Back to Journals » Diabetes, Metabolic Syndrome and Obesity: Targets and Therapy » Volume 15

Diabetes Complications and Related Comorbidities Impair the Accuracy of FreeStyle Libre, a Flash Continuous Glucose Monitoring System, in Patients with Type 2 Diabetes

Authors Wen X , Zeng N, Zhang N, Ou T, Li X, Li X, Li W, Xu K, Du T 

Received 9 July 2022

Accepted for publication 16 September 2022

Published 3 November 2022 Volume 2022:15 Pages 3437—3445

DOI https://doi.org/10.2147/DMSO.S381565

Checked for plagiarism Yes

Review by Single anonymous peer review

Peer reviewer comments 3

Editor who approved publication: Prof. Dr. Juei-Tang Cheng



Xiaofang Wen,1,* Nan Zeng,1,* Ningbo Zhang,1,2,* Tingting Ou,1 Xiaowei Li,1 Xiaoying Li,1 Wangen Li,1 Kang Xu,3 Tao Du1

1Department of Endocrinology, The Second Affiliated Hospital, Guangzhou Medical University, Guangzhou, 510260, People’s Republic of China; 2Department of Endocrinology, Peking University Shenzhen Hospital, Shenzhen, 518036, People’s Republic of China; 3Guangdong Provincial Key Laboratory of Malignant Tumor Epigenetics and Gene Regulation, Medical Research Center, Sun Yat-sen Memorial Hospital, Sun Yat-sen University, Guangzhou, 510120, People’s Republic of China

*These authors contributed equally to this work

Correspondence: Tao Du; Kang Xu, Email [email protected]; [email protected]

Background: Although flash continuous glucose monitoring systems (FCGM) accuracy has been extensively studied in diabetes, its accuracy is still not fully evaluated in type 2 diabetes (T2D) patients in real-world settings. In the present study, we aim to assess the effects of diabetes complications and related comorbidities on FCGM accuracy in T2D patients with diabetes complications and related comorbidities in the real world.
Methods: FCGM data were collected at eight-time points daily (3 AM, 7 AM, 9 AM, 11 AM, 1 PM, 5 PM, 7 PM, and 9 PM) from 742 patients with T2D and compared with simultaneous fingertip capillary blood glucose (reference blood glucose, REF), and the difference was evaluated using Parkes error grid (PEG), surveillance error grid (SEG), and logistic regression analysis.
Results: In total, 25,579 FCGM/REF data pairs were included in the study. The FCGM values were lower than the paired REF values in 75% of the pairs. The maximum bias (− 23.0%) and maximum mean absolute relative difference (24.5%) were observed at 3 AM among eight-time points. SEG analysis also demonstrated the highest percentage of paired readings in moderate and great risk zone (C and D) at 3 AM than PEG analysis (7.33% vs 0.43%, P< 0.001). According to the SEG classification, hypoglycemia, infection, diabetic foot, diabetic ketoacidosis, and hypertension were independent risk factors that impaired FCGM accuracy in patients.
Conclusion: FCGM commonly underestimates blood glucose levels. Compared with PEG, SEG analysis seems more conducive to the analysis of FCGM performance. The present data highlights the impairment of diabetes complications and related comorbidities on the FCGM accuracy in T2D patients.

Keywords: flash continuous glucose monitoring system, type 2 diabetes, surveillance error grid, Parkes error grid, diabetes complications

Introduction

Self-monitoring of blood glucose (SMBG) is an integral component of diabetes management that aims to improve glycemic control and decrease the risk of diabetes-related complications.1–3 Blood glucose monitoring system (BGMS), including real-time continuous glucose monitoring (CGM) and intermittently scanned CGM (also known as flash continuous glucose monitoring systems, FCGM), can provide continuous measurements of glucose levels and thereby aid diabetes patients in the adjustment of their diet, physical activity, and treatment regimen.4,5 Any error in the BGMS, including slight inaccuracies in the measurements, could adversely affect patient care and treatment.6,7 CGM measures the glucose level in the subcutaneous interstitial fluid to estimate blood glucose. Therefore, it is crucial to evaluate the performance and accuracy of CGM based on analytical precision and clinical accuracy.8

Unlike real-time CGM, FCGM is designed to measure glucose levels without the need for calibration with blood glucose samples, which improves convenience and quality of life in diabetes patients. The FCGM accuracy has been extensively studied in type 1 diabetes (T1D) and type 2 diabetes (T2D).9–12 However, its accuracy remains to be fully explored in T2D, particularly in those with diabetes complications and related comorbidities using emerging analysis. In our previous study, we revealed the accuracy of FCGM reading out of range should be cautiously interpreted, especially as reading below the meter range.13 We also reported that T2D patients with hypertension had a higher frequency of readings below the meter range.13 In the present study, we aimed to evaluate the effect of diabetes complications and related comorbidities on the performance of FCGM in T2D patients. In addition, two methods for assessing clinical characteristics affecting FCGM accuracy, the Parkes error grid (PEG) and surveillance error grid (SEG), were compared.

Methods

Patients

The paired values of glucose meter and FCGM from 742 adult patients with T2D were analyzed in the present study. All patients were from the Second Affiliated Hospital of Guangzhou Medical University between January 9, 2018, and January 7, 2020. The patients were hospitalized for newly diagnosed or poorly controlled diabetes and were enrolled for using FCGM. Inpatients were ineligible for application of glucose meter and FCGM: 1) if they had shock, unconsciousness, or other conditions that could affect their cooperation in the study, 2) if they had coagulopathy, infectious diseases, or other disorders, in the opinion of the clinicians, would put the patients at risk. This study complied with the Declaration of Helsinki and was performed according to the protocol approved by the academic ethics review boards of the hospital. Informed consents were obtained from all patients. Our study complies with the Declaration of Helsinki.

Data Collection and Analysis/Study Design

The specialist nurses subcutaneously placed a 4–5 mm linear probe (FCGM sensor, FreeStyle Libre, Abbott Diabetes Care) on the outer side of the left upper arm of the patient. The data were collected the day after the sensor installation and the correction of diabetic ketoacidosis. The indicators of correction of DKA included: 1) negative serum and urinary ketones, 2) normal serum electrolytes and serum osmolality, 3) normal arterial blood gases, and 4) reducing blood glucose and correcting syndromes of dehydration. The specialist nurses performed SMBG and scanned the sensor at the same time. All patients were instructed to consume a diabetic diet based on their weight at fixed meal times: 7 AM, 11 AM, and 5 PM. There were no extra meals or strenuous exercise during hospitalization.

The glucose levels in the fingertip capillary blood were measured at eight-time points (3 AM, 7 AM, 9 AM, 11 AM, 1 PM, 5 PM, 7 PM, and 9 PM) using a blood glucose meter (Accu-Chek Performa, Roche) as the reference blood glucose (REF) to compare with FCGM data at the same time point. The consistency and difference between FCGM readings and REF were evaluated using the mean absolute relative difference (MARD), Bias, Bland-Altman plots, PEG, and SEG.

Error Grid Analysis

PEG was constructed according to the recommendations previously reported by DuBois et al14 and outlined by Pfützner et al.15 SEG analysis was conducted using the Excel macro program available in R software and the Diabetes Technology Society website (www.diabetestechnology.org/SEGsoftware).16

Statistical Analysis

The Statistical Package for the Social Sciences software (version 22.0) was used to perform all statistical analyses. The Shapiro–Wilk test was performed for assessing the normality of data, and nonparametric statistical tests were used for assessing the non-normality of data. Logistic regression analysis was performed to identify the risk factors that affected FCGM accuracy in patients. P<0.05 indicated statistical significance.

Results

Among the 742 patients included in the study (Table 1), elderly patients aged over 60 years accounted for 62.9% of the total cohort, with a median age of 64 years. Twenty-seven patients (3.6%) had diabetic ketoacidosis (DKA), forty-five patients (6.1%) had diabetic foot, and 277 patients (37.3%) had previous cardiovascular and cerebrovascular events. A total of 25,579 paired data in FCGM range from 742 patients were evaluated. The consistency and difference between FCGM readings and REF were evaluated using Bias, Bland-Altman plots, MARD, and CV (Figure 1). The Bland-Altman plots presented a greater error in the lower blood glucose, which gradually decreased as REF increased (Figure 1A); 72.3% of the paired FCGM/REF readings showed FCGM values below REF, even up to 91.5% at the 3 AM time point (Figure 1B). The maximum bias (−23.0%) and maximum MARD value (24.5%) were also observed at the 3 AM time point, which was at least 10% greater than those at other time points (Figure 1C). Moreover, the FCGM accuracy according to the ISO 15197:2013 standard showed the highest percentage (50.0%) of above 20 mg/dL or 20% at 3 AM among eight-time points (Figure 1D).

Table 1 Clinical Characteristics of Patients (n = 742)

Figure 1 The consistency and difference between FCGM readings and REF in patients. Bland-Altman plots (A); comparison of FCGM value and REF value (B); Bias, MARD, CV, 95% limit of agreement (C); difference according to the ISO 15197:2013 standard (D). Bias: mean relative difference between FCGM and REF (FCGM-REF)/REF; MARD: |FCGM-REF|/REF; CV: standard deviation of the relative difference between FCGM and REF. Lower 95% Limit of Agreement: Bias - 1.96 × CV; Upper 95% Limit of Agreement: Bias + 1.96 × CV.

To optimally assess the accuracy of continuous glucose sensors, PEG and SEG were compared for evaluating the FCGM accuracy by plotting the paired data in the error grids (Figure 2, Supplementary Table 1). In contrast to the clear boundaries of a large range of divisions of PEG, the SEG with 15 color-coded risk levels (Supplementary Table 2) shows the progressive and continuous boundaries between the different strata. SEG analysis showed the significantly higher percentage of paired readings in risk zone C and D than PEG analysis (7.33% vs 0.43%, P<0.001) (Figure 2). We also found the highest proportion of paired readings in SEG risk zone C and D at the 3 AM time point (40.3%) compared with other time points. Of 29 paired readings in SEG risk zone D, only two pairs were in Parkes risk D, three pairs in Parkes risk B, and 24 pairs in Parkes risk C (Supplementary Table 3). Notably, 28 FCGM values of 29 FCGM/REF paired readings in SEG risk zone D were below 70 mg/dl, but all 29 paired REF readings were above 110 mg/dl and 9 of them were above 200 mg/dl, which suggested the severe clinical risk of FCGM may be primarily the false alarm of hypoglycemia.

Figure 2 Comparison of the distribution of paired reading in risk zone in SEG and PEG.

Then, to analyze the effects of clinical characteristics on FCGM accuracy, the patients were divided into two groups according to whether their paired values included SEG risk zone C/D distribution. Compared to those without risk zone C/D readings, 415 patients (55.9%) with risk zone C/D readings displayed: longer diabetes duration, higher hypoglycemia incidence, increased 24-h microalbuminuria and decreased eGFR, higher ratio of infection, increased incidences of DKA and hypertension, and higher incidence of cardiovascular and cerebrovascular events (P<0.05) (Table 2). The results of multivariate logistic regression analysis showed that hypoglycemia (OR 1.98 [95% CI 1.344–2.916], P=0.001), infection (OR 2.328 [95% CI 1.553–3.491], P<0.001), diabetic foot (OR 2.658 [95% CI 1.187–5.951], P=0.017), DKA (OR 2.602 [95% CI 1.005–6.735], P=0.017), and hypertension (Grade 2: OR 1.895 [95% CI 1.213–2.961], P=0.005; Grade 3: OR 1.609 [95% CI 1.099–2.355], P=0.014) were independent risk factors that affect FCGM accuracy in hospitalized patients (Table 3).

Table 2 Clinical Characteristics of Patients with SEG C/D Risk Zones

Table 3 Independent Factors Affect the Accuracy of FCGM

Discussion

Although the reliability and credibility of FCGM in glucose monitoring have been extensively discussed, its accuracy is still not fully documented in type 2 diabetes. Furthermore, methods using Clark Error Grid (CEG) and PEG have limited the accuracy in assessing FCGM precision and reliability, which has been reported to be less accurate in the hypoglycemic range.13,17–19 In the present study, 25,579 paired readings from 742 patients were evaluated. We found that FCGM mostly underestimated blood glucose levels, particularly at night with the least accuracy. SEG analysis revealed that diabetes complications and related comorbidities could impair FCGM accuracy.

In our study, the hypoglycemic state in T2D patients affected FCGM accuracy, especially nocturnal hypoglycemia. A study reported MARD of FCGM was generally 9.56–15.4%.11 Bonora et al suggested that CGM accuracy at a MARD of 10–17% was clinically acceptable. In the present study, overall MARD was 14%.9 However, the MARD at 3 AM time point was extremely high, up to 24.5%, which could be a result of the higher frequency of hypoglycemia at 3 AM. Some studies have revealed poor FCGM accuracy at low blood glucose level. For example, Olafsdottir et al found that the MARD was 20.3% when the blood glucose level was <72 mg/dL,18 and Moser et al also reported that the MARD of FCGM was even as high as 31.6% during hypoglycemia.17 Moreover, our results indicated that in patients who experienced hypoglycemia, the high-risk rate of FCGM/REF pairs was twice as high as in patients without hypoglycemia. It is partially because that hypoglycemia could affect the interstitial glucose concentration by affecting blood flow and vascular permeability.20 And according to the ISO 15197:2013 standard, CGM should comply with the 95% accuracy criteria. However, overall FCGM accuracy was only 63.1% and decreased to 34% at 3 AM in our study. These findings demonstrated the importance of interpreting FCGM data and improving FCGM technology in the setting of hypoglycemia.

The error grid is a commonly used tool for evaluating FCGM accuracy per the ISO 15197:2013 standard criteria.21 When the CEG and PEG were developed (in the late 1980s and the early 1990s), SMBG was a relatively new instrument with a relatively low analytical accuracy, and CEG and PEG were effective in assessing its accuracy, although the methods for evaluating SMBG accuracy were not rigorous enough. With a significant improvement in SMBG performance in recent years, the CEG and PEG seem no longer applicable to properly assess the accuracy of new equipment.16 In CEG and PEG, every point within the same discrete risk zones has the same degree of risk. Compared to CEG and PEG, the SEG is a continuous and three-dimensional graph plotted on two dimensions that allow each point to have its risk value. The SEG appears more capable of assessing the clinical risk of device errors than the CEG and PEG.16 In the present study, although PEG showed a similar overall evaluation trend to SEG in the paired blood glucose assessment, the SEG represented detailed stratification and high sensitivity for the paired data with potential clinical risk. Chandnani et al also reported that high-risk levels were more frequent when analyzing paired readings with SEG than with PEG.22 Compared with PEG, our data revealed the SEG should be more conducive to properly evaluating FCGM performance and identifying potential adverse clinical decisions. For example, in Supplementary Table 3, the No. 22 paired reading (FCGM 49 mg/dL vs REF 142 mg/dL) was displayed in SEG risk zone D, but in PEG risk zone B. The risk zone B in PEG is clinically acceptable; however, it should be an unacceptable risk because the clinical decision may lead to adverse consequences according to the FCGM value (49 mg/dL). In the No. 22 case, the condition would be mistaken for hypoglycemia and would increase to a higher level if the patient is given rapid sugar supplementation. Obviously, it completely deviates from the goal of blood glucose self-management. In addition, we found that almost all FCGM values in SEG risk zone D indicated hypoglycemia, whereas paired REF values indicated the absence of hypoglycemia. These results suggest that FCGM, while being highly sensitive in reporting hypoglycemia, is associated with low specificity and potentially high clinical risk.

Diabetes comorbidities and complications are common in T2D patients, but their impact on FCGM accuracy remains to be fully explored. It is known that more than 5% of diabetes patients experience foot ulcers, and the cumulative lifetime rate could be up to 15%,23 which is the main reason for amputation in diabetes patients. DKA is also one of the most common conditions in patients with poorly managed diabetes.24 Both DKA25 and hypertension26 could cause endothelial dysfunction through oxidative stress, which could affect microcirculation and glucose delivery to the interstitium.27 Hypertension also increases the risk of progression of microvascular complications in diabetes patients, such as diabetic nephropathy and retinopathy.28,29 Undergoing hemodialysis in T2D patients with diabetic nephropathy could elevate MRAD and deteriorate the FCGM accuracy.30 These studies suggest the necessity of exploring the effect of diabetes complications and related comorbidities on FCGM accuracy in patients in real-world settings. Reineke et al reported that DKA influenced CGMS accuracy (CGMS Gold, Medtronic MiniMed, Northridge, CA) using CEG and PEG analysis in 13 dogs and 11 cats.31 But the subjects were animals, not patients, which limited its interpretation in the clinical setting. In the present study, we found that infection, diabetic foot, DKA, and hypertension were independent risk factors for a high-risk value in patients using SEG analysis.

Conclusion

This study has both limitations and strengths. The study collected data from patients with T2D only and did not include patients with T1D, which would limit the generalizability of the results in diabetes. Moreover, the reference method was not sophisticated methods such as PCA-hexokinase method and the isotope dilution gas chromatography-mass spectrometry method, nor did the study include the additional blood glucose meter, to obtain a more desirable assessment of the FCGM performance. Nevertheless, our data delivers the valuable evaluation of the FCGM system in real-world settings. Our study indicates the lowest FCGM accuracy at 3 AM compared with other daily timepoints, and diabetes complications impair FCGM accuracy. Moreover, compared with PEG analysis, SEG could more accurately evaluate the FCGM performance. Overall, the present data highlights the importance of properly interpreting FCGM data in T2D patients, especially with diabetes complications or at nocturnal, to avoid improper intervention.

Data Sharing Statement

The data used to support the findings of this study may be released upon application to the Second Affiliated Hospital of Guangzhou Medical University, which can be contacted through Prof. Tao Du (email: [email protected]).

Disclosure of Ethical Statements

This study was approved by the academic ethics review boards of the hospital, the Second Affiliated Hospital of Guangzhou Medical University. Informed consents were obtained from all patients.

Acknowledgments

The authors thank all participants in this study. This work was supported by grants from the Guangzhou Municipal Science and Technology Project (201607010036), and the Guangdong Basic and Applied Basic Research Foundation (2020A1515010060, 2021A1515011333). Xiaofang Wen, Nan Zeng, and Ningbo Zhang are co-first authors for this study.

Disclosure

The authors declare no conflicts of interest in this work.

References

1. Martin S, Schneider B, Heinemann L., et al. Self-monitoring of blood glucose in type 2 diabetes and long-term outcome: an epidemiological cohort study. Diabetologia. 2006;49(2):271–278. doi:10.1007/s00125-005-0083-5

2. Shah VN, DuBose SN, Li Z, et al. Continuous glucose monitoring profiles in healthy nondiabetic participants: a multicenter prospective study. J Clin Endocrinol Metab. 2019;104(10):4356–4364. doi:10.1210/jc.2018-02763

3. Ruan Y, Zhong J, Chen R, et al. Association of body fat percentage with time in range generated by continuous glucose monitoring during continuous subcutaneous insulin infusion therapy in type 2 diabetes. J Diabetes Res. 2021;2021:5551216. doi:10.1155/2021/5551216

4. Kan K, Zhu W, Lu F, et al. Contribution of structured self-monitoring of blood glucose to the glycemic control and the quality of life in both insulin- and noninsulin-treated patients with poorly controlled diabetes. Diabetes Technol Ther. 2017;19(12):707–714. doi:10.1089/dia.2017.0275

5. Malanda UL, Bot SD, Nijpels G. Self-monitoring of blood glucose in noninsulin-using type 2 diabetic patients: it is time to face the evidence. Diabetes Care. 2013;36(1):176–178. doi:10.2337/dc12-0831

6. Breton MD, Kovatchev BP. Impact of blood glucose self-monitoring errors on glucose variability, risk for hypoglycemia, and average glucose control in type 1 diabetes: an in silico study. J Diabetes Sci Technol. 2010;4(3):562–570. doi:10.1177/193229681000400309

7. Budiman ES, Samant N, Resch A. Clinical implications and economic impact of accuracy differences among commercially available blood glucose monitoring systems. J Diabetes Sci Technol. 2013;7(2):365–380. doi:10.1177/193229681300700213

8. Krouwer JS, Cembrowski GS. A review of standards and statistics used to describe blood glucose monitor performance. J Diabetes Sci Technol. 2010;4(1):75–83. doi:10.1177/193229681000400110

9. Bonora B, Maran A, Ciciliot S, Avogaro A, Fadini GP. Head-to-head comparison between flash and continuous glucose monitoring systems in outpatients with type 1 diabetes. J Endocrinol Invest. 2016;39(12):1391–1399. doi:10.1007/s40618-016-0495-8

10. Yaron M, Roitman E, Aharon-Hananel G, et al. Effect of flash glucose monitoring technology on glycemic control and treatment satisfaction in patients with type 2 diabetes. Diabetes Care. 2019;42(7):1178–1184. doi:10.2337/dc18-0166

11. Mancini G, Berioli MG, Santi E, et al. Flash glucose monitoring: a review of the literature with a special focus on type 1 diabetes. Nutrients. 2018;10(8):992. doi:10.3390/nu10080992

12. Elhadd T, Bashir M, Baager KA, et al. Mitigation of hypoglycemia during Ramadan using the flash glucose monitoring system following dose adjustment of insulin and sulphonylurea in patients taking multiple glucose-lowering therapies (The PROFAST-IT Study). Diabetes Res Clin Pract. 2021;172:108589. doi:10.1016/j.diabres.2020.108589

13. Wen X, Ou T, Zeng N, et al. Clinical interpretation of the flash continuous glucose monitoring data out of range in type 2 diabetes: a real-world study. Diabetes Res Clin Pract. 2021;178(8):108989. doi:10.1016/j.diabres.2021.108989

14. DuBois JA, Slingerland RJ, Fokkert M, et al. Bedside glucose monitoring-is it safe? A new, regulatory-compliant risk assessment evaluation protocol in critically ill patient care settings. Crit Care Med. 2017;45(4):567–574. doi:10.1097/CCM.0000000000002252

15. Pfutzner A, Klonoff DC, Pardo S, Parkes JL. Technical aspects of the Parkes error grid. J Diabetes Sci Technol. 2013;7(5):1275–1281. doi:10.1177/193229681300700517

16. Klonoff DC, Lias C, Vigersky R, et al. The surveillance error grid. J Diabetes Sci Technol. 2014;8(4):658–672. doi:10.1177/1932296814539589

17. Moser O, Eckstein ML, McCarthy O, et al. Performance of the Freestyle Libre flash glucose monitoring (flash GM) system in individuals with type 1 diabetes: a secondary outcome analysis of a randomized crossover trial. Diabetes Obes Metab. 2019;21(11):2505–2512. doi:10.1111/dom.13835

18. Olafsdottir AF, Attvall S, Sandgren U, et al. A clinical trial of the accuracy and treatment experience of the flash glucose monitor freestyle libre in adults with type 1 diabetes. Diabetes Technol Ther. 2017;19(3):164–172. doi:10.1089/dia.2016.0392

19. Cao B, Wang R, Gong C, et al. An evaluation of the accuracy of a flash glucose monitoring system in children with diabetes in comparison with venous blood glucose. J Diabetes Res. 2019;2019:4845729. doi:10.1155/2019/4845729

20. Koschinsky T, Heinemann L. Sensors for glucose monitoring: technical and clinical aspects. Diabetes Metab Res Rev. 2001;17(2):113–123. doi:10.1002/dmrr.188

21. Bedini JL, Wallace JF, Pardo S, Petruschke T. Performance evaluation of three blood glucose monitoring systems using ISO 15197: 2013 accuracy criteria, consensus and surveillance error grid analyses, and insulin dosing error modeling in a hospital setting. J Diabetes Sci Technol. 2015;10(1):85–92. doi:10.1177/1932296815609368

22. Chandnani SR, Ramakrishna CD, Dave BA, Kothavade PS, Thakkar AS. A post-marketing surveillance study to evaluate performance of the EXIMO™ blood glucose monitoring system. J Clin Diagn Res. 2017;11(5):Fc29–fc32. doi:10.7860/jcdr/2017/26596.9892

23. Boulton AJ. The diabetic foot: from art to science. The 18th Camillo Golgi lecture. Diabetologia. 2004;47(8):1343–1353. doi:10.1007/s00125-004-1463-y

24. Boulton AJM, Armstrong DG, Kirsner RS, et al. Diagnosis and management of diabetic foot complications. Compendia. 2018. doi:10.2337/db20182-1

25. Chaudhuri A, Umpierrez GE. Oxidative stress and inflammation in hyperglycemic crises and resolution with insulin: implications for the acute and chronic complications of hyperglycemia. J Diabetes Complications. 2012;26(4):257–258. doi:10.1016/j.jdiacomp.2012.04.016

26. Higashi Y, Sasaki S, Nakagawa K, Matsuura H, Oshima T, Chayama K. Endothelial function and oxidative stress in renovascular hypertension. N Engl J Med. 2002;346(25):1954–1962. doi:10.1056/NEJMoa013591

27. Bakker W, Eringa EC, Sipkema P, van Hinsbergh VW. Endothelial dysfunction and diabetes: roles of hyperglycemia, impaired insulin signaling and obesity. Cell Tissue Res. 2009;335(1):165–189. doi:10.1007/s00441-008-0685-6

28. Adler AI, Stratton IM, Neil HA, et al. Association of systolic blood pressure with macrovascular and microvascular complications of type 2 diabetes (UKPDS 36): prospective observational study. BMJ. 2000;321(7258):412–419. doi:10.1136/bmj.321.7258.412

29. Chen G, McAlister FA, Walker RL, Hemmelgarn BR, Campbell NR. Cardiovascular outcomes in Framingham participants with diabetes: the importance of blood pressure. Hypertension. 2011;57(5):891–897. doi:10.1161/hypertensionaha.110.162446

30. Toyoda M, Murata T, Saito N, et al. Assessment of the accuracy of an intermittent-scanning continuous glucose monitoring device in patients with type 2 diabetes mellitus undergoing hemodialysis (AIDT2H) study. Ther Apher Dial. 2021;25(5):586–594. doi:10.1111/1744-9987.13618

31. Reineke EL, Fletcher DJ, King LG, Drobatz KJ. Accuracy of a continuous glucose monitoring system in dogs and cats with diabetic ketoacidosis. J Vet Emerg Crit Care. 2010;20(3):303–312. doi:10.1111/j.1476-4431.2010.00538.x

Creative Commons License © 2022 The Author(s). This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution - Non Commercial (unported, v3.0) License. By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms.