Abstract
Background:
Evidence regarding the implementation of medium-term strategies in advanced hybrid closed-loop (AHCL) system users is limited. Therefore, this study aimed to describe the efficacy and safety of the AHCL system in patients with type 1 diabetes (T1D) on a six-month follow-up in a virtual diabetes clinic (VDC).
Method:
A prospective cohort of adult patients with T1D treated using the AHCL system (Mini Med 780G; Medtronic, Northridge, California) in a VDC follow-up. Standardized training and follow-up were conducted virtually. Clinical data and metabolic control outcomes were reported at baseline, and at three and six months.
Results:
Sixty-four patients (mean age = 42 ± 14.6 years, 65% men, 54% with graduate education) were included. Percentage time in range (%TIR) increased significantly regardless of prior therapy with intermittently scanned continuous glucose monitoring + multiple daily injections and sensor-augmented pump therapy with predictive low-glucose management after starting AHCL and persisted during the follow-up period with no hypoglycemic events. The %TIR 70 to 180 mg/dL according to socioeconomic strata was 73.4% ± 5.3%, 78.1% ± 8.1%, and 84.2% ± 7.5% for the lower, middle, and upper strata, respectively. The sensor was used more frequently in the population with a higher education level. Adherence to sensor use and SmartGuard retention were higher in patients who underwent the VDC follow-up.
Conclusions:
Medium-term follow-up of users of AHCL systems in a VDC contributes to safely achieving %TIR goals. Virtual diabetes clinic follow-up favored adherence to sensor use and continuous SmartGuard use. Socioeconomic strata were associated with a better glycemic profile and education level was associated with better adherence to sensor use.
Introduction
Type 1 diabetes (T1D) is a challenging condition in terms of management for both, individuals and health care systems. Despite its multiple treatment options, only 21% of individuals with T1D achieve adequate metabolic control. 1 Major advances in automated insulin delivery (AID) technologies have occurred in recent years, with improvements in glycemic control, and lower frequency and severity of hypoglycemia. 2 The advanced hybrid closed-loop (AHCL) system includes SmartGuard, a system that continuously anticipates insulin needs and adjusts basal insulin delivery to avoid hyper- and hypoglycemia by automatic adjustments. 3 Various AID devices are used to treat patients with type 1 diabetes. Table 1 provides a summary of the different characteristics of the AID devices, including some of the most frequently found on the market.
AID Systems Main Features.
Abbreviations: AID, automated insulin delivery; PID, proportional integral derivate; MPC, model predictive control; HCP, health care provider.
Both clinical trials and real-world analyses have shown that the AHCL system increases time in range (TIR) to values between 70 and 180 mg/dL, reduces time above range (TAR), prevents severe hypoglycemia (SH), and promotes longer adherence to SmartGuard™, compared with first-generation hybrid closed-loop (HCL) systems.4,5 Most of these studies have included a European population, 6 but the results for a cohort of AHCL users in Latin America were recently published. 7 The results for that cohort were similar to those obtained in real-world studies in Europe, thus suggesting that the AHCL system is effective in diverse populations and may provide similar benefits globally in terms of glucose management and safety.
The implementation of telehealth services for the treatment and management of individuals with T1D has increased in recent years. The advancement of health care technologies has produced various digital tools and skills. Studies on remote training of individuals with T1D using HCL systems have shown that virtual training and education using specific protocols8,9 is safe and helps to achieve %TIR goal. However, evidence on the implementation of this strategy in AHCL system users is limited.
The aim of the present study was to describe the mid-term efficacy and safety of the AHCL system under real-life conditions in T1D patients in a follow-up in a virtual diabetes clinic (VDC).
Methods
This longitudinal real-world study was based on a prospective cohort of patients aged >18 years from different regions of Colombia (Cundinamarca, Valle del Cauca, Bolivar, Atlántico, and Boyacá) diagnosed with T1D who initiated management with the AHCL system (Mini Med 780G, Medtronic, Northridge, California) at the San Ignacio University Hospital in Bogotá, Colombia. Patient recruitment took place between March 2022 and October 2022. Eligible patients were previously managed with multiple daily insulin injections (MDIs), sensor-augmented insulin pump therapy (SAPT) with low-glucose suspend (LGS) (Paradigm VEO, Mini Med; Medtronic) or SAPT with predictive low-glucose management (PLGM) (Mini Med 640G; Medtronic), and HCL systems (Mini Med 670G; Medtronic). Participation in the study and the download of continuous glucose monitoring (CGM) data were authorized by the users by signing an informed consent form. Patients who refused to sign the consent form were excluded. The study was approved by the ethics committee of the San Ignacio University Hospital and the Pontifical Javeriana University, Bogotá, Colombia, with code 0784-22.
All patients were trained in the VDC at San Ignacio University Hospital, which is run by a physician with expertise in diabetes and an education and nutrition team. A schematic representation of the VDC and its operation is provided in Figure 1. Patients received weekly training, for a total of seven sessions, on device features, use in manual and automatic modes, uploading CGM data to Carelink, carbohydrate counting, calibrations, bolus injection, sensor and infusion set changes, hyper- and hypoglycemia protocols, CGM basics, and device alerts. All training sessions were conducted through the Zoom video conferencing platform (Zoom Video Communications, Inc) and Google Meet (Google Inc).
Operation of the virtual diabetes clinic. Representation of the virtual diabetes clinic: (A) Data from the devices are automatically transmitted to a smartphone application (MiniMed Mobile), which provides immediate feedback to the patient. (B) The application transfers the data to the Carelink software where it can be viewed by the endocrinology team. (C) The team accesses the data, which are presented in standardized formats, and makes changes to therapy by recording them in the electronic medical record. (D) The team communicates with the patient to schedule a remote consultation through one of the telemedicine tools.
The device settings were initially programmed according to the manufacturer’s recommendations.10,11 Patients previously treated with SAPT-LGS, SAPT-PGLM, or HCL systems and those who had TIR >70% kept their previous settings. Likewise, for patients with less than 70% time in range (%TIR), adjustments were made to insulin administration, ratios, and correction factor. The Ambulatory Glucose Profile (AGP) was downloaded from the Carelink platform, and modifications were made by the VDC based on observed requirements. Insulin administration, ratios, and correction factor per kilogram of body weight were calculated for patients previously treated with MDI therapy. Patients were instructed to turn on the PLGM feature with a threshold of 60 mg/dL, but a threshold of 70 mg/dL was used in those with a history of SH and hypoglycemia unawareness (HU). The active insulin time (AIT) feature was set to two hours, but a longer time was allowed if the glomerular filtration rate (GFR) was <30 mL/min or if patients had longer AIT time settings in previous technologies. An automatic basal target of 100 mg/dL was initially set. All patients started using the device in manual mode and then switched to SmartGuard mode. This change occurred after 48 hours in patients who had previously used insulin pumps or after one week in those who had been treated with MDI. 11
After completing seven training sessions, one group of patients continued to receive care at the VDC under the supervision of a multidisciplinary team, while another group of patients continued to receive medical care in their health care network. Patients who continued in the VDC had regular monthly follow-ups and reports, which were obtained through Carelink. Those patients who continued with their insurance company were followed according to the protocols of their respective institutions and their metrics were reviewed in the reports obtained by Carelink. Each physician assessing the patient evaluated whether the metrics were off target according to the 2019 TIR consensus. 12 If the metrics were not within the target range, the physician treating the patient would make the necessary adjustments to achieve the appropriate glycemic targets. No interventions were performed on the group being followed by their insurance company. The glycemic control information collected from the Carelink database provided access to all the data collected by the device, both for baseline and three- and six-month post-treatment measurements. At each time point, the CGM data corresponded with the values of the previous 14 days.
The Clarke questionnaire 13 was applied at baseline and at the three-month follow-up by telephone. Severe hypoglycemia events at baseline and after three and six months were also recorded. The population was grouped as lower, middle, and upper strata. Glomerular filtration rate was calculated using the CKD-EPI equation. 14 Severe hypoglycemia was defined as the need for third-party assistance for recovery and HU by scores ≥4 on the Clarke questionnaire. 13 The Diabetes Treatment Satisfaction Questionnaire (DTSQs) was also applied at the end of the initial training on the AHCL system to assess satisfaction with the new treatment. 15 The questionnaire contains eight items, each with seven possible responses scored from 0 (very dissatisfied) to 6 (extremely satisfied). The sum of six of the eight items yields an overall satisfaction score ranging from 0 (lowest possible satisfaction) to 36 (highest possible satisfaction) points. The other two items refer to the patient’s perceived frequency of episodes of hypoglycemia and hyperglycemia, with scores ranging from 0 (never) to 6 (most of the time). These items were analyzed individually and descriptively. 15
Continuous variables are expressed as mean and standard deviation (SD) or as median and interquartile range (IQR), depending on whether the normality assumption was met or not. Categorical variables are presented as absolute and relative frequencies. Time in range (TIR), time below range (TBR) (<54 and <70 mg/dL), TAR (>180 mg/dL and > 250 mg/dL), coefficient of variation (CV), percentage of sensor use, and percentage of SmartGuard retention at baseline and after three and six months were analyzed. The Kruskal-Wallis test 16 was used to determine optimal device use according to socioeconomic strata, education level, and group that continued patient follow-up (multidisciplinary—VDC or only by an endocrinologist). For the analysis of socioeconomic strata, all groups were compared with each other, considering the alternative hypothesis that the median of at least one of the groups was different. The Wilcoxon signed-rank test with continuity correction was used to analyze differences following the change in technology. Changes over time in the Clarke questionnaire scores were assessed using the Stuart-Maxwell marginal homogeneity test. In addition, a subanalysis was performed for elderly patients (defined according to the World Health Organization definition as those aged >60 years), 17 and users in optimal settings (defined as those who spent ≥95% of the time using a glycemic control target of 100 mg/dL and a TIA of two hours). 18 STATA version 16.0 was used for the statistical analysis.
Results
Sixty-four patients (including eight elderly patients) were included in the analysis. Demographic variables are summarized in Table 2. The mean age was 42 ± 14.6 years, with 54.6% having graduate or postgraduate education. The mean baseline HbA1c was 7.5% ± 1.2%, with a mean GFR of 97.5 mL/min/1.73 m2. The most frequent indication to start treatment with integrated systems was the presence of hypoglycemia (40%). In addition, 21% had a Clarke questionnaire score ≥4, and 34% had a history of SH in the previous year.
Baseline Characteristics of the Patients Included in the Study.
Abbreviations: SD, standard deviation; GFR, glomerular filtration rate; HbA1c, glycated hemoglobin; MDI, multiple dose injections; SMBG, self-monitoring blood glucose; isCGM, intermittently scanned continuous glucose monitoring; SAPT-LGS, sensor-augmented pump therapy with low-glucose suspend; SAPT-PLGM, sensor-augmented pump therapy with predicted low-glucose management; HCL, hybrid closed-loop system; DTSQ-s, Diabetes Treatment Satisfaction Questionnaire.
Table 3 and Figure 2 show metabolic control analysis at baseline and at the end of training sessions according to the baseline therapy. The proportions of patients who met the goal of TIR > 70% after completing training were 34/38 (89.0%) and 8/11 (72.0%) for patients previously treated with SAPT-PLGM and MDI + intermittently scanned continuous glucose monitoring (isCGM) Freestyle, respectively. Three patients using HCL systems showed similar trends. Of ten patients previously treated with MDI alone, eight met the goal of TIR >70%.
Metabolic Control Indicators According to Prior Therapy and After Changing to the Advanced Hybrid Closed-Loop System.
Abbreviations: isCGM, intermittently scanned continuous glucose monitoring; MDI, multiple dose injections; SAPT-PLGM, sensor-augmented pump therapy with low-glucose management; AHCL, advanced hybrid closed-loop; TIR, time in range; IQR, interquartile range; TBR, time below range; GMI, glucose management indicator; CV, coefficient of variation; TAR, time above range.
Times in range with the change from device to advanced hybrid closed-loop system. Results are expressed as medians.
Regarding glycemic control goals at the six-month follow-up with the AHCL system, 83.3% of patients achieved a TIR >70%, 96.7% a TAR >180 mg/dL <25%, and 63.3% a TAR >250 mg/dL <5%. Goals of TBR <70 mg/dL <4% and TBR <54 mg/dL <1% were achieved in 90.0% and 80.0% of patients, respectively. Coefficient of variation <36% was achieved by 73.3% of the patients, and glucose management indicator (GMI) <7% by 86.0%. Overall, 48.3% of patients achieved all targets for adequate glycemic control during the six-month follow-up. The ten users in the total cohort who did not achieve a TIR goal >70% had a mean HbA1c of 8.1% before starting training and a mean CV of 36%, sensor use rate of 79%, and SmartGuard retention rate of 80% during training. Most SmartGuard system outputs were user-generated with a mean output of three per week. Two of them had an AIT of four hours.
The glycemic control results after the training and at three- and six-month follow-ups are indicated in Table 4 and Figure 3. The mean TIR at six-month follow-up was 76.5% ± 9.7%, and the mean GMI was 6.7% ± 0.4%. No significant differences were found in TIR results according to education level (P = .532), although sensor use was higher in the population with a higher education level (P = .033). Adherence to the use of the sensor and SmartGuard was maintained during follow-up.
Continuous Glucose Monitoring Data and Adherence to the Device in the Total Sample at the End of Training and at Three- and Six-Month Follow-ups.
Abbreviations: AHCL, advanced hybrid closed-loop; TIR, time in range; TBR, time below range; GMI, glucose management indicator; CV, coefficient of variation; TAR, time above range; SMBG, Self-monitoring blood glucose; SD, Standard deviation.
Time in range (%) at the end of training and at three and six months after starting the AHCL system. The results expressed as means (SD).
Regarding device programming, no statistically or clinically significant differences were found when comparing optimal settings with other device settings. Glycemic control results according to device programming are presented as supplemental data. No statistically significant differences were found in Clarke questionnaire scores at the three-month follow-up compared with the initial scores (P = .1047). No SH events were reported during follow-up. The impact on quality of life and satisfaction with the treatment of TD1 with the Mini Med 780G system determined with the DTSQ-s questionnaire 15 showed good satisfaction with device use with a mean score of 32 and low perception of hypoglycemia and hyperglycemia when the questionnaire was applied at the end of the training sessions.
Analysis According to Socioeconomic Stratum
Different parameters were compared according to socioeconomic stratum: lower, middle, and upper. Mean TIR significantly increased as socioeconomic stratum increased, with values of 73.4% ± 5.3%, 78.1% ± 8.1%, and 84.2% ± 7.5%, respectively (P = .042).
Mean TAR >180 mg/dL was not statistically different between the strata (18.4% ± 4.4%, 15.4% ± 5.6%, and 12.5% ± 4.6%, respectively; P = .103), nor was TAR >250 mg/dL (4.3% ± 2.3%, 3.8% ± 2.8%, and 1.7% ± 2.5%, respectively; P = .095), TBR <70 mg/dL (2.3 ± 1.9, 2.2 ± 1.6, and 1.2 ± 0.9; P = .312), or TBR <54 mg/dL (0.8 ± 1.2, 0.6 ± 0.8, and 0.3 ± 0.4, respectively; P = .642). No statistical differences were found for CV, sensor use time, SmartGuard retention, and GMI between the strata.
Analysis According to Type of Follow-up Comparing VDC With In-Person Follow-up
In total, 27 patients continued follow-up with a multidisciplinary group (endocrinologist, nutrition, and training in VDC) through teleconsultation, and 28 patients continued follow-up only by an endocrinologist in person. Although data were obtained from 55 patients and we had access to the Carelink platform to the metric data of the patients, nine remaining patients could not be contacted to define whether they continued in follow-up by an endocrinologist. Time in range at six-month follow-up was similar between the groups (79.9% ± 8.3% vs 76.0% ± 7.3%, P = .068), but TAR >180 mg/dL was lower for patients who underwent multidisciplinary—VDC follow-up (13.9% ± 5.5% vs 17.2% ± 4.9%, P = .036), while adherence to sensor use (89.3% ± 5.3% vs 83.5% ± 11.2%, P = .041) and SmartGuard retention (93.7% ± 8.1% vs 86.3% ± 14.7%, P = .028) were higher in that same group (Table 5).
Continuous Glucose Monitoring Data According to Six-Month Follow-up Group With Advanced Hybrid Closed-Loop System.
Abbreviations: VDC, virtual diabetes clinic; TIR, time in range; SD, standard deviation; TBR, time below range; CV, coefficient of variation; GMI, glucose management indicator; TAR, time above range.
Analysis of Elderly Patients
Data from the cohort of patients aged >60 years showed consistency with the main cohort data. The mean age of this cohort was 67.4 ± 7.5. Mean time of diabetes duration of 27.5 ± 16.0 years and mean baseline HbA1c was 7.4 ± 0.8, with a mean GFR of 69.9 mL/min/1.73 m2. Two patients were undergoing MDI treatment with isCGM, five with Mini Med 640 G, and one with Mini Med 670 G.
Continuous glucose monitoring data and device use in this population during the six months of follow-up are presented in Table 6. Overall, 90.9% of elderly patients had good sensor use and satisfactory SmartGuard retention. They all had a target glucose level of 100 mg/dL, 75% had an AIT of two hours, 75% met the goal of TIR >70%, all met the goal of TAR 180 mg/dL <25%, 75% met the goal of TAR 250 mg/dL <5%, all met the goal of TBR 70 mg/dL <4%, all met the goal of TBR 54 mg/dL, 87.5% met the goal of CV <36%, and 75% reached a GMI <7%. Six patients (75%) met all goals.
Continuous Glucose Monitoring Data and Device Use in Patients Aged Over 60 Years During the Six-Month Follow-up Using the Advanced Hybrid Closed-Loop System.
Abbreviations: TIR, time in range; SD, standard deviation; TBR, time below range; GMI, glucose management indicator; CV, coefficient of variation; TAR, time above range.
Discussion
This study provides real-world evidence of AHCL users in a VDC follow-up, showing a sustained improvement in glycemic control, without SH events from the end of training to the six-month follow-up period. Socioeconomic status, multidisciplinary follow-up in a VDC, and education level were associated with a better glycemic profile and adherence to technology use. Among the elderly patients, 75% achieved all goals, thus supporting the usefulness of this technology in this population.
The improvement observed during the study regarding glycemic control could be attributed to the different technological advances that allow the development of better algorithms associated with predictive systems, the automation of insulin delivery, and real-time monitoring that all converge for patient benefit. Patients who switched from MDI + isCGM and SAPT-PLGM to AHCL showed an increase in %TIR with a reduction in hypoglycemia, similar to previous reports.5,7,19 Time below range was kept within the targets established by the recommendations of the international consensus on TIR 12 in SAPT-PLGM/HCL users and was maintained after switching to the AHCL system, reinforcing the evidence regarding the safety of this technology.
The findings of this study agree with those of several real-life studies.5,18 A study of 4120 patients from eight countries reported a TIR of 76.2%, a TBR of 2.5%, and a GMI of 6.8%. 5 A study of 50 Spanish patients reported a TIR of 74% and a TBR of 2%. 19 Results of a cohort of 1025 users of integrated systems in Latin America who switched to the AHCL system were recently published. Improvements in metabolic control outcomes were detected with a GMI of 6.7%, TIR of 76.5%, and TBR of 2.7%. 6 The proportions of users who achieved the goals of GMI <7%, TIR >70%, and TBR <4% were 80.8%, 78.1%, and 80.1%, respectively. 6 These results in Latin America are similar to those obtained in Europe, suggesting that the AHCL system is safe and effective across diverse populations.
In our cohort, the upper socioeconomic stratum was significantly associated with a higher %TIR, possibly related to better access to supplies, although our health system fully covers this technology. All strata met the goal of TIR >70%. This goal was also achieved in all education-level categories, although a longer sensor use time was found in patients with postgraduate education. This suggests that automation is crucial in achieving satisfactory glycemic control, independent of sociodemographic factors, as previously reported.7
In this study, patients who underwent VDC follow-up had a lower %TAR. Considering that this is a disease with increasing prevalence and that patients live in remote cities, the use of virtual care strategies favors frequent contact with patients, reducing therapeutic inertia, facilitating adjustments to avoid delays in the intensifying treatment, and reducing the costs associated with chronic disease management for the health system and for patients by reducing the number of visits for medical care. This results in better diabetes control, increased adherence, satisfaction with treatment, and improved quality of life. 20 In addition, patients in the VDC follow-up were more adherent regarding sensor use time and SmartGuard retention, highlighting the success of the virtual follow-up.
Regarding device settings, 85.1% of patients using optimal device settings17 reached the TIR goal with a mean GMI of 6.7%, which agrees with the findings of other publications.7 Of the patients who did not achieve the goal of TIR >70%, two had an AIT of four hours, with low sensor use adherence and SmartGuard retention, underscoring the importance of proper programming and strict follow-up to achieve glycemic control goals.
To our knowledge, this is the first real-life study of T1D patients using AHCL devices who underwent training and follow-up in a VDC for six months, achieving glycemic control outcomes similar to those of other cohorts. Moreover, it included different patient profiles, migration from various technologies, providing information on the performance of AHCL systems in different scenarios. There are some limitations that need to be acknowledged: first, the small number of patients in the study sample, which may explain why we could not detect differences between different subgroups. Subsequent studies with a larger sample size are required to confirm our findings. Second, we evaluated two different monthly follow-up methods after the initial virtual training, the first characterized by virtual follow-up only and the second characterized by in-person follow-up. However, both methods showed similar benefits with respect to TIR. The better TAR outcomes, sensor adherence, and SmartGuard retention observed in the virtual-only group may be related to the benefits of multidisciplinary versus single endocrinologist care, so new studies comparing in-person versus virtual multidisciplinary care are needed. Finally, factors such as health care coverage and differences in the assessment of socioeconomic strata could limit the application of these findings in other populations.
Conclusion
These data suggest that patients diagnosed with T1D who start treatment with the AHCL system and undergo training and follow-up in a VDC can achieve the %TIR goals without SH events, which is maintained during a medium-term follow-up. In addition, socioeconomic stratum is associated with a better glycemic profile, and multidisciplinary follow-up in a VDC is associated with greater adherence to sensor use and SmartGuard retention, which were factors associated with the success of this therapy.
Supplemental Material
sj-docx-1-dst-10.1177_19322968231204376 – Supplemental material for Results From a Virtual Clinic for the Follow-up of Patients Using the Advanced Hybrid Closed-Loop System
Supplemental material, sj-docx-1-dst-10.1177_19322968231204376 for Results From a Virtual Clinic for the Follow-up of Patients Using the Advanced Hybrid Closed-Loop System by Ana María Gómez Medina, Diana Cristina Henao Carrillo, Julio David Silva León, Javier Alberto Gómez González, Oscar Mauricio Muñoz Velandia, Lucia Conde Brahim, Guillermo Andrés Mecón Prada and Martin Rondón Sepúlveda in Journal of Diabetes Science and Technology
Footnotes
Abbreviations
AHCL, advanced hybrid closed-loop; GFR, glomerular filtration rate; HbA1c, glycated hemoglobin; HCL, hybrid closed-loop; isCGM, Intermittently scanned continuous glucose monitoring; MDI, multiple daily injections; SAPT-LGS, sensor-augmented pump therapy with low-glucose suspend; SAPT-PLGM, sensor-augmented pump therapy with predictive low-glucose management; SD, standard deviation.
Declaration of Conflicting Interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: A.M.G.M. has received grants for presentations at conferences for Novo Nordisk, Sanofi, Elli Lilly, Boehringer Ingelheim, Abbott, and Medtronic. D.C.H.C. has received grants for presentations at conferences for Novo Nordisk, Sanofi, and Abbott.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
ORCID iDs
Supplemental Material
Supplemental material for this article is available online.
References
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.
