Accuracy of the Dexcom G7 Continuous Glucose Monitoring Sensors in People with Diabetes Undergoing Hemodialysis (ALPHA-2 Study)

Abstract

The accuracy of the latest generation Dexcom G7 sensors in individuals with diabetes undergoing hemodialysis has not previously been investigated. Participants with diabetes undergoing hemodialysis were recruited, with paired sensor glucose from Dexcom G7 recorded with plasma glucose analyzed in the laboratory, as well as the Freestyle Precision Pro glucometer and EKF Biosen C-Line analyzer. Ten adults (median age 64.0 [58.0–74.5] years) were recruited. Overall percentage (%) mean and median absolute relative differences were 10.4% and 8.5% for matched laboratory pairs, respectively (n = 720). Diabetes Technology Society error grid analysis showed 99.7%, 100%, and 99.9% of pairs within zones A and B for lab, glucometer, and EKF methods, respectively. This, the first Dexcom G7 accuracy study conducted in people on hemodialysis, demonstrates accuracy and safety when compared with lab reference readings. These data support the accessibility of continuous glucose monitoring (CGM) and hybrid closed-loop systems for people with diabetes on hemodialysis.

Introduction

Type 2 diabetes is the primary cause of end-stage kidney disease.¹ There remains a paucity of accuracy studies of continuous glucose monitoring (CGM) systems on hemodialysis. Previous data suggest that the Freestyle Libre 1 (Abbott Diabetes Care, Alameda, CA, USA) sensor was accurate and safe in adults with diabetes on hemodialysis, while the Dexcom G6 (Dexcom, Inc., San Diego, CA, USA) sensor had a less favorable accuracy profile.² Here, we evaluate the accuracy of the Dexcom G7 CGM system (Dexcom, Inc., San Diego, CA, USA) compared with three reference standards: laboratory glucose (Lab), Freestyle Precision Pro Glucometer (POC; Abbott Diabetes Care, Alameda, CA, USA), and EKF Diagnostics³ analyzer (EKF Diagnostics, Cardiff, UK) in people with diabetes undergoing hemodialysis.

Methods

This prospective single-arm observational study was conducted at the St. Charles Haemodialysis Centre, Imperial College Healthcare NHS Trust, UK. The study took place from April to May 2024 with approval from the UK Health Research Authority, London Bromley Research Ethics Committee and Health Research Authority (HRA Reference No. 19/LO/0431; ClinicalTrials.gov NCT03885362). Written consent was obtained from all participants before their inclusion in the study.

This study was performed as an extension phase to the original Accuracy of Continuous Glucose Sensors in People with Diabetes Undergoing Haemodialysis (ALPHA) study² and utilized similar methodology. To summarize, 10 participants underwent 12 study visits coinciding with their routine consecutive hemodialysis sessions. During each visit, venous blood samples were drawn from the dialysis circuit (pre-dialysis limb) at 30-min intervals. In the presence of a fistula, this would be arterialized venous blood. If dialysis was performed through a Tessio line, this would be central venous blood. Participants received their standard dialysis treatment without any additional intervention specifically aimed at altering glucose concentrations. This included no restrictions to food intake.

Samples were analyzed against the reference glucose methods: (1) Abbott Alinity Glucose c laboratory analyzer, (2) Abbott Freestyle Precision Pro point-of-care glucometer (POC), and (3) EKF Biosen C-line glucose analyzer (Supplementary Fig. S1). The laboratory measure is considered the gold standard.^4

–8 The EKF Biosen C-line analyzer was calibrated as per standard protocol in the manufacturer’s instructions.³ Dexcom G7 readings were retrospectively matched to reference glucose readings to ensure proximity within 2.5 min using a similar methodology to the original ALPHA study.²

The primary outcome was the overall mean absolute relative difference (MARD, %) between Dexcom G7 glucose readings and laboratory blood glucose measurements obtained during hemodialysis sessions. Secondary objectives include the MARD for Dexcom G7 readings compared to glucometer and EKF readings. The performance evaluation also included the proportion of the CGM system values that are within ±20% of the reference value at glucose levels >100 mg/dL and ±20 mg/dL of the glucose levels ≤100 mg/dL (hereafter referred to as %20/20), as well as the proportion of the CGM system values that are within ±15% of the reference value at glucose levels >100 mg/dL and ±15 mg/dL of the glucose levels ≤100 mg/dL (hereafter referred to as % 15/15). Other accuracy metrics analyzed included linear regression, Bland–Altman analysis [GraphPad Prism (v10.2.3), www.graphpad.com], as well as the Diabetes Technology Society (DTS) Error Grid⁹ and Surveillance Error Grid (SEG) analysis (www.diabetestechnology.org/seg/). The study design infographic and statistical methods are accessible via the Supplementary Data S1.

Results

Baseline demographics

The study consisted of 10 participants, 7 of whom were male. The median (IQR) age was 64.0 (58.0–74.5) years, HbA1c was 51.5 (46.5–57.5) mmol/mol and body mass index was 28.0 (25.4–30.3) kg/m². Within this cohort, 3 had type 1 diabetes, 6 had type 2 diabetes, and 1 had post-transplant diabetes. All 6 participants with type 2 diabetes used insulin. Supplementary Table S1 demonstrates the participant baseline characteristics.

Dexcom G7 accuracy and precision

During evaluation across all dialysis sessions, there were 720 matched Dexcom G7-Lab, 348 Dexcom G7-POC and 729 Dexcom G7-EKF paired readings (Table 1). The overall MARDs were 10.4%, 8.6%, and 16.5% for lab, POC, and EKF, respectively. Median ARDs were 8.5%, 6.9%, and 16.3% for lab, POC, and EKF, respectively.

Table 1.

Measures of Accuracy for Dexcom G7 Compared with the Reference Blood Glucose

	Dexcom G7-Lab	Dexcom G7-POC	Dexcom G7-EKF	P value
Total matched pairs, n	720	348	729	—
Overall median ARD, %	8.5 (4.2–14.2)	6.9 (2.6–12.2)	16.3 (9.6–23.0)	<0.0001
Overall MARD, %	10.4 (±8.6)	8.6 (±7.6)	16.5 (±9.0)	—
Median ARDs by predefined range, %
<70 mg/dL (<3.9 mmol/L)	23.4 (17.8–38.5)^a	29.4 (27.5–31.4)^a	17.6 (16.1–37.0)^a	—
70–180 mg/dL (3.9–10 mmol/L)	8.9 (4.5–15.9)^b	7.6 (3.2–13.7)^b	13.2 (7.5–21.0)^b	—
>180 mg/dL (>10 mmol/L)	8.0 (3.9–12.5)^c	5.8 (2.4–11.0)^c	19.0 (11.9–23.8)^c	—
>270 mg/dL (>15 mmol/L)	9.4 (5.6–12.5)^d	10.5 (4.9–13.2)^d	22.4 (19.7–24.4)^d	—
%20/20, %	89.3	92.5	64.5	—
%15/15, %	78.8	84.5	44.2	—

Results shown as median (IQR), mean (SD) or n (value or %).

n = 5, n = 2, n = 5 paired Dexcom G7-Lab, POC, EKF samples <70mg/dL, respectively.

n = 341, n = 168, n = 342 paired Dexcom G7-Lab, POC, EKF samples 70-180mg/dL, respectively.

n = 374, n = 178, n = 382 paired Dexcom G7-Lab, POC, EKF samples >180mg/dL, respectively.

n = 34, n = 20, n = 35 paired Dexcom G7-Lab, POC, EKF samples >270mg/dL, respectively.

ARD, absolute relative difference; IQR, interquartile range; MARD, mean absolute relative difference; POC, point of care glucometer; SD, standard deviation.

Least-squares linear regression (Supplementary Fig. S2) showed a gradient of 0.97, 1.01, and 1.14 for Dexcom G7 with lab, POC, and EKF matched pairs, respectively. Bland–Altman plots demonstrated a mean bias of +5.5 mg/dL, −3.3 mg/dL, and +21.9 mg/dL for G7 compared to lab, POC, and EKF, respectively. The 95% limits of agreement for Dexcom G7-lab were −38.7 mg/dL, +49.6 mg/dL, Dexcom G7-POC were −46.2 mg/dL, +39.6 mg/dL, and Dexcom G7-EKF were −22.2 mg/dL, +66.0 mg/dL.

To assess the clinical safety of the Dexcom G7, the DTS error grid analysis was performed with 87.8% of Dexcom-lab pairs fell within the no clinical risk Zone A, while 11.9% were categorized in the slight (lower) risk Zone B (Fig. 1). The total proportion of points within zones A and B were 99.7%, 100%, and 99.9% for lab, POC, and EKF matched pairs, respectively. Data for SEG analyses are available in Supplementary Figure S3.

FIG. 1.

Diabetes Technology Society (DTS) Error Grid analysis for Dexcom G7 compared with reference glucose values. (A) Dexcom G7 compared to Lab glucose (n = 720) (B) Dexcom G7 compared to POC glucose (n = 348) (C) Dexcom G7 compared to EKF glucose (n = 729). DTS grade zones:⁹ (A) green—no risk, (B) yellow—mild risk, (C) orange—moderate risk, (D) red—high risk, (E) purple—extreme risk. Abbreviations: POC, point of care glucometer.

Analysis of Dexcom G7 accuracy across sensor days for lab and POC-matched pairs exhibited no significant difference in median ARD and MARD between day 1 and days 2–10 (Supplementary Table S3). The MARD for lab-matched fistula-derived measurements (n = 383) was 7.0%, while for lab-matched central line-derived measurements (n = 337) was 10.8%.

Safety assessment

Throughout the study period, there were no general or device-related serious adverse events. One participant withdrew from the study at visit 8 for personal reasons. The analysis included all data collected from participants. There were no Dexcom or reference glucose values that exceeded the measurement range.

Conclusions

This is the first study to evaluate the accuracy and clinical safety of the Dexcom G7 in people with diabetes undergoing hemodialysis.

The Dexcom G7 demonstrated good accuracy with a MARD and median ARD of 10.4% and 8.5% against lab comparators, respectively. A lower MARD signifies better sensor accuracy, and MARD of <10% is considered adequate to make insulin dosing decisions without confirmation.¹⁰ The median ARDs showed good accuracy for Dexcom G7 when matched with lab and POC pairs across glycemic ranges 70–180 mg/dL, >180 mg/dL, and >270 mg/dL. There were only 5 readings in the hypoglycemia range <70 mg/dL, and thus no conclusions can be drawn for accuracy in the hypoglycemia range. The %20/20 for lab and POC matched pairs were 89.3% and 92.5%, respectively, meeting the FDA guideline of 87% or greater for sufficient CGM accuracy.¹¹ SEG and DTS error grid analysis demonstrated that the Dexcom G7, when compared to all three references, achieved clinical safety by surpassing a minimum threshold of 97% of points within zones A and B.¹²

Interestingly, CGM glucose compared to EKF glucose had a higher MARD and median ARD compared to lab and POC pairs. The larger bias observed in the Bland–Altman analysis of the G7/EKF readings compared to other reference methods suggests a potential underestimation by the EKF analyzer relative to sensor values. However, this can only be confirmed through a pairwise comparison between the EKF and other analyzers, which we have not presented in this study.

The differences in MARD observed between comparators (i.e., lab, POC, and EKF) may be due to the different enzymes being used to measure blood glucose. While the EKF uses glucose oxidase,³ the Abbott Freestyle Precision Pro glucometer utilizes NAD-glucose dehydrogenase¹³ and the Abbott Alinity c glucose method utilizes hexokinase.¹⁴ Although the EKF demonstrated the highest MARD, it is noteworthy that both the EKF and Dexcom G7 sensor rely on the glucose oxidase enzyme,^3,15 making their readings potentially more comparable. Although the underlying cause of reduced accuracy with the EKF during hemodialysis remains unclear, it may be influenced by fluid shifts or the presence of interferents that occur as part of the dialysis process. Laboratory measurement is considered the gold standard.^7,8 Glucose samples in the study were collected in tubes with sodium fluoride/potassium oxalate, fluoride being an inhibitor of enolase in the glycolytic scheme, so that glucose metabolism is inhibited.¹⁶ Studies assessing the EKF analyzer outside of the dialysis cohort have shown sufficient accuracy compared to reference glucose measures, with correlation coefficients ranging from 0.973 to 0.997 across multiple studies using whole blood samples.^4
–6 Similarly, the Freestyle Precision Pro exhibits a linear regression gradient of 1.01 when paired with laboratory references,¹³ and it is the platform technology used to assess accuracy of both Libre 3 and Dexcom G7.¹⁷

Previous studies have analyzed the accuracy of CGM systems within a hemodialysis population. The Dexcom G5 had a MARD of 15.2% in people with advanced chronic kidney disease.¹⁸ However, the data were not exclusively obtained during hemodialysis sessions and included participants not on dialysis. The %MARD of the Dexcom G6 Pro in individuals on hemodialysis was 14.4% during dialysis,¹⁹ and the original ALPHA study assessed the accuracy of the Dexcom G6 sensor, reporting a MARD of 21.4% with 49% of values achieving %20/20.² Similarly, Narasaki et al. reported MARD values of 20.7% with the Dexcom G6 during hemodialysis with 47.2% of values achieving %20/20.²⁰ However, Bland–Altman analyses showed a lesser degree of bias between CGM and blood glucose in the latter study than the ALPHA study.²⁰ Freestyle Libre 1 was found to have reliable accuracy with overall MARD of 11.3%,² and although the Freestyle Libre 2 sensor was not formally assessed in this study, it has a similar sensor technology as Freestyle Libre 1.

The differences between Dexcom G6 and G7 sensors observed between ALPHA and ALPHA-2 studies may be attributable to improvements in sensor technology. Differences in hardware include a shorter sensor filament inserted into the skin at a steeper angle of 90 degrees, an improved processing algorithm resulting in a reduced lag time of 3.5 min, and a shorter start-up time of 30 min compared with 2 h for the Dexcom G6. Dexcom G7 accuracy studies outside of dialysis populations have reported an improvement in the MARD compared with the Dexcom G6 sensor from 9.0% to 8.2%.^21,22

One of the limitations of the study is that the EKF Biosen C-Line analyzer was used instead of the YSI2300 used in the original study, which has been widely used as a comparator reference system in clinical and analytical studies. This is due to the YSI2300 being discontinued. However, for this reason, we have used multiple reference comparators, and the results demonstrate the importance of an appropriate reference standard. Another limitation is that there are limited data points in the hypoglycemia <70 mg/dL and hyperglycemia >250 mg/dL range, with further studies required to assess sensor performance across the glucose range.

Finally, the observed accuracy varied considerably depending on the chosen reference method, and we noted a decline in measured accuracy when using the EKF reference system. These findings underline the importance of study methodology in evaluating CGM system performance.^23,24 Factors such as the type of reference instrument used, the procedures for matching sensor readings to reference measurements, the source of the reference samples, and whether glucose levels are changing rapidly due to deliberate interventions or food intake can all significantly influence results. Future investigations should carefully consider these methodological elements to ensure that reported accuracy reflects the true performance of CGM technologies.

In conclusion, the Dexcom G7 has demonstrated sufficient accuracy and safety with lab and POC comparators. Establishing the accuracy and safety of CGM sensors such as the Dexcom G7 paves the way for the potential use of closed-loop systems within this cohort, which has the potential to transform diabetes management for those on dialysis.

Footnotes

Acknowledgments

The authors wish to thank all the study participants for their valuable time and commitment. Thank you to the nursing and health care professionals at St Charles’ Hospital Renal Dialysis Units for their invaluable support during the study. Infrastructure support was provided by the NIHR Imperial Biomedical Research Centre and the NIHR Imperial Clinical Research Facility. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care.

Authors’ Contributions

M.R., N.O., and P.A. conceived and designed the study. P.A., N.O., and M.R. wrote the study protocol. R.P. and P.A. acquired, analyzed and interpreted the data. R.P. and P.A. drafted the article. All authors reviewed, edited, and approved the final version of the article. R.P. and P.A. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Author Disclosure Statement

P.A. has received equipment from Dexcom for investigator-initiated studies. M.R. has received honoraria for advisory board participation from Dexcom and Roche Diabetes. N.O. has received honoraria for speaking and advisory board participation from Abbott Diabetes, Dexcom, Medtronic Diabetes, and Roche Diabetes. A.F. has received honoraria for speaking and advisory board participation from Astra Zeneca, Boehringer Inglrheim, Lilly, Napp UK and Vifor Pharma UK. T.T. has received limited non-financial support from Dexcom for investigator-initiated studies.

Funding Information

Dexcom provided sensor supplies. Remaining funding was obtained from the NIHR Imperial Biomedical Research Centre. The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. P.A. is funded by NIHR Grant No. 132960.

Supplementary Material

Supplementary Data S1

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

Supplementary Table S1

Supplementary Table S2

Supplementary Table S3

References

Coresh

, Selvin

, Stevens

, et al. Prevalence of chronic kidney disease in the United States. JAMA, 2007; 298(17):2038–2047; doi: 10.1001/JAMA.298.17.2038

Avari

, Tang

, Jugnee

, et al. The accuracy of continuous glucose sensors in people with diabetes undergoing hemodialysis (ALPHA Study). Diabetes Technol Ther, 2023; 25(7):447–456; doi: 10.1089/DIA.2023.0013

Anonymous. Biosen Lactate and Glucose Analyzer. EKF Diagnostics. n.d. Available from: https://www.ekfdiagnostics.com/biosen-analyzer.html [Last accessed: March 19, 2024 ].

Gijzen

, Moolenaar

DLJ

, Weusten

JJAM

, et al. Is there a suitable point-of-care glucose meter for tight glycemic control? Evaluation of one home-use and four hospital-use meters in an intensive care unit. Clin Chem Lab Med, 2012; 50(11):1985–1992; doi: 10.1515/CCLM-2012-0104/DOWNLOADASSET/SUPPL/CCLM-2012-0104AD.PDF

Luley

. Validation of BIOSEN C-Line glucose/lactate analyser versus HITACHI 917. 2006.

Tsachev

. Clinical evaluation of Biosen C-Line analyzer manufactured by EKF Diagnostics for testing of glucose concentration in whole blood, plasma and serum. 2007.

, Thomson

, Sa

, et al. Glucose monitoring and management in the NICU—how are we doing? Infant, 2017; 13(5):182–186.

, Harris

, Estilong

, et al. Blood glucose measurement in the intensive care unit: What is the best method? J Diabetes Sci Technol, 2013; 7(2):489–499; doi: 10.1177/193229681300700226

Klonoff

, Freckmann

, Pleus

, et al. The diabetes technology society error grid and trend accuracy matrix for glucose monitors. J Diabetes Sci Technol, 2024; 18(6):1346–1361; doi: 10.1177/19322968241275701

10.

Kovatchev

, Patek

, Ortiz

, et al. Assessing sensor accuracy for non-adjunct use of continuous glucose monitoring. Diabetes Technol Ther, 2015; 17(3):177–186; doi: 10.1089/DIA.2014.0272

11.

CFR—Code of Federal Regulations Title 21. n.d. Available from: https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/cfrsearch.cfm?fr=862.1355 [Last accessed: July 18, 2024 ].

12.

Kovatchev

, Wakeman

, Breton

, et al. Computing the surveillance error grid analysis: Procedure and examples. J Diabetes Sci Technol, 2014; 8(4):673–684; doi: 10.1177/1932296814539590

13.

Brazg

, Hughes

, Martin

, et al. Clinical evaluation of the FreeStyle Precision Pro system. Clin Chim Acta, 2013; 421:243–250; doi: 10.1016/J.CCA.2013.03.021

14.

Shaker

, Swift

. Peroxidase-Coupled Glucose Method. StatPearls; 2023.

15.

Huang

. 510(k) substantial equivalence determination decision Summary I Background Information: A 510(k) Number. n.d.

16.

Bonetti

, Carta

, Montagnana

, et al; Italian joint SIBioC-SIPMeL Study Group on Diabetes Mellitus. Effectiveness of citrate buffer-fluoride mixture in terumo tubes as an inhibitor of in vitro glycolysis. Biochem Med (Zagreb), 2016; 26(1):68–76; doi: 10.1161/3/BM.2016.006

17.

Hanson

, Kipnes

, Tran

. Comparison of point accuracy between two widely used continuous glucose monitoring systems. J Diabetes Sci Technol, 2024; 18(3):598–607; doi: 10.1177/19322968231225676

18.

Ólafsdottir

, Andelin

, Saeed

, et al. Performance of dexcom G5 and FreeStyle libre sensors tested simultaneously in people with type 1 or 2 diabetes and advanced chronic kidney disease. World J Clin Cases, 2022; 10(22):7794–7807; doi: 10.1299/8/WJCC.V10.I22.7794

19.

Villard

, Breton

, Rao

, et al. Accuracy of a factory-calibrated continuous glucose monitor in individuals with diabetes on hemodialysis. Diabetes Care, 2022; 45(7):1666–1669; doi: 10.2337/DC22-0073

20.

Narasaki

, Kalantar-Zadeh

, Daza

, et al. Accuracy of continuous glucose monitoring in hemodialysis patients with diabetes. Diabetes Care, 2024; 47(11):1922–1929; doi: 10.2337/DC24-0635

21.

Continuous Glucose Monitoring Dexcom G7 System User Guide. doi: https://dexcompdf.s3.us-west-2.amazonaws.com/en-us/G7-CGM-Users-Guide.pdf n.d.

22.

Garg

, Kipnes

, Castorino

, et al. Accuracy and safety of Dexcom G7 continuous glucose monitoring in adults with diabetes. Diabetes Technol Ther, 2022; 24(6):373–380; doi: 10.1089/DIA.2022.0011

23.

Pleus

, Eichenlaub

, Eriksson Boija

, et al. The need for standardization of continuous glucose monitoring performance evaluation: An opinion by the international federation of clinical chemistry and laboratory medicine working group on continuous glucose monitoring. J Diabetes Sci Technol, 2024; 19322968241296097; doi: 10.1177/19322968241296097

24.

Oliver

, Reddy

, Leelarathna

. Continuous glucose sensor accuracy: Beyond the headline metric. Lancet Diabetes Endocrinol, 2024; 12(12):934–938; doi: 10.1016/S2213-8587(24)00245-6

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.08 MB

0.39 MB

0.49 MB

0.79 MB

0.15 MB

0.14 MB

0.17 MB