What is the impact of an advanced hybrid closed‐loop system on weight control among people with type 1 diabetes?

Despite advances in continuous subcutaneous insulin infusion and continuous glucose monitoring (CGM), many people with type 1 diabetes (PWDs) still struggle to meet recommended glycaemic targets.1, 2 Automated insulin delivery systems (e.g., advanced hybrid closed-loop [aHCL] systems) offer more responsive glucose regulation. These systems use CGM data and predictive algorithms to automatically adjust basal insulin and correction doses in real time, while still requiring the user to give meal boluses.3 They also consistently improve glycaemic control, psychosocial outcomes, satisfaction, and sleep.4 However, their effects on body weight remain unclear. Weight gain in PWDs is a growing concern: 50% of PWDs are currently overweight or obese, adding to the already increased cardiovascular risk in this population.5 Few studies, showing conflicting results, have assessed trends in weight changes following aHCL system initiation (Table S1). Rather than using group average values, our study adopts a different approach by separating individuals who gained weight from those who maintained or lost weight after aHCL initiation, allowing a more detailed examination of associated clinical and glucometric factors. This study was a retrospective observational cohort analysis of PWDs who began aHCL-therapy (Medtronic-Minimed 780G [MM780G]) in routine care (01APR2022–21AUG2024) at Diabeter, the Netherlands. At this clinic aHCL therapy is initiated following a standardized multidisciplinary pathway, including structured education, device training, and follow-up. To maximize body mass index (BMI) and glycated haemoglobin A1c (HbA1c) data availability, broad windows were defined for baseline and follow-up. Baseline BMI was the measurement closest to automode initiation, within 180 days before or 14 days after initiation. Users were required to have ≥3 months up to 12 months of follow-up. For HbA1c, values closest to the boundaries of the two periods were selected if within ±45 days. For system-derived metrics, baseline reflected MM780G use prior to automode activation, while post-aHCL reflected the period after automode initiation. Inclusion criteria were: PWDs receiving care at the four satellite locations of the Diabeter NL clinic; type 1 diabetes diagnosis (interpretation of clinical/historical data by internationally accepted guidelines6; Table S2); disease duration ≥1 year; MM780G use; consent for CareLink Personal and EHR data use; and ≥10 days of valid sensor glucose data in both periods. To account for growth-related differences in body composition, participants ≤20 years were analysed using BMI z-scores derived from age- and sex-specific reference tables.7 Those ≤20 years at baseline but >20 years post-aHCL (n = 20) were excluded because BMI z-score and BMI cannot be analysed interchangeably. The primary outcome was change in BMI z-score /BMI, categorized as No increase/Increase using established thresholds: ≤0.05 versus >0.05 for BMI z-score8-10 and ≤0.2 kg/m2 versus >0.2 kg/m2 for BMI.11, 12 This approach quantified proportions of individuals who did or did not experience weight gain, rather than mean changes. Secondary outcomes included HbA1c, system glucometrics (TIR, TITR, TAR1–2, TBR1–2,13 glucose management indicator, mean sensor glucose, total daily insulin [TDD] dose/kg), % time in automode, and optimal settings use yes/no (≥80% of sensor time with a glucose target of 100 mg/dL [5.6 mmol/L] and an active insulin time of 2 h). Patient data were extracted from electronic health records and MM780G data from CareLink Personal. The study complied with the Declaration of Helsinki. The study was exempt from further approval procedures as the pseudonymized participants were not subjected to any study-specific interventions, actions, or restrictions and are followed in regular care. Baseline/post-aHCL data were summarized using standard descriptive statistics. Logistic regression modelled Increase/No increase in BMI z-score/BMI, with demographic, anthropometric, and glycaemic variables as predictors, analysed separately for baseline values and for changes between Baseline and follow-up. To adjust for variation in follow-up duration, this variable was entered as a covariate in all regression analyses. Given the exploratory design, no adjustments for multiple comparisons were applied. Only complete cases were included. Collinearity between variables of interest and length of observation period (days) was minimal (max Pearson r = 0.188). Of 1672 PWDs using a MM780G aHCL system, 496 met the inclusion criteria (≤20 years: n = 210; >20 years: n = 286; Figure S1). Among those ≤20 years, 66% had an increased BMI z-score, with a median (IQR) increase of −0.30 (0.17–0.49) (Figure S2). In this age group, 60% of participants with baseline BMI z-score >1 (i.e., overweight) gained further weight (Table 1). Among adults >20 years, 52% showed increased BMI, with a median (IQR) increase of 1.05 (0.63–1.70) kg/m2; 55% of those with baseline BMI >25 kg/m2 (i.e., overweight) gained further weight. Sex and diabetes duration were not associated with weight gain in either age group. In logistic regression using baseline variables (see Table S3), higher baseline BMI z-score in the ≤20 group was associated with lower odds of weight gain (OR = 0.61, p = 0.007; i.e., 1-unit increase in baseline BMI z-score corresponded to a 39% reduction in the odds of weight gain). Baseline HbA1c showed a trend toward increasing odds of weight gain in youth (OR = 1.21, p = 0.083; i.e., 21% increase in odds of weight gain per 1% higher HbA1c; Figure 1). For the >20 years group, baseline HbA1c showed a significant positive association (OR = 1.27, p = 0.024; i.e., for each 1%-point increase in baseline HbA1c, the odds of weight gain increased by 27%) and baseline TIR showed a negative association (OR = 0.99, p = 0.047; i.e., for each 1%-point increase in TIR, the odds of weight gain decreased by 1.5%). In regression models using changes from baseline to post-aHCL, both age groups showed significant associations between weight gain and larger reductions in HbA1c for both ≤20 years (OR = 0.73, p = 0.028; i.e., 27% higher odds of weight gain per 1 HbA1c %-point greater drop in HbA1c) and >20 years (OR = 0.62, p = 0.005; i.e., 38% higher odds of weight gain per HbA1c %-point greater drop in HbA1c). Increases in TTD were also associated with weight gain (≤20 years: OR = 5.83, p = 0.005; >20 years: OR = 5.04, p = 0.022), with each 0.1 U/kg increase corresponding to ~12% and ~19% higher odds, respectively (Figure 1). Only 35% used optimal settings throughout follow-up, with no associations between optimal use and weight gain (Table S4). The primary objectives of this study were to determine the prevalence of weight gain following aHCL initiation and to identify associated clinical and system-derived factors in a real-world setting. Existing literature has typically reported mean weight changes, averaging individuals who gain, lose, or maintain weight, thereby underestimating the true magnitude of weight gain. Stratified into stable/decreased versus increased weight, two thirds of PWDs ≤20 years and half of those >20 years gained weight after starting automode. Notably, 60% of overweight youth and 55% of overweight adults showed further weight increases. Higher baseline HbA1c predicted weight gain across age groups. In youth, lower baseline BMI z-score and in adults lower baseline TIR were associated with weight gain. Larger reductions in HbA1c and increases in TDD/kg following automode initiation predicted weight gain in both age groups. Weight gain with insulin intensification is well documented, particularly among individuals with high initial HbA1c and substantial HbA1c reductions.14 aHCL represents another form of intensification, with automated correction boluses increasing TDD, consistent with our findings. Among youth, lower baseline BMI z-score has previously been associated with greater weight increases,15 while this pattern is less evident in adults,14 aligning with our results. Clinically, these findings highlight the importance of (1) structured education and lifestyle counselling and (2) more frequent review of system settings, particularly for individuals initiating aHCL with elevated HbA1c or pre-existing overweight. More than half of those already overweight at baseline gained additional weight, compounding cardiometabolic risk. Additionally, recent evidence suggests that body weight variability also represents a risk factor for CVD in PWDs.16 Identifying psychosocial and dietary factors influencing susceptibility to weight gain remains an important research priority.17, 18 Strengths include a relatively large, longitudinal cohort and stratified analysis. Limitations include the retrospective observational design, low use of optimal settings, lack of dietary and lifestyle data, single-centre setting with a relatively young population, and introducing potential residual confounding. Concluding, stratified analyses revealed a sizeable high-risk subgroup, particularly individuals already overweight. Lifestyle counselling should accompany aHCL initiation to maximize glycaemic benefits while minimizing weight-related risk. SK contributed to the conception and design of the study, data collection, interpretation of the findings, and writing of the manuscript. PD was involved in the study conception and design, data collection and analysis, interpretation of the findings, and was primarily responsible for drafting and writing the manuscript. EB contributed to the conception and design of the study, performed a part of the data analysis, and contributed to data interpretation and critical review of the manuscript. TvdH participated in data collection and provided intellectual input to the manuscript. KvB contributed to the interpretation of the data within the context of clinical practice and provided critical intellectual input to the manuscript. JC, HV, and OC contributed to the critical review and refinement of the manuscript. H-JA conceived and designed the study, contributed to data interpretation, supported the writing process, and critically reviewed the manuscript. We thank Lars Boers for data extraction of aHCL system glucometrics and system data and Theo Sas for critically reviewing the manuscript. During manuscript preparation, ChatGPT (OpenAI, San Francisco, CA; September 2025 version) was used to support shortening and refining. The authors reviewed, edited, and take full responsibility for the content. No AI tools were used for data analysis, interpretation, or drawing scientific conclusions. None. TvdH, JC, and OC are full-time employees of Medtronic. Diabeter Netherlands is an independent clinic, which was acquired by Medtronic. The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/dom.70410. Data can be provided upon reasonable request. Table S1. Available studies on AID use and showing conflicting results. Table S2. Diagnostic criteria for diabetes and features suggestive of type 1 diabetes. Table S3. Associations between ‘Increased BMI (z-score)’ and parameter values stratified by age-group (≤20 years/>20 years). Table S4. Percentages of PWDs on optimal settings, stratified for ≤20 and >20 years PWDs, both substratified for ‘No increase in BMI (z-score)’ and ‘Increased BMI (z-score)’. Figure S1. Flow chart of inclusion. Figure S2. Median (IQR) differences between baseline and post-aHCL for the ≤20 and >20 years group, stratified for ‘No increase in BMI (z-score)’ and ‘Increased BMI (z-score)’. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.