Abstract
Background:
Automated insulin delivery (AID) systems improve glycemic control in people with type 1 diabetes (PwT1D), but evidence on effectiveness and safety in routine clinical practice over 2 years remains limited.
Methods:
The Observatoire de la Boucle Fermée en France is a nationwide prospective observational study evaluating AID use in children and adults with type 1 diabetes in real-world conditions. The primary end point was time in range (TIR, 70–180 mg/dL). The primary analysis assessed the noninferiority of TIR at 24 months (M24) compared with 12 months (M12). Secondary analyses evaluated the superiority of glycemic outcomes at M24 versus baseline (M0), including HbA1c, continuous glucose monitoring metrics, and safety outcomes. A sensitivity analysis using a linear mixed-effects model examined the change in TIR between M0 and M24 according to baseline HbA1c (<8% vs. ≥8%), adjusted for age, sex, and diabetes duration.
Results:
Among 2741 PwT1D who initiated AID therapy, 2225 (81.1%) had available data at M24. Median TIR at M24 was noninferior to M12 (70.0% [62.0–77.0] at both time points). Compared with baseline, TIR increased by 12.0% at M24 (P < 0.0001), and HbA1c decreased from 7.6% to 7.1% (P < 0.0001). AID discontinuation at M24 occurred in 2.1% of participants. The proportions of participants experiencing at least one episode of severe hypoglycemia over the preceding 12 months were lower at M24 than at baseline. A significant interaction between time of AID use and baseline HbA1c category was observed (P < 0.001). At M24, TIR remained higher in participants with baseline HbA1c <8% compared with those with baseline HbA1c ≥8% (P < 0.0001).
Conclusions:
In a large nationwide real-world cohort, AID use was associated with noninferior TIR between 12 and 24 months, significant improvements in glycemic outcomes compared with baseline, low rates of treatment discontinuation, and favorable safety outcomes.
Keywords
Introduction
Automated insulin delivery (AID) systems have profoundly changed the management of type 1 diabetes by combining continuous glucose monitoring (CGM) with algorithm-driven insulin administration. 1 Evidence from multiple randomized controlled trials has demonstrated that hybrid closed-loop systems improve glycemic control, increasing time in range (TIR) by about 10% while significantly lowering the time spent both in hyperglycemia and hypoglycemia compared with sensor-augmented pump therapy.2–4 These benefits have been further confirmed in real-world settings, as observational studies and a recent meta-analysis have shown that the improvements of glucose control by AID systems persist in routine clinical practice. 5 However, most randomized and observational studies remain limited to short-term follow-up and rarely extend beyond 1 year.
The Observatoire de la Boucle Fermée en France (OB2F) is a nationwide program designed to assess the use of several AID systems in a large, unselected population of adults and children with type 1 diabetes under routine care conditions. Initial results after 1 year demonstrated sustained improvements in glycemic outcomes, including a 10%–12% rise in TIR, a reduction in HbA1c from 7.6% to 7.0%, and a substantial reduction in the incidence of severe hypoglycemia (SH) and diabetic ketoacidosis (DKA) after 12 months of AID therapy. 6 The objective of the present study was to further investigate these findings by assessing the effectiveness, safety, follow-up modalities, and adherence to AID therapy at 24 months in this large real-world cohort.
Materials and Methods
Study design and population
OB2F is a large nationwide survey conducted across 79 centers. In France, AID systems have been fully and largely reimbursed since 2021 for patients with type 1 diabetes who do not achieve optimal glycemic control despite optimized therapy, with access granted through a shared decision between the diabetologist and the patient. Previously, OB2F evaluated the evolution of glucose control in people with type 1 diabetes (PwT1D) over the first 12 months (M0 to M12) following initiation of an AID system. 6 In this article, we present the 24-month follow-up of the same cohort, using the same data collection procedures as previously described at M0, M3, M6, and M12. For the current analysis, only M0, M12, and M24 are considered. The following data were collected by the participating centers at initiation of AID (M0), at 12 months (M12), and at 24 months (M24). At M0, demographic and clinical characteristics included age, sex, body mass index (BMI), HbA1c level, diabetes duration, history of diabetic complications (retinopathy, nephropathy, neuropathy), prior cardiovascular events, previous insulin therapy (MDI or CSII), and history of severe SH or DKA in the past 12 months.
CGM-derived metrics—percentage of time below target range (TBR, <3.9 mmol/L), time in range (TIR, 3.9–10 mmol/L), time above target range (TAR, >10.0 mmol/L), glucose management indicator (GMI), coefficient of variation (%CV), Glycemic Risk Index (GRI), and CGM index (COGI)—were extracted from 15-day ambulatory glucose profile reports generated by platforms such as MyDiabby, CareLink, and Glooko XT and transcribed into the electronic case report form.
At M12 and M24, follow-up assessments included HbA1c level, the percentage of time spent with the AID system activated, history of SH or DKA over the preceding 12 months, and the same CGM-derived metrics as at initiation.
In addition, AID continuation status was recorded, along with the reason for discontinuation if applicable. DKA and SH events were retrospectively extracted from medical records completed by physicians and adjudicated by an independent study monitoring committee, based on clinical information obtained from structured telephone interviews conducted with each center at baseline, 12 months, and 24 months.
Endpoints and statistical hypotheses
The primary end point was the noninferiority of the percentage of TIR at M24 compared with M12. The primary hypothesis was that glycemic control achieved at M12 would be maintained at M24 within a predefined noninferiority margin, using a linear regression model adjusted for age, sex, baseline HbA1c, and duration of diabetes. Secondary endpoints were analyzed according to a predefined hierarchical order and included HbA1c, %TBR (percentage of time below 70 mg/dL), %TBR 2 (percentage of time below 54 mg/dL), %TAR (percentage of time above 180 mg/dL), %TAR 2 (percentage of time above 250 mg/dL), %GMI, %CV, GRI, and COGI. For each end point, noninferiority of M24 versus M12 was tested sequentially, conditional on the success of the preceding end point in the hierarchy, using the same adjustment variables. Noninferiority margins were defined a priori based on clinical relevance and existing literature 6 : 5% for %TIR, 0.4% for HbA1c and GMI, 1% for %TBR and %TBR 2, 3% for %TAR and %TAR 2, 5% for %CV, 5 for GRI, and 5 for COGI.
In addition, for all continuous endpoints (%TIR, HbA1c, %TBR, %TAR, %GMI, and %CV), superiority analyses comparing M24 to baseline were performed using adjusted linear regression models including the same covariates (age, sex, baseline HbA1c, and duration of diabetes), to assess the long-term improvement in glycemic outcomes after initiation of AID.
Exploratory secondary analyses were conducted for categorical glycemic outcomes, including achievement of HbA1c <7%, occurrence of at least one SH event, occurrence of at least one acute complication, achievement of %TIR ≥70%, achievement of %TBR <4%, and the composite outcome at M24 combining %TBR <4% and %TIR ≥70%. For these categorical endpoints, noninferiority of M24 compared with M12 was evaluated first, followed by superiority analyses comparing M24 with baseline. Noninferiority was assessed either by calculating the relative difference in proportions between M24 and M12: (M24 − M12)/M12 or by calculating the absolute difference in proportions between M24 and M12: (M24 − M12). Predefined noninferiority margins were based on clinical relevance and scaled according to the type of end point. For HbA1c and CGM-derived outcomes (%TIR ≥70%, %TBR <4%, %TBR <4%, and %TIR ≥70%), a relative allowable decrease of 5% of the M12 proportion was applied; for rare safety events, including SH and DKA, a clinically relevant absolute increase of 0.5% compared with M12 was used. Noninferiority was considered demonstrated if the observed proportion at M24 remained above the lower margin (for efficacy endpoints) or below the upper margin (for rare safety events).
In addition, for all categorical glycemic endpoints (including HbA1c <7%, %TIR ≥70%, %TBR <4%, the composite end point, and occurrence of acute complications), superiority analyses comparing M24 with baseline (M0) were performed using logistic mixed-effects models including time as a fixed effect and participant as a random intercept to account for repeated measurements, adjusted for age, sex, baseline HbA1c, and duration of diabetes, to evaluate long-term improvements in glycemic outcomes following initiation of AID.
As a secondary exploratory approach, within-subject changes between M12 and M24 were evaluated among participants with available measurements at both time points. Paired comparisons were performed using Wilcoxon signed-rank tests. Individual trajectories and distributions of changes were examined to provide a complementary description of treatment response at the patient level.
In a sensitivity analysis, we evaluated whether the change in %TIR from baseline to M24 differed from baseline glycemic control. Participants were stratified according to baseline HbA1c into two groups (<8% vs. ≥8%). The change in %TIR over time was analyzed using a linear mixed-effects model including time (M0 vs. M24), baseline HbA1c group, and their interaction, with adjustment for age, sex, diabetes duration, and baseline HbA1c value. To further explore the relationship between baseline glycemic control and subsequent changes, we generated two additional graphs to visualize the evolution of HbA1c and TIR over time, stratified by baseline values.
All analyses were conducted at a two-sided alpha level of 0.05.
Results
Characteristics of the study population at AID initiation
Among the 2741 PwT1D who initiated AID therapy, 2225 (81.1%) had a visit at 24 months (±1 month), allowing data collection. There were no clinically relevant differences between these 2225 participants and the 517 who were lost to follow-up, and baseline HbA1c, %TIR, %TAR, and %TBR were similar between the two groups. However, some differences were found among those lost to follow-up: age, which was slightly higher, duration of diabetes slightly longer, and BMI a little higher (Supplementary Table S1).
The M24 cohort comprised 287 children aged 4–13 years, 373 adolescents and young adults aged 14–24 years, 1395 adults aged 25–64 years, and 170 elderly individuals aged 65 years or older. The population included a slight majority of women (55.3%), with a median diabetes duration of 19 years (interquartile range: 10–29).
At the time of AID initiation, the median HbA1c level was 7.6% (7.0–8.2), and the median TIR was 58.0% (48–68). The full baseline clinical and CGM-derived characteristics of the overall population and of each age group are presented in Table 1.
Baseline Characteristics of the Study Population and Age-Stratified Subgroups
Continuous variables are presented as median [Q1–Q3], and categorical variables as number (percentage).
AID, automated insulin delivery; BMI, body mass index; CGM, continuous glucose monitoring; DKA, diabetic ketoacidosis; GMI, glucose monitoring index; PAD, peripheral arterial disease; SH, severe hypoglycemia; TAR, time above target range; TBR, time below range.
Treatment exposure and AID continuation
At M24, 47 patients (2.1%) had discontinued AID therapy, with treatment burden being the main reason for discontinuation (24, 50%). Among participants who were still using an AID system at M24, the median proportion of time with the system activated was 95.0% [90.8–98.0].
At M24, 47 patients (2.1%) had discontinued AID therapy, with treatment burden being the main reason for discontinuation (24, 50%); discontinuations occurred in 28 Tandem-CIQ, 15 Medtronic-780G, and 4 DBLG1 users. Among participants who were still using an AID system at M24, the median proportion of time with the system activated was 95.0% [90.8–98.0]. In addition, 84 patients had changed their AID systems between M0 and M24: notably, 15 users of Medtronic-780G and 17 of Tandem-CIQ switched to Cam APS, while 6 users of Medtronic-780G and 8 of Tandem-CIQ switched to Omnipod 5; full details are provided in Supplementary Figure S1.
Primary and secondary outcomes
Glycemic outcomes improved markedly after AID initiation and were maintained over 24 months of follow-up.
Primary outcome—Time in Range (TIR)
Median %TIR increased from 58.0% [48.0–68.0] at baseline (M0) to 70.0% [62.6–77.0] at 12 months (M12) and remained stable at 24 months (M24: 70.0% [62.0–77.0]) (Fig. 1). The adjusted comparison between M24 and M12 demonstrated noninferiority, with a between-time difference of −0.7%, 95% CI [−1.39; −0.01]. In addition, % TIR at M24 was significantly higher than at baseline (P < 0.0001), confirming sustained superiority versus M0 (Supplementary Table S2A).
Evolution of glucose control following AID initiation: HbA1c and CGM variables. TBR1, time below range (<70 mg/dL); TBR2, time below range (<54 mg/dL); TIR, time in range; TAR1, time above range (>180 mg/dL); TAR2, time above range (>250 mg/dL); %CV, coefficient of variation; GRI, glycemic risk index; COGI, continuous glucose monitoring index. Statistical significance is indicated as follows: **** p value <0.0001.
Secondary continuous CGM-derived outcomes
The median HbA1c decreased from 7.6% [7.0–8.2] at M0 to 7.0% [6.6–7.5] at M12 and remained stable at M24 (7.1% [6.7–7.6]) (Fig. 1). The adjusted difference between M24 and M12 was +0.07%, 95% CI [0.02; 0.11], supporting noninferiority, while HbA1c at M24 remained significantly lower than at baseline (P < 0.0001) (Supplementary Table S2A).
Similar patterns were observed for other CGM metrics:
%TBR decreased from 2.0% [1.0–4.0] at M0 to 1.5% [0.9–3.0] at M12 and further to 1.4% [0.8–3.0] at M24 (Fig. 1). The adjusted difference between M24 and M12 was −0.17% (95% CI: [−0.37; 0.03]), supporting noninferiority, with significant improvement compared with M0 (P < 0.0001). %TAR declined from 38.8% [27.7–49.0] at M0 to 27.7% [20.6–35.3] at M12 and remained stable at M24 (28.0% [21.0–36.0]) (Fig. 1). The adjusted difference between M24 and M12 was +0.93% (95% [0.22; 1.63]), confirming noninferiority, with sustained improvement compared with M0 (P < 0.0001). GMI, %CV, GRI, %TBR 2, % TAR 2, and COGI also showed sustained improvements from M0 to M24, with noninferiority at M24 compared with M12 and significant superiority versus baseline (all P < 0.0001) (Fig. 1 and Supplementary Table S2A).
Secondary categorical outcomes
The proportion of participants with HbA1c <7% decreased from 50.7% at M12 to 40.0% at M24. The observed M24 proportion (40.0%) fell below the predefined noninferiority margin of 48.2% (i.e., a 5% relative decrease from 50.7% at M12), indicating that noninferiority was not met (Fig. 2). In contrast, the proportions of participants achieving %TIR ≥70% (49.5% vs. 48.5%), %TBR <4% (82.7% vs. 84.4%), and the combined target of %TIR ≥70% and %TBR <4% (39.9% vs. 42.0%) all remained above their respective 5% relative decrease margins (47.0%, 78.6%, and 37.9%), demonstrating noninferiority (Fig. 2). Superiority analyses comparing M24 with baseline showed significant improvements for all these endpoints (Fig. 2 and Supplementary Table S2B).
Evolution of glycemic outcomes following AID initiation: proportions of participants achieving HbA1c, TIR, and TBR targets, and incidence of acute events. TIR, time in range. Statistical significance is indicated as follows: **** p value <0.0001.
For rare safety events, the proportion of participants experiencing at least one SH event increased slightly from 0.8% at M12 to 1.1% at M24. This value remained below the predefined noninferiority margin of 1.3% (i.e., a 0.5% absolute increase from M12), indicating that noninferiority was maintained. Similarly, for DKA, the proportion rose from 0.4% at M12 to 0.9% at M24, exactly reaching the noninferiority threshold of 0.9%. This indicates that noninferiority for DKA was achieved, but only marginally. Superiority analyses comparing M24 with baseline demonstrated a sustained reduction in the incidence of these rare events (Fig. 2 and Supplementary Table S2B).
Complementary paired analyses
Paired analyses using Wilcoxon signed-rank tests showed no significant differences between M12 and M24 for %TIR, %TBR, %TBR <54 mg/dL, GRI, and GMI (all P > 0.05). Significant differences were observed for HbA1c (P < 0.001), %TAR (P < 0.05), %TAR >250 mg/dL (P < 0.01), %CV (P < 0.05), and COGI (P < 0.05). Individual trajectories and cumulative distributions of within-subject changes are presented in Supplementary Figures S2 and S2.
Subgroup analyses of study outcomes
In a sensitivity analysis, participants were stratified according to baseline HbA1c (<8% vs. ≥8%). At baseline, 1510 participants had HbA1c <8% and 715 had HbA1c ≥8%. Median %TIR was higher in the <8% group (62% [53–71]) compared with the ≥8% group (47.1% [36.9–56.7]; P < 0.0001). Median age and diabetes duration were similar between groups (age: 37 [19–51] vs. 39 [23–52], ns; diabetes duration: 19 [9–30] vs. 19 [10.5–29], ns). Sex distribution was 43.8% (661) males and 45.2% (849) females in the <8% group and 46.7% (334) males and 53.3% (382) females in the ≥8% group (ns). Additional baseline characteristics are presented in Supplementary Table S3.
At M24, %TIR remained significantly higher in participants with baseline HbA1c <8% compared with those with HbA1c ≥8% (71.9% [64–78.6] vs. 65% [56.6–72]; P < 0.0001) (Supplementary Table S3).
The evolution of glycemic metrics over time, stratified by baseline HbA1c categories, is illustrated in Supplementary Figure S4A and S4B. Supplementary Figure S4A highlights the regression-to-the-mean phenomenon, where participants with higher baseline HbA1c exhibit larger absolute reductions in HbA1c over time, although their final HbA1c levels remain higher compared with those with lower baseline values. Similarly, Supplementary Figure S4B reveals that participants with higher baseline HbA1c (≥8%) achieve substantial improvements in TIR, though their final TIR values remain lower than those of participants with baseline HbA1c <8%. These findings confirm that while both groups benefit from the intervention, quantitative differences persist between baseline HbA1c strata.
A linear mixed-effects model was then used to assess the change in TIR between M0 and M24 according to baseline HbA1c category, including a random intercept for participants and adjusting for age, sex, and diabetes duration. A significant interaction between time and baseline HbA1c category was observed (P < 0.001), indicating a greater improvement in TIR among participants with baseline HbA1c ≥8%.
Discussion
The sustained follow-up of data from the French national observatory OB2F demonstrates that, after 2 years of AID use, efficacy on glucose control is durably maintained, with good safety and excellent adherence. Notably, 97.9% of participants were still using AID at 24 months.
Prolonged evaluation of treatments for individuals with type 1 diabetes is essential, particularly for technological interventions, as their effectiveness may diminish over time and may be associated with worsening glycemic control, including a rise in HbA1c levels. 7 To date, few studies have assessed the sustained outcomes of AID systems, which became commercially available only recently, in 2021.
Recently, prolonged changes in glycemic control following the transition from nonautomated insulin therapy to various AID systems in children and adolescents have been reported.8–9 In Bismuth et al., 9 a sustained improvement of glucose control was reported over 3 years with the Control-IQ system, including during the pubertal period, with no occurrence of SH and a single case of DKA. In the other study, AID systems significantly improved key glycemic outcomes and reduced hypoglycemia compared with previous therapies. However, after 2 years, the initial improvements in HbA1c tended to attenuate. Of note, a 27.2% subset of the study population used the first-generation 670G AID system, already reported as losing sustainability of adherence over time. 10
In adults, a recent study involving 81 individuals using open-source AID systems reported sustained improvements in HbA1c and hypoglycemia awareness, without an increase in hospitalizations, DKA, or SH over a mean follow-up of 1.7 years. 11
To our knowledge, the present study is the first to evaluate AID use in a large, unselected population with a two-year follow-up. Our findings show that the efficacy on glucose control observed at 1 year—particularly in respect to HbA1c and time in range (TIR)—remains excellent at 2 years, with average near attainment of recommended glycemic targets. Safety also remained favorable, with reductions in acute events including severe hypoglycemia and DKA, although these were not completely eliminated.
Several complementary metrics such as TAR1 and 2, GRI, and COGI capture distinct and complementary dimensions of glycemic control, including hyperglycemia burden, hypoglycemia exposure, glycemic variability, and overall glycemic risk. Recent work has suggested that composite CGM-derived metrics may provide increased sensitivity to detect treatment effects compared with traditional measures such as time in range alone, particularly in the context of therapeutic interventions. 12 In line with this rationale, the consistent improvements observed across all CGM-derived metrics in our study further reinforce the robustness of the glycemic benefits associated with AID initiation in routine clinical practice. Importantly, these findings should not be interpreted as a formal comparison of the relative responsiveness or sensitivity of individual CGM metrics, which was beyond the scope of the present study.
While the primary analysis demonstrated the noninferiority of M24 versus M12 at the population level, our complementary exploratory paired analyses suggest that individual-level trends may nuance this conclusion. Specifically, certain glycemic metrics—such as %TAR, %TAR >250 mg/dL, and %CV—showed slight deterioration at M24 compared with M12, potentially reflecting heterogeneity in patient responses or adaptive behaviors over time. Notably, the HbA1c increase at M24 versus M12, though not clinically significant, appears discordant with GMI and TIR trends, which remained stable. This discrepancy may highlight the complementary but distinct information provided by these metrics, with HbA1c reflecting longer-term glycemic exposure, while GMI and TIR capture more granular, real-time CGM-derived dynamics. These exploratory findings underscore the value of multimetric evaluations to fully characterize the individual and population-level effects of AID systems.
We do not compare the two AID systems used in this study, firstly because this was a retrospective, nonrandomized design, and secondly because differences in CGM sensors (Dexcom G6 and Guardian 4) may influence reported TIR values.13,14
Of note, a significant decrease in the proportion of participants achieving HbA1c <7% was observed, whereas the percentage achieving TIR >70% did not differ between 1 and 2 years of follow-up. This discrepancy must take into account the difference, recently highlighted, between CGM measurements according to used devices. 13 Since our study population used two different AID systems, including different CGM devices, the relationship between %TIR and HbA1c levels, may have been impacted according to the used AID system. In addition, when comparing the distributions of HbA1c at one and 2 years of follow-up, they approximate Gaussian curves centered around 7%. Notably, a substantial proportion of patients with HbA1c values between 6.5% and 7.0% at 12 months shift to slightly higher values, typically between 7.0% and 7.5%, at 24 months (Supplementary Fig. S5). This shift around the 7% threshold likely contributes to the observed statistical difference, while remaining of limited clinical significance.
Besides, nearly 60% of our study population failed to reach recommended CGM targets, indicating that despite encouraging overall results, glycemic control remains suboptimal for a substantial proportion of patients. Limitations of efficacy could result from the suboptimal management of nonautomated components of the AID systems, that is, meal and exercise announcements. These findings underscore the importance of ongoing education and structured follow-up, particularly for individuals who do not achieve glycemic goals.
Overall adherence to AID use was nevertheless excellent, with only 2.1% of patients permanently discontinuing the therapy, most often due to perceived burden or system complexity.
Some limitations should be acknowledged. Only two AID systems—MiniMed™ 780G and Tandem Control-IQ™—were evaluated, as they were the only models available in France at the time of study initiation. The most recently available AID systems should therefore be evaluated in real-world settings to provide a broader assessment of the long-term sustainability of glycemic outcomes associated with AID therapy. In addition, the increasing availability of tubeless systems may further expand access to insulin pump therapy, particularly among individuals with higher baseline glucose levels who might otherwise have been reluctant to initiate treatment with tubed systems. 15 In addition, a substantial number of patients could not be assessed after 2 years because some centers discontinued their participation in the study. However, comparisons between patients who were followed-up and those lost to follow-up revealed no significant baseline differences, suggesting the absence of significant selection bias.
We also need to point out that acute metabolic events (DKA and SH) were only retrospectively ascertained from medical records, potentially introducing reporting bias and partly accounting for their relatively low incidence compared with similar cohorts.16–17 In the T1D Exchange Registry, 17 severe hypoglycemia was self-reported via questionnaires, whereas in our study, events were collected from medical records and adjudicated, which may reduce reporting bias.
In addition, the structured implementation of AID in France, supported by early national guidelines, 18 likely contributed to improved outcomes through standardized initiation and follow-up. Nevertheless, acute events were not fully eliminated, highlighting that a subset of patients remains at risk. For these individuals, islet or stem cell transplantation may represent a relevant therapeutic option.
Socioeconomic status was not assessed in this retrospective study, precluding evaluation of whether these positive outcomes are consistent across different socioeconomic and clinical backgrounds. However, previous studies have suggested that access to and benefits from AID systems are observed across a broad range of socioeconomic strata. 19
The principal strength of this study remains in the very large number of patients assessed nationwide, 2 years after AID initiation under real-world conditions.
In conclusion, this study illustrates that efficacy, safety, and acceptance of AID systems are maintained over 2 years after initiation, confirming the transformative potential of this therapeutic innovation for preventing long-term complications of type 1 diabetes. In view of achieving this goal, continued follow-up and sustained support by the healthcare team—particularly for individuals who do not achieve optimal glycemic outcomes—remain essential. Ongoing technological advances, including newly developed algorithms that minimize users’ participation and improved infusion tools, are expected to further enhance glycemic control while also increasing adoption of AID and reducing the burden of the disease.
Authors’ Contributions
J.-P.R., É.R., and J.-F.G. conceived and designed this study. J.-P.R. and J.-B.J. had full access to all the study data and were responsible for the integrity of the data and the accuracy of the data analyses. They curated the data, performed the data analysis, and created the figures and tables. J.-B.J., É.R., M.J., J.-F.G., and J.-P.R. analyzed the results. All authors participated in the collection or interpretation of the data. All authors critically revised the article for important intellectual content and gave final approval for publication. J.-P.R. 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.
Supplemental Material
sj-docx-1-dtt-10.1177_15209156261465478 — Supplemental material for Sustained Follow-up of Automated Insulin Delivery in a Real-World Setting: Results at 2 Years of the French Nationwide Observatory OB2F
Supplemental material, sj-docx-1-dtt-10.1177_15209156261465478 for Sustained Follow-up of Automated Insulin Delivery in a Real-World Setting: Results at 2 Years of the French Nationwide Observatory OB2F by Jean-Baptiste Julla, Chloé Amouyal, Sandrine Lablanche, Agnès Sola Gazagnes, Michael Joubert, Samy Hadjadj, Candace Ben Signor, Elise Bismuth, Yannis Chartier, Nathalie Garrec, Joëlle Dupont, Emma Carreira, René Valéro, Gaëtan Prévost, Rosine Guintrand, Florent Verdier, Carole Legagneur, Jacques Beltrand, Anne-Cécile Paepegaey, Alina Radu, Jean-François Gautier, Éric Renard, and Jean-Pierre Riveline
Footnotes
Acknowledgments
Author Disclosure Statement
J.-B.J. has received consulting fees from Sanofi and lecture fees from Eli Lilly, Novo Nordisk, and Sanofi. S.L. is an advisory panel member for Sanofi, Abbott, Insulet, Dexcom, Medtronic, Ypsomed, and Roche. S.L. reports lecture fees from Dinn’Tech. C.A. is an advisory panel member for Sanofi, Dexcom, and Novonordisk and reports lecture fees from Dexcom and Abbott. ASG declares consultant and/or speaker fees and/or research support from Medtronic, Dexcom, Abbott, Lilly, Sanofi, Novo Nordisk, Astrazeneca, and Sanofi. M.J. declares consultant and/or speaker fees and/or research support from Abbott, Amgen, Astrazeneca, Boehringer–Ingelheim, Dexcom, Glooko, Insulet, Lifescan, Lilly, Medtronic, Novonordisk, Roche Diabetes, Sanofi, Tandem, and Ypsomed. S.H. reports receiving grants from Asdia, Asten, AstraZeneca, Homeperf, LVL, Nestle Home Care, Pierre Fabre, and VitalAire; consulting fees from AstraZeneca, Bayer, Boehringer Ingelheim, Eli Lilly, Novo Nordisk, Sanofi, Servier, and Valbiotis; speaking fees from Abbott, AstraZeneca, Boehringer Ingelheim, Bayer, Dino Santé, Eli Lilly, Novartis, Novo Nordisk, Pierre Fabre, Sanofi, Servier, and Valbiotis; and meeting invitations from AstraZeneca, Abbott, Dino Santé, Eli Lilly, Novo Nordisk, and Sanofi. G.P. reports grants, consulting fees, honoraria for lectures and formation, support for attending meetings, and (co)investigating research from Abbott, Amgen, Asten, Astra Zeneca, Boehringer Ingelheim, Lilly, ISIS diabete, Medtronic, Novartis, Novo-Nordisk, and Sanofi. E.B. has received consulting fees from Abbott, Insulet, and Sanofi and lecture fees from Eli Lilly, Abbott, Insulet, and Sanofi. J.-F.G. reports lecture fees from AstraZeneca, Bayer, Bristol-Myers Squibb, Eli Lilly, Gilead, Novo Nordisk, Pfizer, and Sanofi. He received consulting fees from AstraZeneca, Pfizer, and Sanofi and nonfinancial support from AstraZeneca, Novo Nordisk, and Sanofi. É.R. declares consultant/speaker fees from A. Menarini Diagnostics, Abbott, Air Liquide SI, Astra-Zeneca, Becton-Dickinson, Boehringer-Ingelheim, Cellnovo, Dexcom Inc., Eli-Lilly, Hillo, Insulet Inc., Johnson & Johnson (Animas, LifeScan), Medtronic, Medirio, Novo-Nordisk, Roche, and Sanofi-Aventis and research support by Abbott, Dexcom Inc., Insulet Inc., Roche, and Tandem Diabetes Care. J.-P.R. is an advisory panel member for Sanofi, MSD, Eli Lilly, Novo Nordisk, Abbott, Alphadiab, Air Liquide, Insulet, Dexcom, and Medtronic and has received research funding from and provided research support to Eli Lilly, Abbott, Air Liquide, Sanofi, Novo Nordisk, Insulet, Dexcom, and Medtronic. F.V. has received consulting fees from Sanofi and nonfinancial support from Air Liquide, AstraZeneca, Eli Lilly, Novo Nordisk, and Sanofi. J.B. received consulting fees from Sanofi, Yspomed Lilly, and Insulet and conference fees from Sanofi, Novo, Lilly, Abbott, Insulet, and Medtronic. A.-C.P. has received consulting fees from Lilly and nonfinancial support from Asten, AstraZenecca, NovoNordisk, Sanofi, Dexcom, Medtronic, Abott, NHC, and Lilly. C.L. received consulting fees from Lilly, Sanofi, and Air liquide and nonfinancial support from NHC, Novo, Medtronic, Abbott, and Insulet. E.C. received speaker fees from Abbott and nonfinancial support from Nelli Medical. R.V. reports grants, consulting fees, honoraria for lectures and formation, support for attending meetings, and (co)investigating research from Akcea, Amarin, Amgen, Arrowhead, AstraZeneca, Daiichi Sankyo, Dinno Santé, Ionis, MSD, Novartis, Pfizer, Sanofi-Regeneron, and Servier. A.R. has received financial support for conference invitations and training courses from Vitalaire, AstraZeneca, Nelli Medical, Insulet, and Novo Nordisk.
Funding Information
OB2F has received financial support from the
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.
