standardization
.
ISO 15197:2013: in vitro diagnostic test systems—requirements for blood glucose monitoring systems for self testing in managing diabetes mellitus
.
Available from https://www.iso.org/standard/54976.html. Accessed 8 August 2022
22.
U.S. Food and Drug Administration
.
Self-monitoring of blood glucose test systems for over-the-counter use
.
Available from https://www.fda.gov/media/87721/download. Accessed 4 May 2022
23.
U.S. Food and Drug Administration
.
Medical devices; clinical chemistry and clinical toxicology devices; classification of the integrated continuous glucose monitoring system
.
24.
U.S. Food and Drug Administration
.
Blood glucose monitoring test systems for prescription point-of-care use: guidance for industry and Food and Drug Administration staff
.
25.
Clinical and Laboratory Standards Institute
.
Point-of-Care Testing in Acute and Chronic Care Facilities: Approved Guidelines
. 3rd ed.
Wayne, PA
,
Clinical and Laboratory Standards Institute
,
2013
26.
Heinemann
L
,
Schoemaker
M
,
Schmelzeisen-Redecker
G
, et al
.
Benefits and limitations of MARD as a performance parameter for continuous glucose monitoring in the interstitial space
.
J Diabetes Sci Technol
2020
;
14
:
135
150
27.
Eversense
.
User Guide
.
Available from https://www.ascensiadiabetes.com/eversense/user-guides. Accessed 1 April 2022
28.
Danne
T
,
Nimri
R
,
Battelino
T
, et al
.
International consensus on use of continuous glucose monitoring
.
Diabetes Care
2017
;
40
:
1631
1640
29.
Freckmann
G
,
Pleus
S
,
Grady
M
,
Setford
S
,
Levy
B
.
Measures of accuracy for continuous glucose monitoring and blood glucose monitoring devices
.
J Diabetes Sci Technol
2019
;
13
:
575
583
30.
Klonoff
DC
,
Parkes
JL
,
Kovatchev
BP
, et al
.
Investigation of the accuracy of 18 marketed blood glucose monitors
.
Diabetes Care
2018
;
41
:
1681
1688
31.
Garingarao
CJ
,
Buenaluz-Sedurante
M
,
Jimeno
CA
.
Accuracy of point-of-care blood glucose measurements in critically ill patients in shock
.
J Diabetes Sci Technol
2014
;
8
:
937
944
32.
Pardo
S
,
Simmons
DA
.
The quantitative relationship between ISO 15197 accuracy criteria and mean absolute relative difference (MARD) in the evaluation of analytical performance of self-monitoring of blood glucose (SMBG) systems
.
J Diabetes Sci Technol
2016
;
10
:
1182
1187
33.
Pleus
S
,
Stuhr
A
,
Link
M
,
Haug
C
,
Freckmann
G
.
Variation of mean absolute relative differences of continuous glucose monitoring systems throughout the day
.
J Diabetes Sci Technol
2022
;
16
:
649
658
34.
Clarke
WL
,
Cox
D
,
Gonder-Frederick
LA
,
Carter
W
,
Pohl
SL
.
Evaluating clinical accuracy of systems for self-monitoring of blood glucose
.
Diabetes Care
1987
;
10
:
622
628
35.
Parkes
JL
,
Slatin
SL
,
Pardo
S
,
Ginsberg
BH
.
A new consensus error grid to evaluate the clinical significance of inaccuracies in the measurement of blood glucose
.
Diabetes Care
2000
;
23
:
1143
1148
36.
Klonoff
DC
,
Lias
C
,
Vigersky
R
, et al.;
Error Grid Panel
.
The surveillance error grid
.
J Diabetes Sci Technol
2014
;
8
:
658
672
37.
Trajanoski
Z
,
Brunner
GA
,
Gfrerer
RJ
,
Wach
P
,
Pieber
TR
.
Accuracy of home blood glucose meters during hypoglycemia
.
Diabetes Care
1996
;
19
:
1412
1415
38.
Joshi
R
,
Snell
V
,
Good
K
,
Wert
Y
,
Helmuth
A
.
Continuous glucose monitoring (CGM) in ICU patients during COVID-19: a pilot study [Abstract]
.
Diabetes
2021
;
70
(
Suppl. 1
):
61-LB
39.
Chow
KW
,
Kelly
DJ
,
Rieff
MC
, et al
.
Outcomes and healthcare provider perceptions of real-time continuous glucose monitoring (rtCGM) in patients with diabetes and COVID-19 admitted to the ICU
.
J Diabetes Sci Technol
2021
;
15
:
607
614
40.
Faulds
ER
,
Boutsicaris
A
,
Sumner
L
, et al
.
Use of continuous glucose monitor in critically ill COVID-19 patients requiring insulin infusion: an observational study
.
J Clin Endocrinol Metab
2021
;
106
:
e4007
e4016
41.
Agarwal
S
,
Mathew
J
,
Davis
GM
, et al
.
Continuous glucose monitoring in the intensive care unit during the COVID-19 pandemic
.
Diabetes Care
2021
;
44
:
847
849
42.
Sadhu
AR
,
Serrano
IA
,
Xu
J
, et al
.
Continuous glucose monitoring in critically ill patients with COVID-19: results of an emergent pilot study
.
J Diabetes Sci Technol
2020
;
14
:
1065
1073
43.
Bichard
LK
,
Rushworth
RL
,
Torpy
DJ
.
Flash glucose monitoring compared to capillary glucose levels in patients with diabetic ketoacidosis: potential clinical applications
.
Endocr Pract
2021
;
27
:
813
818
44.
Davis
GM
,
Faulds
E
,
Walker
T
, et al
.
Remote continuous glucose monitoring with a computerized insulin infusion protocol for critically ill patients in a COVID-19 medical ICU: proof of concept
.
Diabetes Care
2021
;
44
:
1055
1058
45.
Davis
GM
,
Spanakis
EK
,
Migdal
AL
, et al
.
Accuracy of Dexcom G6 continuous glucose monitoring in non-critically ill hospitalized patients with diabetes
.
Diabetes Care
2021
;
44
:
1641
1646
46.
Wright
JJ
,
Williams
AJ
,
Friedman
SB
, et al
.
Accuracy of continuous glucose monitors for inpatient diabetes management
.
J Diabetes Sci Technol
.
Online ahead of print on 7 February 2022 (doi:10.1177/19322968221076562)
47.
Murray-Bachmann
R
,
Leung
TM
,
Myers
AK
, et al
.
Reliability of continuous glucose monitoring system in the inpatient setting
.
J Clin Transl Endocrinol
2021
;
25
:
100262
48.
Gómez
AM
,
Umpierrez
GE
,
Muñoz
OM
, et al
.
Continuous glucose monitoring versus capillary point-of-care testing for inpatient glycemic control in type 2 diabetes patients hospitalized in the general ward and treated with a basal bolus insulin regimen
.
J Diabetes Sci Technol
2015
;
10
:
325
329
49.
Sweeney
AT
,
Pena
S
,
Sandeep
J
, et al
.
Use of a continuous glucose monitoring system in high-risk hospitalized noncritically ill patients with diabetes after cardiac surgery and during their transition of care from the intensive care unit during COVID-19: a pilot study
.
Endocr Pract
2022
;
28
:
615
621
50.
Baker
M
,
Musselman
ME
,
Rogers
R
,
Hellman
R
.
Practical implementation of remote continuous glucose monitoring in hospitalized patients with diabetes
.
Am J Health Syst Pharm
2022
;
79
:
452
458
51.
Reutrakul
S
,
Genco
M
,
Salinas
H
, et al
.
Feasibility of inpatient continuous glucose monitoring during the COVID-19 pandemic: early experience
.
Diabetes Care
2020
;
43
:
e137
e138
52.
Price
C
,
Ditton
G
,
Russell
GB
,
Aloi
J
.
Reliability of inpatient CGM: comparison to standard of care
.
J Diabetes Sci Technol
.
Online ahead of print on 15 December 2021 (doi:10.1177/19322968211062168)
53.
Longo
RR
,
Elias
H
,
Khan
M
,
Seley
JJ
.
Use and accuracy of inpatient CGM during the COVID-19 pandemic: an observational study of general medicine and ICU patients
.
J Diabetes Sci Technol
.
Online ahead of print on 10 May 2021 (doi:10.1177/19322968211008446)
54.
Migdal
AL
,
Spanakis
EK
,
Galindo
RJ
, et al
.
Accuracy and precision of continuous glucose monitoring in hospitalized patients undergoing radiology procedures
.
J Diabetes Sci Technol
2020
;
14
:
1135
1136
55.
Sugiyama
Y
,
Wakabayashi
R
,
Urasawa
M
,
Maruyama
Y
,
Shimizu
S
,
Kawamata
M
.
Perioperative characteristics of the accuracy of subcutaneous continuous glucose monitoring: pilot study in neurosurgery and cardiac surgery
.
Diabetes Technol Ther
2018
;
20
:
654
661
56.
Tripyla
A
,
Herzig
D
,
Joachim
D
, et al
.
Performance of a factory-calibrated, real-time continuous glucose monitoring system during elective abdominal surgery
.
Diabetes Obes Metab
2020
;
22
:
1678
1682
57.
Perez-Guzman
MC
,
Shang
T
,
Zhang
JY
,
Jornsay
D
,
Klonoff
DC
.
Continuous glucose monitoring in the hospital
.
Endocrinol Metab (Seoul)
2021
;
36
:
240
255
58.
Nair
BG
,
Dellinger
EP
,
Flum
DR
,
Rooke
GA
,
Hirsch
IB
.
A pilot study of the feasibility and accuracy of inpatient continuous glucose monitoring
.
Diabetes Care
2020
;
43
:
e168
e169
59.
Thomas
C
,
Welsh
JB
,
Lu
S
,
Gray
JM
.
Safety and functional integrity of continuous glucose monitoring components after simulated radiologic procedures
.
J Diabetes Sci Technol
2021
;
15
:
781
785
60.
Takatsu
Y
,
Shiozaki
T
,
Miyati
T
,
Asahara
M
,
Tani
Y
.
Are the recorded data of flash glucose monitoring systems influenced by radiological examinations?
Radiol Phys Technol
2019
;
12
:
224
229
61.
Espinoza
J
,
Xu
NY
,
Nguyen
KT
,
Klonoff
DC
.
The need for data standards and implementation policies to integrate CGM data into the electronic health record
.
J Diabetes Sci Technol
.
Online ahead of print on 20 November 2021 (doi:10.1177/19322968211058148)
62.
Bruttomesso
D
,
Boscari
F
,
Lepore
G
, et al
.
A “slide rule” to adjust insulin dose using trend arrows in adults with type 1 diabetes: test in silico and in real life
.
Diabetes Ther
2021
;
12
:
1313
1324
63.
Aleppo
G
,
Laffel
LM
,
Ahmann
AJ
, et al
.
A practical approach to using trend arrows on the Dexcom G5 CGM system for the management of adults with diabetes
.
J Endocr Soc
2017
;
1
:
1445
1460
64.
Klonoff
DC
,
King
F
,
Kerr
D
.
New opportunities for digital health to thrive
.
J Diabetes Sci Technol
2019
;
13
:
159
163
65.
Boom
DT
,
Sechterberger
MK
,
Rijkenberg
S
, et al
.
Insulin treatment guided by subcutaneous continuous glucose monitoring compared to frequent point-of-care measurement in critically ill patients: a randomized controlled trial
.
Crit Care
2014
;
18
:
453
66.
Dillmann
C
,
Amoura
L
,
Fall Mostaine
F
,
Coste
A
,
Bounyar
L
,
Kessler
L
.
Feasibility of real-time continuous glucose monitoring telemetry system in an inpatient diabetes unit: a pilot study
.
J Diabetes Sci Technol
2022
;
16
:
955
961
67.
Faulds
ER
,
Jones
L
,
McNett
M
, et al
.
Facilitators and barriers to nursing implementation of continuous glucose monitoring (CGM) in critically ill patients with COVID-19
.
Endocr Pract
2021
;
27
:
354
361
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.