A Multicenter Study of an Investigational Extended Wear Insulin Infusion Set in Adults with Type 1 Diabetes

Abstract

Objective:

To evaluate the function and safety of the SteadiSet^™ infusion set over a continuous 7-day wear period in adults with type 1 diabetes.

Research Design and Methods:

Participants used a SteadiSet infusion set with a t:slim X2^™ insulin pump with Control-IQ^™ technology, and either insulin aspart or insulin lispro, for a target of 7 days for 12 consecutive wear periods. Each set removed before 7 days was adjudicated by an independent committee to assess the cause of early removal and to determine whether criteria for primary or key secondary endpoints were met.

Results:

There were 260 participants who inserted 3028 infusion sets. For the primary endpoint, the Kaplan–Meier 7-day survival estimate was 95% (95% CI 94% to 96%). For the key secondary endpoint, which expanded the reasons for a failed infusion set, the Kaplan–Meier 7-day survival estimate was 84% (95% CI 82% to 86%). For both endpoints, the P value was <0.001 compared with a prespecified survival rate of 75%. Time-in-range 70–180 mg/dL (TIR) increased through day 3 of set wear and then decreased through day 7, with the average TIR being 70.6% over the entire wear period across all infusion sets, including those that failed. An increase in total daily insulin was observed from day 4 through day 7.

Conclusions:

The 7-day survival of the SteadiSet infusion set was very high for the primary endpoint and at an acceptable level with respect to early removal for any reason related to the device.

Introduction

Insulin pump utilization by individuals requiring insulin replacement continues to grow globally, with a market value of $6.1 billion in 2024.¹ Despite significant uptake, the infusion set which delivers insulin from the pump to the body remains the “Achilles Heel” of continuous subcutaneous insulin infusion.² A 2020 survey of the T1D Exchange Online Community with 707 respondents on insulin pump or their caregivers revealed that nearly all (97%) had 1 or more infusion set failures per year and 41% had 1 or more failures per month. More frequent infusion set failure was associated with diabetes burnout and pump discontinuation.³

Most commercially available infusion sets per their labeling require replacement after 2–3 days of wear, and infusion set performance can vary over the lifespan of the infusion set.⁴ This creates considerable burden for users. Availability of infusion sets lasting longer without a degradation in performance would be of considerable benefit to people using an insulin pump. The Medtronic Extended^™ infusion set (EIS) was cleared by the Food and Drug Administration to use for up to 7 days in 2021, becoming the first set labeled for use for longer than 3 days.⁵

Driven by increased understanding of infusion set science, the SteadiSet^™ infusion set (Tandem Diabetes Care, San Diego, CA) was developed.^6,7 The initial design and development was performed by Capillary Biomedical, which Tandem Diabetes Care acquired in 2022. The infusion set, which received clearance by the Food and Drug Administration for 3-day wear in May 2025 but is not yet commercially available, contains a coil-reinforced soft polymer indwelling cannula with one distal and three proximal holes. This angled cannula is deployed into subcutaneous tissue via an integrated single-use insertion device (inserter) that is meant for one-handed easy insertion (Fig. 1).

FIG. 1.

SteadiSet Infusion Set. The SteadiSet features a single-use, integrated inserter designed for straightforward, one-handed insertion. The cannula is inserted at a 30-degree angle to support consistent subcutaneous placement and reduce the risk of flow interruptions.

We conducted a multicenter study to evaluate the function and safety of the SteadiSet during a wear period of up to 7 days in adults with type 1 diabetes (T1D) using a Tandem t:slim X2^™ insulin pump with Control-IQ^™ technology (Control-IQ) with either insulin aspart or insulin lispro. Both insulins were evaluated to ensure that there would not be insulin-specific differences in outcomes.

Methods

The study was conducted at 15 centers in the United States (clinicaltrials.gov # NCT06273124). The protocol and informed consent form were approved by a central institutional review board. Informed consent was obtained from each participant.

Major eligibility criteria included age 18 to <81 years old; T1D for at least 12 months; HbA1c <9%; and use of an insulin pump for at least 6 months, with use of a Tandem pump for at least 3 months and use of Control-IQ automated insulin delivery with either insulin aspart or insulin lispro for at least 1 month. Exclusions included current use of a noninsulin glucose lowering medication other than metformin, pregnancy, and severe hypoglycemia event or diabetes-related ketoacidosis (DKA) within the prior 6 months (see Supplementary Table S1 for a complete listing of the inclusion and exclusion criteria).

After consent was signed, eligibility was assessed. The screening visit and subsequent visits could be conducted virtually or in-clinic. Eligible participants were provided with a study Control-IQ pump and continued using their personal Dexcom G6 or G7 continuous glucose monitor (CGM) system (only Dexcom sensors were used to have consistency with the same company’s sensor used by all participants). They were trained on insertion of the SteadiSet infusion set and provided with a blood glucose meter (Ascensia CONTOUR NEXT) and a blood ketone meter (Abbott Precision Xtra).

The study protocol consisted of using a SteadiSet infusion set with 43-inch tubing for a target of 7 days (removal between 168 and 192 h) for 12 consecutive wear periods. Sets that failed in the first 8 h of use were replaced and did not count toward the target of 12 infusion sets worn. Participants were given specific decision-tree instructions on how to manage prolonged episodes of hyperglycemia (CGM ≥ 250 mg/dL for ≥1 h at least 3 h after the most recent meal/snack ingestion, in the absence of illness) and steps to take before replacing an infusion set, including contacting the site before removing an infusion set before 7 days. The CGM high threshold alert was to be set at 250 mg/dL, with a repeat setting of 60 min. Participants were advised to change the insulin cartridge every 48–72 h, and the pump’s “low insulin alert” was to be set at ≥20 units to reduce risk of running the pump on an empty cartridge.

Participants received a daily reminder text indicating the target date and time of infusion set removal. The daily text provided a link to a questionnaire for the participant to complete for each infusion set insertion and removal to provide the date/time and other information about each infusion set change.

Virtual or in-clinic visits occurred at 3 days, 2 weeks, and 6 weeks after initiation of the first infusion set, with a final visit within 7 days following the final wear period. Insulin, glucose, and ketone data were uploaded before or during each visit. Adverse events were solicited at each visit. At the infusion set initiation visit and final visit, a blood sample was obtained for central laboratory HbA1c determination. The 10-item System Usability Scale⁸ was completed at screening and at the 2-week and final visits.

Each set removed prior to 7 days was adjudicated by an independent clinical events committee to determine primary and key secondary endpoints for the analyses. The primary endpoint was the percentage of successfully inserted SteadiSet infusion sets (i.e., not removed within 8 h of insertion) that were not removed prior to 7 days (168 h) due to one of the following three criteria: (1) blood glucose value ≥250 mg/dL with ketones >1.0 mmol/L in the absence of illness or other physiological stress, (2) blood glucose value ≥250 mg/dL for at least 60 min duration at least 3 h following a snack or meal event that failed to respond to up to 2 adequate corrective boluses delivered by the pump with a fall of at least 50 mg/dL, or (3) investigator-advised infusion set replacement to assure participant’s safety.

The key secondary endpoint expanded the reasons for a failed infusion set to encompass, in addition to the primary endpoint criteria, any removal that could be ascribed to the investigational device’s performance, including: (1) removal during the first 8 h meeting the criteria otherwise specified for the primary endpoint, (2) evidence of infusion set site infection defined as requiring treatment or at the investigator’s judgment, (3) cannula dislodgment from the subcutaneous space (excluding accidental removal events), (4) nonresolvable pump occlusion alarm, (5) other device malfunction (e.g., inability to pierce skin, bending, or other malformation that might impact insulin infusion), (6) presence of pain of sufficient severity to prompt removal, and (7) any other event that could be ascribed to the investigational device performance that resulted in failure of insulin delivery. Infusion set removals prior to 168 h due to miscalculation of date and/or time, along with other removals that could not be ascribed to the investigational device performance, were not considered to be failures in the calculation of the key secondary endpoint.

Other endpoints included CGM metrics, HbA1c, insulin metrics, participant assessment of pain on infusion set insertion, and scores on the Usability, Satisfaction, and Preference Survey. Safety endpoints included DKA, severe hypoglycemia, and other serious adverse events.

Statistical methods

The sample size was planned to be a minimum of 240 participants completing the study (defined as at least 6 infusion set wears), with at least 120 using insulin lispro and 120 using insulin aspart. If the true success probability for each infusion set was 83% and the within-participant correlation 0.4, then a sample size of 120 patients per insulin type with 12 infusion set wears per subject would give 87% power for a two-sided test at alpha = 0.05 to reject the null hypothesis that the success rate was 75%.

Analyses were conducted overall and separately for insulin lispro and insulin aspart. Data from one infusion set for which both insulin lispro and insulin aspart were used during the same wear period were not included in the efficacy analyses. As the results were similar with both insulins, except for the primary analysis, only the combined results are reported. Infusion set survival estimates were calculated for the primary endpoint, key secondary endpoint, and for removal due to any reason using the Kaplan–Meier method. The data of infusion sets removed for a reason not meeting one of the endpoint-specific failure criteria were censored at the time of removal. A point estimate and two-sided 95% confidence interval (95% CI) for the 7-day survival rate were calculated with a bootstrap to account for the correlated data from having each participant wearing multiple infusion sets. The corresponding P values calculated for the primary endpoint and key secondary endpoint were based on rejecting the null hypothesis that the true survival probability was 75%.

Paired t tests were used to compute CIs and P values for change from baseline in CGM metrics, insulin metrics, HbA1c, and survey scores, as well as comparisons of metrics during different wear periods.

No multiplicity correction was made for the primary endpoint analysis or the key secondary endpoint analysis. For other analyses, the false discovery rate was controlled using the adaptive Benjamini–Hochberg procedure. SAS version 9.4 was used for the analyses.

Results

Between March 2024 and September 2024, 272 participants were screened for the trial, with 260 meeting eligibility criteria and initiating infusion set wear (128 using insulin lispro, 132 using insulin aspart). Mean age was 44 ± 17 years, T1D duration was 27 ± 14 years, 58% were female, 92% White, and 4% of Hispanic ethnicity. Mean HbA1c was 6.7 ± 0.7% and mean time-in-range 70–180 mg/dL (TIR) was 73.0 ± 12.1% at baseline (Table 1).

Table 1.

Participant Characteristics for Participants Who Initiated Use of the Study Infusion Set

	Overall N = 260
Age at enrollment (years)—mean ± SD	43.9 ± 17.3
Range	18–78
Diabetes duration (years)—mean ± SD	26.5 ± 14.3
Range	1–58
Sex (Female)—N (%)	150 (58%)
Race—N (%)
White	240 (92%)
Black/African American	4 (2%)
Asian	9 (3%)
More than one Race	6 (2%)
Hispanic ethnicity—N (%)	11 (4%)
Highest level of education—N (%)
<Bachelor’s Degree	68 (26%)
Bachelor’s Degree	125 (48%)
Advanced Degree	65 (25%)
Annual household income—N (%)
<$50K	21 (8%)
$50K to <$100K	47 (18%)
≥$100K	136 (52%)
Unknown/Does not wish to provide	56 (22%)
Health insurance—N (%)
Private	231 (89%)
Medicare	19 (7%)
Medicaid	6 (2%)
Other Government	3 (1%)
None	1 (<1%)
Baseline HbA1c (%)—mean ± SD	6.7 ± 0.7
Range	4.9–8.7
CGM metrics—mean ± SD
% Time in range 70–180 mg/dL	73.0% ± 12.1%
Total daily insulin (U)—mean ± SD	48.1 ± 21.2
Range	12.0–156.0
Body mass index (kg/m2)—mean ± SD	26.6 ± 3.6
Range	18.9–34.7

Among the 260 participants, 3028 infusion sets were inserted (Fig. 2), with 2997 (98.9%) being used for at least 8 h (referred to as “successfully inserted”). Twelve successfully inserted infusion sets (with the same insulin throughout the wear period) were worn by 236 (91%) of 260 participants and 6 or more infusion sets by 250 (96%). The lower or upper abdomen was the most common insertion location (64%) followed by side/flank 12%, buttocks 12%, thigh 7%, back 4%, and arm or other location 2%. Some type of tape or skin adhesive barrier wipe was used during 44% of infusion set wears. A G7 sensor was used by 135 (52%) participants, a G6 sensor by 94 (36%) participants, and G6 and G7 sensors at different times by 31 (12%) participants.

FIG. 2.

Flowchart of Enrollment and Infusion Set Use.

Infusion set survival

Among the 3027 inserted infusion sets in the efficacy analysis, 653 (21.6%) were removed before 168 h as assessed by the clinical events committee, with 137 (4.5%) meeting primary failure criteria and 488 (16.1%) removed for other reasons related (directly or indirectly) to the infusion set (Table 2).

Table 2.

Reasons for the Removal of Infusion Sets Before 168 H as Determined by the Clinical Events Committee

	N = 3027a Infusion sets inserted
Primary endpoint criteria met	137 (4.5%)
Hyperglycemia and failure to respond to boluses (without ketones)	92 (3.0%)
Hyperglycemia with ketones >1.0 mmol/L	24 (0.8%)
Investigator advised (without meeting hyperglycemia/ketone criteria)	21 (0.7%)
Key secondary endpoint criteria metb	460 (15.2%)
Primary endpoint criteria met (for successful insertions)	137 (4.5%)
Criteria for primary endpoint met in first 8 h after insertion	3 (0.1%)
Infusion set site infection requiring treatment or investigator’s judgment	13 (0.4%)
Cannula dislodgment from subcutaneous spacec	59 (1.9%)
Occurrence of a nonresolvable pump occlusion alarm	55 (1.8%)
Other device malfunctiond	101 (3.3%)
Pain of sufficient severity to prompt early removal of infusion set	78 (2.6%)
Removal for unresolvable hyperglycemia or ketosis, but not meeting primary or secondary endpoint criteria	214 (7.1%)
Other event related to infusion set resulting in failure of insulin delivery	50 (1.7%)
Set removed for other reasonsb	193 (6.4%)
Early removal due to miscalculation of time/date	48 (1.6%)
Cannula accidentally pulled out	96 (3.2%)
Participant dissatisfaction with glucose control, but not meeting primary or secondary endpoint critera	9 (0.3%)
Other infusion-set-related reason not related to impaired insulin deliverye	12 (0.4%)
Other reason related to infusion set or pump, and excluded from infusion set-related survival evaluationf	28 (0.9%)

^aDuring one infusion set wear, the participant switched from NovoLog to Humalog, so this wear was excluded from efficacy analyses.

^bWhen the primary endpoint criteria were not met, Clinical Events Committee members could specify multiple reasons for the failure of an infusion set, so the sum of the rows below is greater than the number to the right.

^cWith or without liquid leakage at the cannula insertion site, not including accidental removal, such as tubing caught on doorknob.

^dInability to pierce skin, bending, or other malformation that might impact insulin infusion, securement failure, or similar occurrence.

^eIncludes 5 removed for an MRI or CT scan; 4 removed for convenience in conjunction with a cartridge change when running low on insulin; 1 removed to avoid a set change during an upcoming trip; 1 removed for temporary reversion to personal insulin pump after study pump malfunction; and 1 accidentally removed by participant due to confusion about study protocol.

^fIncludes 25 removed at site request after participant inserted additional set beyond the maximum specified per protocol; and 3 removed at site request as a precaution after discovery of improper shipping procedures.

For the primary endpoint, as defined in the Methods, the Kaplan–Meier 7-day survival estimate was 95% (95% CI 94% to 96%, Fig. 3) for the 2997 successfully inserted sets, with a 7-day survival estimate of 95% (95% CI 94% to 97%) for the 1485 infusion sets with insulin lispro and 95% (95% CI 93% to 96%) for the 1512 infusion sets with insulin aspart. For the overall 7-day survival estimate, as well as for the insulin lispro and insulin aspart 7-day survival estimates, the P value was <0.001 compared with a prespecified survival rate of 75% as described in the Statistical Methods. The criteria met for the primary endpoint were hyperglycemia and failure to respond to correction boluses through the pump in 67%, hyperglycemia with ketone level >1.0 mmol/L in 18%, and investigator judgment without meeting hyperglycemia/ketone criteria in 15%. The 7-day survival rate varied from 96% for the 1911 lower or upper abdomen insertions to 89% for the 105 back insertions (Supplementary Table S2). Among the 236 participants with 12 successfully inserted infusion sets (worn ≥8 h), 155 (66%) had no primary endpoint failures.

FIG. 3.

SteadiSet Infusion Set Survival Curves Over 7 Days. The Kaplan–Meier method was used to estimate the survival probabilities over 7 days for the primary endpoint criteria, key secondary endpoint criteria, and removal for any reason related to the infusion set. The table below the figure shows the # at risk at the beginning of each one-day interval, with the number that failed during the interval in parenthesis.

For the key secondary endpoint as defined in the Methods, the Kaplan–Meier 7-day survival estimate was 84% (95% CI 82% to 86%, P < 0.001 compared with survival rate of 75%).

With respect to infusion sets removed before 7 days for any reason related to the infusion set, the 3-day survival rate was 96% and the 7-day survival rate 79% (Fig. 3).

Glycemic outcomes

Mean TIR was 70.6% over the entire wear period across all infusion sets, including those that failed before 7 days (Table 3), and was 71.9% for sets used for 7 days (Table 4). TIR significantly improved from baseline (72.9%) through day 3 (75.9%) and then gradually decreased through day 7 (64.6%) (Table 4 and Supplementary Table S3). Over the entire 7 days across all sets (including failures), there was a small decrease in TIR from baseline (difference = −2.4%, 95% CI −3.2% to −1.5%, P < 0.001). Other metrics reflective of hyperglycemia showed similar small changes from baseline except for the prolonged hyperglycemia event rate per week, which was unchanged from baseline (1.6 ± 1.8 versus 1.6 ± 1.5, 95% CI −0.1 to 0.2, P = 0.41) (Table 3). CGM metrics showed a similar pattern during daytime and nighttime (Supplementary Table S4).

Table 3.

Glycemic and Insulin Outcomes During the Entire Follow-Up Period for All Infusion Sets^a

	Baseline N = 260	Follow-up period N = 258a	Difference 95% confidence interval [P value]b
% Time in range 70–180 mg/dL	73.0% ± 12.1%	70.6% ± 11.5%	−2.4% (−3.2%, −1.5%) [<0.001]
% Time in tight range 70–140 mg/dL	47.6% ± 14.0%	43.3% ± 13.0%	−4.3% (−5.2%, −3.3%) [<0.001]
Mean glucose (mg/dL)	152.7 ± 20.2	157.1 ± 18.7	4.3 (3.0, 5.6) [<0.001]
% Time >180 mg/dL	25.2% ± 12.7%	27.8% ± 12.0%	2.6% (1.7%, 3.4%) [<0.001]
% Time >250 mg/dL	6.4% ± 6.5%	6.8% ± 5.8%	0.5% (0.001%, 0.9%) [0.0495]
% Time <70 mg/dL	1.9% ± 1.7%	1.7% ± 1.5%	−0.2% (−0.4%, −0.1%) [<0.001]
% Time <54 mg/dL	0.4% ± 0.5%	0.4% ± 0.5%	−0.02% (−0.1%, 0.03%) [0.42]
CGM prolonged hyperglycemic event rate per weekc	1.6 ± 1.8	1.6 ± 1.5	0.1 (−0.1, 0.2) [0.41]
CGM hypoglycemic event rate per weekd	0.8 ± 1.1	0.7 ± 1.0	−0.1 (−0.2, 0.05) [0.29]
Total daily insuline	48.1 ± 21.2	52.2 ± 22.1	4.0 (3.0, 5.0) [<0.001]

^aTwo participants did not have sufficient CGM data (≥168 h, as specified in the Statistical Analysis Plan) to include in the analysis; for the remaining 258 participants, data associated with all infusion sets worn (including those worn for <8 h) are analyzed.

^bP-value and confidence interval derived from a paired t test. Difference is follow-up minus baseline.

^cA CGM-measured prolonged hyperglycemic event is defined as at least two sensor values >250 mg/dL that are 120 or more min apart plus no intervening values ≤250 mg/dL; at least two sensor values ≤180 mg/dL that are 15 or more min apart with no intervening values >180 mg/dL are required to define the end of an event, at which point the study participant becomes eligible for a new event.

^dA CGM-measured hypoglycemic event <54 mg/dL is defined as at least two sensor values <54 mg/dL that are 15 or more min apart plus no intervening values ≥54 mg/dL; at least two sensor values ≥54 mg/dL that are 15 or more min apart with no intervening values <54 mg/dL are required to define the end of an event, at which point the study participant becomes eligible for a new event.

^eFour participants did not have sufficient insulin data (≥20 days, as specified in the Statistical Analysis Plan) to include in the analysis.

For total daily insulin, N = 260 at baseline and N = 256 during follow-up.

Table 4.

CGM-Measured Glycemic Outcomes by Day of Infusion Set Wear for Sets Worn ≥7 Days

	N	% Time in range 70–180 mg/dL	% Time in tight range 70–140 mg/dL	Mean glucose (mg/dL)	% Time >180 mg/dL	% Time >250 mg/dL	% Time <70 mg/dL	% Time <54 mg/dL	CGM hyperglycemic event rate per week a	CGM hypoglycemic event rate per week b
Baseline	257c	72.9% ± 12.1%	47.5% ± 14.0%	152.8 ± 20.3	25.2% ± 12.7%	6.4% ± 6.5%	1.9% ± 1.8%	0.4% ± 0.5%	1.6 ± 1.8	0.8 ± 1.1
Day 1	2304	73.6% ± 15.2%	50.3% ± 16.9%	148.2 ± 23.3	23.4% ± 15.2%	6.0% ± 8.3%	3.0% ± 4.0%	0.7% ± 1.8%	1.6 ± 3.5	1.3 ± 3.6
Day 2	2294	77.5% ± 14.5%	52.0% ± 18.1%	144.9 ± 21.7	20.3% ± 14.6%	3.9% ± 6.6%	2.3% ± 3.4%	0.5% ± 1.4%	0.9 ± 2.5	0.9 ± 3.0
Day 3	2294	75.9% ± 15.7%	49.9% ± 18.9%	148.0 ± 23.2	22.1% ± 16.0%	4.5% ± 7.2%	2.0% ± 3.3%	0.4% ± 1.2%	1.1 ± 2.8	0.8 ± 2.8
Day 4	2291	73.8% ± 16.8%	45.8% ± 19.9%	153.2 ± 25.2	24.7% ± 17.1%	5.3% ± 8.4%	1.5% ± 2.7%	0.3% ± 1.1%	1.3 ± 3.1	0.6 ± 2.4
Day 5	2304	70.4% ± 18.2%	41.4% ± 20.3%	159.1 ± 27.4	28.4% ± 18.6%	6.7% ± 10.0%	1.2% ± 2.4%	0.2% ± 1.0%	1.6 ± 3.3	0.5 ± 2.2
Day 6	2294	67.8% ± 20.1%	38.1% ± 21.0%	163.5 ± 29.4	31.3% ± 20.5%	7.7% ± 11.0%	0.9% ± 2.3%	0.2% ± 0.9%	1.8 ± 3.6	0.4 ± 2.2
Day 7	2302	64.6% ± 21.0%	34.3% ± 21.1%	167.9 ± 30.1	34.5% ± 21.4%	8.6% ± 11.3%	0.9% ± 2.2%	0.2% ± 0.9%	2.1 ± 3.9	0.4 ± 1.9
Days 1–7	2338	71.9% ± 12.9%	44.5% ± 14.9%	155.1 ± 20.6	26.5% ± 13.4%	6.1% ± 6.4%	1.7% ± 1.9%	0.4% ± 0.7%	1.5 ± 1.9	0.7 ± 1.5

^aA CGM-measured prolonged hyperglycemic event is defined as at least two sensor values >250 mg/dL that are 120 or more min apart plus no intervening values ≤250 mg/dL; at least two sensor values ≤180 mg/dL that are 15 or more min apart with no intervening values >180 mg/dL are required to define the end of an event, at which point the study participant becomes eligible for a new event.

^bA CGM-measured hypoglycemic event <54 mg/dL is defined as at least two sensor values <54 mg/dL that are 15 or more min apart plus no intervening values ≥54 mg/dL; at least two sensor values ≥54 mg/dL that are 15 or more min apart with no intervening values <54 mg/dL are required to define the end of an event, at which point the study participant becomes eligible for a new event.

^cThree participants had no set wears of duration for at least 7 days and were excluded from the baseline and in-study analysis. Baseline CGM data were collected from the 14 days prior to the screening visit.

Results appeared similar using Dexcom G6 or G7 sensors (data not shown). Mean HbA1c was 6.7 ± 0.7% at baseline and 6.8 ± 0.7% at study completion (P < 0.001).

Insulin outcomes

For all infusion sets, including those that failed before 7 days, total daily insulin increased from 48.1 ± 21.2 units at baseline to 52.2 ± 22.1 units during the study period (difference = 4.0 units, 95% CI 3.0 to 5.0, P < 0.001, Table 3). Total daily insulin was 51.8 ± 22.8 units over the first 7 days for the subset of infusion sets worn for at least 7 days (Supplementary Table S6). For those sets, basal insulin increased from 22.9 ± 10.5 units at baseline to 24.4 ± 11.3 units during the 7 days and bolus insulin increased from 25.4 ± 13.9 units to 27.4 ± 14.4 units. As seen in Table 5 and Supplementary Table S3, the daily increase started with day 4 and grew through day 7, with the mean total daily insulin being 48.3 units at both baseline and day 3 and then gradually increasing to 57.6 units on day 7 (P < 0.001 comparing days 1–3 versus days 4–7) while the basal–bolus ratio remained constant.

Table 5.

Insulin Outcomes by Day of Infusion Set Wear for Sets Worn ≥7 Days

Day	N	Total daily insulin (U)	Basal (U)	Bolus (U)	Bolus/basal
Baseline	257a	48.3 ± 21.2	22.9 ± 10.5	25.4 ± 13.9	1.2 ± 0.7
Day 1	2369	49.2 ± 23.5	22.6 ± 10.9	26.6 ± 16.1	1.3 ± 0.7
Day 2	2367	47.4 ± 22.2	22.5 ± 10.6	24.9 ± 15.0	1.2 ± 0.7
Day 3	2366	48.3 ± 23.5	22.9 ± 11.2	25.4 ± 15.5	1.2 ± 0.7
Day 4	2370	50.6 ± 24.2	23.9 ± 11.5	26.6 ± 16.1	1.2 ± 0.7
Day 5	2370	53.5 ± 26.0	25.1 ± 12.0	28.4 ± 17.7	1.2 ± 0.7
Day 6	2371	55.6 ± 26.6	26.2 ± 12.7	29.5 ± 17.8	1.2 ± 0.7
Day 7	2370	57.6 ± 26.6	27.3 ± 13.0	30.3 ± 17.6	1.2 ± 0.7
Days 1–7	2370	51.8 ± 22.8	24.4 ± 11.3	27.4 ± 14.4	1.2 ± 0.6

^aThree participants had no set wears of duration for at least 7 days and were excluded from the baseline and in-study analysis. Baseline data were collected from the 7 days before the screening visit.

Safety and survey outcomes

There were 55 adverse events reported for 43 participants (Supplementary Table S5). There was only one serious adverse event as follows: a severe hypoglycemia event unrelated to the study device. No cases of DKA or other serious adverse events occurred. There were 42 adverse events related to infusion site reactions from 42 infusion sets worn by 34 participants among the 3028 inserted infusion set wears (1.4% of infusion sets), with 12 of the events reported after 168 h of set wear.

Pain on insertion was generally none or minimal. On a scale of 0 to 100, the median pain score across all insertions was 5 (interquartile range 0–13), with only 91 (3.0%) associated with a pain score of 40 or higher. The pain scores did not vary across the 12 insertions (Supplementary Table S6).

On the System Usability Scale survey, scores were unchanged between baseline (79.9 ± 14.9) and the end of study (80.5 ± 17.6) (Supplementary Table S7).

Discussion

This study evaluated the SteadiSet infusion set in 260 adults with T1D using Control-IQ during 3027 infusion set wears. The 7-day infusion set survival rate was 95% for the primary endpoint criteria and 84% for the expanded key secondary endpoint criteria, substantially exceeding the prespecified minimum survival of 75%. The primary endpoint survival rate was ≥89% for all insertion set locations and 96% for abdomen insertions, which comprised most insertion sites. The frequency of adverse events was low, and there were no serious device-related adverse events. Few participants had more than minimal discomfort on infusion set insertion and most reported little or no pain. System Usability Scale scores indicated no difference between participants’ perception of their prestudy infusion set and the study infusion set in terms of overall usability.

Glycemic metrics reflective of hyperglycemia, such as TIR and mean glucose, improved during the first 3 days of set wear and then gradually decreased through 7 days, despite an increase in total daily insulin from day 4 through day 7. The initial improvement after day 1 has been reported previously and is consistent with improvement seen after the infusion site has time to recover from the trauma of catheter insertion.⁴ Previously published data show that insulin pharmacokinetics are accelerated through 7 days of wear, but with decreased insulin exposure.^9,10 This is likely attributable to increased eosinophilic inflammation, vessel density, and fibrosis at biopsied infusion sites,¹¹ leading to progressive increases in subcutaneous resistance to fluid flow.¹² Even with the increase of glycemic levels near the end of the infusion set wear period, mean TIR over the study period even when including failed infusion sets decreased by only 2.4%, which is less than the international consensus level of 3% for a meaningful change in TIR,¹³ and was still >70% over the entire 7 days. Thus, this small change seems clinically inconsequential with respect to risk of chronic diabetic complications. In addition, HbA1c was virtually unchanged from baseline to the end of study (6.7% versus 6.8%), supporting the minimal effect on glucose levels over-all. Total daily insulin dose increased by about 4 units over the 7-day period, with the increase observed beginning with day 4 of infusion set wear. The basal–bolus ratio did not change, suggesting that the Control-IQ automation was able to adjust insulin as needed.

These findings are comparable to the earlier evaluations of the SteadiSet. A pharmacokinetics and pharmacodynamics study of an earlier version of this infusion set, which was similar but not identical to the SteadiSet, demonstrated more rapid insulin pharmacokinetics compared with a conventional Teflon-angled insulin infusion set, but there was a 24% decrease in the area under the insulin concentration curve comparing day 0 to day 7.¹⁰ In a feasibility study with 20 adults with T1D who used the prototype infusion set for two 7-day test periods, the infusion set was worn for an average of 6.6 days, with 88% worn for 7 days.⁷ Five infusion sets failed before 7 days, 4 due to unresolvable hyperglycemia or ketosis and 1 due to infection. Glucose levels and total daily insulin dose increased, particularly over days 4–7.

For the current trial, when conservatively considering all infusion sets removed before 168 h due to any infusion-set related reason as failures (including accidental pullouts and time miscalculations), the 7-day survival rate was 79% for the 2999 analyzed infusion sets. This survival rate is similar to the 7-day survival rate of 78% reported in the Medtronic EIS pivotal trial.⁵ However, the Medtronic survival rate did not include 8% of infusion sets removed early either inadvertently or due to insulin depletion or pump replacement. Excluding such failures (all “other reasons” in Table 2 other than participant dissatisfaction) in the current study would increase the 7-day survival rate for the SteadiSet to 84% compared with 78% with the Medtronic EIS. This difference could be due to the more frequent use of supplemental adhesives with the SteadiSet than the Medtronic EIS since adhesive selection may directly affect the duration of infusion set wear.¹⁴ In the Medtronic trial, daily mean glucose levels followed a similar pattern of decreasing initially and then increasing during days 5–7. However, there was little change in total daily insulin over the 7 days in contrast to the small increase observed in the current study with the SteadiSet. It is difficult to compare both the relative frequency of removal reasons and insulin use across studies due to differences in the protocols with respect to the categorization of failures—self-reported by participants with later review by study team in the Medtronic trial versus per-protocol assessment by the study team before set removal plus adjudication by an independent endpoint committee in the current trial. As the studies also reported different subcategories for secondary and other failure reasons, and in some cases included more than one reason for removal (Table 2), it is not possible to compare the reasons for removal directly.

The strengths of the study included a large sample size recruited at multiple clinical sites, a structured protocol for management of hyperglycemia potentially due to infusion set failure, including specific procedures for participants to follow before an infusion set was removed, and standardization of definitions of infusion set failure with an independent committee adjudicating all instances of infusion set removal before 7 days. The main limitation with respect to interpretation of the results is that the study cohort included only adults who were experienced pump users already using Control-IQ and who were 92% White with a mean HbA1c of 6.6% and mean body mass index of 26.6 kg/m². Results might differ in pump-naive individuals, in youth, or in adults with other characteristics. In addition, it is possible that results in a real-world setting could be more variable than the results in the controlled environment of a study.

In summary, the primary endpoint of the study was met for the 7-day survival of the SteadiSet infusion set for both insulin lispro and insulin aspart without safety concerns. Glycemic control during the first 3 days of SteadiSet infusion set wear was similar or better than the commercial 3-day wear infusion sets used during the baseline period. While there was a slight deterioration of glycemic control and increase in insulin needs during days 4–7 of SteadiSet wear, participants maintained TIR over 70% for the 7-day wear period. The results of the study support the 7-day SteadiSet infusion set being a viable option for people using an insulin pump once it receives regulatory clearance.

Supplementary Material

Funding Information

Study support was provided by Tandem Diabetes Care.