Real-World Evidence of Omnipod® 5 Shows Glycemic Outcomes That Indicate Effective Use of the System in a Non-Clinical Setting
In February of 2024, Forlenza et al. published a retrospective analysis titled “Real-World Evidence of Omnipod® 5 Automated Insulin Delivery System Use in 69,902 People with Type 1 Diabetes” in Diabetes Technology & Therapeutics. Prior to this study, pivotal trials of the tubeless Omnipod® 5 Automated Insulin Delivery (AID) System showed improved glycemic control for adult and pediatric patients with type 1 diabetes. In one trial, Omnipod 5 improved time in range (TIR) by 9.3% in adults/adolescents and by 15.6% in children.1 In another, it improved TIR of very young children (aged 2-5.9) by 10.9%.2 Following these clinical trials, Forlenza and colleagues sought to assess effectiveness of the system in a real-world setting. They found that the “glycemic outcomes from a large and diverse sample of nearly 70,000 children and adults demonstrate effective use of the Omnipod 5 System under real-world conditions.”3
Study Overview
The Omnipod 5 System automatically and securely uploads patient data to Insulet Corporation’s secure database. Because of this feature, the selection bias was minimized4 and the initial dataset encompassed all Omnipod 5 users in the US. The primary analysis included all users who met the following criteria as of July 31, 2023:
- at least 2 years old
- self-reported type 1 diabetes
- at least 90 days of CGM data (it was a requirement that 75% of the days where CGM data were available had at least 220 readings per day/76% of possible daily readings)
- using ≥5 units of insulin per day
Using these criteria, a total of 69,902 patients (22,162 aged 2-17, 47,740 aged 18+) across 7696 US clinics were included for analysis, representing about 17.7 million days of Omnipod 5 use.5 Additional analyses were conducted for patients using an average target glucose of 110 mg/dL, which included 37,640 patients.
Results Summary
Omnipod 5 allows users to customize their target glucose from 110mg/dL to 150mg/dL in increments of 10mg/dL. There were differences in results when researchers assessed average target glucose, as shown in Table 1.
Table 1: Glycemic Measures Based on Average Target Glucose
| 110 mg/dL | 120 mg/dL | 130-150 mg/dL | |
| Median TIR (70-180 mg/dL) | 68.8% | 61.3% | 53.6% |
| Median time below range (<70 mg/dL) | 1.12% | 0.87% | 0.71% |
| Median time below range (<54 mg/dL) | 0.19% | 0.15% | 0.13% |
The full population studied had a median TIR of 64.2% and a median time below range of 0.97%. The 2-17 age cohort had a median TIR of 60.6% and time below range of 1.21%. The 18+ cohort had a median TIR of 66.0% and 0.86% time below range.
It’s worth noting that the younger cohort was more likely to use the customizable target glucose feature and more likely to choose the higher target glucose settings. Regardless of the chosen target glucose, time in hypoglycemia was low across the whole cohort.
Cohort Using 110mg/dL Average Target Glucose
The analyses examining only the cohort using an average 110mg/dL target glucose (n=37,640) yielded the following results:
Bolusing:
Patients bolusing four or more times per day had a median TIR of 72.4% and 1.28% time in hypoglycemia. Those bolusing fewer than four times a day had a median TIR of 59.9% and 0.85% time in hypoglycemia. Though both groups had low time below range, bolusing behavior had a significant impact on TIR results.
Prior Omnipod DASH® Users:
Data were available for 3582 previous Omnipod DASH users, showing significant TIR and hypoglycemia improvements after Omnipod 5 initiation.
For children and adolescents median TIR increased from 60.5% to 69.3% with Omnipod 5. Time in hypoglycemia declined from 2.12% to 1.54% for <70mg/dL while time <54 mg/dL declined from 0.32% to 0.27%.
In adults the median TIR increased from 65.1% to 73.3%. Time <70 mg/dL decreased from 2.02% to 1.06%, and time <54 mg/dL declined from 0.27% to 0.18%.
Prior MDI Users:
Data were available for 6525 self-reported MDI users using an average 110mg/dL target glucose. This group also showed favorable outcomes on Omnipod 5.
Overall, their median TIR was 70.8% and time <70mg/dL was 0.96%. The child/adolescent cohort of this group had a median TIR of 68.9% and time below range of 1.15%. For the adult cohort, it was 71.3% and 0.90%, respectively. These results suggest that prior MDI users can be successful on Omnipod 5.
Patients on Medicare and Medicaid:
Similar to the other subgroups, patients with Medicare and Medicaid insurance using an average target glucose of 110mg/dL showed promising outcomes on Omnipod 5.
The Medicaid insurance cohort (n = 1064; median age 21 years) had a median TIR of 62.2% and time <70 mg/dL of 1.12%
The Medicare insurance cohort (n = 1503; median age 67 years) had a median TIR of 72.3% and time <70 mg/dL of 0.91%.
Limitations of this study include exclusion of patients who discontinued Omnipod 5 use within 90 days, self-reported user demographics, the lack of a control group, and the exclusion of off-label use. Overall, the study also has major strengths, such as the large data set and the fact that the automatic data uploads via SIM card or wireless internet minimized selection bias.4 The authors note that “such a sizeable sample provides evidence of Omnipod 5 System use in a more diverse population, including nearly 6,000 users with public health insurance and users that may not be considered typical candidates for AID.”
Key Takeaways
Omnipod 5 was found to be used effectively outside of a clinical setting and achieved outcomes that were comparable to those of the pivotal trial.6 On the population level, patients who used the lowest target glucose also saw the highest TIR outcomes. However, the authors note that the high incidence of young patients and the 40% of adult patients using the customizable glucose targets “emphasizes that having the option to customize glucose targets can be beneficial to optimize outcomes and personalize care.”
Though the authors listed several other studies of real-world evidence for AID, they stressed that there are multiple factors impacting the results of a real-world study, including where the study cohort is from. For example, “national registry data have demonstrated people with T1D in the US have higher HbA1c levels on average compared to people with T1D across Europe.”7 Due to these confounding variables, the authors caution comparing real-world evidence outcomes between studies.
Finally, the authors write that “future investigations can help understand the challenges faced by users who are not achieving consensus targets, the underlying differences between those with and without sufficient data, and reasons for attrition.”