ABSTRACT
Aims
To conduct indirect treatment comparisons (ITCs) of insulin efsitora alfa (efsitora) versus insulin icodec (icodec) in insulin‐experienced individuals with Type 2 Diabetes Mellitus.
Methods
Efficacy, safety and patient reported outcomes were compared via matching‐adjusted ITCs using results from randomised controlled trials: QWINT‐3 versus ONWARDS 2 (basal‐insulin experienced individuals) and QWINT‐4 versus ONWARDS 4 (basal‐bolus insulin‐experienced individuals).
Results
In basal insulin‐experienced participants, efsitora was comparable to icodec for change from baseline (CfB) in HbA1c (%) (efficacy estimand mean difference [MD] [95% CI]: 0.137 [−0.058, 0.333]), proportion of participants (PoP) discontinuing treatment due to adverse event (risk difference: −0.0002 [−0.0223, 0.0220]) and CfB in Diabetes Treatment Satisfaction Questionnaire‐Change (DTSQc) (MD: 1.01 [−0.243, 2.27]). Efsitora exhibited significantly reduced weight gain (MD: −1.52 kg [−2.60, −0.447]) and had lower event rates of Level 2 or 3 hypoglycaemia (rate ratio: 0.436 [0.241, 0.786]) versus icodec, despite no significant difference in PoP with Level 2 or 3 hypoglycaemia (odds ratio [OR]: 0.581 [0.294, 1.14]). In basal‐bolus insulin‐experienced participants, efsitora was comparable to icodec: CfB in HbA1c (%) (MD: −0.0447 [−0.252, 0.163]); CfB in body weight (kg) (MD: −0.316 [−1.45, 0.817]); PoP with Level 2 or 3 hypoglycaemia (OR: 1.10 [0.698, 1.73]); event rate of Level 2 or 3 hypoglycaemia (rate ratio [95% CI]: 1.03 [0.759, 1.39]).
Conclusions
These indirect comparisons suggest that efsitora had lower event rates of hypoglycaemia and less weight gain compared to icodec in basal insulin‐experienced participants. Efsitora was comparable to icodec for all other endpoints.
Keywords: basal insulin, glycaemic control, hypoglycaemia, type 2 diabetes, weight control
1. Introduction
Type 2 diabetes mellitus (T2D) is a chronic condition characterised by hyperglycaemia, which increases the risk of a range of complications. These can place a substantial burden on individuals with T2D and their quality of life, and can also result in significant economic impact [1, 2]. T2D results from a combination of resistance to insulin action, progressive pancreatic beta cell failure and insufficient insulin secretory response [3]. As a progressive disorder, T2D is characterised by the need to intensify therapy to achieve and maintain optimal glycaemic control [4]. In addition to lifestyle modification, pharmacotherapies recommended for T2D include metformin, dipeptidyl peptidase‐4 inhibitors (DPP‐4i), sodium‐glucose co‐transporter 2 inhibitors (SGLT‐2i), glucagon‐like peptide‐1 receptor agonists (GLP‐1 RA), dual glucose‐dependent insulinotropic polypeptide (GIP)/GLP‐1 RA, sulphonylureas (SU), thiazolidinediones (TZD) and insulin‐based treatments [5].
Insulin regulates glucose levels by facilitating glucose uptake into cells, so in T2D, more insulin (whether endogenous or exogenous) is required to compensate for the body's reduced sensitivity to insulin [6]. Insulin therapies are administered subcutaneously and can be broadly categorised as basal insulins (longer‐term effects, usually designed to be taken daily) or bolus insulins (shorter‐acting, designed to rapidly prevent or correct hyperglycaemia) [7]. Participants with different insulin backgrounds, such as those treated with basal and/or bolus insulins, exhibit distinct physiological profiles, therapeutic needs, and behavioural responses [8]. As such, clearly defining and stratifying these populations allows for more precise assessment of treatment efficacy, safety, and real‐world applicability; investigational studies in T2D therefore often employ insulin experience as a key stratification factor.
Most current basal insulins require daily administration, which represents a barrier to adherence and persistence to therapy in individuals with T2D [9]. Evidence has shown that reducing dosing frequency can improve treatment adherence and persistence in T2D [10, 11, 12]. PRO instruments (treatment related impact measure for diabetes [TRIM‐D]; Diabetes Treatment Satisfaction Questionnaire‐Change [DTSQc]; Simplicity of Diabetes Treatment Questionnaire [SIM‐Q]) have demonstrated that once‐weekly insulin regimens were associated with greater treatment compliance and satisfaction, reduced treatment burden and improved overall health state compared to daily regimens [12, 13, 14]. As such, once‐weekly insulin regimens provide a more convenient alternative to daily regimens, providing similar glycaemic reduction, with the potential to improve adherence, persistence, and ultimately, treatment outcomes for individuals with T2D [9, 15].
Insulin efsitora alfa (efsitora) is an innovative basal insulin designed to be administered once weekly [16]. Efsitora is a large molecule (64.1 kDa) with reduced affinity for human insulin receptor binding, comprised of a single‐chain insulin receptor analogue, an inter‐domain linker, and the fragment crystallisable (Fc) domain of antibody IgG2 [16]. These features reduce receptor‐mediated and renal clearance, permitting recycling and creation of a systemic reservoir, while retaining similar functional properties to human insulin [16]. This supports once‐weekly dosing through a flat pharmacokinetic profile with an extended half‐life of approximately 17 days, a peak bloodstream level (t max) at 4 days, and associated sustained glucose reduction [16].
Efsitora has been evaluated versus daily basal insulins (insulin degludec [degludec] and insulin glargine [glargine]) in both insulin‐naïve and insulin‐experienced T2D populations in a series of head‐to‐head Phase 3, treat‐to‐target clinical trials: the QWINT programme [17]. These trials demonstrated non‐inferiority of efsitora to comparators in adults with T2D for multiple outcomes, including change from baseline in HbA1c and rate of hypoglycaemic episodes, including within subgroup analyses [18, 19, 20, 21].
Insulin icodec (icodec) is another innovative basal insulin designed to be administered once weekly, with a half‐life of approximately 8 days and a t max of 16 h [22]. Icodec has been evaluated in the ONWARDS programme [23, 24, 25, 26], a series of head‐to‐head Phase 3, treat‐to‐target clinical trials comparing with daily basal insulins, degludec and glargine. Icodec demonstrated non‐inferiority to daily basal insulins in change from baseline (CfB) in HbA1c in all trials and superiority in ONWARDS 1–3 and 5. In ONWARDS 1, icodec also demonstrated superiority to glargine in percentage time in range (TIR) [23, 24, 25, 26].
No head‐to‐head trials directly comparing efsitora with icodec are available. Consequently, an indirect treatment comparison (ITC) comparing efsitora with icodec on key safety and efficacy outcomes in insulin‐experienced participants from the QWINT and ONWARDS programmes was conducted.
Objective: To estimate the efficacy and safety of efsitora in comparison with icodec in two populations of adults with T2D in an anchored ITC: those who were previously treated with basal insulin or basal‐bolus insulin regimens.
2. Methods
2.1. Data Sources
The QWINT and ONWARDS trials were selected as pivotal evidence across different insulin‐experienced populations for efsitora and icodec, respectively. Two QWINT‐ONWARDS pairs, matched on insulin experience, were identified for comparison in respective ITCs (Table S2): QWINT‐3 versus ONWARDS 2 (basal insulin‐experienced) and QWINT‐4 versus ONWARDS 4 (basal‐bolus insulin‐experienced) [19, 20, 24, 26]. Individual participant data (IPD) were available for the QWINT trials and aggregate data for the ONWARDS trials.
2.2. Feasibility Assessment
ITCs assume similarity and transitivity, which require similarity of study designs, including similarly defined outcomes (and estimands) at comparable timepoints, and comparability of participant populations in terms of treatment effect modifiers (TEMs). Population adjustment methodology should be applied to resolve imbalances in TEMs [27, 28, 29]. Therefore, a feasibility assessment was conducted to identify TEMs, and assess participant and trial characteristics, endpoint definitions and reporting through elicitation of expert opinion, with reference to published literature and trial data, as recommended by current guidance [30, 31].
2.3. Population Refinement and Safety Endpoint Re‐Analysis
For some characteristics, ONWARDS eligibility criteria were applied to the QWINT IPD to generate refined populations more aligned with the respective ONWARDS trials (Table S1). The additional ONWARDS 2 exclusion criteria applied to the QWINT‐3 population were: HbA1c ≥ 6.5% and < 7.0%; T2D onset less than 180 days prior to screening; body mass index (BMI) > 40 mg/m2; estimated glomerular filtration rate (eGFR) ≥ 20 and eGFR < 30; bilirubin > 1.5 × ULN at screening; myocardial infarction (MI), stroke, hospitalisation for transient ischemic attack (TIA) within 180 days prior to the day of screening; systolic blood pressure (BP) ≥ 180 or diastolic BP ≥ 110 at screening. The additional ONWARDS 4 exclusion criteria applied to the QWINT‐4 population were: T2D onset less than 180 days prior to screening; BMI > 40 mg/m2; total bilirubin > 1.5 × ULN at screening; MI, stroke, hospitalisation for TIA within 180 days prior to screening; systolic BP ≥ 180 or diastolic BP ≥ 110 at screening.
For safety endpoints, QWINT outcomes were reanalysed at Week 31 using the refined IPD, to match ONWARDS reporting.
2.4. Analyses Conducted
Specification of analyses was driven by the findings of the feasibility assessment (see Section 3). The base case analyses were matching‐adjusted indirect comparisons (MAICs) for both, basal insulin‐ and basal‐bolus insulin‐experienced populations, adjusting for baseline HbA1c in both cases. MAIC was selected as the population adjustment methodology to deal with between‐study variation, aligning with current guidance [29]. The common comparators connecting the pairwise networks were insulin degludec and insulin glargine, in the basal insulin‐ and basal‐bolus insulin‐experienced comparisons, respectively (Figures S1 and S2).
Several sensitivity analyses were conducted. An unadjusted Bucher ITC was conducted in parallel to MAICs for all analyses, using the refined QWINT populations but without weighting [32]. Although not identified as TEMs, there were substantial imbalances in race/ethnicity between trials in the basal (proportion of Asian and White participants; Table 2) and basal‐bolus populations (proportion of Latino/Hispanic participants; Table 3). The differences observed in the basal‐bolus population warranted inclusion in sensitivity analysis; MAICs adjusting for race/ethnicity only, and for race/ethnicity and baseline HbA1c were therefore also conducted for the basal‐bolus population. The differences in the basal population were substantially smaller than those observed in the basal‐bolus population, and additional subgroup analysis from ONWARDS 2 and QWINT‐3 (Table S3) showed no statistically significant interaction between treatment and ethnicity for change in HbA1c [33]. On the basis of these data, no adjustment for race/ethnicity in the basal insulin‐experienced population was made.
TABLE 2.
Participant characteristics of studies included in basal insulin comparison.
| Baseline characteristics | Treatment arm | QWINT‐3 (prior to refinement and weighting) | QWINT‐3 (after refinement, prior to weighting) | QWINT‐3 (after refinement and weighting) | ONWARDS 2 |
|---|---|---|---|---|---|
| Age (years), mean (SD) | Intervention | 60.64 (10.17) | 60.1 (10.2) | 60 (10.2) | 62.3 (9.8) |
| Comparator | 60.81 (9.97) | 60.9 (9.8) | 60.7 (9.6) | 62.6 (8.4) | |
| Sex–F, % | Intervention | 42.6 | 43.8 | 44.1 | 38 |
| Comparator | 45.92 | 45.3 | 46.2 | 47 | |
| Weight (kg), mean (SD) | Intervention | 84.96 (18.8) | 82.46 (16.85) | 82.8 (16.6) | 83.7 (18.4) |
| Comparator | 85.22 (17.88) | 83.08 (16.00) | 83.2 (15.8) | 81.5 (17.1) | |
| BMI (kg/m2), mean (SD) | Intervention | 30.22 (5.71) | 29.27 (4.84) | 29.4 (4.83) | 29.5 (5.2) |
| Comparator | 30.61 (5.36) | 29.85 (4.54) | 29.9 (4.51) | 29.2 (4.9) | |
| Baseline HbA1c (%), mean (SD) | Intervention | 7.8 (0.87) | 7.97 (0.82) | 8.17 (0.858) | 8.17 (0.77) |
| Comparator | 7.8 (0.83) | 7.97 (0.78) | 8.1 (0.796) | 8.1 (0.77) | |
| Duration of Diabetes (years), mean (SD) | Intervention | 15.43 (8.13) | 15.27 (7.84) | 15.3 (7.74) | 16.5 (8.4) |
| Comparator | 15.14 (7.75) | 15.53 (7.82) | 15.4 (7.75) | 16.9 (7.9) | |
| FSG (mg/dL), mean (SD) | Intervention | 135.47 (41.89) | 138.95 (43.04) | 144 (46.2) | 152.2 (47.5) |
| Comparator | 137.01 (45.23) | 139.97 (45.03) | 143 (47.3) | 150.7 (40.9) | |
| Race, % | |||||
| American Indian or Alaska Native | Intervention | 0.31 | 0.2 | NR | < 1% |
| Comparator | 0 | 0 | NR | 0 | |
| Asian | Intervention | 27.48 | 30.1 | 29.1 | 33 |
| Comparator | 27.79 | 27.5 | 26.4 | 42 | |
| Black or African American | Intervention | 4.89 | 3.5 | 4.02 | 4 |
| Comparator | 6.04 | 5.7 | 5.93 | 5 | |
| White | Intervention | 66.9 | 65.6 | 66.1 | 61 |
| Comparator | 65.9 | 66.8 | 67.6 | 52 | |
| Multiple | Intervention | 0.31 | 0.4 | NR | NR |
| Comparator | 0.3 | 0 | NR | NR | |
| Native Hawaiian or other Pacific Islander | Intervention | 0.15 | 0.2 | NR | NR |
| Comparator | 0 | 0 | NR | NR | |
| Ethnicity, % | |||||
| Hispanic or Latino | Intervention | 27.3 | 26.6 | NR | 6.1 |
| Comparator | 27.1 | 29.1 | NR | 6.1 | |
| Non‐Hispanic or Latino | Intervention | 72.7 | 73.4 | NR | 93.9 |
| Comparator | 72.9 | 70.9 | NR | 93.9 | |
| Concomitant therapies at baseline, % | |||||
| GLP1‐RA | Intervention | 28.9 | 29.2 | 28.9 | 26 |
| Comparator | 29.3 | 25.7 | 24.4 | 26 | |
| SU | Intervention | 0 | 0 | 0 | 23 |
| Comparator | 0 | 0 | 0 | 22 | |
| SGLT2i | Intervention | 40.5 | 39.6 | 38.4 | 34 |
| Comparator | 38.4 | 38.5 | 38.7 | 32 | |
Abbreviations: BMI, body mass index; FSG, fasting serum glucose; GLP1‐RA, glucagon‐like peptide‐1 receptor agonist; HbA1c, glycated haemoglobin; NR, not reported; SD, standard deviation; SGLT2i, sodium‐glucose co‐transporter inhibitor; SU, sulphonylureas.
TABLE 3.
Participant characteristics of studies included in basal‐bolus insulin comparison.
| Baseline characteristics | Treatment Arm | QWINT‐4 (prior to refinement and weighting) | QWINT‐4 (after refinement, prior to weighting) | QWINT‐4 (after refinement and weighting) | ONWARDS 4 |
|---|---|---|---|---|---|
| Age (years), mean (SD) | Intervention | 58.3 (10.47) | 58.4 (10.5) | 58.2 (10.5) | 59.7 (10.1) |
| Comparator | 59.43 (10.52) | 59.2 (10.5) | 59 (10.5) | 59.9 (9.9) | |
| Sex–F, % | Intervention | 52.88 | 54.7 | 54.6 | 47 |
| Comparator | 48.22 | 47.4 | 48.6 | 48 | |
| Weight (kg), mean (SD) | Intervention | 87.79 (19.93) | 84.67 (16.92) | 84.6 (16.9) | 85.5 (17.6) |
| Comparator | 88.38 (19.72) | 85.95 (18.17) | 85.5 (18.1) | 83.1 (17.3) | |
| BMI (kg/m2), mean (SD) | Intervention | 31.85 (5.48) | 30.91 (4.51) | 30.9 (4.51) | 30.5 (5) |
| Comparator | 31.84 (5.48) | 30.95 (4.74) | 30.9 (4.77) | 30 (5) | |
| Baseline HbA1c (%), mean (SD) | Intervention | 8.17 (0.83) | 8.20 (0.82) | 8.29 (0.853) | 8.29 (0.86) |
| Comparator | 8.18 (0.79) | 8.19 (0.82) | 8.31 (0.816) | 8.31 (0.9) | |
| Duration of diabetes (years), mean (SD) | Intervention | 16.55 (8.84) | 16.29 (8.64) | 16.2 (8.61) | 18 (9.1) |
| Comparator | 16.94 (9.02) | 16.87 (8.88) | 16.8 (8.83) | 16.3 (7.7) | |
| FSG (mg/dL), mean (SD) | Intervention | 148.28 (49.41) | 147.45 (49.83) | 149 (50.7) | 167 (54) |
| Comparator | 152.92 (54.7) | 149.84 (52.41) | 152 (53.2) | 173 (63) | |
| Race, % | |||||
| American Indian or Alaska Native | Intervention | 12.6 | 13.0 | 13 | 0 |
| Comparator | 12.3 | 13.2 | 13.3 | 0 | |
| Asian | Intervention | 14.79 | 16.3 | 16.6 | 32.65 |
| Comparator | 13.97 | 15.3 | 16.4 | 31.96 | |
| Black or African American | Intervention | 5.48 | 4.8 | 4.61 | 4.47 |
| Comparator | 3.01 | 2.4 | 2.51 | 2.75 | |
| White | Intervention | 67.12 | 65.9 | 65.8 | 62.89 |
| Comparator | 70.68 | 69.9 | 67.8 | 64.26 | |
| Ethnicity, % | |||||
| Hispanic | Intervention | 55.1 | 55.3 | 56 | 17.87 |
| Comparator | 55.6 | 54.4 | 54.9 | 18.21 | |
| Non‐Hispanic | Intervention | 44.7 | 44.4 | 43.7 | 82.13 |
| Comparator | 44.1 | 45.3 | 44.8 | 81.44 | |
| Concomitant therapies at baseline, % | |||||
| GLP1‐RA | Intervention | 17.3 | 15.4 | 15.1 | 13 |
| Comparator | 18.6 | 17.1 | 16.7 | 12 | |
| SU | Intervention | 0 | 0 | 0 | 10 |
| Comparator | 0 | 0 | 0 | 5 | |
| SGLT2i | Intervention | 31.5 | 33.5 | 33.2 | 28 |
| Comparator | 26.6 | 30.0 | 29.5 | 30 | |
Abbreviations: BMI, body mass index; FSG, fasting serum glucose; GLP1‐RA, glucagon‐like peptide‐1 receptor agonist; HbA1c, glycated haemoglobin; NR, not reported; SD, standard deviation; SGLT2i, sodium‐glucose co‐transporter inhibitor; SU, sulphonylureas.
Subgroup analyses were also specified for both insulin experience populations. The subgroups of interest comprised concomitant medications, SGLT‐2i use and GLP‐1 RA use for both populations. Subgroup analyses were conducted where outcome data were available.
Efficacy endpoints analysed were CfB in HbA1c, proportion of participants (PoP) reaching HbA1c ≤ 6.5%, PoP reaching HbA1c < 7% and CfB in percent of TIR (CGM 70–180 mg/dL) (Table 1). Safety endpoints were PoP with and event rates of Level 2 or 3 hypoglycaemia, CfB in body weight and PoP with treatment discontinuations due to adverse events (AEs). Composite endpoints were PoP with HbA1c ≤ 6.5% without Level 2 or 3 hypoglycaemia and PoP with < 7% without Level 2 or 3 hypoglycaemia. The PRO endpoint, CfB in DTSQc score was available only for the basal insulin‐experienced population. Safety endpoints were analysed at Week 31 and all other endpoints at Week 26. Driven by availability for ONWARDS trials, the principal estimand for analysis was the treatment regimen estimand, which included all study‐eligible participants, regardless of treatment discontinuation or initiation of rescue medication. The efficacy estimand (study‐eligible participants who adhered to randomised treatment without intercurrent events during the study treatment period) was also available for CfB in HbA1c for the basal insulin‐experienced population.
TABLE 1.
Efficacy, PRO and safety endpoints of interest.
| Endpoint type | Outcome of interest | Estimand | Timepoint of interest | Population |
|---|---|---|---|---|
| Efficacy | CFB in HbA1c (%) | Treatment‐regimen | Week 26 | Both |
| PoP with HbA1c (%) < 7% | Week 26 | |||
| CFB in FSG (mg/dl) | Week 26 | |||
| CFB in body weight (kg) | Week 26 | |||
| PoP with HbA1c < 7% at Week 26 without Level 2 or 3 hypoglycaemia within the 12 weeks prior to Week 26 | Week 26 (Weeks 14–26) | |||
| PoP with HbA1c ≤ 6.5% | Week 26 | Basal insulin‐experienced only | ||
| PoP with HbA1c ≤ 6.5% at Week 26 without Level 2 or Level 3 hypoglycaemia in the prior 12 weeks | Week 26 (Weeks 14–26) | |||
| TIR (%) | Week 26 | Both | ||
| CFB in HbA1c (%) | Efficacy | Week 26 | Basal insulin‐experienced only | |
| PRO | CFB in DTSQc | NA | Week 26 | Basal insulin‐experienced only |
| Safety | PoP with, and event rate of, Level 2 hypoglycaemic events | NA | Week 31 | Both |
| PoP with, and event rate of, severe (Level 3) hypoglycaemic events | Week 31 | Both | ||
| PoP with, and event rate of, Levels 2 and 3 hypoglycaemic events | Week 31 | Both | ||
| PoP with, and event rate of, nocturnal Level 2 and Level 3 hypoglycaemic events | Week 31 | Both | ||
| PoP discontinuing treatment due to AEs | Week 26 | Both | ||
| PoP with injection site reactions | Week 26 | Both |
Abbreviations: AE, adverse event; CFB, change from baseline; DTSQc, Diabetes Treatment Satisfaction Questionnaire‐Change; FSG, fasting serum glucose; HbA1c, glycated haemoglobin; PoP, proportion of participants; PRO, patient reported outcome; TIR, time in range.
2.5. Statistical Methodology
MAICs were conducted according to the methodology of Signorovitch et al. [34] and applied to match the refined population IPD for each QWINT trial to the corresponding ONWARDS trial population [29]. Briefly, participants in each refined QWINT population were weighted through a logistic model to create a pseudo‐population closely resembling the aggregated baseline characteristics reported in the corresponding ONWARDS trial. Adjusted aggregate outcomes from each QWINT trial were then used in the indirect comparison versus the respective ONWARDS trial. The validity of each MAIC was assessed through evaluation of weight distributions for skew and extreme values. For MAIC results to be considered reliable, participant weights should follow an approximately Normal distribution centred close to one. MAIC validity was also assessed based on effective sample size (ESS), a measure of the size of the weighted population after population adjustment. Minimal reductions in ESS compared with the original sample size indicate that a large proportion of the information in the sample is retained after population adjustment and hence, these comparisons are considered more reliable.
Bucher ITCs were conducted using established methodology [32].
Analyses were conducted in R 4.3.2, using meta and metafor packages for Bucher ITCs [35, 36], and boot, survey, twang, caret, lpSolve, and MASS packages for MAIC [37, 38, 39, 40, 41, 42, 43].
3. Results
3.1. Feasibility Assessment
Definitions for analysed outcomes were considered to be comparable between trials. For each pair of trials, trial designs and participant characteristics identified as TEMs, including age, sex, BMI, fasting serum glucose (FSG), eGFR and baseline comorbidities, were also considered similar (Tables 2 and 3). However, in both the basal insulin‐experienced population and basal‐bolus insulin‐experienced populations, baseline HbA1c was identified as a TEM with sufficient imbalance to require adjustment. Although not identified as a TEM, substantial between‐trial differences in race/ethnicity in the basal‐bolus insulin‐experienced population was identified and targeted for adjustment in sensitivity analyses. Differences in bolus insulin between QWINT‐4 (insulin lispro) and ONWARDS 4 (insulin aspart) were not considered to be clinically meaningful.
Differences in insulin titration algorithm between each of the paired trials were identified, resulting in differences in insulin dosages at Week 26 (Tables S4 and S5). In the basal insulin‐experienced population, efsitora and icodec had mean weekly doses of 3.71 U/kg/week and 3.3 U/kg/week, respectively, while degludec U100 had mean doses of 0.58 U/kg/day (equivalent to 4.06 U/kg/week) and 0.43 U/kg/day (3.01 U/kg/week) from QWINT‐3 and ONWARDS 2, respectively. In the basal‐bolus insulin‐experienced population, the mean basal insulin doses were 4.3 U/kg/week for efsitora and 3.6 U/kg/week for icodec, and 0.67 U/kg/day (4.69 U/kg/week) and 0.47 U/kg/day (3.29 U/kg/week) for glargine U100 in QWINT‐4 and ONWARDS 4, respectively. The mean bolus insulin doses were 2.1 U/kg/week for the efsitora arm versus 2.3 U/kg/week for the icodec arm, with glargine U100 treatment arms at 0.38 U/kg/day (2.66 U/kg/week) and 0.44 U/kg/day (3.08 U/kg/week) from QWINT‐4 and ONWARDS 4, respectively. However, these differences could not be adjusted in an ITC due to unavailability of IPD for ONWARDS trials.
3.2. ITC Results
3.2.1. Basal Insulin‐Experienced Populations
Results of ITCs are presented in Figure 1 and in Table S7. The refinement and weighting procedures resolved differences between participant populations (Table 2), diagnostics were considered to be appropriate for reliable interpretation of results, and reweighting resulted in minimal reductions in ESS from the unweighted sample for the base case and subgroup analyses (treatment regimen estimand: reduction of 4.3% for efsitora and 3.8% for degludec; efficacy estimand: reduction of 4.6% for efsitora and 3.9% for degludec; Table S6).
FIGURE 1.
Forest plot of base case results for basal‐insulin and basal‐bolus insulin‐experienced participants for (A) CfB in HbA1C (%); (B) CfB in body weight (kg); (C) CfB in DSTQc; (D) PoP with Level 2 or 3 hypoglycaemia; (E) Event rate of Level 2 or 3 hypoglycaemia. CfB, change from baseline; CI, confidence interval; DTSQc, Diabetes Treatment Satisfaction Questionnaire‐Change; HbA1c, glycated haemoglobin; kg, kilogram; PoP, proportion of participants.
Efsitora was comparable to icodec for efficacy outcomes, with CfB in HbA1c (%) (treatment regimen estimand mean difference [MD] [95% CI]: 0.166 [−0.0361, 0.369]; efficacy estimand MD [95% CI]: 0.137 [−0.058, 0.333]) and comparable CfB in percent of TIR (CGM 70–180 mg/dL) (MD [95% CI]: −3.41 [−7.89, 1.08]). The PoP reaching HbA1c ≤ 6.5% was also comparable between efsitora and icodec (odds ratio [OR] [95% CI] 0.618 [0.311, 1.23]).
CfB in body weight (kg) (MD [95% CI] –1.52 [−2.60, −0.447]) was significantly lower for those treated with efsitora compared to those receiving icodec; efsitora participants experienced a smaller increase in body weight. The PoP with Level 2 or 3 hypoglycaemia was comparable between efsitora and icodec (OR [95% CI]: 0.581 [0.294, 1.14]) whilst the event rates of Level 2 or 3 hypoglycaemia were significantly lower for those treated with efsitora compared to those receiving icodec (rate ratio RR [95% CI]: 0.436 [0.241, 0.786]); event rates per patient‐year equivalent (PYE) for Level 2 or 3 hypoglycaemia are available in Table S9. The PoP with treatment discontinuations due to an AE was comparable between efsitora and icodec (RD [95% CI]: −0.0002 [−0.0223, 0.0220]). For the PRO outcome, CfB in total score of the DTSQc was comparable between efsitora and icodec (MD [95% CI]: 1.01 [−0.243, 2.27]).
Outcomes were comparable for composite endpoints including the PoP with HbA1c ≤ 6.5% without Level 2 or 3 hypoglycaemia within the 12 weeks prior to Week 26 (OR [95% CI]: 0.879 [0.445, 1.73]).
A subgroup analysis for GLP‐1 RA users at baseline for CfB in HbA1c showed comparable results between efsitora and icodec (MD [95% CI]: −0.100 [−0.471, 0.270]). In the GLP‐1 RA‐naïve subgroup, efsitora showed statistically significantly smaller reductions in HbA1c compared with icodec (MD of 0.277 [95% CI: 0.0311, 0.524]).
Results of sensitivity Bucher ITCs were in broad agreement with MAIC results.
3.2.2. Basal‐Bolus Insulin‐Experienced Populations
ITC results are presented in Figure 1 and Table S8. The weighting and refinement procedures resolved differences between participant populations (Table 3), the diagnostics for this MAIC were considered to be appropriate for reliable interpretation of results, and reweighting resulted in minimal reductions in ESS from the unweighted sample for the base case and subgroup analyses (reduction of 1.7% for efsitora and 1.5% for glargine; Table S6). Sensitivity analyses, other than Bucher ITC results, were associated with problematically distributed weights and substantial reductions in ESS; consequently, these results were not considered to be robust and are not reported for this ITC.
For efficacy outcomes, efsitora was comparable to icodec for CfB in HbA1c (%) (MD [95% CI]: −0.0447 [−0.252, 0.163]), the PoP reaching HbA1c < 7% (OR [95% CI]: 1.27 [0.813, 1.98]) and the CfB in percent of TIR (CGM 70–180 mg/dL) (MD [95% CI]: 2.41 [−1.95, 6.77]).
For safety outcomes, efsitora was comparable to icodec for the CfB in body weight (kg) (MD [95% CI]: −0.316 [−1.45, 0.817]); PoP with Level 2 or 3 hypoglycaemia (OR [95% CI]: 1.10 [0.698, 1.73]) and the event rate of Level 2 or 3 hypoglycaemia (rate ratio [95% CI]: 1.03 [0.759, 1.39]); event rates per PYE for Level 2 or 3 hypoglycaemia are provided in Table S10. The PoP with treatment discontinuations due to an AE was comparable between efsitora and icodec (OR [95% CI]: 4.98 [0.688, 36.1]).
The PoP with HbA1c < 7% without Level 2 or 3 hypoglycaemia within the 12 weeks prior to Week 26 in efsitora‐treated participants was comparable to that in participants treated with icodec (OR [95% CI]: 0.944 [0.567, 1.57]).
Subgroup analyses for CfB in HbA1c was comparable between efsitora and icodec for GLP‐1 RA use at baseline (MD [95% CI]: −0.144 [−0.668, 0.401]) and GLP‐1 RA‐naïve at baseline (MD of −0.0223 [95% CI: −0.250, 0.206]).
Results of sensitivity Bucher ITCs were in broad agreement with MAIC results.
4. Discussion
4.1. Key Study Findings
In this indirect comparison study, in the basal insulin‐experienced population, outcomes were comparable between efsitora and icodec for a range of endpoints, including CfB in HbA1c, although a significant difference was observed in the GLP‐1 RA‐naïve subgroup. Other comparable outcomes included the PoP reaching HbA1c ≤ 6.5%, the PoP with Level 2 or 3 hypoglycaemia, total DTSQc score, the PoP with HbA1c ≤ 6.5% without Level 2 or 3 hypoglycaemia within the 12 weeks prior to Week 26 and the PoP with treatment discontinuation due to AEs. In this population, participants treated with efsitora experienced a significantly smaller increase in body weight compared to icodec and significantly lower event rates of Level 2 or 3 hypoglycaemia. While statistically significant, it should be noted that event rates of Level 2 or 3 hypoglycaemia were quite low overall; the estimated relative effect may have been driven by the very low absolute event rates in the degludec arm of ONWARDS 2 (Table S9).
These findings, which suggest benefits of efsitora for the safety of diabetes management in the basal insulin‐experienced population, should be considered hypothesis‐generating and may be further explored in head‐to‐head RCTs [5]. In the subgroup analysis, participants without prior GLP‐1 RA showed significantly smaller reductions in HbA1c when treated with efsitora compared to icodec.
In the basal‐bolus population, all outcomes were comparable between efsitora and icodec: CfB in HbA1c (including in subgroup analysis of those with or without prior GLP‐1 RA use), the PoP reaching HbA1c < 7%, CfB in percent of TIR, the PoP with, and event rate of Level 2 or 3 hypoglycaemia, CfB in body weight, the PoP with HbA1c < 7% without Level 2 or 3 hypoglycaemia within the 12 weeks prior to Week 26 and the PoP with treatment discontinuation due to AEs.
4.2. Strengths
To ensure the validity of trial comparisons, a rigorous feasibility assessment approach was employed, aligned with current technical recommendations, which addressed the statistical assumptions of ITCs [29, 30, 31]. MAIC, a well‐established approach for population adjusted indirect comparisons recommended by the National Institute for Health and Care Excellence (NICE), was used to successfully address between‐trial differences in TEMs and analyses followed published guidelines including detailed assessment of diagnostic results [29, 34, 44]. Furthermore, reanalysis and refinement of QWINT IPD permitted matching timepoints to, and closer alignment of participant populations with, the ONWARDS trials. Lastly, the international scope, high‐quality Phase 3 evidence, and stratification of comparisons based on participants' insulin experience support the generalisability of the results within these specific populations.
4.3. Limitations
Despite application of population‐adjusted methods, indirect comparisons are subject to theoretical assumptions that may not be fully met despite application of appropriate methodology, such as between‐trial variation in unmeasured effect modifiers, and consequently results may be subject to more bias than randomised head‐to‐head evidence. Specific to this analysis, mean insulin doses differed between compared trials due to dissimilar insulin dosing algorithms (see Section 3), which could not be addressed in the absence of patient level data for ONWARDS trials. While race/ethnicity was not considered to be effect modifying based on clinical opinion, it was not possible to assess the impact of imbalances in this characteristic between basal bolus‐experienced populations due to unreliable MAIC results.
4.4. Conclusions
Overall, this indirect comparison analysis reports comparability of efsitora and icodec over a range of key endpoints in both basal insulin‐experienced and basal‐bolus insulin‐experienced populations. Additionally, our analysis suggests that efsitora led to significantly less weight gain and lower hypoglycaemic event rates in the basal insulin‐experienced population.
In the absence of direct evidence via an RCT, these analyses, conducted in line with technical guidance, support the value of efsitora as an effective once‐weekly basal insulin option, offering increased choice for patients, healthcare professionals, and payers in using basal insulin in the management of T2D. Future research could involve generation of direct comparative evidence through randomised, head‐to‐head trials for efsitora and icodec.
Author Contributions
Harpreet S. Bajaj: interpretation of data, writing – review and editing. Marvin Fischer: design, analysis and Interpretation of data, writing – review; Michael B. Davidson: design, interpretation of data, writing – review. Abdul M. Tarafder: design, interpretation of data, writing – review. Christopher J. Michaels: design, acquisition and analysis of data, writing – review and editing. Rachel H. Tao: design, acquisition and analysis of data, writing – review and editing. Alex Pashley: acquisition and analysis of data, writing – review and editing. Jess Amis: interpretation of data, writing – review and editing. Josefine Redig: design, acquisition and interpretation of data, writing – review.
Funding
This study was funded by Eli Lilly and Company. The sponsor was involved in the study design, data collection, analysis, and interpretation, as well as in the preparation, review, and approval of the manuscript. Medical writing and analytical support were provided by Costello Medical and funded by Eli Lilly and Company.
Conflicts of Interest
Harpreet S. Bajaj has received funded grants/clinical trial payments made to his institution from Abbott, Amgen, Arrowhead, AstraZeneca, Boehringer Ingelheim, Eli Lilly, Ionis, Novartis, Novo Nordisk, Pfizer, Roche, and Vertex; and advisory/speakers' compensation from Eli Lilly, Novo Nordisk, Roche, and Sandoz. Marvin Fischer, Michael B. Davidson and Josefine Redig are employees and shareholders of Eli Lilly and Company. Abdul M. Tarafder was employed by Eli Lilly and Company at the time of manuscript submission and remains a shareholder of Eli Lilly and Company. Christopher J. Michaels, Rachel H. Tao, Alex Pashley and Jess Amis are emloyees of Costello Medical, which received payment from Eli Lilly and Company for writing and analytical services for this study.
Supporting information
Table S1: Inclusion and exclusion criteria for all four trials included in analysis.
Table S2: Study characteristics of studies included in ITCs.
Table S3: QWINT‐3 subgroup analysis by race and ethnicity for CFB in HbA1c.
Table S4: Insulin titration algorithms of studies included in basal insulin.
Table S5: Insulin titration algorithms of studies included in basal‐bolus insulin.
Table S6: Estimated sample sizes of base case and subgroup analyses.
Table S7: Base case results for basal insulin‐experienced participants.
Table S8: Base case results for basal‐bolus insulin‐experienced participants.
Table S9: Event rates per PYE of Level 2 or 3 hypoglycaemia to Week 26 (basal insulin‐experienced population).
Table S10: Event rates per PYE of Level 2 or 3 hypoglycaemia to Week 31 (basal‐bolus insulin‐experienced population).
Figure S1: Network diagram for QWINT‐3 and ONWARDS 2 (basal insulin‐experienced population).
Figure S2: Network diagram for QWINT‐4 and ONWARDS 4 (basal‐bolus insulin‐experienced population).
Acknowledgements
The authors would like to thank Amrit Heir (Costello Medical) for scientific writing in preparation of the manuscript.
Bajaj H. S., Fischer M., Davidson M. B., et al., “Insulin Efsitora Demonstrates Similar Efficacy and Safety Profiles to Insulin Icodec in Insulin‐Experienced Individuals With Type 2 Diabetes Mellitus: Matching‐Adjusted Indirect Comparisons of QWINT‐3 Versus ONWARDS 2, and QWINT‐4 Versus ONWARDS 4,” Diabetes, Obesity and Metabolism 28, no. 11 (2026): 10725–10735, 10.1111/dom.71272.
Handling Editor: Richard Donnelly
Data Availability Statement
Eli Lilly and Company provides access to all individual participant data collected during the trial, after anonymisation, with the exception of pharmacokinetic or genetic data. Data are available to request 6 months after the indication studied has been approved in the US and EU and after primary publication acceptance, whichever is later. No expiration date for data requests is currently set once data are made available. Access is provided after a proposal has been approved by an independent review committee identified for this purpose and after a receipt of a signed data‐sharing agreement. Data and documents, including the study protocol, statistical analysis plan and blank or annotated case report forms, will be provided in a secure data‐sharing environment. For details on submitting a request, see the instructions provided at www.vivli.org.
References
- 1. “Diabetes Facts & Figures,” accessed June, 2021, https://idf.org/about‐diabetes/diabetes‐facts‐figures/.
- 2. Diabetes UK , “Life With Diabetes,” accessed November 2025, https://www.diabetes.org.uk/living‐with‐diabetes/life‐with‐diabetes.
- 3. Lu X., Xie Q., Pan X., et al., “Type 2 Diabetes Mellitus in Adults: Pathogenesis, Prevention and Therapy,” Signal Transduction and Targeted Therapy 9, no. 1 (2024): 262. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Cahn A. and Cefalu W. T., “Clinical Considerations for Use of Initial Combination Therapy in Type 2 Diabetes,” Diabetes Care 39, no. Supplement_2 (2016): S137–S145. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Davies M. J., Aroda V. R., Collins B. S., et al., “Management of Hyperglycaemia in Type 2 Diabetes, 2022. A Consensus Report by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD),” Diabetologia 65, no. 12 (2022): 1925–1966. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Rahman M. S., Hossain K. S., Das S., et al., “Role of Insulin in Health and Disease: An Update,” International Journal of Molecular Sciences 22, no. 12 (2021): 6403. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. NICE , “Type 2 Diabetes in Adults: Management,” (2022), https://www.nice.org.uk/guidance/ng28/resources/type‐2‐diabetes‐in‐adults‐management‐pdf‐1837338615493.
- 8. Galdón Sanz‐Pastor A., Justel Enríquez A., Sánchez Bao A., and Ampudia‐Blasco F. J., “Current Barriers to Initiating Insulin Therapy in Individuals With Type 2 Diabetes,” Frontiers in Endocrinology 15 (2024): 1366368. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Wang X., Xiao W., Liang Z., Li S., and Tang Q., “Efficacy and Safety of Once‐Weekly Basal Insulin Versus Once‐Daily Basal Insulin in Patients With Type 2 Diabetes: A Systematic Review and Meta‐Analysis,” Medicine 102, no. 52 (2023): e36308. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Polonsky W. H., Arora R., Faurby M., Fernandes J., and Liebl A., “Higher Rates of Persistence and Adherence in Patients With Type 2 Diabetes Initiating Once‐Weekly vs Daily Injectable Glucagon‐Like Peptide‐1 Receptor Agonists in US Clinical Practice (STAY Study),” Diabetes Therapy 13, no. 1 (2022): 175–187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Weeda E. R., Muraoka A. K., Brock M. D., and Cannon J. M., “Medication Adherence to Injectable Glucagon‐Like Peptide‐1 (GLP‐1) Receptor Agonists Dosed Once Weekly vs Once Daily in Patients With Type 2 Diabetes: A Meta‐Analysis,” International Journal of Clinical Practice 75, no. 9 (2021): e14060. [DOI] [PubMed] [Google Scholar]
- 12. Miller E., Davidson M. B., Bajaj H. S., et al., “Evaluation of Overall Health State, Treatment Burden, and Satisfaction With Insulin Efsitora Alfa (Efsitora) vs. Daily Comparator in Adults With Type 2 Diabetes in the QWINT Clinical Trial Program,” Diabetes Therapy 17 (2026): 431–447. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Bergenstal R. M., Philis‐Tsimikas A., Rodbard H. W., et al., Insulin Efsitora Alfa (Efsitora) Demonstrates Better Overall Health State vs. Daily Degludec in Basal‐Switch Adults with T2D in QWINT‐3 (America Diabetes Association, 2025), 820. [Google Scholar]
- 14. Hs B., Ch W., Lingvay I., et al., “821‐P: Insulin Efsitora Alfa (Efsitora) Demonstrates Better Overall Health State vs. Daily Degludec in Insulin‐Naive Adults With T2D in QWINT‐2,” Diabetes 74, no. Supplement_1 (2025): 821‐P. [Google Scholar]
- 15. Bue‐Valleskey J. M., Kazda C. M., Ma C., et al., “Once‐Weekly Basal Insulin fc Demonstrated Similar Glycemic Control to Once‐Daily Insulin Degludec in Insulin‐Naive Patients With Type 2 Diabetes: A Phase 2 Randomized Control Trial,” Diabetes Care 46, no. 5 (2023): 1060–1067. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Heise T., Chien J., Beals J. M., et al., “Pharmacokinetic and Pharmacodynamic Properties of the Novel Basal Insulin fc (Insulin Efsitora Alfa), an Insulin Fusion Protein in Development for Once‐Weekly Dosing for the Treatment of Patients With Diabetes,” Diabetes, Obesity & Metabolism 25, no. 4 (2023): 1080–1090. [DOI] [PubMed] [Google Scholar]
- 17. Bergenstal R. M., Philis‐Tsimikas A., Wysham C., et al., “Once‐Weekly Insulin Efsitora Alfa: Design and Rationale for the QWINT Phase 3 Clinical Development Programme,” Diabetes, Obesity and Metabolism 26, no. 8 (2024): 3020–3030. [DOI] [PubMed] [Google Scholar]
- 18. Rosenstock J., Bailey T., Connery L., et al., “Weekly Fixed‐Dose Insulin Efsitora in Type 2 Diabetes Without Previous Insulin Therapy,” New England Journal of Medicine 393 (2025): 325–335. [DOI] [PubMed] [Google Scholar]
- 19. Blevins T., Dahl D., Manghi F. C. P., et al., “Once‐Weekly Insulin Efsitora Alfa Versus Once‐Daily Insulin Glargine U100 in Adults With Type 2 Diabetes Treated With Basal and Prandial Insulin (QWINT‐4): A Phase 3, Randomised, Non‐Inferiority Trial,” Lancet 405 (2025): 2290–2301. [DOI] [PubMed] [Google Scholar]
- 20. Philis‐Tsimikas A., Bergenstal R. M., Bailey T. S., et al., “Once‐Weekly Insulin Efsitora Alfa Versus Once‐Daily Insulin Degludec in Adults With Type 2 Diabetes Currently Treated With Basal Insulin (QWINT‐3): A Phase 3, Randomised, Non‐Inferiority Trial,” Lancet 405 (2025): 2279–2289. [DOI] [PubMed] [Google Scholar]
- 21. Wysham C., Bajaj H. S., Del Prato S., et al., “Insulin Efsitora Versus Degludec in Type 2 Diabetes Without Previous Insulin Treatment,” New England Journal of Medicine 391, no. 23 (2024): 2201–2211. [DOI] [PubMed] [Google Scholar]
- 22. Hövelmann U., Brøndsted L., Kristensen N. R., et al., “237‐OR: Insulin Icodec: An Insulin Analog Suited for Once‐Weekly Dosing in Type 2 Diabetes,” Diabetes 69, no. Supplement_1 (2020): 237‐OR. [Google Scholar]
- 23. Rosenstock J., Bajaj H. S., Janež A., et al., “Once‐Weekly Insulin for Type 2 Diabetes Without Previous Insulin Treatment,” New England Journal of Medicine 383, no. 22 (2020): 2107–2116. [DOI] [PubMed] [Google Scholar]
- 24. Philis‐Tsimikas A., Asong M., Franek E., et al., “Switching to Once‐Weekly Insulin Icodec Versus Once‐Daily Insulin Degludec in Individuals With Basal Insulin‐Treated Type 2 Diabetes (ONWARDS 2): A Phase 3a, Randomised, Open Label, Multicentre, Treat‐To‐Target Trial,” Lancet Diabetes & Endocrinology 11, no. 6 (2023): 414–425. [DOI] [PubMed] [Google Scholar]
- 25. Lingvay I., Asong M., Desouza C., et al., “Once‐Weekly Insulin Icodec vs Once‐Daily Insulin Degludec in Adults With Insulin‐Naive Type 2 Diabetes: The ONWARDS 3 Randomized Clinical Trial,” Journal of the American Medical Association 330, no. 3 (2023): 228–237. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Mathieu C., Ásbjörnsdóttir B., Bajaj H. S., et al., “Switching to Once‐Weekly Insulin Icodec Versus Once‐Daily Insulin Glargine U100 in Individuals With Basal‐Bolus Insulin‐Treated Type 2 Diabetes (ONWARDS 4): A Phase 3a, Randomised, Open‐Label, Multicentre, Treat‐To‐Target, Non‐Inferiority Trial,” Lancet 401, no. 10392 (2023): 1929–1940. [DOI] [PubMed] [Google Scholar]
- 27. Dias S., Sutton A. J., Welton N. J., and Ades A. E., “NICE DSU Technical Support Document 3: Heterogeneity: Subgroups, Meta‐Regression, Bias and Bias‐Adjustment,” (2011), http://www.nicedsu.org.uk. [PubMed]
- 28. Dias S., Welton N. J., Sutton A. J., and Ades A. E., “NICE DSU Technical Support Document 2: A Generalised Linear Modelling Framework for Pairwise and Network Meta‐Analysis of Randomised Controlled Trials,” (2016). [PubMed]
- 29. Phillippo D. M., Ades A., Dias S., Palmer S., Abrams K. R., and Welton N. J., “NICE DSU Technical Support Document 18: Methods for Population‐Adjusted Indirect Comparisons In Submissions to NICE,” (2016).
- 30. JCA , “Practical Guideline for Quantitative Evidence Synthesis: Direct and Indirect Comparisons (Pursuant to Article 3(7), Point (d), of Regulation (EU) 2021/2282 on Health Technology Assessment),” (2024).
- 31. JCA , “Methodological Guideline for Quantitative Evidence Synthesis: Direct and Indirect Comparisons; Regulation (EU) 2021/2282 on Health Technology Assessment,” (2024).
- 32. Bucher H. C., Guyatt G. H., Griffith L. E., and Walter S. D., “The Results of Direct and Indirect Treatment Comparisons in Meta‐Analysis of Randomized Controlled Trials,” Journal of Clinical Epidemiology 50, no. 6 (1997): 683–691. [DOI] [PubMed] [Google Scholar]
- 33. Lingvay I., Bache J. K., Desouza C. V., et al., “Efficacy and Hypoglycemia Profile of Once‐Weekly Insulin Icodec vs Once‐Daily Comparators Across Demographic Subgroups,” Journal of Clinical Endocrinology & Metabolism 110, no. 11 (2025): e3596–e3606. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Signorovitch J. E., Sikirica V., Erder M. H., et al., “Matching‐Adjusted Indirect Comparisons: A New Tool for Timely Comparative Effectiveness Research,” Value in Health 15, no. 6 (2012): 940–947. [DOI] [PubMed] [Google Scholar]
- 35. Balduzzi S., Rücker G., and Schwarzer G., “How to Perform a Meta‐Analysis With R: A Practical Tutorial,” Evidence‐Based Mental Health 22 (2019): 153–160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Viechtbauer W., “Conducting Meta‐Analyses in R With the Metafor Package,” Journal of Statistical Software 36, no. 3 (2010): 1–48. [Google Scholar]
- 37. Canty A. and Ripley B., “Boot: Bootstrap Functions,” R Package Version 1.3–32, (2025), https://CRAN.R‐project.org/package=boot.
- 38. Davison A. and Hinkley D., Bootstrap Methods and Their Applications (Cambridge University Press, 1997). [Google Scholar]
- 39. Lumley T., “Survey: Analysis of Complex Survey Samples,” R Package Version 4.4, (2024).
- 40. Cefalu M., Ridgeway G., McCaffrey D., Morral A., Griffin B. A., and Burgette L., “Package ‘twang’,” Toolkit for Weighting and Analysis of Nonequivalent Groups, (2024).
- 41. Kuhn M., “Building Predictive Models in R Using the Caret Package,” Journal of Statistical Software 28, no. 5 (2008): 1–26. [Google Scholar]
- 42. Csárdi G. and Berkelaar M., “Package ‘lpSolve’,” Interface to ‘Lp_solve’ v. 5.5 to Solve Linear/Integer Programs, (2024).
- 43. Venables W. and Ripley B. D., Modern Applied Statistics With S, Fourth ed. (Springer, 2002). [Google Scholar]
- 44. National Institute for Health and Care Excellence , “NICE DSU Technical Support Document 18: Methods for Population‐Adjusted Indirect Comparisons in Submissions to NICE,” (2016), https://sheffield.ac.uk/nice‐dsu/tsds/population‐adjusted.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1: Inclusion and exclusion criteria for all four trials included in analysis.
Table S2: Study characteristics of studies included in ITCs.
Table S3: QWINT‐3 subgroup analysis by race and ethnicity for CFB in HbA1c.
Table S4: Insulin titration algorithms of studies included in basal insulin.
Table S5: Insulin titration algorithms of studies included in basal‐bolus insulin.
Table S6: Estimated sample sizes of base case and subgroup analyses.
Table S7: Base case results for basal insulin‐experienced participants.
Table S8: Base case results for basal‐bolus insulin‐experienced participants.
Table S9: Event rates per PYE of Level 2 or 3 hypoglycaemia to Week 26 (basal insulin‐experienced population).
Table S10: Event rates per PYE of Level 2 or 3 hypoglycaemia to Week 31 (basal‐bolus insulin‐experienced population).
Figure S1: Network diagram for QWINT‐3 and ONWARDS 2 (basal insulin‐experienced population).
Figure S2: Network diagram for QWINT‐4 and ONWARDS 4 (basal‐bolus insulin‐experienced population).
Data Availability Statement
Eli Lilly and Company provides access to all individual participant data collected during the trial, after anonymisation, with the exception of pharmacokinetic or genetic data. Data are available to request 6 months after the indication studied has been approved in the US and EU and after primary publication acceptance, whichever is later. No expiration date for data requests is currently set once data are made available. Access is provided after a proposal has been approved by an independent review committee identified for this purpose and after a receipt of a signed data‐sharing agreement. Data and documents, including the study protocol, statistical analysis plan and blank or annotated case report forms, will be provided in a secure data‐sharing environment. For details on submitting a request, see the instructions provided at www.vivli.org.