OBJECTIVE

The adoption of continuous glucose monitors (CGMs) in inpatient settings in the pediatric population has been slow because of a scarcity of data on their reliability in hospitalized children.

RESEARCH DESIGN AND METHODS

We retrospectively reviewed the accuracy of the Dexcom G6 CGM system in pediatric patients with diabetes admitted to our academic children’s hospital from March 2018 to September 2023. We cross-referenced the Dexcom Clarity database against an internal database of inpatient admissions to identify all children with CGM data admitted to the hospital. We recorded sensor glucose readings from Clarity and values for point-of-care (POC) glucose, blood urea nitrogen (BUN), and pH from the electronic medical record. CGM accuracy and clinical reliability were measured by mean absolute relative difference (MARD) and Clarke error grid (CEG) analyses.

RESULTS

There were 3,200 admissions of children with diabetes in this period, of which 277 (from 202 patients age 2–18 years) had associated CGM data. Paired CGM and POC measurements (n = 2,904) were compared, resulting in an MARD of 15.9%, with 96.6% of the values in zones A and B of the CEG analysis. Approximately 62% of paired values fell within a 15% or 15 mg/dL difference, whichever was larger (15%/15 mg/dL range), 74% within 20%/20, and 88% within 30%/30. Serum pH, sodium, and BUN had no impact on CGM values or absolute relative difference in linear regression analysis.

CONCLUSIONS

CGMs demonstrated acceptable accuracy in hospitalized children with diabetes. CGM data should be integrated into hospital electronic records to optimize management.

Type 1 diabetes is a chronic metabolic disease characterized by persistent hyperglycemia resulting from the body’s inability to produce sufficient insulin. Although much progress has been made, treatment goals are still not met in many patients with type 1 diabetes. The advent of continuous glucose monitor (CGM) systems represents a significant advancement in mitigating these challenges. The Dexcom G6 (Dexcom, Inc., San Diego, CA), a CGM device approved by the U.S. Food and Drug Administration (FDA) in March 2018 for individuals age ≥2 years with type 1 diabetes in ambulatory settings, facilitates real-time glucose monitoring, offering critical alerts for glucose fluctuations. It comprises a wearable sensor that measures glucose levels in the interstitial fluid and a transmitter that wirelessly relays data to a digital display device, thereby generating a comprehensive continuous glucose profile.

Whereas the efficacy of the Dexcom G6 in outpatient settings has been well documented, its accuracy in inpatient environments is less well established. The hospital environment may affect CGM accuracy. Metabolic stressors, including acidosis, fluctuating patient activity levels, changes in peripheral circulation from dehydration, and potential interference from pharmacologic agents commonly administered in hospital settings (e.g., acetaminophen), could influence CGM performance. Although not approved for inpatient use, in April 2020, Dexcom received notification from the FDA that the FDA would exercise enforcement discretion (1). This meant that the FDA would not oppose the use of CGM systems in hospitals and other health care facilities as a supportive measure for health care initiatives during the COVID-19 pandemic. Although this has led to an increasing body of literature on the safety, accuracy, and reliability of CGMs in hospitalized adults, the pediatric literature remains limited, especially in patients with type 1 diabetes (studies summarized in Table 1).

Table 1

CGM studies in hospitalized patients

StudyYearDesignCGM usedPaired values, nMARD, %CEG zone A + B, %
Adults       
 Davis et al. (152021 Retrospective Dexcom G6 4,067 12.8 98.7 
 Murray-Bachmann et al. (162021 Prospective Libre 2 467 15.6 NA 
 Wright et al. (172022 Prospective Libre 1  21.4 98.8 
     Libre 2  17.7  
 Finn et al. (102023 Retrospective Dexcom G6 2,744 POC glucose 15.1 96.5 
       Noncritical illness 13.8  
       Critical illness 16.3  
      3,705 Laboratory glucose 11.4 99.1 
       Noncritical illness 9.8  
       Critical illness 12.1  
Children       
 Prabhudesai et al. (132015 Prospective NA 103 20 96 
 Pott et al. (82022 Prospective Dexcom G6 620 NA 95.6 
 Zeng et al. (142023 Prospective  399  98.5 
     Guardian 3  13.4  
     Dexcom G6  11.2  
     Libre 1  11.3  
 Waterman et al. (12), Cobry et al. (182024 Retrospective Dexcom G6 1,120 DKA 11.8 97.6–98.6 
 Severe DKA 8.9  
       Nonsevere DKA 14.3  
       Non-DKA 11.7  
       i.v. insulin 13.4  
      1,120 Overall (POC) 11.8 98 
       Medical floor 13.5  
       ICU 7.9  
      288 Overall (laboratory) 6.5  
StudyYearDesignCGM usedPaired values, nMARD, %CEG zone A + B, %
Adults       
 Davis et al. (152021 Retrospective Dexcom G6 4,067 12.8 98.7 
 Murray-Bachmann et al. (162021 Prospective Libre 2 467 15.6 NA 
 Wright et al. (172022 Prospective Libre 1  21.4 98.8 
     Libre 2  17.7  
 Finn et al. (102023 Retrospective Dexcom G6 2,744 POC glucose 15.1 96.5 
       Noncritical illness 13.8  
       Critical illness 16.3  
      3,705 Laboratory glucose 11.4 99.1 
       Noncritical illness 9.8  
       Critical illness 12.1  
Children       
 Prabhudesai et al. (132015 Prospective NA 103 20 96 
 Pott et al. (82022 Prospective Dexcom G6 620 NA 95.6 
 Zeng et al. (142023 Prospective  399  98.5 
     Guardian 3  13.4  
     Dexcom G6  11.2  
     Libre 1  11.3  
 Waterman et al. (12), Cobry et al. (182024 Retrospective Dexcom G6 1,120 DKA 11.8 97.6–98.6 
 Severe DKA 8.9  
       Nonsevere DKA 14.3  
       Non-DKA 11.7  
       i.v. insulin 13.4  
      1,120 Overall (POC) 11.8 98 
       Medical floor 13.5  
       ICU 7.9  
      288 Overall (laboratory) 6.5  

DKA, diabetic ketoacidosis; ICU, intensive care unit; NA, not applicable.

Our study retrospectively analyzed the accuracy of the Dexcom G6 system in hospitalized children with type 1 diabetes. By comparing sensor-derived glucose readings with reference glucose values, we aimed to determine device reliability and performance in the hospital setting.

Setting

Children’s Medical Center Dallas is a large urban pediatric teaching hospital licensed for 487 beds, with a 72-bed satellite facility 22 miles (35 km) north. Outpatients with diabetes are seen in both locations and may be admitted to either facility if ill. Approximately 94% of our patients with type 1 diabetes with at least two visits a year currently use CGMs.

Study Design

This was a retrospective comparison of blood glucose values during hospital admissions with time-matched Dexcom G6 sensor-derived glucose values. The study was approved by the University of Texas Southwestern Medical Center Institutional Review Board.

Population

We included children age 2–18 years with a history of type 1 diabetes admitted to Children’s Medical Center from March 2018 (date of approval of the Dexcom G6 system) to September 2023 who were sharing their Dexcom G6 sensor data with our diabetes clinic through the Dexcom Clarity portal and who continued to wear their sensors during their admission.

Data Collection

We downloaded individual patient CGM data from the Dexcom Clarity cloud-based data warehouse and cross-referenced those data against an internal database of all hospital admissions for children with diabetes to identify all inpatients for whom we had matched CGM and reference glucose values. We collected demographic data and primary admission diagnosis, point-of-care (POC) blood glucose, serum sodium, blood urea nitrogen (BUN), and pH information using SAP Analytics (Walldorf, Germany) to interrogate a clinical results repository extracted from our electronic medical record (Epic Systems, Madison, WI). We excluded capillary blood glucose readings outside the reporting range for Dexcom G6 devices (40–400 mg/dL).

Although not a part of the interactive reports used by most clinicians, Dexcom CGMs generate time-stamped glucose values every 5 min, which are uploaded to the Clarity database when patients have the necessary app open on their phones. These time-stamped raw data points are available for download via the Export All Data function in Dexcom Clarity.

POC glucose values from the glucometer (Freestyle Precision Pro; Abbott Laboratories, Chicago, IL) are transmitted to the patient’s electronic medical record by placing the glucose meter on the docking station. The time stamp for the glucose readings reflects when the test is performed, and the glucometer clock is calibrated each time it is docked.

Each time-stamped POC glucose value was paired in Microsoft Excel with the Dexcom G6 values obtained immediately after it (within 5 min) and at 5-min intervals thereafter for 60 min. We did not attempt to correlate POC readings with CGM readings before each POC measurement because interstitial fluid glucose is known to lag blood glucose (2). Additionally, each blood glucose value was paired with time-matched values, when available, for pH, sodium, and/or BUN obtained within 1 h before or after the time of the glucose reading.

Statistical Analysis

We pooled all blood glucose readings from hospital encounters for which we had time-matched blood glucose and CGM data and calculated several measures assessing CGM accuracy.

The mean absolute relative difference (MARD) was calculated to represent the average of the absolute differences between the CGM readings and the reference glucose measurements relative to the reference values (3). This metric is expressed as a percentage, with lower values indicating greater accuracy of CGM systems.

Clarke error grid (CEG) analyses plotted the CGM readings against the reference glucose values on a grid divided into five zones (A, B, C, D, and E), each representing a different level of clinical risk or treatment error based on the discrepancy between the CGM and reference readings (4). Readings falling into zones A and B are considered clinically acceptable, either because they are accurate (zone A) or lead to benign or no treatment intervention (zone B). Zone C indicates values that may lead to unnecessary corrections and poor outcomes. Values in zone D represent a dangerous failure to detect and treat. Values in zone E represent erroneous treatment (4).

Bland-Altman plots were constructed to offer a visual and analytic method to evaluate the accuracy and reliability of CGM systems across the range of glucose concentrations. The plots highlighted the average difference (or bias) between the CGM readings and reference measurements, providing an indication of systematic over- or underestimation by devices. They illustrate limits of agreement, defined as the mean difference ±1.96 SDs, which encompass the range within which 95% of the differences between the CGM and reference values lie (5).

We also calculated the percentage of CGM values that fell within a 15% margin of the POC/serum values when glucose levels were >100 mg/dL or within 15 mg/dL when levels were ≤100 mg/dL (15%/15 mg/dL range), alongside similar agreement rates of 20%/20 and 30%/30 mg/dL (6). Using one-way ANOVA, we assessed how these values and absolute relative difference (ARD) were affected by acidosis and high sodium or BUN levels.

Finally, we conducted linear regression analyses to evaluate the correlation between POC glucose, pH, sodium, and BUN and CGM values. In all analyses, sodium levels were corrected for hyperglycemia using the following formula: measured Na + ([glucose level − 100] × 0.016).

Of the 3,200 admissions of patients with diabetes to our hospital system from March 2018 to September 2023, there were 277 admissions of patients age 2–18 years with type 1 diabetes with Dexcom data available during the admission, representing 202 unique patients. Patient characteristics are listed in Table 2.

Table 2

Patient characteristics

Characteristicn (%)
Median (IQR) age, years 12.31 (8.85, 15.72) 
Sex  
 Male 144 (52) 
 Female 133 (48) 
Race  
 Black 51 (18) 
 White 170 (61) 
 Hispanic 48 (17) 
 Other 8 (3) 
Median (IQR) time since diagnosis, years 3.72 (1.81, 6.54) 
Payer class  
 Commercial 187 (65) 
 Noncommercial 109 (35) 
Grouped admission diagnosis  
 Diabetes related 164 (60.0) 
 DKA 111 (40.0) 
 Infectious 13 (4.7) 
 Neurologic 29 (10.5) 
 Gastrointestinal 38 (13.7) 
 ENT 8 (2.8) 
 Mental health 3 (1.0) 
 Other  22 (7.9) 
Characteristicn (%)
Median (IQR) age, years 12.31 (8.85, 15.72) 
Sex  
 Male 144 (52) 
 Female 133 (48) 
Race  
 Black 51 (18) 
 White 170 (61) 
 Hispanic 48 (17) 
 Other 8 (3) 
Median (IQR) time since diagnosis, years 3.72 (1.81, 6.54) 
Payer class  
 Commercial 187 (65) 
 Noncommercial 109 (35) 
Grouped admission diagnosis  
 Diabetes related 164 (60.0) 
 DKA 111 (40.0) 
 Infectious 13 (4.7) 
 Neurologic 29 (10.5) 
 Gastrointestinal 38 (13.7) 
 ENT 8 (2.8) 
 Mental health 3 (1.0) 
 Other  22 (7.9) 

DKA, diabetic ketoacidosis; ENT, ear nose throat; IQR, interquartile range.

Accuracy

We analyzed 2,904 paired glucose values from the 277 admissions. Our analyses revealed an MARD of 15.8% between reference and CGM-derived glucose values obtained within 5 min of each other. MARD values progressively increased, with greater intervals between reference and CGM values (only 0–5-, 5–10-, and 10–15-min intervals are shown) (Table 3). This presumably represented an actual change in blood glucose over time.

Table 3

MARD, accuracy assessment percentage, and CEG analysis for agreement between sensor glucose and POC (reference) blood glucose

nMARD, %Accuracy assessment percentage/mg/dL range, %CEG, %
%15/15%20/20%30/30ABA + B
Time interval, min         
 0–5 2,904 15.8 61.4 73.1 88.3 72.3 24.3 96.6 
 5–10 2,903 16.7 60.6 72.0 87.5 71.4 25.0 96.4 
 10–15 2,881 19.0 57.7 69.2 84.8 30.1 42.8 72.9 
pH         
 <7.1 18.5 33.3 77.8 88.9 77.8 22.2 100.0 
 7.1–7.3 284 15.9 51.1 63.0 81.3 68.3 31.0 99.3 
 >7.3 460 16.4 60.0 76.3 94.8 70.2 28.3 98.5 
BUN, mg/dL         
 >20 77 17.3 61.0 72.7 89.6 72.7 26.0 98.7 
 ≤20 506 16.1 57.1 72.1 89.3 70.2 28.4 98.6 
Na, mmol/L         
 >145 26 13.9 61.5 73.1 92.3 73.1 26.9 100 
 <145 557 16.4 57.4 72.2 89.2 70.4 28.2 98.6 
BG range, mg/dL         
 40–70 185 26.0 62.2 74.1 84.3 64.9 0.0 64.9 
 71–180 1,333 16.9 59.9 70.7 85.8 69.2 30.2 99.4 
 181–250 763 14.0 62.4 74.4 89.8 74.4 24.1 98.6 
 251–400 623 12.8 66.9 78.5 92.8 78.5 19.3 97.8 
nMARD, %Accuracy assessment percentage/mg/dL range, %CEG, %
%15/15%20/20%30/30ABA + B
Time interval, min         
 0–5 2,904 15.8 61.4 73.1 88.3 72.3 24.3 96.6 
 5–10 2,903 16.7 60.6 72.0 87.5 71.4 25.0 96.4 
 10–15 2,881 19.0 57.7 69.2 84.8 30.1 42.8 72.9 
pH         
 <7.1 18.5 33.3 77.8 88.9 77.8 22.2 100.0 
 7.1–7.3 284 15.9 51.1 63.0 81.3 68.3 31.0 99.3 
 >7.3 460 16.4 60.0 76.3 94.8 70.2 28.3 98.5 
BUN, mg/dL         
 >20 77 17.3 61.0 72.7 89.6 72.7 26.0 98.7 
 ≤20 506 16.1 57.1 72.1 89.3 70.2 28.4 98.6 
Na, mmol/L         
 >145 26 13.9 61.5 73.1 92.3 73.1 26.9 100 
 <145 557 16.4 57.4 72.2 89.2 70.4 28.2 98.6 
BG range, mg/dL         
 40–70 185 26.0 62.2 74.1 84.3 64.9 0.0 64.9 
 71–180 1,333 16.9 59.9 70.7 85.8 69.2 30.2 99.4 
 181–250 763 14.0 62.4 74.4 89.8 74.4 24.1 98.6 
 251–400 623 12.8 66.9 78.5 92.8 78.5 19.3 97.8 

BG, blood glucose.

On CEG analysis, 72.3% of the data fell within zone A, whereas zone B accounted for 24.3%, with a total of 96.6% in the clinically acceptable range (Fig. 1 and Table 3). Approximately 61% of paired values fell within 15%/15, 73% within 20%/20, and 88% within a 30%/30 mg/dL range. Accuracy of all measures progressively decreased with CGM values obtained 5–10 or 10–15 min after the reference values, compared with values within 0–5 min of each reference value (Table 3).

Figure 1

A: CEG analysis for POC vs. CGM values obtained with 5 min of each POC value (n = 2,904). Zone A (72.3%) indicates values that lie within 20% of the reference sensor and are considered clinically accurate. Zone B (24.2%) indicates values that lie outside of 20% but are not likely to lead to inappropriate treatment and are considered benign errors. Zone C (0.5%) indicates values that may lead to unnecessary corrections and poor outcomes. Zone D (2.9%) values represent a dangerous failure to detect and treat. Zone E (0.1%) values represent erroneous treatment. B: Bland-Altman plot. Scatter plot of paired glucose values from CGM and POC testing, with the position of each point on the x-axis representing the average of CGM and POC values and the y-axis representing the difference. A regression line is included with 95% confidence limits. This plot shows that the average difference (or bias) between the CGM readings and POC is +5.1 mg/dL (CGM trending higher).

Figure 1

A: CEG analysis for POC vs. CGM values obtained with 5 min of each POC value (n = 2,904). Zone A (72.3%) indicates values that lie within 20% of the reference sensor and are considered clinically accurate. Zone B (24.2%) indicates values that lie outside of 20% but are not likely to lead to inappropriate treatment and are considered benign errors. Zone C (0.5%) indicates values that may lead to unnecessary corrections and poor outcomes. Zone D (2.9%) values represent a dangerous failure to detect and treat. Zone E (0.1%) values represent erroneous treatment. B: Bland-Altman plot. Scatter plot of paired glucose values from CGM and POC testing, with the position of each point on the x-axis representing the average of CGM and POC values and the y-axis representing the difference. A regression line is included with 95% confidence limits. This plot shows that the average difference (or bias) between the CGM readings and POC is +5.1 mg/dL (CGM trending higher).

Close modal

The Bland-Altman plot showed minimum bias with a mean of +5.1 mg/dL (95% CI −66 to +76.2 mg/dL; CGM trending higher) across the range of glucose comparisons (Fig. 1).

Accuracy Across Glucose, pH, Sodium, and BUN Ranges

The MARD varied across the glycemic range, showing lower levels (greater accuracy) as glucose levels increased. In the CEG analysis, >95% of the values were in zones A and B across all glycemic ranges, except for hypoglycemia (glucose 40–70 mg/dL).

MARD values and CEG and 15%/15, 20%/20, and 30%/30 mg/dL proportions did not differ (tested by one-way ANOVA) based on severity of acidosis or presence of serum Na or BUN values above their respective reference ranges (Table 3).

In an alternative approach, linear regression analysis was used to determine the effect of POC glucose values, pH, sodium, and BUN on CGM values. POC glucose values and CGM values were almost perfectly correlated (regression coefficient 1.02; R2 = 0.97). There were no correlations between CGM glucose and pH, sodium, or BUN values when POC glucose was included as a covariate or between pH, sodium, or BUN values and ARD.

With the growing adoption of technology among patients with diabetes, it is inevitable that an increasing number will use CGMs in the hospital setting. Whereas CGMs are not officially sanctioned for guiding medical decisions at our hospital, patients with type 1 diabetes who are admitted and already using CGMs are allowed to continue wearing their sensors. This creates challenges, because CGM data are not currently captured in the clinical database but instead are available in cloud-based databases with which these sensors often communicate. The increasing reliance among individuals with type 1 diabetes on automated insulin delivery systems, which are entirely dependent on CGM data, will only add to the challenge of reconciling these different data streams. Consequently, understanding the limitations of CGM use becomes crucial, not only in outpatient settings but also in the inpatient environment.

There are limited data about CGM accuracy in pediatric patients with type 1 diabetes admitted to the hospital (Table 1). Our study demonstrated that the Dexcom G6 has acceptable accuracy in hospitalized pediatric patients with type 1 diabetes, with an MARD of 15.8% when compared with standard-of-care POC glucose monitoring. Although the Clinical Laboratory Standards Institute does not provide a minimally acceptable MARD for CGM (7), MARD values >18% are considered to represent poor accuracy (8). Our study involved sensors placed at home by patients without professional oversight of the insertion technique and without standardized protocols for site maintenance once the children were hospitalized. Therefore, our accuracy results should be compared cautiously with prospective studies (9,10) using sensors placed in a hospital setting under professional supervision.

MARD is widely used as the predominant method for evaluating CGM accuracy because of its simplicity as a single easy-to-understand numeric measure. However, it is not without its drawbacks (11). There is substantial variability in reported MARD values across various studies, populations, and settings (Table 1). Therefore, we also examined the CEG. For the overall analysis and all subanalyses, except for blood glucose in the 40–70 mg/dL range, >95% of the values were in zones A and B, consistent with other studies (Table 1). This further supports the use of CGM for decision-making in a hospital setting.

Considering that acidosis, hypernatremia, and elevated BUN can be associated with dehydration, we investigated their potential impact on CGM accuracy. However, no correlation was observed between these factors and CGM values (when POC glucose was included as a covariate) or between these factors and the ARD between CGM and POC glucose values. This is consistent with the limited existing data showing no effect of acidosis on CGM values in children with type 1 diabetes (9,12,13).

Our study is larger than previous pediatric studies and provides further evidence supporting the use of CGM in hospitals (Table 1). It is also one of the few studies examining the impact of acidosis and hypernatremia on CGM values. Limitations included the retrospective and single-center nature of the study and the fact that data were not available on duration of wear or site of placement of each sensor. The study did not assess the potential influence of medications on CGM accuracy. Finally, the study included a limited number of critically ill patients (only nine of 277 encounters had a pH <7.1), making it challenging to evaluate the performance of the system in such cases. However, CGMs are not designed to measure blood glucose >400 regardless, so the paucity of performance data in severe diabetic illness is a relatively minor limitation.

In conclusion, our study suggests that despite the unique challenges of the inpatient environment, the Dexcom G6 demonstrates acceptable accuracy in hospitalized children with type 1 diabetes. The use of CGMs in hospitals should be encouraged, particularly because of the growing adoption of automated insulin delivery systems, which are more effective than conventional insulin therapy in managing type 1 diabetes in children and adolescents (14). CGM data should be integrated into hospital electronic records to optimize management. Further prospective research is required to validate these findings and establish standard protocols for CGM use in pediatric inpatient settings.

Acknowledgments. The authors thank Marconi Abreu, University of Texas Southwestern, Dallas, Texas, for critical reading of this manuscript and providing helpful insight.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. N.G. performed the literature review, acquired the data, and wrote the first draft of the manuscript. N.G. and S.A. wrote the research proposal. K.L. and P.C.W. analyzed the data. P.C.W. and S.A. conceptualized the study. All authors contributed to the study design, reviewed and edited the manuscript, and approved the final version of the manuscript. N.G. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. This study was presented in part as a poster at the 84th American Diabetes Association Scientific Sessions, Orlando, FL, 21–24 June 2024.

Handling Editors. The journal editors responsible for overseeing the review of the manuscript were Elizabeth Selvin and M. Sue Kirkman.

1.
U.S. Food and 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
. Accessed 30 April 2022. Available from https://www.fda.gov/regulatory-information/search-fda-guidance-documents/enforcement-policy-non-invasive-remote-monitoring-devices-used-support-patient-monitoring
2.
Basu
A
,
Dube
S
,
Veettil
S
, et al
.
Time lag of glucose from intravascular to interstitial compartment in type 1 diabetes
.
J Diabetes Sci Technol
2015
;
9
:
63
68
3.
Reiterer
F
,
Polterauer
P
,
Schoemaker
M
, et al
.
Significance and reliability of MARD for the accuracy of CGM systems
.
J Diabetes Sci Technol
2017
;
11
:
59
67
4.
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
5.
Dogan
NO
.
Bland-Altman analysis: a paradigm to understand correlation and agreement
.
Turk J Emerg Med
2018
;
18
:
139
141
6.
Galindo
RJ
,
Migdal
AL
,
Davis
GM
, et al
.
Comparison of the FreeStyle Libre Pro Flash continuous glucose monitoring (CGM) system and point-of-care capillary glucose testing in hospitalized patients with type 2 diabetes treated with basal-bolus insulin regimen
.
Diabetes Care
2020
;
43
:
2730
2735
7.
Vigersky
RA
,
Shin
J
.
The myth of MARD (mean absolute relative difference): limitations of MARD in the clinical assessment of continuous glucose monitoring data
.
Diabetes Technol Ther
2024
;
26
:
38
44
8.
Wernerman
J
,
Desaive
T
,
Finfer
S
, et al
.
Continuous glucose control in the ICU: report of a 2013 round table meeting
.
Crit Care
2014
;
18
:
226
9.
Pott
T
,
Jimenez-Vega
J
,
Parker
J
,
Fitzgerald
R
.
Continuous glucose monitoring in pediatric diabetic ketoacidosis
.
J Diabetes Sci Technol
2024
;
18
:
899
903
10.
Finn
E
,
Schlichting
L
,
Grau
L
,
Douglas
IS
,
Pereira
RI
.
Real-world accuracy of CGM in inpatient critical and noncritical care settings at a safety-net hospital
.
Diabetes Care
2023
;
46
:
1825
1830
11.
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
12.
Waterman
LA
,
Pyle
L
,
Forlenza
GP
, et al
.
Accuracy of a real-time continuous glucose monitor in pediatric diabetic ketoacidosis admissions
.
Diabetes Technol Ther
2024
;
26
:
626
632
13.
Prabhudesai
S
,
Kanjani
A
,
Bhagat
I
,
Ravikumar
KG
,
Ramachandran
B
.
Accuracy of a real-time continuous glucose monitoring system in children with septic shock: a pilot study
.
Indian J Crit Care Med
2015
;
19
:
642
647
14.
Zeng
B
,
Gao
L
,
Yang
Q
,
Jia
H
,
Sun
F
.
Automated insulin delivery systems in children and adolescents with type 1 diabetes: a systematic review and meta-analysis of outpatient randomized controlled trials
.
Diabetes Care
2023
;
46
:
2300
2307
15.
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
16.
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
17.
Wright
JJ
,
Williams
AJ
,
Friedman
SB
, et al
.
Accuracy of continuous glucose monitors for inpatient diabetes management
.
J Diabetes Sci Technol
2023
;
17
:
1252
1255
18.
Cobry
EC
,
Pyle
L
,
Waterman
LA
, et al
.
Accuracy of a continuous glucose monitor during pediatric type 1 diabetes inpatient admissions
.
Diabetes Technol Ther
2024
;
26
:
119
124
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.