https://orcid.org/0000-0002-6649-8644
Figures
Abstract
Background
Lowering glucose levels is a complex task for patients with type 1 diabetes, and they often lack contact with health care professionals. Intermittently scanned continuous glucose monitoring (isCGM) has the potential to aid them with blood glucose management at home. The aim of this study was to investigate the long-term effect of isCGM on HbA1c in type 1 diabetes patients with poor glycaemic control in a region-wide real-world setting.
Methods
All patients with type 1 diabetes receiving an isCGM due to poor glycaemic control (≥70 mmol/mol [≥8.6%]) in the period of 2020–21 in Region North Denmark (“T1D-CGM”) were compared with all type 1 diabetes patients without isCGM (“T1D-NOCGM”) in the same period. A multiple linear regression model adjusted for age, sex, diabetes duration and use of continuous subcutaneous insulin infusion was constructed to estimate the difference in change from baseline HbA1c between the two groups and within subgroups of T1D-CGM.
Results
A total of 2,527 patients (T1D-CGM: 897; T1D-NOCGM: 1,630) were included in the study. The estimated adjusted difference in change from baseline HbA1c between T1D-CGM vs T1D-NOCGM was -5.68 mmol/mol (95% CI: (-6.69 to -4.67 mmol/mol; p<0.0001)). Older patients using isCGM dropped less in HbA1c.
Citation: Jensen MH, Cichosz SL, Gustenhoff P, Nikontovic A, Hejlesen O, Vestergaard P (2022) Long-term glucose-lowering effect of intermittently scanned continuous glucose monitoring for type 1 diabetes patients in poor glycaemic control from Region North Denmark: An observational real-world cohort study. PLoS ONE 17(10): e0274626. https://doi.org/10.1371/journal.pone.0274626
Editor: Francesca D’Addio, Ospedale San Raffaele, ITALY
Received: May 4, 2022; Accepted: August 31, 2022; Published: October 14, 2022
Copyright: © 2022 Jensen et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Data cannot be published because they contain identifying and sensitive patient information. Access to the patient data through a remote desktop connection can be gained after acceptance of a research/quality/development project application by the local approval body at Aalborg University Hospital or North Denmark Regional Hospital. Contact information: Forskningsdata og Statistik, Forskningens Hus, Sdr. Skovvej 15 9000 Aalborg, Denmark, e-mail: forskningsanmeldelse@rn.dk.
Funding: The author(s) received no specific funding for this work.
Competing interests: Author Jensen holds shares in Novo Nordisk and has received consultant fees from Abbott Laboratories A/S. Author Gustenhoff has received consultant fees from Abbott Laboratories A/S. Author Vestergaard has received unrestricted grants from MSD and Servier and travel grants from Amgen, Eli Lilly, Novartis, Sanofi Aventis, and Servier. This does not alter our adherence to PLOS ONE policies on sharing data and materials. There are no patents, products in development or marketed products associated with this research to declare.
Abbreviations: T1D, Type 1 diabetes; isCGM, intermittently scanned continuous glucose monitoring; HbA1c, glycated hemaglobin A1c; SD, standard deviation; CSII, continuous subcutaneous insulin infusion
Introduction
Lowering glucose levels is not a trivial task for patients with type 1 diabetes (T1D) and includes a complex interplay between several factors in situations without close contact with health care professionals [1, 2]. Several studies have indicated that visualization of interstitial glucose from intermittently scanned continuous glucose monitoring (isCGM) leads to optimized glycaemic control by providing the users with a deeper insight into their own glucose metabolism unveiling some of the factors’ interplay [3–11]. Furthermore, the addition of real-time alarms in newer isCGM devices might lead to further improvement [12]. However, these studies are typically of a shorter duration of 3–6 months and with strict inclusion/exclusion criteria [13–15]. It is suspected that digital interventions, such as isCGM, have a great impact on glycaemic control in the first half year with a less pronounced or even no effect at later stages, and studies with longer durations in real-world settings without close contact with health care professionals are therefore needed [16]. Furthermore, several indications for the use of isCGM are sometimes present in the patient groups under investigation, which may blur glucose-lowering treatment effects [17]. To sum up, we hypothesized that the effect of isCGM in patients with T1D is greatest in the first half year and is thereafter reduced. To assess the real-world long-term effect of isCGM including both devices with and without alarms is novel and a needed research study for the benefit of patients, clinicians, and decision makers.
The aim of this study was to investigate the long-term effect of isCGM on HbA1c in patients with T1D in a region-wide real-world setting.
Methods
Study design
This was a region-wide real-world cohort study of T1D patients receiving an isCGM device (Freestyle Libre 1 or 2, Abbott Laboratories, Illinois, US) due to poor glycaemic control in the period of 2020–21 in Region North Denmark. isCGM devices for patients with T1D are reimbursed based on nationally defined indications, such as poor glycaemic control (HbA1c ≥ 70 mmol/mol [≥ 8.6%]), pregnancy and hypoglycaemia unawareness, but the focus in this study was patients with poor glycaemic control. The baseline period was defined as the period 12 months before to the date of receiving the isCGM (the index date). The inclusion criteria were adults (≥19 years old at index date) with at least one HbA1c value above 70 mmol/mol in the baseline period to exclude patients receiving isCGM for other indications. All patients with T1D receiving an isCGM (“T1D-CGM”) were compared with all T1D patients who did not receive an isCGM device (“T1D-NOCGM”) with respect to HbA1c. CGM metrics were not available. Patients in the T1D-NOCGM group inherited the date of initiation of isCGM as an index date from the T1D-CGM patient with the closest year of birth.
Education
In Region North Denmark, patients receive education in the use of isCGM. At the first outpatient visit at the Department of Endocrinology, patients in groups of eight receive the isCGMs and receive a 2-hour training program with diabetes nurses on how to use the device and how to interpret the data. Then, a half-hour individual session is held with a diabetes nurse. Three months after, an outpatient visit with a similar structure follows for the same group of patients. Between visits, patients are encouraged to contact the Department of Endocrinology in case of questions or technological challenges.
Follow-up and outcome
Patients in the cohort were followed from 12 months before the index date to 24 months after in timepoints of quarterly intervals: -12, -9, -6,…,18, 21, 24. HbA1c values were extracted in the 3-year follow-up period, and each value was carried forward to the nearest timepoint. For example, an HbA1c value 45 days after the index date was carried forward to month 3. If a patient had multiple HbA1c values within a quarter the average was carried forward. The outcome of the study was change from baseline in HbA1c. Baseline HbA1c for a patient was calculated as the average of all HbA1c values for the given patient within the baseline period. This baseline HbA1c was chosen instead of using the value at the index date to reduce bias from randomly high HbA1c levels at the time of consideration of isCGM as treatment intensification by the diabetes specialist.
Statistical analysis
Descriptive statistics are presented as means with standard deviations (SDs) or counts and percentages of patients. Mann–Whitney U tests, t tests and χ2 tests were used to present statistical differences in patient characteristics. Age and diabetes duration were calculated from the date of birth and the date of diabetes diagnosis, respectively, to the index date. Mean and standard error of the mean of change from baseline HbA1c are presented for T1D-CGM vs T1D-NOCGM in a graph together with statistical significance notation derived from a multiple linear regression model of change from baseline HbA1c as dependent variable and timepoints as a categorical independent variable. The reference was the baseline HbA1c added as a dummy timepoint. Another change from baseline was calculated as the difference between baseline HbA1c and the mean of all HbA1c values after the index date. The mean and 95% confidence interval for this change from baseline HbA1c are presented in a table for T1D-CGM and T1D-NOCGM together with p-values from t-tests. Furthermore, the adjusted difference between this change from baseline for T1D-CGM versus T1D-NOCGM is shown and was derived from a multiple linear regression model adjusted for sex, age, diabetes duration and use of continuous subcutaneous insulin infusion (CSII), and the analysis was repeated within the T1D-CGM group for dichotomous stratifications of HbA1c at baseline, age, sex, and diabetes duration. The stratification cutoff for the numeric variables was objectively defined as mean + SD of the variable (rounded to the nearest whole number).
All analyses were conducted in SAS 9.4 (SAS Institute Inc., Cary, North Carolina, US).
Data sources
Data were extracted from the Region North Denmark Diabetes Dataplatform. The Diabetes Dataplatform is a real-time quality and research database collecting data from multiple hospital systems, including demographics, laboratory, and prescription data for all patients with diabetes who have had any contact with the hospitals in Region North Denmark. HbA1c values were extracted from the local laboratories where blood samples undergo state-of-the-art analysis. Blood sampling was performed by diabetes nurses, lab technicians or endocrinologists at the patients’ outpatient visits, The validity of the HbA1c values in this study is therefore deemed high.
Results
In total, 3,426 people had T1D in the Region North Denmark, and 1,424 received an isCGM, see Fig 1. After applying the inclusion criteria, the final cohort consisted of 2,527 patients with T1D (T1D-CGM: 897; T1D-NOCGM: 1,639). The baseline characteristics of the final cohort are shown in Table 1. In Table 2, the mean change in baseline HbA1c and number of hospital visits can be seen for T1D-CGM and T1D-NOCGM. Fig 2 shows the mean change in baseline HbA1c for T1D-CGM and T1D-NOCGM during the three years follow-up, and Fig 3 shows the mean change in baseline HbA1c for subgroups within the T1D-CGM group in the follow-up period. Table 3 shows the estimated adjusted difference in change from baseline HbA1c between T1D-CGM and T1D-NOCGM and between subgroups within T1D-CGM. In Fig 2, HbA1c levels drop with approx. 7 mmol/mol after receiving an isCGM, and the drop remains fairly stable for the following 24 months. In Fig 3, baseline HbA1c level is one of the most clinically significant predictors of the effect of isCGM, and the effect on patients with baseline HbA1c above 86 mmol/mol is greater than for patients with baseline HbA1c below 86 mmol/mol. Age is also a clinically significant predictor where isCGM for patients above 67 years seems to have a small effect on HbA1c.
The statistically significant asterisk notation is from a multiple linear regression model with change from baseline HbA1c as the dependent variable, time as the independent variable and HbA1c at baseline as the reference, i.e. the statistical significant notation refers to baseline within the group. Available HbA1c values in the quarter before each time point contributed to the means.
The upper left graph (Fig 3A) shows patients with HbA1c ≤ 86 mmol/mol at baseline versus patients with HbA1c > 86 mmol/mol at baseline. The upper right graph (Fig 3B) shows patients aged ≤ 68 years versus patients aged > 68 years. The bottom left graph (Fig 3C) is females versus males. The bottom right graph (Fig 3D) shows patients with a diabetes duration ≤ 21 years versus patients with a diabetes duration > 21 years. The estimates were derived from general linear models with change from baseline in HbA1c as the dependent variable, time as the independent variable and HbA1c at baseline as the reference. Available HbA1c values in the quarter before each time point contributed to the estimates. Statistically significant changes are marked with an asterisk and refers to baseline within the group.
Estimates were derived from t tests. p value for hospital visits is for test of difference between groups, whereas p values for change from baseline HbA1c is difference before and after baseline within each group.
Estimates and p values are from multiple linear regression models adjusted for age, sex, diabetes duration and use of CSII (Y/N).
Discussion
This study sought to investigate the real-world long-term effect of isCGM on HbA1c in patients with T1D with poor glycaemic control. We found that patients using isCGM experienced clinically significantly lower HbA1c levels, which persisted for the 24-month period after initiation, and our hypothesis is therefore rejected based on the results of this study.
In an observational study by Lameijer et al. [18], a similar but lower sustained HbA1c improvement of 3.5 mmol/mol was observed among 342 patients with type 1 or 2 diabetes in a 24-month period after initiation of isCGM. Selection bias might have occurred in the study as only half of the invited patients accepted a follow-up. Furthermore, HbA1c values were self-reported, which decreases validity. In another observational study by Nathanson et al. [17], a large cohort of Swedish T1D users of isCGM were compared by propensity score matching with a group of T1D without isCGM, and they found an estimated difference in change from baseline HbA1c of -2.5 mmol/mol for patients with baseline HbA1c ≥ 70 mmol/mol. The indications for isCGM were not clear from the work, and whether the included patients received isCGM for indications other than poor glycaemic control is unknown. If HbA1c ≥ 70 mmol/mol was an indication for reimbursement, matching on HbA1c values might be problematic as patients from the control group with HbA1c ≥ 70 mmol/mol who did not receive isCGM are probably not representative of the general population of T1D. In our study, we found a slightly higher estimated difference between T1D-CGM and T1D-NOCGM, which may be caused by the varying study designs. However, the thorough education in the use of isCGM in Region North Denmark might also have affected the results [13, 19].
Several limitations in our study exist. First, the study was not randomized, and the patients initiating isCGM may have been more motivated to handle their own disease, and other interventions could have had the same glucose-lowering effect. Since the indication HbA1c ≥ 70 mmol/mol for reimbursement of isCGM is objective, we believe this selection bias has been minimized though. Another issue with the lack of randomization is that our T1D-NOCGM group is not comparable with the T1D-CGM group. For example, T1D-CGM patients have much higher HbA1c at baseline due to the inclusion criterion of HbA1c ≥ 70 mmol/mol, but it was not possible to use the same criterion on T1D-NOCGM patients, because patients who did not initiate isCGM despite of the indication are not representative of T1D patients in general. Another issue with the study design is that other interventions/changes in the treatment of patients with diabetes in Region North Denmark in the same period could influence our results. However, our addition of a control group with follow-up in the same period makes it possible to assess the effect of any other glucose-lowering interventions for T1D patients. In the same line, data were extracted in the period of COVID-19, which may in general have affected the results. Previous published results show improved glycaemic control during the COVID-19 pandemic [20]. Finally, we lack important information, such as, body mass index, insulin dosing, type of CGM device (Freestyle Libre 1 or 2), type of CSII and device use, which could have strengthened our analyses. For example, compared with Libre 1, the Libre 2 device has real-time alarms to warn about hypo- and hyperglycaemia, which could influence glycaemic control. Furthermore, studies have shown that use of different CSII systems lead to significantly different glycaemic control [21]. Adjustment for the exact type of CSII would thus have strengthened our results.
An inherent risk in observational studies is selection bias, and especially matched designs may be prone to selection bias, because the choice of matched variables is subjective or based on available information. Other important factors, such as motivation, are not available in the data and may potentially be confounding factors. We tried to avoid matching to enable observation of all patients with T1D in the region, which we assess as a strength of the study. Another related strength is the high number of patients included and the regionwide setup. Finally, a strength is that the Diabetes Dataplatform used in this study has undergone significant quality control. For example, the disease codes for all patients in the database have been validated by screening patient journals.
In conclusion, our results indicate that patients with T1D in poor glycaemic control in Region North Denmark in general benefitted from using isCGM with a sustained 24-month improvement in HbA1c. However, older patients seem to have less effect of isCGM on HbA1c. We believe these results are important and should be taken into consideration by decision makers, but the results should be interpreted with care asour study was not randomized.
Acknowledgments
We would like to thank diabetes nurses at Aalborg University Hospital who provided us with the information about education in isCGM in Region North Denmark.
References
- 1. Davidson M, Penney E, Muller B, Grey M. Stressors and self-care challenges faced by adolescents living with type 1 diabetes. Appl Nurs Res. 2004;17(2):72–80. pmid:15154119
- 2. Ramchandani N, Way N, Melkus GD, Sullivan-Bolyai S. Challenges to Diabetes Self-Management in Emerging Adults With Type 1 Diabetes. Diabetes Educ. 2019;45(5):484–97. pmid:31304878
- 3. Secher AL, Pedersen-Bjergaard U, Svendsen OL, Gade-Rasmussen B, Almdal T, Raimond L, et al. Flash glucose monitoring and automated bolus calculation in type 1 diabetes treated with multiple daily insulin injections: a 26 week randomised, controlled, multicentre trial. Diabetologia. 2021;64(12):2713–24. pmid:34495375
- 4. Dicembrini I, Cosentino C, Monami M, Mannucci E, Pala L. Effects of real-time continuous glucose monitoring in type 1 diabetes: a meta-analysis of randomized controlled trials. Acta Diabetol. 2021;58(4):401–10. pmid:32789691
- 5. Szypowska A, Ramotowska A, Dzygało K, Golicki D. Beneficial effect of real-time continuous glucose monitoring system on glycemic control in type 1 diabetic patients: Systematic review and meta-analysis of randomized trials. Eur J Endocrinol. 2012;166(4):567–74. pmid:22096111
- 6. Pickup JC, Freeman SC, Sutton AJ. Glycaemic control in type 1 diabetes during real time continuous glucose monitoring compared with self monitoring of blood glucose: Meta-analysis of randomised controlled trials using individual patient data. Bmj. 2011;343(7815):1–14. pmid:21737469
- 7. Jensen MH, Vestergaard P, Hirsch IB, Hejlesen O. Use of Personal Continuous Glucose Monitoring Device Is Associated With Reduced Risk of Hypoglycemia in a 16-Week Clinical Trial of People With Type 1 Diabetes Using Continuous Subcutaneous Insulin Infusion. J Diabetes Sci Technol. 2020.
- 8. Lind M, Polonsky W, Hirsch IB, Heise T, Bolinder J, Dahlqvist S, et al. Continuous glucose monitoring vs conventional therapy for glycemic control in adults with type 1 diabetes treated with multiple daily insulin injections the gold randomized clinical trial. JAMA—J Am Med Assoc. 2017;317(4):379–87. pmid:28118454
- 9. Beck RW, Riddlesworth T, Ruedy K, Ahmann A, Bergenstal R, Haller S, et al. Effect of continuous glucose monitoring on glycemic control in adults with type 1 diabetes using insulin injections the diamond randomized clinical trial. JAMA—J Am Med Assoc. 2017;317(4):371–8. pmid:28118453
- 10. Tyndall V, Stimson RH, Zammitt NN, Ritchie SA, McKnight JA, Dover AR, et al. Marked improvement in HbA1c following commencement of flash glucose monitoring in people with type 1 diabetes. Diabetologia. 2019;62(8):1349–56. pmid:31177314
- 11. Stimson RH, Dover AR, Ritchie SA, Wright RJ, McKnight JA, Zammitt NN, et al. HbA1c response and hospital admissions following commencement of flash glucose monitoring in adults with type 1 diabetes. BMJ Open Diabetes Res Care. 2020;8(1). pmid:32719078
- 12. d. Al Hayek AA, Robert AA, Al Dawish MA. Effectiveness of the freestyle libre 2 flash glucose monitoring system on diabetes-self-management practices and glycemic parameters among patients with type 1 diabetes using insulin pump. Diabetes Metab Syndr Clin Res Rev. 2021;15(5). pmid:34488057
- 13. Cowart K, Updike W, Bullers K. Systematic Review of Randomized Controlled Trials Evaluating Glycemic Efficacy and Patient Satisfaction of Intermittent-Scanned Continuous Glucose Monitoring in Patients with Diabetes. Diabetes Technol Ther. 2020;22(5):337–45. pmid:31859531
- 14. Castellana M, Parisi C, Di Molfetta S, Di Gioia L, Natalicchio A, Perrini S, et al. Efficacy and safety of flash glucose monitoring in patients with type 1 and type 2 diabetes: a systematic review and meta-analysis. BMJ open diabetes Res care. 2020;8(1). pmid:32487593
- 15. Evans M, Welsh Z, Ells S, Seibold A. The Impact of Flash Glucose Monitoring on Glycaemic Control as Measured by HbA1c: A Meta-analysis of Clinical Trials and Real-World Observational Studies. Diabetes Ther. 2020;11(1):83–95. pmid:31673972
- 16. Klonoff DC, Kerr D. Overcoming Barriers to Adoption of Digital Health Tools for Diabetes. J Diabetes Sci Technol. 2018;12(1):3–6. pmid:28969460
- 17. Nathanson D, Svensson A-M, Miftaraj M, Franzén S, Bolinder J, Eeg-Olofsson K. Effect of flash glucose monitoring in adults with type 1 diabetes: a nationwide, longitudinal observational study of 14,372 flash users compared with 7691 glucose sensor naive controls. Diabetologia. 2021;64:1595–603. pmid:33774713
- 18. Lameijer A, Fokkert MJ, Edens MA, Gans ROB, Bilo HJG, Van Dijk PR. Two-year use of flash glucose monitoring is associated with sustained improvement of glycemic control and quality of life (FLARE-NL-6). BMJ Open Diabetes Res Care. 2021;9(1):1–7. pmid:34521652
- 19. Broos B, Charleer S, Bolsens N, Moyson C, Mathieu C, Gillard P, et al. Diabetes knowledge and metabolic control in type 1 diabetes starting with continuous glucose Monitoring: FUTURE-PEAK. J Clin Endocrinol Metab. 2021;106(8):E3037–48. pmid:33740049
- 20. Van Der Linden J, Welsh JB, Hirsch IB, Garg SK. Real-Time Continuous Glucose Monitoring during the Coronavirus Disease 2019 Pandemic and Its Impact on Time in Range. Diabetes Technol Ther. 2021;23(S1):S1–7. pmid:33470892
- 21. Lunati ME, Morpurgo PS, Rossi A, Gandolfi A, Cogliati I, Bolla AM, et al. Hybrid Close-Loop Systems Versus Predictive Low-Glucose Suspend and Sensor-Augmented Pump Therapy in Patients With Type 1 Diabetes: A Single-Center Cohort Study. Front Endocrinol (Lausanne). 2022;13. pmid:35498423