{
  "abstract": "Background Tubeless insulin pumps, introduced in France in 2016, have been associated with higher user satisfaction than tubed pumps, primarily due to reduced interference with daily tasks. Whether this translates to greater treatment persistence remains uncertain. This study evaluates persistence among people with type 1 diabetes initiating tubed or tubeless pump therapy in France.Methods This real-world, retrospective study used data from the French National Health Data System, a comprehensive medical claims database for the entire French population from 2016 to 2021. Treatment persistence was assessed using Kaplan-Meier curves and Cox models. Changes from one pump type to another were compared across calendar years using χ 2 tests.Results Among 1 241 624 people with insulin claims, 233 874 had type 1 diabetes. Of these, 39% (n=90 362) initiated insulin therapy during the study period; 32% started and remained on multiple daily injections (excluded), and 68% initiated pump therapy. Among pump users (n=61 251), 51% used a tubed pump, 67% a tubeless pump, and 18% initiated both types sequentially. People initiating tubeless compared with tubed pumps had higher persistence rates (unadjusted HR 0.60, 95% CI 0.58 to 0.61) and a 40% lower risk of discontinuation (p<0.0001), consistent with the adjusted model (HR 0.59, 95% CI 0.57 to 0.61). Tubeless pumps were also associated with fewer changes from one therapy to another within 12 months. Treatment changes rose for tubed pump users after tubeless pumps entered the French market.Conclusions Higher persistence and lower therapy changes among tubeless pump users underscore its sustained benefit for diabetes management.",
  "authors": [
    {
      "affiliations": [
        "Department of Diabetes, Metabolic Disease and Nutrition, Toulouse University Hospital, University of Toulouse, France, Toulouse, Occitanie, France"
      ],
      "name": "Helene Hanaire"
    },
    {
      "affiliations": [
        "Real World Evidence, Public Health Expertise, Paris, France"
      ],
      "name": "Alexandre Vimont"
    },
    {
      "affiliations": [
        "Insulet France SAS, Paris, France"
      ],
      "name": "Alice Bonin"
    },
    {
      "affiliations": [
        "Insulet France SAS, Paris, France"
      ],
      "name": "Gabriel Guigand"
    },
    {
      "affiliations": [
        "Insulet International, London, UK"
      ],
      "name": "Colin Hopley"
    },
    {
      "affiliations": [
        "Department of Endocrinology ad Metabolic Diseases, Hôpital Saint-André, Bordeaux University Hospital, Bordeaux, France, Pessac, Nouvelle-Aquitaine, France"
      ],
      "name": "Bogdan Catargi"
    }
  ],
  "full_text": "WHAT IS ALREADY KNOWN ON THIS TOPIC Insulin pump therapy, both tubed and tubeless, is associated with improved glycemic control and quality of life compared with multiple daily injections, but tubed pumps are linked to wear-related discomfort that may lead to discontinuation.Tubeless pumps offer wearability and convenience advantages, yet comparative real-world data on persistence and complication rates between tubed and tubeless pumps remain limited, especially in large populations.WHAT THIS STUDY ADDS This study provides real-world evidence from a nationwide study of around 61 000 patients that tubeless insulin pumps are associated with a 40% lower discontinuation risk and fewer therapy changes within 12 months compared with tubed pumps, across all age groups.It also reveals higher incidence rates of acute metabolic complications among tubed-pump users, while demonstrating stable persistence and lower complication rates for tubeless pumps.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY These findings justify further investigation into the long-term outcomes of pump therapy and the support for personalized therapy recommendations toward access to tubeless pumps, especially for populations where wearability is a barrier.Introduction Type 1 diabetes affects individuals of all ages globally, with its prevalence steadily rising each year. 1 In France, an estimated 320 943 people were living with type 1 diabetes in 2023, including over 21 000 individuals under the age of 18, and an increasing number of young people (<20 years) are diagnosed each year.2 3 Effective insulin therapy is essential for people living with type 1 diabetes to maintain blood glucose levels within normal ranges and reduce glycemic variability, which, if done safely, can slow or prevent diabetes-related microvascular, macrovascular, and acute metabolic complications.4 5Much of the research on the persistence of insulin therapy has focused on the discontinuation of continuous subcutaneous insulin infusion, or insulin pump therapy, in favor of multiple daily injections (MDI).6 Discontinuation of pump therapy is relatively uncommon (median rate of 7%)6 and can be driven by a number of factors, including personal comfort, challenges meeting glycemic targets, and psychosocial factors.6 The most commonly reported reasons for discontinuation are wear-related concerns with insulin pumps, like discomfort, interference with activities, and body image concerns.6 The reasons for discontinuing therapy can vary both between and within countries, highlighting the importance of localized data.6Insulin pump use has been increasing,7 particularly in light of evidence demonstrating an association with improved glycemic outcomes (eg, lower glycated hemoglobin (HbA1c)), lower insulin requirements, and heightened quality of life compared with MDI.8–11 Insulin pumps can be broadly characterized as tubed or tubeless.12 Tubed pumps deliver insulin through a thin tube connected to an infusion set inserted into the subcutaneous tissue, while tubeless pumps deliver insulin directly through a built-in cannula.12 Tubeless pumps are more discreet and waterproof, offering added convenience for some users, but typically have smaller insulin reservoirs than their tubed counterparts.12 It is possible that the convenience and wearability advantages offered by a tubeless pump might positively influence treatment persistence rates, as it directly addresses some wearability complaints associated with tubed pumps (eg, tube snagging, need to disconnect during showering or swimming).13 However, research focused on persistence rates in terms of changes between different forms of insulin pumps is limited.Tubeless insulin pumps (Omnipod Insulin Management System and Omnipod DASH Insulin Management System) were introduced in France in 2016, adding to the existing range of tubed pumps already available. Recent evidence from France suggests higher user satisfaction with tubeless pumps compared with tubed pumps, largely due to reduced interference with daily tasks.14 However, whether this satisfaction translates to improved treatment persistence remains to be seen. This study aimed to evaluate the treatment persistence in people with type 1 diabetes in France initiating tubed or tubeless insulin pump therapy and to estimate the incidence of diabetes complications in this group.Methods Study design and data source This study was a real-world, retrospective, observational study of the French National Health Data Information System (Système National des Données de Santé (SNDS) 15) of people with type 1 diabetes. Participant information was obtained from the SNDS, which is an exhaustive database of all medical claims made for the entire French population, including demographics, medical data (eg, long-term disease coverage), information related to in-hospital admissions (eg, procedures), reimbursements (eg, drug prescriptions), and date of death.15 16 SNDS is made possible through a compulsory public insurance scheme in France for all individuals, which covers most healthcare costs. This study used 6 years of SNDS data (2016–2021) to address the study objective. Data from 2011 to 2015 was used only to obtain the medical history of individuals.Participants All people in the SNDS with type 1 diabetes being treated with insulin therapy were eligible for inclusion. The SNDS data does not include diagnoses of type 1 diabetes; as such, people with type 1 diabetes were identified using the inclusion and exclusion criteria described in the ENTRED study. 17 Briefly, a person was classified as having type 1 diabetes if they submitted a claim for insulin therapy, had no claims for antidiabetic medications, and were under the age of 45 at the start of the study (2016), at insulin initiation, or when placed on long-term disease coverage for diabetes (ICD-10 codes E10 and E11). To minimize misclassification, individuals using MDI were additionally excluded if they had fewer than six insulin reimbursements, discontinued insulin therapy for at least 6 months, or initiated insulin therapy less than a year before the study ended (2021). Individuals with gestational diabetes were also excluded.People were divided into treatment groups according to the type of insulin pump therapy each person initiated during the study period (ie, tubeless or tubed). The SNDS database includes an insulin pump procedure code to identify the specific insulin pump used. Individuals not included in the two insulin pump groups were assumed to be using MDI, defined as those with a record of insulin and/or syringe or pen reimbursements without any recorded insulin pump codes. This approach was done to prevent the misclassification of insulin injections as MDI when insulin injections are used in combination with pumps (bolus insulin) or as a backup measure during pump failures. Temporary interruptions of less than 1 month and replacement of the same pump model due to device failure could not be captured by the database, and hybrid closed-loop (HCL) systems were not on the market at the time of the study.Outcomes The primary outcome of interest was treatment persistence for tubed and tubeless insulin pump therapy. Time to next treatment (TTNT) was assessed as the duration of insulin pump use from the initiation of therapy to the initiation of a different insulin therapy. End of follow-up, death, or rescue therapy for insulin pumps was censored. A change in glucose sensor was not considered a change in treatment. Secondary endpoints included the proportion of changes from one therapy to another within 12 months of initiating therapy with an insulin pump, the order of insulin therapies used, and the average duration of each therapy. Exploratory objectives included the incidence of diabetes-related complications during therapy (acute metabolic complications (hypoglycemia, ketoacidosis, and ketoacidosis with coma), macrovascular (cardiovascular, cerebrovascular, and peripheral vascular), and microvascular (retinopathy, nephropathy, and neuropathy)). Complications were identified through hospitalization records in the SNDS, and only those severe enough to require hospitalization were included.Statistical analyses Most statistical analyses were limited to people who started insulin pump therapy during the study period (people who remained on the same insulin therapy from before the study were excluded). The exception to this was the overall description of people with type 1 diabetes, which included the full analysis set of people meeting the inclusion criteria. Descriptive statistics were used to summarize characteristics of people at baseline or the time of insulin pump initiation and the medical history of people overall and by treatment group, including demographics (ie, age, sex, long-term disease coverage, solidarity coverage for people with limited resources), duration of long-term disease coverage for diabetes, and people new to insulin therapy. Medical history was obtained from the SNDS historical data (2011–2015), including history of diabetes complications (detailed above) and comorbidities (eg, chronic obstructive pulmonary disease, hypertension, dyslipidemia, coronary disease, history of stroke, heart failure, active cancer (within the last year), history of cancer (within the last 5 years), psychiatric disorder, neurovascular disease, and rare diseases).Survival analysis was conducted to analyze the TTNT for people initiating tubed and tubeless insulin pump therapy using Kaplan-Meier curves (non-parametric), overall, and by age group, as well as unadjusted and adjusted Cox models (parametric) to obtain HRs with 95% CI. HRs from the unadjusted model were compared with a likelihood ratio test; HRs from the adjusted model were compared with χ2 tests. Covariate selection for the adjusted Cox model was performed using a backward selection process, with an entry p-value threshold of 0.05, to identify significant predictors and assess potential interactions. The model was adjusted for the following demographic and clinical variables: age, sex, pregnancy, duration of long-term disease coverage for diabetes, concomitant medication, year of treatment initiation, and history of diabetes complications by complication type.The proportion of people who changed from one insulin therapy type to another within 12 months of initiation of a pump was reported by initial insulin pump type and initiation year. The χ2 tests were used to compare therapies and calendar years of therapy changes by type of insulin pump. Incidence of diabetes complications was calculated as the number of complications occurring during each treatment over the study period, expressed as the number of events per 1000 person-years. To minimize differences in baseline risk, crude and age-sex standardized incidences were reported for all people with type 1 diabetes, with their associated 95% CI (Wald method), and by type of insulin therapy being used at the time.This study was conducted under the French SNDS governance framework.Results Patient characteristics and medical history A total of 233 874 people living with type 1 diabetes were identified among 1 241 624 people with claims for insulin over the 2016–2021 period ( figure 1). Among people with type 1 diabetes, 38.6% (n=90 362) initiated insulin therapy during the study period, with 32.1% starting and remaining on multiple daily injections, and 68.2% initiating pump therapy. Among people starting on a pump (n=61 251), 51.2% used a tubed pump, 67% used a tubeless pump, and 18.3% (11,193) used both types sequentially (figure 1). Patients under 18 years or over 75 years comprised 28.3% and 0.4% of people initiating tubed insulin pumps, and 25.8% and 0.4% of people initiating tubeless insulin pumps, but 16.2% and 6.3% of all people with type 1 diabetes were included in the study (table 1). Few people initiating tubed or tubeless insulin pump therapy were on solidarity coverage (17.3% and 13.5%, respectively) and nearly all had long-term disease coverage for diabetes (98.6% and 98.9%, respectively) with an average duration of 14.9±10.8 years and 15.5±11.2 years, respectively. Patients new to insulin treatment made up 15.2% of people initiating a tubed pump and 8.3% of people initiating a tubeless pump. Among people initiating insulin pumps during pregnancy, 10.5% initiated a tubed pump, and 3.9% initiated a tubeless pump. Additional demographic characteristics of people overall and at the time of insulin pump initiation are presented in table 1.Figure 1Study flowchart. MDI, multiple daily injection; T1D, type 1 diabetes.Table 1Demographic characteristics of people with T1D overall at baseline (2016) and for those initiating a tubed or tubeless insulin pump therapy during the study period (2016–2021)VariablePeople initiating a tubed pump(n=31 377)People initiating a tubeless pump(n=41 067)People with T1D in 2016(n=2 33 874)Sex: male, n (%)13 933 (44.4%)17 643 (43.0%)127 911 (54.7%)Age (years)30.69 (18.60)32.69 (17.41)34.60 (18.81)Age group, n (%) <188798 (28%)10 593 (25.8%)37 934 (16.2%) 18–349089 (29%)11 904 (29%)55 104 (23.6%) 35–543799 (12.1%)13 459 (32.8%)71 027 (30.4%) 55–749574 (30.5%)4931 (12%)55 002 (23.5%) ≥75117 (0.4%)180 (0.4%)14 807 (6.3%)People with solidarity healthcare coverage, n (%)5186 (17.3%)5330 (13.5%)9929 (13.5%)People with long-term disease coverage for diabetes, n (%)29 612 (98.6%)39 120 (98.9%)204 341 (94.5%)Duration of long-term disease coverage for diabetes*14.94 (10.80)15.52 (11.15)18.27 (11.87)People new to insulin, n (%)†4759 (15.2%)3397 (8.3%)26 223 (11.2%)Pregnant at the time of initiation3314 (10.5%)1591 (3.9%)7135 (3%)Unless otherwise noted, values presented are mean (SD).*Limited to people with long-term disease coverage for diabetes.†People with no insulin treatment at baseline (2016) or before insulin therapy initiation during the study period (2016–2021).‡Represents the insulin pump therapy first initiated by a person during the study period.T1D, type 1 diabetes.A history of diabetes complications and comorbidities among people overall and among those starting tubed or tubeless insulin therapy is presented in table 2. In all people with type 1 diabetes, those who initiated tubed insulin pumps, and those who initiated tubeless insulin pumps, the proportions with a history of acute metabolic complications were 9.6%, 18.5%, and 15.4%, respectively, and of microvascular complications were 5.8%, 12.5%, and 12.5%, respectively (table 2). The proportion with macrovascular complications in those initiating pump therapy was similar to the overall sample (table 2). Few people in each group had a history of comorbidities, with hypertension and dyslipidemia being the most common and seemingly lower proportions of most comorbidities among those initiating insulin pump therapies than the overall sample (table 2).Table 2History of diabetes complications and comorbidities for people with T1D overall at baseline (2016) and for those initiating a tubed or tubeless insulin pump therapy during the study period (2016–2021)People initiating a tubed pump(n=31 377)People initiating a tubeless pump(n=41 067)People with T1D in 2016(n=2 33 874)Diabetes complications, n (%)* Acute metabolic complications6559 (18.5%)6188 (15.4%)21 737 (9.6%) Macrovascular complications1372 (3.9%)1419 (3.5%)6925 (3.1%) Microvascular complications4418 (12.5%)5028 (12.5%)13 089 (5.8%)Comorbidities, n (%)† Chronic obstructive pulmonary disease3710 (6.8%)2138 (5.6%)17 353 (8%) Hypertension6455 (21.4%)6870 (17.4%)78 141 (35.4%) Dyslipidemia5800 (19.3%)6389 (16.2%)66 743 (30.3%) Coronary disease3416 (6.1%)1406 (3.6%)19 092 (8.7%) History of stroke852 (1.5%)430 (1.1%)5424 (2.5%) Heart failure714 (1.3%)223 (0.6%)7013 (3.2%) Active cancer (within previous 1 year)797 (1.4%)443 (1.1%)4405 (2%) History of cancer (within past 5 years)1026 (1.8%)609 (1.5%)5642 (2.6%) Psychiatric disorder3016 (5.4%)1949 (4.9%)37 830 (17.2%) Neurovascular disease1216 (2.2%)690 (1.7%)6703 (3%) Rare disease‡1630 (2.9%)1054 (2.7%)6682 (3%)*History of diabetes complications identified using historical data (2011 to study initiation). Acute metabolic complications include hypoglycemia, ketoacidosis, and ketoacidosis with coma; macrovascular complications include cardiovascular, cerebrovascular, and peripheral vascular complications; microvascular complications include retinopathy, nephropathy, and neuropathy.†Comorbidities present at the initiation of insulin pump therapy.‡Rare diseases include HIV, hemophilia, cystic fibrosis, hereditary metabolic diseases, amyloidosis, and others.T1D, type 1 diabetes.Time to next treatment Patients initiating tubeless insulin pumps had higher persistence rates compared with people initiating tubed pumps, with an unadjusted HR of 0.60 (95% CI 0.58 to 0.61), representing a 40% lower risk of treatment discontinuation in people initiating therapy with a tubeless pump (p<0.0001; figure 2). Subgroup analysis by age group demonstrates tubeless insulin pump initiation was associated with a significantly lower risk of treatment discontinuation than tubed pumps in all age groups (online supplemental figure S1). Results were of similar extent for patients below 18 years of age (HR 0.58, 95% CI 0.55 to 0.62) and above 18 years of age (HR 0.61, 95% CI 0.57 to 0.64). In the adjusted model, people initiating tubeless insulin pumps also had higher persistence rates than those initiating tubed pumps, with an adjusted HR of 0.59 (95% CI 0.57 to 0.61) associated with a 41% lower risk of discontinuation for other treatment (online supplemental Table S1).SP110.1136/bmjdrc-2025-005860.supp1Supplementary dataFigure 2Kaplan-Meier curves of time to next treatment for people initiating tubed (black line) versus tubeless (blue line) insulin pump therapy over a follow-up period of up to 60 months.Treatment changes within 12 months Compared with tubed insulin pumps, tubeless pump therapy was consistently associated with fewer changes from one therapy to another within 12 months of initiation, regardless of initiation year (p<0.0001) ( figure 3). The proportion of changes in the tubed pump group ranged from 14.2% in 2019 to 17.6% in 2017 and from 7.4% in 2016 to 8.6% in 2020 for the tubeless pump group (figure 3). The proportion of changes each year for tubeless pumps was stable, with a slight increase from 2017 to 2018. The proportion of changes each year for tubed pumps increased from 2016 to 2017 and decreased from 2018 to 2019 (figure 3).Figure 3Proportion of people with type 1 diabetes changing treatments within 12 months of initiating tubed (n=31 377) or tubeless (n=41 067) pump therapy by year of treatment initiation (n=61 251). Asterisk (*) indicates a statistically significant difference in the proportion of changes compared with the previous year (p<0.05).Treatment therapy sequences 70% of people who initiated insulin pump therapy did not change treatments during the study period. This value comprised 37% of people initiating tubeless insulin pumps and 33% initiating tubed insulin pumps ( figure 4). 19% of people changed from tubed to tubeless insulin pumps after an average of 1.8 years of treatment, and 4% changed from tubeless to tubed insulin pumps after an average of 2.2 years (figure 4). 6% of people changed to MDI after using insulin pumps (tubed and tubeless) for around 1 year (figure 4).Figure 4Proportion of participants and mean length of time using each treatment for various treatment sequences observed during the study period (2016–2021). Tubeless pump use indicated with blue bars; tubed pump use indicated with black bars, and MDI indicated with white bars. MDI, multiple daily injection.Of the 41 067 people in the tubeless pump group, 87% remained on tubeless insulin pumps, 55% of whom were previously treated with MDI and 32% with a tubed pump. Of the remaining, 9% changed to tubed pumps and 6% transitioned to MDI. Of the 31 377 people in the tubed pump group, 63% remained on tubed insulin pumps, of whom 54% were previously treated with MDI and 9% with a tubeless pump. Of the remaining people, 25% changed to a tubeless pump and 5% transitioned to MDI. Pump-to-pump transitions may reflect technological upgrading, patient preference, or evolving clinical needs.Incidence of complications Age-sex standardized incidences of acute metabolic complications were estimated at 11.9, 12.8, and 19.8 events per 1000/person-years for the overall sample of people and tubeless and tubed pump groups, respectively, mostly from ketoacidosis (9.3, 10.8, and 16.6 events per 1000/person-years, respectively; table 3). Microvascular complications were more frequent in the groups initiating insulin pump therapy than in the overall sample of people with type 1 diabetes, while macrovascular complications were similar in all groups (table 3). Retinopathy was a main microvascular complication with an incidence rate of 8.9, 14.9, and 16.4 events per 1000/person-years for the overall sample, tubeless, and tubed pump groups, respectively.Table 3Incidence of diabetes complications* for people with T1D overall at baseline (2016) and for those using tubed or tubeless insulin pump therapy† during the study period (2016–2021)VariablePeople with T1D (n=2 33 874)Tubed pump(n=31 377)Tubeless pump(n=41 067)Median treatment duration (months)‡55.730.332.5Patient-year1 085 56579 128111 307Crude incidence Acute metabolic complications11.9 (11.7 to 12.1)22.8 (21.7 to 23.8)14.9 (14.2 to 15.7) Macrovascular complications14.8 (14.5 to 15)13.0 (12.2 to 13.8)8.7 (8.1 to 9.2) Microvascular complications19.1 (18.8 to 19.3)30.2 (29 to 31.4)23.0 (22.2 to 23.9)Age-sex standardized incidence Acute metabolic complications11.9 (11.7 to 12.1)19.8 (18.8 to 20.8)12.8 (12.2 to 13.5) Macrovascular complications14.8 (14.5 to 15)15.5 (14.6 to16.3)14.8 (14.1 to 15.5) Microvascular complications19.1 (18.8 to 19.3)31.2 (29.9 to 32.4)26.7 (25.7 to 27.6)Values are events per 1000 person-years (95% CI), unless otherwise noted.*Acute metabolic complications include hypoglycemia, ketoacidosis, and ketoacidosis with coma; macrovascular complications include cardiovascular, cerebrovascular, and peripheral vascular complications; microvascular complications include retinopathy, nephropathy, and neuropathy.†Groups based on the type of insulin pump therapy being used at the time the complication was recorded.‡Consecutive months.T1D, type 1 diabetes.Discussion This is the first real-world study to analyze the TTNT, the prevalence of changes from one therapy to another, the order of insulin therapies used, and the incidence of diabetes complications for people with type 1 diabetes initiating insulin pump therapy in France. The use of the comprehensive SNDS database enabled nearly every individual with type 1 diabetes in France to be considered for the analysis, garnering an extraordinarily generous and generalizable sample with which to meet our research objectives (n=90 362). Results from the primary analysis show that treatment with tubeless pumps, which in this case was limited to either the Omnipod or Omnipod DASH insulin management systems, was associated with higher persistence compared with tubed pumps for all age groups (unadjusted) with fewer changes from one treatment to another within the first 12 months. Pump-to-pump transitions may reflect technological upgrading, patient preference, or evolving clinical needs and should not be interpreted solely as discontinuation. This aligns with a real-world cross-sectional study comparing user experiences between different insulin delivery systems in France, with 84.6% of tubeless pump users reporting they would retain the use of tubeless pumps if changing to HCL systems; only 43.3% of tubed pump users reported the same. 14 Tubeless pump users in that study also reported the lowest interference with activities compared with HCL, tubed pump, and MDI users, with interference scores similar to one another.14 Very few people initiating insulin pump therapies in our study changed to MDI, which also aligns with research investigating the discontinuation of insulin pumps in favor of MDI, showing a median of 7% underwent this change.The proportion of people changing therapies within 12 months of insulin pump initiation was significantly higher for tubed pump users in 2017 (17.6%) compared with 2016 (14.9%), followed by a significant decline from 16.4% in 2018 to 14.2% in 2019. This trend coincides with the introduction of the Omnipod and Omnipod DASH systems into the French market in 2016, suggesting an initial wave of changes from tubed to tubeless pumps in the early years, followed by a decline in 2019 and beyond as early adopters had already transitioned. Our results showing 20% of people who initiated tubed insulin pumps changed to tubeless pumps during the study period (with only 1% changing back) support this interpretation.This study, with a 60-month follow-up period, which resulted in a median consecutive treatment duration of 32.5 months for tubeless and 30.3 months for tubed pumps, likely provides sufficient time for people to evaluate their treatment experience and consider potential therapy changes. This timeline is supported by research of a German and Austrian registry suggesting discontinuation of therapy with a tubeless insulin pump (Omnipod), though uncommon in the first year (7.4%), occurred after 0.9 ± 1.2 years, and the retention rate remained high (>90%) after 1 year of use.18 That rate of discontinuation in the first year is echoed in the proportion of changes observed for tubeless pump therapy in our study (7.4%–8.6%). Similarly, it also supports the finding that without discontinuation, people remained on their insulin pumps for about 4 years (3.7 years and 4.1 years for tubeless and tubed pumps, respectively) during our 5-year study. A generally high retention rate has also been found in other studies of insulin pump users.19 Our results also corroborate with a recent French real-world study that revealed sustained adherence (80%–89%) over a median duration of use of 15–16 months regardless of age group and type of diabetes.20Results suggest that more people initiating either tubed or tubeless pump therapy had a history of acute metabolic complications and microvascular complications compared with the overall sample of people with type 1 diabetes. This is likely reflecting real-world practice in France, where insulin pumps are indicated for patients not achieving optimal glycemic control with MDI. Findings also indicate that fewer people initiating insulin pumps had a history of certain comorbidities (hypertension, dyslipidemia, and coronary disease) than the overall sample. These differences may reflect the younger age distribution of people in the insulin pump treatment groups, with a seemingly higher proportion under 18 years old, fewer over 75 years old, and, in the tubeless pump group specifically, fewer people aged 55–75 years.The age-sex-standardized incidence of several diabetes complications during the study period appeared much higher in the tubed pump group (for acute metabolic complications and microvascular complications) and the tubeless pump group (for microvascular complications) than in the overall sample of people with type 1 diabetes. The tubed pump group had incidence rates that appeared even higher than those in the tubeless pump group. This may, in part, be explained by the apparently higher proportion of people new to insulin therapy in the tubed pump group (15.2%) compared with the tubeless pump group (8.3%) and the overall sample of people with type 1 diabetes (11.2%).20 21 The age-sex standardized incidence of ketoacidosis in the overall sample and for both pump groups is in line with a systematic review that reported the incidence of diabetic ketoacidosis to be between zero and 56 events per 1000 person-years.22 23 These results should be interpreted with caution since this study was not designed to support causal inference or comparative effectiveness, and this objective was exploratory. In addition, complications were identified only when resulting in hospitalization, capturing the most severe events and potentially reflecting differences in healthcare-seeking behavior, access to care, or admission thresholds across groups rather than differences attributable to pump type.A large real-world study of 4738 children, adolescents, and adults with type 1 diabetes initiating treatment with the Omnipod DASH system found improved HbA1c and fewer hypoglycemic events in adults changing from MDI or tubed insulin pumps and among children changing from MDI.23 The glycemic benefits of the Omnipod or Omnipod DASH systems compared with MDI have been demonstrated on several occasions, yet the glycemic benefits compared with conventional tubed pumps have not been consistently found.24–27 The results from this study can inform clinical practice by helping healthcare providers align insulin pump therapy choices more closely with individual needs and expectations, ensuring greater treatment satisfaction, persistence, and personalized diabetes management goals.28The results of the present study should be interpreted in light of potential limitations. The nature of the SNDS data meant that we did not have complete medical records with glycemic data, socioeconomic status, treatment center characteristics, or information on the reason for initiating or changing insulin therapies. Some of these covariates may have residual confounding effects in multivariate analysis of persistence. The identification of diabetes-related complications was limited to those severe enough to warrant hospitalization, so the results cannot be generalized to mild or moderate complications or complications managed in an outpatient setting. Medical claims recorded in the SNDS, including diagnostic codes, are subject to inaccuracies that may have impacted the distinction between type 1 and type 2 diabetes. The differentiation between type 1 and type 2 diabetes was restricted on the basis of age, which is consistent with other research using SNDS data, such as the ENTRED study and the annual national health insurance report, but allows for potential misclassification, especially of people with adult-onset type 1 diabetes or early-onset type 2 diabetes.29 30 Excluding onset autoimmune diabetes after 45 years from the study was expected to have limited impact on persistence results since it concerned a limited number of patients, and results by age class demonstrated a consistent persistence effect. Also, persistence was measured with TTNT, which is a reimbursement-based proxy and did not capture clinical reasons for switching, temporary interruptions, or replacement failure, which is not captured by the database. Moreover, pump-to-pump switching may represent upgrading or changes in patient needs rather than poor persistence, whereas pump-to-MDI transitions more closely reflect discontinuation of pump therapy.Since conducting the analyses presented in this paper, significant advancements have been made in utilizing artificial intelligence to differentiate between type 1 and type 2 diabetes within the SNDS database.31 Future research may benefit from incorporating these emerging methodologies.31 Treatment changes were identified through records of medical claims (eg, device or insulin claims), which may not accurately reflect the actual therapy being used. HCL systems were not available on the market in France until 202132 and were therefore not incorporated into this research. Future studies can investigate if the findings found here translate to people initiating therapy with tubed or tubeless HCL pumps. The individuals excluded from the present study because they were identified as having type 2 diabetes are included in a forthcoming complementary study. This study benefits from a large sample size and a comprehensive real-world setting, capturing nearly all people initiating insulin pump therapy in France. Consequently, these results have strong external validity and are likely generalizable to similar populations.Conclusions This study leveraged the SNDS, one of the largest health administrative databases globally, providing comprehensive real-world insights into insulin pump therapy use across the French population. Over a 60-month study period, people initiating tubeless insulin pumps demonstrated higher treatment persistence and fewer changes to other therapies, with a 40% lower risk of treatment discontinuation than those initiating therapy with tubed pumps. These findings highlight the potential benefits of tubeless pump therapy in diabetes management. These results provide valuable evidence to guide clinical decision-making and counseling, ultimately supporting more informed choices in insulin pump therapy.",
  "title": "Tubed or tubeless insulin pumps? A retrospective real-world analysis of national French health data",
  "uid": "e3c8dbdc-b5de-5e69-b7eb-8fdac26835eb"
}