Continuous glucose monitoring (CGM) technology provides real-time glucose data and the ability to achieve improved glycemic control; however, widespread adoption has been limited (1). Among the challenges to CGM uptake are cost and the inconvenience of changing the sensor.
In this prospective, multicenter study, the Dexcom G4 PLATINUM system was analyzed over 14 days of continuous sensor wear. Each subject wore two sensors simultaneously. After 7 days, subjects left their original sensors in place and began a second 7-day sensor session. This procedure was then repeated with two new sensors for a second 2-week period.
Blood glucose (BG) readings from Bayer CONTOUR NEXT meters served as reference (2), with absolute relative difference (ARD) defined as the percent error between sensor and matched BG values. The Mann-Whitney rank sum test was used for comparison of sensor accuracy for week 1 versus week 2. The data were analyzed in a generalized linear model to account for the within-patient correlation. Sensor survival is presented in Fig. 1 with Kaplan-Meier survival curves for failures due to adhesive failure and for all causes of failure. Sensor accuracy in Fig. 1 is presented as median (interquartile range [IQR] 25th, 75th centile).
Figure 1.
The left y-axis shows the MARD of sensor readings for each day of wear (black circle) with IQR. The right y-axis shows the portion of sensors functioning on each day of wear, with the gray line representing sensor loss related to adhesive failures and the black line representing all sensor failures.
Fifty-seven subjects completed the study (46% male, mean ± SD age 28.7 ± 8.7 years and HbA1c 7.4 ± 0.8% [57 ± 6.6 mmol/mol]). In total, there were 222 2-week sensor sessions available for analysis, and 56% of sensors functioned for the full 2 weeks. Sensor failures occurred later in the 2-week period, and the main cause of sensor removal was related to failure of the tape adhesive (falling off or accidental dislodgment) (n = 65, 29%). Only 10% of sensors were removed for “sensor failure” and 3% for “loss of signal.” Sensors tended to be well tolerated with minimal erythema or induration. There was one sensor site infection, which occurred on day 3.
The median ARD (MARD) across all glucose ranges for week 1 of sensor life (n = 6,639 paired values) was 11.2% (IQR 5.1, 20.5) compared with 10.8% (IQR 5.0, 20.0) during week 2 (n = 4,185 paired values) (P = 0.08). Sensors were more accurate in the hypoglycemic range (BG <70 mg/dL) during week 2 (n = 219, MARD 15.6% [IQR 7.5, 31.3]) compared with week 1 (n = 282, MARD 20.8% [IQR 10.2, 38.2]) (P = 0.007). Accuracy was similar between weeks 1 and 2 in the euglycemic range (BG 70–180 mg/dL) (P = 0.23) and hyperglycemic range (BG >180 mg/dL) (P = 0.30). On day 8, the increase in MARD and variability was due to sensor recalibration, which is intrinsic to the Dexcom calibration algorithm.
Extending CGM sensor life offers a convenience to patients and may result in cost savings. In our study, the majority of sensors lasted the full 14 days, and accuracy was similar between weeks 1 and 2.
Our data suggest that CGM sensors that remain in place for 14 days may be as accurate in the second week and could be used in closed-loop systems, especially if algorithms for sensor failure are available (3). Future studies should analyze the accuracy of newer sensors (4) over 14 days.
Article Information
Acknowledgments. This work could not have been done without the altruistic efforts of our study subjects.
Funding. The authors would like to acknowledge JDRF (grant 17-2013-471) and the Clinical and Translational Science Awards Program funded by the National Center for Advancing Translational Sciences (grant UL1 TR000093) at the National Institutes of Health. For his postdoctoral research, D.J.D. received grant support from the Child Health Research Institute (CHRI) at Stanford University and a JDRF postdoctoral fellowship (grant 2-PDF-2014-114-A-N). Dexcom provided sensors at a research discount for this study.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. D.J.D. and B.A.B. were responsible for study design, acquisition of data, data analysis, and writing the manuscript. T.T.L., R.P.W., L.M., and D.M.M. were responsible for acquisition of data, data analysis, and writing of the manuscript. E.W., D.G., and S.H. assisted with data collection and data analysis, and R.v.E. provided statistical analysis of the data. B.A.B. 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.
Footnotes
Clinical trial reg. no. NCT02199028, clinicaltrials.gov.
References
- 1.Miller KM, Foster NC, Beck RW, et al.; T1D Exchange Clinic Network . Current state of type 1 diabetes treatment in the U.S.: updated data from the T1D Exchange clinic registry. Diabetes Care 2015;38:971–978 [DOI] [PubMed] [Google Scholar]
- 2.DeSalvo DJ, Shanmugham S, Ly TT, Wilson DM, Buckingham BA. Accuracy evaluation of blood glucose monitoring systems in children on overnight closed-loop control. J Diabetes Sci Technol 2014;8:969–973 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Bequette BW. Fault detection and safety in closed-loop artificial pancreas systems. J Diabetes Sci Technol 2014;8:1204–1214 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Laffel L. Improved accuracy of continuous glucose monitoring systems in pediatric patients with diabetes mellitus: results from two studies. Diabetes Technol Ther 2016;18(Suppl. 2):S223–S233 [DOI] [PMC free article] [PubMed] [Google Scholar]