A Randomized Trial of Closed-Loop Insulin Delivery Postpartum in Type 1 Diabetes

Achieving target glycemia for individuals with type 1 diabetes requires considerable effort. The postpartum period has unique challenges that make attention to glucose levels particularly difficult. The demands of caring for a newborn, breastfeeding, change in daily routine, and the commonly accompanied sleep deprivation result in little time or energy to direct toward diabetes self-care. Women with type 1 diabetes have reported that fear of hypoglycemia related to breastfeeding and changing insulin sensitivity in the postpartum period is a substantial concern for them that has contributed to some stopping breastfeeding sooner than they had desired (1). Others have commented that postpartum is “a time when your mental space to manage diabetes is at its lowest” (2).

Hybrid closed-loop insulin delivery automatically adjusts the amount of insulin delivered based on continuous glucose monitoring (CGM) data. Studies of hybrid closed-loop insulin delivery show that this technology is effective in improving HbA_1c, time spent with glucose in and below target range, and quality of life among children, adolescents, adults, and pregnant women (3–10). However, data are limited on the use of closed-loop insulin delivery postpartum (6), and we are aware of no previously published studies of postpartum or lactating women randomized to closed-loop insulin delivery. The aim of this study was to evaluate the impact of hybrid closed-loop insulin delivery with the MiniMed 670G/770G system on postpartum maternal glycemia, the burden of diabetes self-care, and infant feeding practices.

This pilot, parallel-arm, open-label, randomized controlled trial was conducted in outpatient diabetes in pregnancy clinics of four Canadian academic centers (Calgary Foothills, Toronto Mt. Sinai, Centre Hospitalier de l’Université Laval, and Health Sciences Centre Manitoba). Participants were randomized within the first postpartum week to begin hybrid closed-loop (MiniMed 670G/770G in automode) or continue sensor-augmented pump therapy (MiniMed 670G/770G in manual mode) from 1 week postpartum to 11 weeks, 6 days postpartum. A continuation phase followed in which all participants were to use closed-loop insulin delivery from 12 to 24 weeks postpartum. This study was approved by the Conjoint Health Research Ethics Board, University of Calgary (study ID: REB19-1470) and the ethics boards of participating centers.

Pregnant individuals were eligible if they were age 18–45 years, between 12 and 32 weeks gestation, had >2 years duration of type 1 diabetes, were on intensive insulin therapy (multiple daily insulin injections or insulin pump), had an HbA_1c during pregnancy <85 mmol/mol (9.9%) at study entry, and had a viable singleton pregnancy confirmed by ultrasound. Participants were required to speak and understand English or French and have an average total daily insulin use of 8–250 units/day at the time of randomization. Exclusion criteria included current use of a closed-loop insulin delivery device or drugs known to interfere with glucose metabolism and the presence of advanced diabetes complications or other physical or psychological disease judged by the investigators as likely to interfere with conduct of the study.

Participants were randomized after the birth of their child in a 1:1 ratio to start closed-loop insulin delivery 1 week postpartum or to remain with sensor-augmented pump therapy until 12 weeks postpartum when they were to start closed-loop. Those randomized to start closed-loop insulin delivery 1 week postpartum were to continue with it until study completion (24 weeks postpartum). Randomization occurred via a web-based system (Research Electronic Data Capture) that used a computer-generated randomization list. The trial coordinating team and investigators had no access to the randomization schedule. The study biostatistician remained masked to the different trial arms until analyses were completed.

After providing written informed consent, participants were continued on or provided with the MiniMed 670G or 770G insulin pump to use in manual mode during pregnancy. Those new to the MiniMed 670G or 770G insulin pump received individual online or in-person training lasting ∼2 hours. The MiniMed 770G insulin pump uses the same control algorithm and target glucose options (120 mg/dL or 150 mg/dL [temporary target]) as the MiniMed 670G. The MiniMed 770G became available during the study, so active trial participants who were using the MiniMed 670G were changed to the MiniMed 770G pump, and those starting after the 770G pump was available used it for the duration of the study. Two participants who switched to the 770G pump during the study were out of closed-loop insulin delivery for 6 days prior to reinitiating closed-loop delivery as a requirement of the pump prior to closed-loop reinitiation. All participants received Guardian Link 3 transmitters and Guardian 3 CGM sensors for use until 24 weeks postpartum.

All participants continued to receive routine medical care for diabetes in pregnancy, which included frequent follow-up with an endocrinologist, diabetes nurse educators, dietitians, and an obstetrician. Contact with the endocrinologist occurred a minimum of monthly by phone or in person. Phone contact with the diabetes nurse educators every 1–2 weeks was encouraged during pregnancy. During pregnancy, which was prior to randomization, participants were advised to reduced insulin pump basal rates and use less aggressive insulin-to-carbohydrate ratios and insulin sensitivity factors starting 1–2 h prior to childbirth. Specific pump adjustments were tailored to the individual. Principles of postpartum insulin dosing below prepregnancy insulin dosing guided these adjustments. Participants were told that some women with type 1 diabetes report hypoglycemia with breastfeeding and that a small carbohydrate snack during breastfeeding could be consumed to prevent hypoglycemia.

Study recruitment began in July 2020. Follow-up of study participants was completed in February 2023. Participants were provided a 1-h training session on the use of closed-loop insulin delivery, usually by 34 weeks gestation, but there was one participant for whom this occurred after 34 weeks gestation for her personal convenience. The strength of basal insulin modulation with the MiniMed 670G/770G closed-loop system is influenced by the total daily dose of insulin used in the previous 6 days. Since insulin sensitivity is much greater in the first few days after birth compared with the week prior to birth, closed-loop insulin delivery was not used in the first postpartum week (11) because of concerns that the basal modulation could be too aggressive in the first postpartum week. Closed-loop was initiated 1 week postpartum in order to allow the system to use a daily total that was more appropriate for the postpartum period. All participants were provided brief refresher training (≤30 min) on the use of closed-loop insulin delivery at the start of closed-loop initiation, either immediately after randomization for those randomized to the closed-loop group or at 12 weeks postpartum as part of the continuation phase.

Study visits were originally scheduled at 12 and 24 weeks postpartum ± 1 week. However, after the first two participants completed their 12-week study visit, the decision to add a study visit at 4–8 weeks postpartum was made to capture more information on breastfeeding. Device deficiencies, adverse events, and episodes of severe hypoglycemia (defined as an event requiring third-party assistance to treat hypoglycemia) or diabetic ketoacidosis were queried and recorded at study visits. Diabetic ketoacidosis was defined as an episode with elevated plasma ketones categorized as mild/self-treated (plasma ketones 5–10 mg/dL [0.5–1.0 mmol/mol]), moderate/self-treated (plasma ketones >10 mg/dL and resolving without hospital admission), or severe (plasma ketones >10 mg/dL and requiring hospital admission and treatment with intravenous insulin infusion). Probable diabetic ketoacidosis was defined as in the judgement of the investigator diabetic ketoacidosis was probable, but not enough information was available to categorize the event as meeting the above criteria.

Coronavirus 2019–related restrictions at the time of the study resulted in most research and clinical visits occurring remotely. As a result, participant weights in this trial were largely collected by self-report and are missing for individuals who did not have a home scale. Type and frequency of additional telephone, e-mail, or clinical contacts with participants was recorded in the electronic data capture forms. CGM measures were obtained using the real-time Guardian 3 sensors. The insulin pump and CGM data were uploaded to Carelink and used for therapeutic insulin adjustments as recommended by local clinical staff. Participants allocated to sensor-augmented pump therapy were permitted to use the SmartGuard suspend before low feature as they and their care teams deemed indicated. HbA_1c measurements were to be done at the local site laboratory at 12 and 24 weeks postpartum. Pandemic supply chain shortages and laboratory restrictions interfered with the collection and timing of HbA_1c measurements. All HbA_1c levels were measured using validated assays standardized to the National Glycohemoglobin Standardization Program-Diabetes Control and Complications Trial (DCCT) reference.

Attitudes toward breastfeeding were collected at 34 weeks gestation and 12 weeks postpartum using the Iowa Infant Feeding Scale (12). Diabetes distress was measured in participants using the Diabetes Distress Screening Scale at 34 weeks gestation and 12 and 24 weeks postpartum (13), and fear of hypoglycemia was assessed using the Hypoglycemia Fear Survey II at these same time points (14). Sleep quality was assessed using the Pittsburgh Sleep Quality index (15). A child food and liquid intake questionnaire and infant feeding diaries were completed at 12 and 24 weeks postpartum (16).

The primary outcome was the percent time in the target glucose range of 70–180 mg/dL (3.9–10.0 mmol/L). Key secondary glucose outcomes included percent times below range (<70 mg/dL [3.9 mmol/L] and <54 mg/dL [3.0 mmol/L]), above range (>180 mg/dL [10.0 mmol/L]), and mean glucose. Additional outcomes included CGM and closed-loop use and the participant-reported outcome surveys described above.

Statistical analyses were prespecified and performed on an intention-to-treat basis and included all available CGM data for all participants regardless of sensor use per day and/or time in closed-loop. Available data were used with no imputation for missing data. Generalized linear mixed-effects models were fitted to the repeated-measures data to assess between-group differences in glucose measures from median start time of closed-loop insulin delivery (8 days) to 11 weeks, 6 days postpartum and were adjusted for early pregnancy HbA_1c (obtained at 5–15 gestational weeks). Normally distributed random effects were fitted to the interindividual differences in change over the postpartum period (slope) and starting point (intercept). A continuous autoregressive function was used to fit within-patient correlation over time as described by Murphy et al. (17). Percent time in range was analyzed for the 24-h period, as well as for daytime (0701–2259 h) and overnight periods (2300–0700 h) as defined by Stewart et al. (5). Additionally, prespecified per-protocol analyses were performed, which excluded all CGM data from any participant who used closed-loop insulin delivery <75% of the time when they were to do so. Additional glucose metrics were compared between randomization groups using the same analysis methods.

For the continuation phase of the study, the differences between the first 12 and last 12 weeks postpartum for each trial arm were compared using linear mixed-effects regression models for time in range for the first 12 weeks through until the last 12 weeks postpartum as the dependent variable. The model adjusted for early pregnancy HbA_1c as a fixed effect and participant as a random effect. This analysis was repeated for time >180 mg/dL, time <70 mg/dL, and time <54 mg/dL. These analyses were conducted separately within each study group to determine the effect of closed-loop insulin delivery from randomized phase (day 8 to week 11, day 6 postpartum) to continuation phase (week 12, day 0, to week 24, day 0 postpartum). Questionnaire scores for participant-reported outcomes were analyzed to compare the two study groups on the change in the total score and subscale scores for each questionnaire from 34 weeks during pregnancy to 12 and 24 weeks postpartum using mixed models regression analysis. Breastfeeding or breastmilk pumping was compared between trial arms at 6, 12, and 24 weeks postpartum using a χ² test.

This pilot study had a target recruitment of 20 participants. We did not perform a power calculation. A two-sided significance level of 5% was used for all comparisons without adjustments for multiple comparisons. Statistical analyses were performed using Stata 17 (StataCorp, College Station, TX) and SPSS version 25 (IBM Corporation, Armonk, NY).

Study data are available by request from the corresponding author after ethics review and approval for the proposed study is provided and approved by the steering committee.

Twenty women were recruited. Two withdrew prior to randomization (one because she feared that she would be too busy postpartum to participate, and one preferred a different CGM system). The remaining 18 participants completed 12 weeks of postpartum follow-up in their assigned randomization groups (Fig. 1). During the continuation phase, one participant who was randomized to initial use of sensor-augmented pump therapy only used closed-loop insulin delivery for 2 weeks before she returned to using manual mode from 14–24 weeks postpartum. Two participants used the MiniMed 670G pump throughout the study, 5 changed from the MiniMed 670G to the 770G pump during the study, and 11 used the MiniMed 770G pump for the duration of the study. The suspend before low feature was discussed with participants when deemed clinically indicated by participants’ care providers. One participant (number 12) implemented this feature from time of giving birth until 12 weeks postpartum. Care providers recommended suspend before low for participants 6 and 8, but these participants chose not to use it because of reported experiences with rebound glucose elevations after having suspensions for predicted hypoglycemia when they used this feature prior to randomization. The baseline mean age of randomized participants was 32 ± 3.5 years, mean duration of diabetes was 22 ± 7.3 years, and mean early pregnancy HbA_1c was 52 ± 6.8 mmol/mol (6.9 ± 0.9%) (Table 1). Gestational age at enrollment ranged from 13 to 33 weeks gestation. Average total daily insulin doses for the first postpartum week ranged from 14.4 to 85.0 units/day.

The primary outcome of percent time in range (70–180 mg/dL) did not differ between the closed-loop and sensor-augmented pump groups (79.2 ± 8.7% vs. 78.2 ± 6.0%; P = 0.41) (Table 2). Participants randomized to the closed-loop group spent less time <70 mg/dL and <54 mg/dL than those randomized to the sensor-augmented pump group (1.7 ± 0.8% vs. 5.5 ± 3.3%; P < 0.001 and 0.3 ± 0.2% vs. 1.1 ± 0.9%; P = 0.008, respectively). Time >180 mg/dL was not different between groups (18.7 ± 8.8% vs. 15.9 ± 7.7%; P = 0.21). Mean HbA_1c was 49.5 ± 0.7 mmol/mol (6.7 ± 0.1%; n = 5) and 48.7 ± 3.7 mmol/mol (6.6 ± 0.5%; n = 7; P = 0.40) for the closed-loop and sensor-augmented pump groups, respectively (Supplementary Table 1). CGM metrics during the daytime and overnight were examined. Those randomized to the closed-loop group spent less time <70 mg/dL and <54 mg/dL during the daytime and overnight compared with those randomized to the sensor-augmented pump group (Supplementary Table 1). The mean number of episodes with ≥15 min of glucose <70 mg/dL was less in the group randomized to closed-loop insulin delivery (43.6 ± 17.3 vs. 104.3 ± 66.2; P = 0.026). The rate of change in all reported CGM metrics, apart from the SD of CGM values, differed between the closed-loop and senor-augmented pump groups by postpartum week. Figure 2 shows how percent time in and below range varied by postpartum week for individual participants. There were no statistically significant differences in percent CGM use (95.0 ± 3.4% vs. 90.9 ± 11.7%; P = 0.33) for the closed-loop and sensor-augmented pump groups or number of contacts with clinic or research staff between groups during the randomized phase (Supplementary Table 2). Closed-loop insulin delivery was used a mean of 89.5 ± 5.6% of the time among those randomized to the closed-loop group, and no participants randomized to the closed-loop group used it <75% of the time. As a result, the per-protocol analysis for the randomized phase of the study is identical to the intention-to-treat analysis.

In the continuation phase of the study, percent time in range did not differ for either group from the randomized phase to the continuation phase (79.2 ± 8.7% vs. 78.0 ± 6.8% [P = 0.44] and 78.2 ± 6.0% vs. 78.1 ± 6.1% [P = 0.66] for those initially randomized to the closed-loop or sensor-augmented pump group, respectively) (Table 2). The group initially randomized to sensor-augmented pump therapy had less time <70 mg/dL after initiation of closed-loop insulin delivery compared with during the randomized phase (5.5 ± 3.3% vs. 3.3 ± 2.2%; P = 0.039) (Table 2). The closed-loop group maintained similar CGM metrics in both study phases (Table 2).

Closed-loop use did not differ between study phases among those randomized to close-loop use (mean percent closed-loop time 89.5 ± 5.6% vs. 88.7 ± 8.1%; P = 0.77). No participants randomized to the closed-loop group used it <75% of the time, so the per-protocol analysis for those randomized to use closed-loop insulin delivery during both study phases did not differ from the intention-to-treat analysis. Those randomized to sensor-augmented pump therapy used closed-loop insulin delivery a mean of 75.8 ± 25.7% of the time (median 86% [95% CI 70.1, 95.8%]) in the continuation phase. Although two participants were eliminated in the per-protocol analysis, the results were similar to the main analysis (Supplementary Table 3).

Participant 8 had an increase in time <70 mg/dL (3.9 mmol/L) after initiation of closed-loop insulin delivery. This pattern was not evident in any of the other participants (Fig. 2).

There were no reportable serious adverse events during the study. No episodes of diabetic ketoacidosis or severe hypoglycemia occurred in either group in the randomized or continuation phases of the study (Supplementary Table 4). No device deficiencies (pump infusion set or sensor failure) (Supplementary Table 5) resulted in severe hypoglycemia or other adverse clinical outcomes. Two of the adverse events were contact dermatitis from the CGM sensor tape in one participant. No other adverse events were due to device deficiencies.

Diabetes Distress Scale and subscores, the Pittsburgh Sleep Quality Index scores, and the Hypoglycemia Fear Survey II total and behavior subscale scores did not differ significantly between randomization groups at 12 or 24 weeks postpartum or change significantly during the study (Supplementary Tables 6–8). However, the scores on the Hypoglycemia Fear Survey II worry subscale were significantly less, indicating less worry, at 24 weeks compared with 12 weeks for the group of participants who switched from sensor-augmented pump therapy to closed-loop insulin delivery in the continuation phase of the study (Supplementary Table 8).

The frequency of participants breastfeeding or pumping breastmilk did not differ between randomization groups at 6 weeks postpartum (P = 0.52) (Fig. 2 and Supplementary Table 9). However, fewer participants randomized to the closed-loop group were breastfeeding or pumping breastmilk at 12 weeks postpartum compared with the sensor-augmented pump participants (4 [44%] vs. 8 [88.9%]; P = 0.046) (Fig. 2 and Supplementary Table 9). One participant, who was subsequently randomized to the closed-loop group, indicated that she had no intention to breastfeed (participant 13). The remaining participants had positive or neutral scores toward breastfeeding on the Iowa Infant Feeding Scale (Supplementary Table 9). Another participant, who was randomized to the closed-loop group, experienced a ruptured appendix at 5 weeks postpartum (participant 1) that required separation from her infant, as coronavirus 2019 pandemic restrictions at that time would not allow her infant to join her in hospital. Breastfeeding status at 24 weeks postpartum was missing for participant 17. All remaining participants who were breastfeeding or pumping breastmilk at 12 weeks postpartum continued to do so at 24 weeks postpartum.

This randomized trial of hybrid closed-loop insulin delivery in the early postpartum period found that women randomized to hybrid closed-loop insulin delivery from 1 to 12 weeks postpartum had less hypoglycemia than those randomized to sensor-augmented pump therapy. This difference in hypoglycemia exposure was evident over the 24-h period and during the daytime and overnight. Once they were switched to hybrid closed-loop therapy, those initially randomized to sensor-augmented pump therapy spent less time below range. Those who were randomized to hybrid closed-loop insulin delivery maintained similar CGM metrics from 12 to 24 weeks postpartum when they remained on the closed-loop system. Time in range was high for both groups and did not differ between participants randomized to closed-loop or sensor-augmented pump therapy. There were no safety concerns for either group.

Published data on the use of hybrid closed-loop insulin delivery postpartum are limited. Stewart et al. (6) showed in the nonrandomized continuation phase of their study to 6 weeks postpartum that among 12 participants who chose to continue using a hybrid closed-loop system (Florence D2A; University of Cambridge, Cambridge, U.K.) after the birth of their babies, there were no safety concerns, and participants had a high time in range of 70–180 mg/dL (77.1%) and low time <70 mg/dL (2.3%) in the first 6 weeks postpartum. In their study, there was no comparison group using sensor-augmented pump therapy in the postpartum period. In our study, we found a high time in range in both randomization groups but less time exposed to hypoglycemia among participants randomized to hybrid closed-loop insulin delivery rather than sensor-augmented pump therapy. A potential explanation for the lack of difference in time-in-range glucose between groups in our study is that all participants achieved a remarkably high time in range, minimizing the potential for further improvement with hybrid closed-loop insulin delivery.

Our study provides CGM data for up to 24 weeks postpartum. There are two other published studies with CGM data to 6 months postpartum (18,19). Thirty-three women with type 1 diabetes, of whom 79% were breastfeeding, wore a masked CGM for 6 days at 1 month postpartum. Median percent time <72 mg/dL (4.0 mmol/L) for breastfeeding and formula-feeding mothers was 5.1% (range 0–19.8%) and 4.6% (range 0–10.0%), respectively (19). Among mothers who continued to breastfeed, masked CGM was applied again at 2 and 6 months postpartum; median percent time <72 mg/dL was 4.5% (range 0–22.8%) and 3.8% (range 0–29.9%), and mean percent time in range (72–180 mg/dL) was 63.3 ± 18.4% and 68.3 ± 14.8% (19). In addition to evaluating closed-loop insulin delivery, our study benefits from the detailed CGM data throughout the 6-month postpartum period in women with type 1 diabetes and adds to the limited information available on glycemia in the postpartum period among women with type 1 diabetes.

The sensor-augmented pump randomization group trended toward a lower mean glucose than the closed-loop group (P = 0.066) (Table 2). Given that both groups reached the target postpartum time-in-range glucose and that this trend was achieved with more time-below-range glucose, this lower mean glucose may not have been a benefit. Data on individual participants (Fig. 2) show that participant 8 experienced more time below range after switching to closed-loop insulin delivery compared with her time on sensor-augmented pump therapy. The reasons for this finding are unclear, but it does highlight that not all who began closed-loop therapy at 12 weeks postpartum experienced a reduction in time-below-range glucose.

The lack of change in most patient-reported outcome measures may be a result of our small sample size and/or the inability of the tools we used to adequately capture patient-reported outcome measures of greatest importance to postpartum women (1). A recent qualitative inquiry study showed that although individuals with type 1 diabetes are fearful of hyperglycemia in pregnancy, postpartum hypoglycemia is feared much more, and for some, this contributed to cessation of breastfeeding (1). Thus, our finding of less hypoglycemia exposure with postpartum use of hybrid closed-loop insulin delivery will be of importance to people living with type 1 diabetes.

Although there were statistical differences in the frequency of participants breastfeeding or pumping breastmilk by 12 weeks postpartum, the impact of hybrid close-loop insulin delivery on breastfeeding practices cannot be thoroughly evaluated in this trial because of the small sample size. The higher number of preterm births and neonatal intensive care unit admissions (Supplementary Table 11) in the closed-loop randomization group may have contributed to this lower frequency of breastfeeding. Furthermore, with fewer women breastfeeding in the closed-loop randomization group, it is possible that this influenced the study’s glycemic findings. However, initiation of closed-loop therapy in the continuation phase of the study resulted in significantly less time below range among those initially randomized to sensor-augmented pump therapy, despite most of those participants continuing to breastfeed until the end of the continuation phase. The 72% rate of any breastfeeding in our cohort at 12 weeks postpartum is similar to the 61% of women with type 1 diabetes with any breastfeeding at 4 months postpartum found by others (20). Since most participants fully or partially breastfed or pumped breastmilk for the full 24-week duration of the study, there are insufficient data to assess the impact of lactation cessation on CGM metrics.

Our study findings will need to be considered in the context of other groups of participants and other types of hybrid closed-loop insulin delivery systems to determine their generalizability once the results of the postpartum phases of ongoing trials of closed-loop insulin delivery in pregnancy become available (21) (ClinicalTrials.gov identifiers NCT04902378 and NCT03774186). Most participants were from one site, so clinic location may impact the generalizability of results; however, clinical practice is similar across participating sites. Meanwhile, those with a particular concern or fear of hypoglycemia may be reassured that using the 120 mg/dL (6.7 mmol/L) closed-loop target option postpartum reduced the time spent in hypoglycemia for most women in this trial from 1 to 24 weeks postpartum. The results of this study cannot be generalized to the first postpartum week since the closed-loop system was not used in the first postpartum week as part of the study design. This was because of concerns that basal insulin modulation with the MiniMed 670G/770G closed-loop system could be too strong for the first postpartum week (when insulin requirements drop dramatically compared with the final week of pregnancy) since the MiniMed 670G/770G algorithm is strongly influenced by total daily dose of insulin used in the previous 6 days.

Strengths of this study include the randomized design, the real-world nature that did not remotely monitor or restrict activities of participants, the duration of postpartum follow-up, and details surrounding infant feeding. The lack of differences in clinical or research contact with participants or in sensor use between randomization groups are additional strengths. Importantly, this study includes women who were in a postpartum setting, many of whom breastfed; this group is historically underrepresented or excluded from closed-loop insulin delivery studies. There are, however, some limitations. First, we cannot be certain whether the intergroup differences in exposure to hypoglycemia would have been found had all participants randomized to sensor-augmented pump therapy chosen to use the suspend before low option of the pump. We also cannot be certain that time in range would have been the same between randomization groups had the suspend before low option been used in all participants randomized to sensor-augmented pump therapy. Second, the open-label nature of this study could not be overcome. To mitigate this limitation, all statistical analyses were performed by an analyst who was masked to randomization group, and the study investigators remained masked to the randomization group results until after review of trial findings. Finally, this was a pilot study with a small number of women with a high level of formal education (Table 1) and a high percentage of time-in-range glucose that may not be reflective of the larger population of people with type 1 diabetes.

In conclusion, hybrid closed-loop insulin delivery postpartum resulted in less hypoglycemia, no difference in time-in-range glucose, and no safety concerns when used from 1 to 24 weeks postpartum. Hybrid closed-loop insulin delivery with a target glucose of 120 mg/dL appears to be a safe option for use in the first 24 weeks postpartum among individuals living with type 1 diabetes.

Acknowledgments. The authors thank those who participated in this research, as well as research staff, including Carolyn Oldford and Catherine S. Young at the University of Calgary, Alana Galper at the University of Toronto, Marie-Christine Dubé at Laval University, and Michelle Chrisp at the University of Manitoba.

Funding. This study was funded by the Calgary Health Trust. Medtronic provided in-kind support of study supplies, including Guardian Link 3 transmitters and Guardian 3 sensors and loan of study devices MiniMed 670G and 770G insulin pumps and transmitter docks.

The funders of the study had no role in the study design, data collection, data analysis, or data interpretation. Medtronic reviewed the manuscript prior to submission for correctness of the description of its products.

Duality of Interest. L.E.D. reported in-kind donations and reduced cost for study supplies for investigator-initiated trials from Medtronic, Dexcom, Tandem Diabetes Care, and Inter-analytics. J.M.Y. reported in-kind donations and reduced cost for study supplies for investigator-initiated trials from Medtronic and Abbott. D.S.F. reported in-kind donations for study supplies for investigator-initiated trials from Medtronic, Dexcom, Tandem Diabetes Care, and Inter-analytics, membership on advisory boards for Novo Nordisk, and honoraria from Sanofi and Novo Nordisk. H.R.M. sits on the Medtronic European scientific advisory board and reported receiving speaker honoraria from Dexcom, Abbott, Medtronic, and Novo Nordisk. R.J.S. reported research support from Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. L.E.D. oversaw the conduct of the trial and wrote the first draft of the manuscript. L.E.D., D.S.F., P.L., H.R.M., R.C.B., R.J.S., J.H., H.V., S.C., and J.M.Y. contributed to the statistical analysis plan. L.E.D., D.S.F., P.L., and J.M.Y. enrolled participants and provided their clinical care at their respective institutions. L.E.D., R.C.B., R.J.S., and J.M.Y. designed the study protocol. J.H. advised on infant outcomes. All authors reviewed and provided critical revisions to the manuscript. L.E.D. and S.C. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in oral abstract form at the 83rd Scientific Sessions of the American Diabetes Association, San Diego, CA, 21–24 June 2023, and at the 2023 Diabetes Canada/CSEM Professional Conference, Montreal, Canada, 25–29 October 2023.