Managing Impending Nonsevere Hypoglycemia With Oral Carbohydrates in Type 1 Diabetes: The REVERSIBLE Trial Free

The Diabetes Control and Complications Trial, the results of which were published in the early 1990s, provided invaluable evidence of the favorable effect of tight blood glucose (BG) management on long-term complications of diabetes for individuals living with type 1 diabetes (T1D) using insulin therapy (1). However, an important proportion of those with T1D experience nonsevere (NS) hypoglycemia under this regimen, with a reported mean frequency of two symptomatic episodes per week (2). Recent data from the BETTER Canadian T1D registry showed that >80% of individuals with T1D experienced at least one episode of NS hypoglycemia (<54 mg/dL) in the past month, with an average of 6.47 ± 7.67 episodes per month (3). Hypoglycemia is associated with multiple short- and long-term physical and psychosocial consequences, such as acute and unpleasant adrenergic symptoms, neurocognitive dysfunction, risk of fall, and fear, all of which lead to reduced quality of life. It also interferes with many aspects of daily life (e.g., driver’s license, work productivity, and employment) (4,5). Repeated episodes lead to impaired hypoglycemia awareness with increased risk of severe hypoglycemic events during which patients are unable to treat themselves (6). In addition, hypoglycemic episodes limit the ability of patients to reach glucose targets, thus indirectly contributing to hyperglycemia and some chronic complications (2). Therefore, iatrogenic hypoglycemia remains one of the biggest challenges in T1D management.

Current guidelines recommend the treatment of NS hypoglycemia with 15 g rapidly absorbed carbohydrates (CHO) at 15-min intervals until the resolution of an episode (7,8). This step-by-step protocol is designed to rapidly normalize BG and attenuate hypoglycemic symptoms while avoiding rebound hyperglycemia and excessive caloric intake. In practice, only 32–50% of patients follow this recommendation (9,10). Savard et al. (11) showed that 73% of patients tended to consume more CHO than the recommended quantity, with an initial mean intake of 32 ± 24 g. These data suggest that the current 15-15 rule may be ineffective for the prompt treatment of NS hypoglycemia and merits review.

In recent years, real-time continuous and intermittent glucose monitoring (CGM) has increasingly been adopted by individuals with T1D. Real-time CGM technology offers programmable alerts that warn patients of current and/or impending hypoglycemia (12). Overall, CGM technology improves glucose management, namely, hypoglycemia reduction (13). By gaining insights into glycemic fluctuations with this technology, the community of individuals with T1D is witnessing a clear shift in trends from hypoglycemia treatment to hypoglycemia prevention. In a group of individuals with T1D widely using this technology, >25% reported taking CHO before reaching the recommended glycemic level of <70 mg/dL (3.9 mmol/L) when foreseeing a hypoglycemic event (3). To date, no recommendation has been published on the proactive management of impending hypoglycemia (i.e., a BG drop that will likely lead to hypoglycemia if left untreated). We aimed to study the efficacy of 16 g oral CHO in NS hypoglycemia prevention when consumed at a plasma glucose (PG) level of ≤80 (4.5) or ≤90 mg/dL (5.0 mmol/L). We hypothesized that earlier treatment of impending hypoglycemia could reduce the frequency and duration of hypoglycemic events in individuals with T1D.

This was an open-label three-intervention crossover study. After an inpatient insulin-induced BG decrease, 16 g oral CHO was administered at three PG levels (<70, ≤80, and ≤90 mg/dL). Eligible participants were adults (age ≥18 years) with a clinical diagnosis of T1D for at least 1 year who were treated with multiple daily insulin injections (MDI) or continuous subcutaneous insulin infusion (CSII) and had glycated hemoglobin (HbA_1c) ≤86 mmol/mol (<10%). Exclusion criteria included one or more clinically significant microvascular complications, recent (<3 months) acute macrovascular event, known significant cardiac rhythm or neurological abnormality, ongoing pregnancy or active breastfeeding, severe hypoglycemic episode in the past month, or known uncorrected hypokalemia (<3.5 mEq/L) in the 3 months before enrollment. Participants were recruited at the Montreal Clinical Research Institute through the T1D clinic and from the pool of participants in previous studies who had expressed interest in future trials. The study was announced on the websites of Diabète Québec, JDRF, and the BETTER registry as well as on the Connect1d platform. The study was approved by the Montreal Clinical Research Institute ethics committee (Montréal, Québec, Canada) and conducted according to the Helsinki declaration. All participants provided written informed consent.

At the admission visit, demographic data, diabetes history, HbA_1c, potassium level, and resting electrocardiogram were collected. The physician investigator completed a medical assessment, and a certified nurse provided training on the installation of a glucose sensor. Block-balanced randomization was used to determine the intervention sequence. Randomization envelopes were opened once the participant’s eligibility was confirmed, and informed consent was obtained. The interventions were separated by a minimum washout period of 7 days. For the duration of the trial, CSII users’ basal rates remained unchanged, and minimal basal insulin adjustments (<10%) were made for participants receiving MDI therapy.

The day before each intervention, participants were asked to install a glucose sensor (Dexcom G6; Dexcom, San Diego, CA) and refrain from physical activity and alcohol consumption. The glucose sensor was exclusively used to identify any instances of nocturnal hypoglycemia before the intervention. To minimize the risk of nocturnal hypoglycemia, CSII users were asked to decrease their nocturnal basal rate by 10% at midnight on the day of the intervention. Automated insulin delivery systems were changed for open-loop systems at least 2 h before the intervention. MDI users reduced their long-acting insulin by 10% the day before or their ultra-long-acting insulin by 10% 2 days before the intervention.

Participants arrived at the testing center in the fasting state at 7:00 a.m. The research team confirmed that adjustments of basal insulin doses and basal insulin rates in pump settings were carried out correctly before the intervention. Glucose sensor data were verified for any significant nocturnal hypoglycemic event (BG <63 mg/dL for ≥20 min on two separate occasions or one single event of ≥40 min). PG was measured at arrival for any significant hyperglycemia (BG >270 mg/dL) or hypoglycemia (<70 mg/dL). If any of the abovementioned conditions were not met or if the participant exhibited significant nocturnal hypoglycemia, hypoglycemia, or hyperglycemia at arrival, the intervention was cancelled and rescheduled. In the case of significant hyperglycemia, blood ketones were measured to ensure patient safety and were subsequently managed by the on-site physician as needed. If a participant’s BG upon arrival fell below the randomized condition (e.g., randomly assigned to the ≤90 mg/dL intervention, but arrival BG was 85 mg/dL), randomization was broken. The day’s intervention was reassigned to an appropriate arm (e.g., to <70 mg/dL or ≤80 mg/dL). If a participant had already completed that intervention, the day would be rescheduled. Otherwise, a venous catheter for blood sampling was installed before beginning the intervention.

At each intervention, a subcutaneous insulin bolus was administered to induce a BG decrease by using the participant’s usual rapid insulin analog. The insulin dosage was calculated based on the participant’s body weight and PG at arrival: 0.13, 0.1, or 0.08 units/kg for PG between 180 and 270, 126 and 179, or 70 and 125 mg/dL, respectively. Once the treatment level was reached (PG <70, ≤80, or ≤90 mg/dL), 16 g oral CHO (four tablets of Dex4) was administered. The PG concentration at the end of the glucose tablet consumption was considered the PG at time 0 min. If PG remained <70 mg/dL, additional administrations of 16 g oral CHO were given at 20-min intervals to allow sufficient time for CHO absorption.

During each intervention, blood samples were collected every 15 min from the time of the insulin bolus to the time PG concentration reached <100 mg/dL. Subsequently, PG was measured every 5 min until the treatment level was reached. After the initial 16 g oral CHO consumption, blood samples were collected at 5, 10, 15, 20, 30, 45, and 60 min. If needed, additional samples were drawn every 5 min until hypoglycemia recovery (PG ≥70 mg/dL). All blood samples were centrifuged immediately, and PG was measured using a YSI 2300 STAT Plus analyzer (Yellow Springs, OH). Plasma samples were then stored at −80°C for subsequent immunological assay measurement of insulin concentrations (Millipore, Billeria, MA). No significant cross-reactivity was observed for any insulin tested.

Upon completion of the intervention, participants were supplied with a standardized meal containing 65 g carbohydrates for men and 50 g for women. Their regular insulin bolus was administered accordingly. Participants were permitted to leave the testing center if their BG level remained ≥70 mg/dL for a minimum of 90 min after the meal.

The primary outcome was time spent in hypoglycemia (<70 mg/dL) after a correction with 16 g oral CHO at a PG level of <70, ≤80, or ≤90 mg/dL. Secondary outcomes were the percentage of participants reaching hypoglycemia after receiving 16 g oral CHO, the glucose nadir reached after the initial correction, the time and range of glucose decrease from the initial CHO treatment to the glucose nadir, the number of treatments required to reach or maintain normoglycemia (≥70 mg/dL), and the percentage of participants experiencing rebound hyperglycemia (>180 mg/dL) within the first hour after the initial oral CHO intake.

We assumed that the difference in hypoglycemia duration between the PG <70 and ≤80 mg/dL intervention arms would be 9.31 ± 14.92 min (14). To detect this difference with 90% power and a 5% significance level, a minimum of 29 participants would need to complete the study. With an estimated dropout rate of 15%, a recruitment sample size of 34 participants was calculated for this study.

Descriptive statistics for participants’ baseline characteristics are reported as mean ± SD with minimum and maximum for continuous variables and as count and percentage for categorical variables. PG level at the time of treatment, insulin bolus units, and insulin concentrations were analyzed using a parametric one-way ANOVA test. A mixed random effects model, adjusted for study period, randomized intervention sequence, interventions, and baseline covariates, was used to compare the other continuous outcomes. To compensate for the required end points assumption in a mixed random effect, a nonparametric bootstrap resampling approach was used to estimate the parameters and their CIs; 500 resamples of participants’ collected data in the study with replacement were conducted in this approach. All statistical analyses were performed using R software (version 4.2.2). A two-sided 5% significance threshold was used to declare statistical significance.

All data generated or analyzed during this study are included in the published article. No applicable resources were generated or analyzed during the current study.

From May 2021 to December 2022, 34 participants were enrolled in the study. Five participants dropped out after random assignment and did not complete all interventions: one for vasovagal reaction resulting from i.v. catheter insertion, one for difficult venous access, and three for subsequent unavailability to participate in the trial. As such, 87 interventions completed by 29 participants were included in the final analysis (Supplementary Fig. 1). Participant characteristics were as follows: 52% male, age 46.8 ± 16.3 years, BMI 26.4 ± 3.9 kg/m², diabetes duration 26.2 ± 15.9 years, A1C 53.0 ± 8.1 mmol/mol (7.0 ± 0.7%), 62% CSII users, and 45% ultra-rapid insulin (Fiasp) users.

For each intervention (PG <70, ≤80, or ≤90 mg/dL), the participants’ plasmatic insulin concentration at arrival was 220.2 ± 188.7, 215.3 ± 196.2, or 218.8 ± 198.8 pmol/L, respectively (P = 0.997). This was followed by an insulin bolus of 6.9 ± 3.9, 7.7 ± 3.5, or 6.6 ± 2.9 units (P = 0.511), respectively, to induce a glucose fall. At the time of oral CHO treatment, the mean PG levels were 64.1 ± 5.4, 73.1 ± 5.4, and 82.8 ± 4.3 mg/dL for the <70, ≤80, and ≤90 mg/dL arms (P < 0.0001), respectively. These PG levels corresponded to plasmatic insulin concentrations of 359.8 ± 191.0, 377.7 ± 205.6, and 354.8 ± 205.0 pmol/L (P = 0.872) (Table 1).

When comparing the <70 (control group) with the ≤80 and ≤90 mg/dL treatment levels, 100 vs. 86 (P = 0.1201) vs. 34% (P < 0.001) of participants had hypoglycemia. Specifically, 31 vs. 14 (P = 0.1237) vs. 3% (P = 0.021) had clinically significant hypoglycemia (<54 mg/dL). These hypoglycemic episodes lasted on average 26.0 ± 12.6 vs. 17.9 ± 14.7 (P = 0.026) vs. 7.1 ± 11.8 min (P = 0.002). After the initial 16 g oral CHO correction, PG continued to drop for 9.4 ± 7.7 vs. 11.9 ± 8.9 (P = 0.284) vs. 9.7 ± 5.9 min (P = 0.942), corresponding to a PG decrease of 7.6 ± 7.6 vs. 9.4 ± 6.9 (P = 0.326) vs. 9.4 ± 8.0 mg/dL (P = 0.370), before reaching a nadir of 56.5 ± 9.9 vs. 63.5 ± 7.9 (P = 0.008) vs. 73.4 ± 9.7 mg/dL (P = 0.002), respectively. In the control group, 69% of participants required more than one treatment to reach or maintain normoglycemia compared with 52% in the ≤80 mg/dL (P = 0.232) and 31% in the ≤90 mg/dL group (P = 0.028), with no significant rebound hyperglycemia within the first hour (Table 2). Figure 1 illustrates the PG changes during each intervention arm.

The Prevention of Mild-to-Moderate Hypoglycemia in Type 1 Diabetes (REVERSIBLE) trial studied a new approach to impending NS hypoglycemia prevention and correction in which 16 g oral CHO was administered in the fasting state at a higher PG level of ≤80 or ≤90 mg/dL in lieu of the usual <70 mg/dL cutoff after an insulin-induced PG decrease. Our data show that an earlier oral CHO correction at a PG level of ≤90 mg/dL decreases hypoglycemia (<70 mg/dL) incidence by 66% and clinically significant hypoglycemia (<54 mg/dL) incidence by 28%. In comparison with the control group, individuals spent on average 8 min less in hypoglycemia when corrected at ≤80 mg/dL and 18 min less when corrected at ≤90 mg/dL (Table 2). These findings confirm our hypothesis: initiating correction of impending hypoglycemia at higher BG levels than that currently recommended could mitigate the burden of some NS hypoglycemic episodes for individuals with T1D.

After the initial correction with 16 g oral CHO, PG continued to decrease for ∼10 min at an average rate of 0.9 mg/dL/min (comparable to one arrow down on the Dexcom G6) before reaching its nadir, irrespectively of treatment level. A plausible explanation for this response delay could be the absorption rate and biokinetics of oral CHO in the context of iatrogenic hypoglycemia. The night before each intervention, participants followed a standardized protocol for basal insulin adjustments. As a result, insulin concentrations at arrival were comparable between intervention periods. After a standardized subcutaneous insulin bolus, similar insulin concentrations at the time of treatment were also obtained. In this manner, the insulin delivery rates were homogeneous among all intervention arms, thus reducing this potential cofounding factor (Table 1). In other words, an oral CHO intake at a higher PG level does not reduce the time or magnitude of an insulin-induced PG fall. Rather, an earlier treatment shifts the glucose excursion curve upward, thus increasing the glucose nadir (Fig. 1). As shown by our data, the glucose nadir was 56.5 ± 9.9 mg/dL in the control group, compared with 63.5 ± 7.9 and 73.4 ± 9.7 mg/dL in the ≤80 and ≤90 mg/dL groups, respectively. This finding further corroborates our hypothesis: initiating correction of impending hypoglycemia at a higher PG level could reduce the severity of some NS hypoglycemic episodes in individuals with TID.

Our study comes at a time of need for more efficacious NS hypoglycemia treatment algorithms. In recent years, many experts have challenged the current guidelines that recommend the correction of NS hypoglycemia with 15 g rapidly absorbed CHO at 15-min intervals until the resolution of an episode (6,7). One central argument put forward is that this recommendation originates from two studies conducted in the 1980s, when the landscape of diabetes management was significantly different (15,16). In fact, contemporary studies have shown that with current intensive insulin therapies, an initial correction with 15 g oral CHO may be inadequate to promptly correct NS hypoglycemia. Larsen et al. (17) suggested that in a group of patients with TID treated with MDI insulin therapy, 30% of hypoglycemic episodes (<63 mg/dL) remained unresolved at 30 min after 10 to 20 g oral CHO. In individuals with T1D treated with CSII, Gingras et al. (18) showed that 16 g CHO corrected 38% of NS hypoglycemic episodes after 15 min, with a time to normoglycemia recovery of 19.5 ± 12.0 min. A secondary analysis on closed-loop therapies conducted by Taleb et al. (19) also demonstrated that 15 g oral CHO corrected 45% of hypoglycemic events at 15 min. These findings are on par with the data from our control group, which show 69% of hypoglycemic events remained unresolved at 20 min after 16 g oral CHO (Table 2). As new diabetes therapies and technologies are emerging, it may be time to reform the current one-size-fits-all NS hypoglycemia protocol. For instance, individuals with T1D are increasingly using real-time CGM. With its programmable alerts for current and/or impending hypoglycemia and trend arrows, this technology offers better insight into glycemic fluctuations, shifting the hypoglycemia management goal from reactive treatment to proactive prevention. This new paradigm is the rationale of our study.

Many efforts have already been made in developing alternative methods to correct NS hypoglycemia when it occurs. McTavish et al. (20) suggested that carbohydrates based on body weight (0.3 g/kg; e.g., 21 g for a 70-kg adult) were more efficient in treating NS hypoglycemia than the conventional 15 g. In their study, the rise in BG level was 27 mg/dL after 10 min with a weight-based dose, compared with 22 mg/dL with 15 g. A recent report by Taleb et al. (14) also showed that 32 g CHO at a PG level of 54–63 mg/dL was more efficient than 16 g for the treatment of NS hypoglycemia, with mean times to correction of 23.7 ± 13.6 and 31.9 ± 15.2 min, respectively. However, 32 g CHO for hypoglycemia treatment represents 4–8% of daily caloric intake (11). Because obesity is rapidly rising in individuals with T1D, limiting overall CHO intake to treat or prevent NS hypoglycemia is an important goal. To minimize this risk, a mini dose of glucagon has been proposed as an alternative. Haymond et al. (21) showed that a mini dose of 150 μg nonaqueous glucagon increased BG by 1.8 mg/dL after 20 min. Not only is this effect minimal, but glucagon injections also come with limitations, such as product stability and handling, discomfort at the injection site, nausea, and cost (22). Therefore, other alternative methods are needed for the management of NS hypoglycemia. To this day, all efforts have been concentrated on the treatment of established hypoglycemia. To our knowledge, the REVERSIBLE trial is the first study to explore the proactive treatment of impending hypoglycemia.

From a pathophysiological perspective, earlier initiation of NS hypoglycemia treatment is intriguing. In normal physiology, falling BG concentrations elicit a sequence of endogenous responses to prevent or rapidly correct hypoglycemia. To increase hepatic glucose production, the first physiological defense mechanism decreases pancreatic islet β-cell insulin secretion as PG concentrations decline to a level <84.6 mg/dL, followed by an increase in α-cell glucagon secretion. However, in the setting of T1D and exogenous insulin therapy, the ability to rapidly reduce circulating insulin is lost, and counterregulatory mechanisms are frequently impaired. This inability to increase endogenous glucose production results in hypoglycemia (2). Per current recommendations, hypoglycemia treatment should be initiated when BG is <70 mg/dL, a concentration at which all physiological mechanisms are expected to have already been activated with increased hepatic glucose production. Thenceforth, oral CHO intake at a higher BG level could better mimic the initial physiological response and more efficiently counteract falling BG in individuals with T1D. Specifically, post hoc analysis of our data showed that a treatment level of ≤90 is superior to that of ≤80 mg/dL. The former resulted in fewer hypoglycemic events (34 vs. 86%; P < 0.001) of shorter duration (7.1 ± 11.8 vs. 17.9 ± 14.7 min; P < 0.001). In fact, the PG concentration at the time of correction in the ≤90 mg/dL intervention group was 82.8 ± 4.3 mg/dL, corresponding to the BG concentration at which initial physiological adaptation is expected (Table 1).

A potential concern with a more proactive treatment of NS hypoglycemia is the risk of increased CHO consumption possibly leading to rebound hyperglycemia and weight gain. Above all, CHO should only be consumed when the BG level is decreasing toward impending hypoglycemia if untreated (e.g., when CGM is displaying downward arrows). Initiating a correction at a higher BG level might entail more punctual treatment, but individuals consume fewer CHO overall during each episode, because fewer repetitive treatments are needed to reach and maintain normoglycemia. In our study, 69% of participants in the control group required more than one treatment, with 24% requiring three treatments (48 g CHO). In comparison, 52 (P = 0.232) and 31% (P = 0.028) of participants in the ≤80 and ≤90 mg/dL groups, respectively, needed more than one treatment. In addition, initiating a correction at a higher BG concentration did not cause significant rebound hyperglycemia within the hour after the initial treatment (Table 2). Therefore, the minimal risk of increased CHO consumption, along with the benefits of NS hypoglycemia prevention and rapid treatment, makes this new approach an interesting one for individuals with T1D.

Our study presents limitations. Firstly, the intervention sequence depended on the participant’s BG upon arrival, with two of 29 participants deviating from random assignment. This exemption could have affected the results, because lower BG levels may indicate more insulin on board, with greater CHO needs. However, the plasma insulin concentrations at arrival and at the time of treatment were comparable among the intervention arms. Secondly, we induced a PG decrease with a subcutaneous insulin bolus. This iatrogenic glucose fall induced by an insulin bolus without food represents only a fraction of real-life events, because hypoglycemia can occur in different situations (e.g., after correction bolus, overestimation of meal carbohydrate content, and physical activity). Other factors could also affect the amount of CHO treatment required (e.g., insulin sensitivity, ratio of basal to prandial insulin, time of day, coexisting gastroparesis, and celiac disease). Although future trials in free-life conditions are needed, studying this new strategy in standardized conditions is a crucial first step. Thirdly, because of the rapid glucose drop, the PG level at which oral CHO correction was initiated was lower than the predefined treatment level. Nevertheless, PG at the time of treatment remained statistically different among the intervention arms (Table 1). A delay of ∼3–5 min from the time of blood sample retrieval to the time of sample centrifugation, PG analysis, and oral CHO treatment was expected. This treatment delay is also expected in the real-life setting, because CGM presents a lag time of ∼5–10 min, particularly when BG is rapidly changing (23). Fourthly, 16 g oral CHO was administered in the three intervention arms to compare the effect of PG treatment level on hypoglycemia outcomes. Additional studies must be conducted regarding the minimum amount of CHO required at a higher PG level. Fifthly, PG level was monitored for 60 min after the initial CHO correction. Certain instances of delayed rebound hyperglycemia might not have been detected within this timeframe, because the average rate of PG rise after initial treatment was 0.06 mg/dL/min in the three intervention arms. Finally, 62% of our study participants were CSII users in open-loop insulin delivery systems at the time of the intervention. Extension of the scope of this study to closed-loop systems is warranted in the future.

In conclusion, NS hypoglycemia remains an important and possibly underestimated challenge for individuals living with T1D (24). Contemporary studies have shown that the current 15-15 rule at a BG level of <70 mg/dL is suboptimal for the prompt correction of NS hypoglycemic events. New algorithms must be developed to reflect the rapidly changing landscape of diabetes management. To our knowledge, the REVERSIBLE trial is the first study to evaluate the efficacy of 16 g oral CHO at higher BG levels to prevent some NS hypoglycemic episodes in individuals with TID. In this study using a model of insulin-induced iatrogenic NS hypoglycemia in the fasting state, earlier correction at PG levels of ≤80 and ≤90 mg/dL decreased hypoglycemia incidence, shortened the time spent in hypoglycemia, increased the glucose nadir, and reduced the need for repetitive treatments. Therefore, for some insulin-induced impending NS hypoglycemic events, individuals with T1D could benefit from oral CHO intake at a higher BG level. This work must be confirmed in other situations in which NS hypoglycemia can occur.

Acknowledgments. The authors thank all the participants in our clinical trials for their valuable time and acceptance of repeated induced hypoglycemic episodes, as well as the nurses and clinical research personnel at the Montreal Clinical Research Institute for their invaluable work and efforts. Z.W. thanks the Canadian Institutes of Health Research (CIHR) and the Fonds de Recherche du Québec-Santé (FRQS).

Funding. This study was supported by grants from JDRF Canada (4-SRA-2018-651-Q-R) and the CIHR Strategy for Patient-Oriented Research (JT1-157204). Z.W. is supported by a postdoctoral fellowship from CIHR and FRQS.

Duality of Interest. R.R.-L. has received research grants from AstraZeneca, Eli Lilly, Merck, Novo Nordisk, and Sanofi; has been a consultant or member of advisory panels of Abbott, Amgen, AstraZeneca, Boehringer Ingelheim, Carlina Technology, Eli Lilly, Janssen, Medtronic, Merck, Neomed, Novo Nordisk, Roche, Sanofi, and Takeda Pharmaceutical; has received honoraria for conferences from Abbott, AstraZeneca, Eli Lilly, Janssen, Medtronic, Merck, Novo Nordisk, and Sanofi; has received in-kind contributions related to closed-loop technology from Animas, Medtronic, and Roche; benefits from unrestricted grants for clinical and educational activities from Eli Lilly, LifeScan, Medtronic, Merck, Novo Nordisk, and Sanofi; and holds intellectual property in the fields of type 2 diabetes risk biomarkers, catheter life, and the closed-loop system. R.R.-L. and V.M. have received purchase fees from Eli Lilly in relation to closed-loop technology. R.C. has received an educational grant from Novo Nordisk. N.T. has received consultant fees from Viatris and honoraria for conferences from Eli Lilly and Dexcom. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. R.C. drafted the manuscript. R.C., N.T., V.M., and R.R.-L. designed the study. R.C., N.T., and R.R.-L. helped with data interpretation. R.C., Z.W., D.B., V.P., M.-L.L.-M., and R.L.-L. conducted data collection. R.C., M.-J.L., and C.G. conducted data analysis. N.T., Z.W., D.B., V.P., M.-L.L.-M., V.B., V.M., and R.R.-L. critically revised the manuscript. All authors revised the final draft of the manuscript and approved its content. R.R.-L. 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.

Prior Presentation. Results from this study were presented at the 83rd Scientific Sessions of the American Diabetes Association, San Diego, CA, 23–26 June 2023.