Advertisement

Greater Adherence and Persistence with Injectable Dulaglutide Compared with Injectable Semaglutide at 1-Year Follow-up: Data from US Clinical Practice

Open AccessPublished:March 06, 2022DOI:https://doi.org/10.1016/j.clinthera.2022.01.017

      Abstract

      Purpose

      Greater medication adherence and persistence have been associated with improved glycemic control in patients with type 2 diabetes mellitus. This study compared adherence, persistence, and treatment patterns among patients naïve to glucagon-like peptide 1 receptor agonists initiating once-weekly injectable treatment with dulaglutide versus semaglutide over 6-month (6M) and 12-month (12M) follow-up periods.

      Methods

      This retrospective, observational cohort study used administrative claims data from three IBM MarketScan research databases. Data from adult patients with type 2 diabetes newly initiating treatment with dulaglutide or semaglutide between January 2018 and January 2020 (index date was defined as the earliest fill date), without evidence of glucagon-like peptide 1 receptor agonist use in the 6M baseline period, and with continuous enrollment in the 6M baseline and 6M or 12M follow-up period were included. Dulaglutide initiators were propensity score–matched, in a 1:1 ratio, to semaglutide initiators in each 6M and 12M follow-up cohort (26,284 and 13,837 pairs, respectively).

      Findings

      In the matched cohorts, baseline characteristics were balanced; the mean age was 53 years, and 50% of patients were women. Compared to semaglutide initiators, dulaglutide initiators were more adherent (6M, 63.4% vs 47.8%; 12M, 54.4% vs 43.3%; both, P < 0.0001), more persistent on therapy (6M, 72% vs 62%, 12M, 55.5% vs 45.3%, both, P < 0.001), and had more mean days of persistence (6M, 145 vs 132, 12M, 254.3 vs 220.7; both, P < 0.001).

      Implications

      At both 6M and 12M follow-up, dulaglutide initiators had significantly greater adherence and greater persistence compared with matched semaglutide initiators.

      Key words

      Introduction

      Medication adherence and persistence are important for the management of type 2 diabetes mellitus (T2D). Greater medication adherence and persistence are associated with improved glycemic control through reduced hemoglobin (Hb) A1c, and subsequent reductions in health care resource utilization. The ability to maintain medication adherence for a longer period of time may have an impact on long-term glycemic control.
      • Pollack M
      • Chastek B
      • Williams S.
      PDB36 Impact of treatment complexity on adherence and glycemic control: an analysis of oral anti-diabetic agents.
      • Farr AM
      • Sheehan JJ
      • Curkendall SM
      • et al.
      Retrospective analysis of long-term adherence to and persistence with DPP-4 inhibitors in US adults with type 2 diabetes mellitus.
      • Alatorre C
      • Fernández Landó L
      • Yu M
      • et al.
      Treatment patterns in patients with type 2 diabetes mellitus treated with glucagon-like peptide-1 receptor agonists: higher adherence and persistence with dulaglutide compared with once-weekly exenatide and liraglutide.
      • Mody R
      • Grabner M
      • Yu M
      • et al.
      Real-world effectiveness, adherence and persistence among patients with type 2 diabetes mellitus initiating dulaglutide treatment.
      • Polonsky WH
      • Henry RR.
      Poor medication adherence in type 2 diabetes: recognizing the scope of the problem and its key contributors.
      Poor medication adherence remains an area of concern in the management of T2D, which is a lifelong, chronic condition. Published estimates of medication adherence in patients with T2D have varied, ranging from 39% to 93% with oral antihyperglycemic agents, and 38% to 61% with injectables such as glucagon-like peptide 1 receptor agonists (GLP-1 RAs).
      • Alatorre C
      • Fernández Landó L
      • Yu M
      • et al.
      Treatment patterns in patients with type 2 diabetes mellitus treated with glucagon-like peptide-1 receptor agonists: higher adherence and persistence with dulaglutide compared with once-weekly exenatide and liraglutide.
      • Mody R
      • Grabner M
      • Yu M
      • et al.
      Real-world effectiveness, adherence and persistence among patients with type 2 diabetes mellitus initiating dulaglutide treatment.
      ,
      • Cramer JA
      • Benedict A
      • Muszbek N
      • et al.
      The significance of compliance and persistence in the treatment of diabetes, hypertension and dyslipidaemia: a review.
      • Mody R
      • Huang Q
      • Yu M
      • et al.
      Adherence, persistence, glycaemic control and costs among patients with type 2 diabetes initiating dulaglutide compared with liraglutide or exenatide once weekly at 12-month follow-up in a real-world setting in the United States.
      • Yu M
      • Xie J
      • Lando LF
      • et al.
      Liraglutide versus exenatide once weekly: persistence, adherence, and early discontinuation.
      • Krass I
      • Schieback P
      • Dhippayom T.
      Adherence to diabetes medication: a systematic review.
      Numerous once-daily and once-weekly injectables and oral GLP-1 RAs are available in the United States for the treatment of patients with T2D.

      Novo Nordisk, Inc. Victoza (liraglutide) [package insert] [U.S. Food and Drug Administration website]. Revised August 2017. Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2017/022341s027lbl.pdf. Accessed May 24, 2021.

      Novo Nordisk, Inc. Ozempic (semaglutide) [package insert] [U.S. Food and Drug Administration website]. Revised December 2017. Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2017/209637lbl.pdf. Accessed May 24, 2021.

      Eli Lilly and Company. Trulicity (dulaglutide) [package insert] [U.S. Food and Drug Administration website]. Revised January 2017. Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2017/125469s007s008lbl.pdf. Accessed May 24, 2021.

      Amylin Pharmaceuticals, Inc. Bydureon (exenatide extended-release) [package insert] [U.S. Food and Drug Administration website]. Revised January 2012. Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2012/022200s000lbl.pdf. Accessed May 24, 2021.

      GLP-1 RAs differ in regard to route of administration (oral vs injectable), dosing regimen (eg, weekly vs daily), and device features (eg, single-dose vs multidose devices, needle handling, and ease of use), all of which may play a role in treatment adherence, persistence, and effectiveness in clinical practice.
      • Pollack M
      • Chastek B
      • Williams S.
      PDB36 Impact of treatment complexity on adherence and glycemic control: an analysis of oral anti-diabetic agents.
      • Farr AM
      • Sheehan JJ
      • Curkendall SM
      • et al.
      Retrospective analysis of long-term adherence to and persistence with DPP-4 inhibitors in US adults with type 2 diabetes mellitus.
      ,
      • Cramer JA
      • Benedict A
      • Muszbek N
      • et al.
      The significance of compliance and persistence in the treatment of diabetes, hypertension and dyslipidaemia: a review.
      ,
      • Ingersoll KS
      • Cohen J.
      The impact of medication regimen factors on adherence to chronic treatment: a review of literature.
      Among GLP-1 RAs, dulaglutide and semaglutide are the most widely used agents. Although published studies have compared adherence and persistence among GLP-1 RAs,
      • Alatorre C
      • Fernández Landó L
      • Yu M
      • et al.
      Treatment patterns in patients with type 2 diabetes mellitus treated with glucagon-like peptide-1 receptor agonists: higher adherence and persistence with dulaglutide compared with once-weekly exenatide and liraglutide.
      • Mody R
      • Grabner M
      • Yu M
      • et al.
      Real-world effectiveness, adherence and persistence among patients with type 2 diabetes mellitus initiating dulaglutide treatment.
      ,
      • Mody R
      • Huang Q
      • Yu M
      • et al.
      Adherence, persistence, glycaemic control and costs among patients with type 2 diabetes initiating dulaglutide compared with liraglutide or exenatide once weekly at 12-month follow-up in a real-world setting in the United States.
      • Yu M
      • Xie J
      • Lando LF
      • et al.
      Liraglutide versus exenatide once weekly: persistence, adherence, and early discontinuation.
      there is limited comparative evidence between dulaglutide and semaglutide.
      • Mody R
      • Yu M
      • Nepal B
      • et al.
      Adherence and persistence among patients with type 2 diabetes initiating dulaglutide compared with semaglutide and exenatide BCise: 6-month follow-up from US real-world data.
      ,
      • Uzoigwe C
      • Liang Y
      • Whitmire S
      • et al.
      Semaglutide once-weekly persistence and adherence versus other GLP-1 RAs in patients with type 2 diabetes in a real-world setting.
      While medication adherence is multifactorial, there is also limited evidence regarding the differences in adherence and persistence with GLP-1 RAs, based on demographic and clinical characteristics, for a longer follow-up period.
      The present study extends previously published observations by comparing adherence, persistence, and treatment patterns among patients naïve to GLP-1 RAs initiating treatment with once-weekly injectable dulaglutide versus once-weekly injectable semaglutide over 6-month (6M) and 12-month (12M) follow-up periods. The study also evaluated the consistency of adherence and persistence end points between cohorts within subgroups based on age, index dose, baseline obesity, and baseline insulin use.

      Materials and Methods

      Data Source

      This retrospective, observational cohort study used US data from three IBM MarketScan databases: Commercial Claims and Encounters, Medicare Supplemental and Coordination of Benefits, and Early View. The Commercial and Medicare databases contain the inpatient, outpatient, and outpatient prescription drug experiences of employees and their dependents, covered by a variety of fee-for-service and managed care health plans, including exclusive provider organizations, preferred-provider organizations, point-of-service plans, indemnity plans, and health maintenance organizations, as well as of individuals with Medicare supplemental insurance paid for by employers. Both databases provide detailed data on cost, use, and outcomes, from health care services performed in both inpatient and outpatient settings. The MarketScan Early View database maximized available follow-up data and reflected a pharmacy claims completion rate of 97%, with all claims paid and adjudicated prior to inclusion in the database.
      All database records were deidentified and Health Insurance Portability and Accountability Act compliant. Institutional review board approval was not required to conduct this study as only deidentified patient data were used.

      Patient Selection

      Adult patients with at least one pharmacy claim for injectable dulaglutide or semaglutide (index date was defined as the earliest fill date) between January 1, 2018, and January 30, 2020 (index period), continuously enrolled in the 6M preindex (baseline), 6M postindex (6M cohort), and 12M postindex (12M cohort) and with at least one inpatient claim or nondiagnostic outpatient claim for T2D during baseline were selected for inclusion in the study (see Supplemental Figure S1). Patients in the 12M cohort included a subset of patients with data from a full year of follow-up available. Patients with at least one medical claim for type 1 diabetes over the baseline period or at least one pharmacy claim for any GLP-1 RA or GLP-1 RA/insulin fixed-ratio combination during the baseline period or at least one pharmacy claim for a GLP-1 RA other than dulaglutide (dulaglutide cohort) or semaglutide (semaglutide cohort), including any GLP-1 RA/insulin fixed-ratio combination on the index date, were excluded from the study. Patients with at least one pharmacy claim for liraglutide 3.0 mg (dose approved for an indication of obesity) during baseline or on the index date were also excluded from the study. Additionally, patients were excluded if they had evidence of pregnancy, gestational diabetes, or bariatric surgery during the entire study period (see Supplemental Figure S2).

      Patient Characteristics at Baseline

      Baseline characteristics included age, sex, geographic region, payer type, provider type, and comorbidities. Comorbidities were captured using International Classification of Diseases, 10th Edition—Clinical Modification (ICD-10-CM) diagnosis codes from medical claims and included anxiety, asthma, atherosclerotic cardiovascular disease, cancer, cerebrovascular disease, chronic kidney disease, depression, dyslipidemia, gout, heart failure, hypertension, myocardial infarction, nephropathy, obesity, osteoarthritis, polycystic ovarian syndrome, and retinopathy. The adapted Diabetes Complications Severity Index score was calculated to describe the comorbidity burden.
      • Chang HY
      • Weiner JP
      • Richards TM
      • et al.
      Validating the adapted diabetes complications severity index in claims data.
      Patients were also stratified based on initiation of treatment at the following daily doses during the study index period: dulaglutide, 0.75 mg (low dose) or 1.5 mg (high dose); semaglutide, 0.25 or 0.50 mg (low dose) or 1.0 mg (high dose).

      End Point Measures

      The primary end point of this study was adherence to the index drug. Secondary end points included persistence, index drug dose-related patterns, and other diabetes-related drug treatment patterns.

      Adherence

      Adherence was determined using the proportion of days covered (PDC), calculated as the number of days in the follow-up period during which a patient had the index drug on hand, divided by the number of days in each follow-up period. Overlapping days covered by two consecutive prescriptions of the index drug were not counted twice in the PDC calculation. Adherence was defined as a PDC of ≥0.80, as based on the wide use of this definition in the literature and on clinical relevance.
      • Cramer JA
      • Benedict A
      • Muszbek N
      • et al.
      The significance of compliance and persistence in the treatment of diabetes, hypertension and dyslipidaemia: a review.
      ,
      • Mody R
      • Huang Q
      • Yu M
      • et al.
      Adherence, persistence, glycaemic control and costs among patients with type 2 diabetes initiating dulaglutide compared with liraglutide or exenatide once weekly at 12-month follow-up in a real-world setting in the United States.
      ,
      • Mody R
      • Yu M
      • Nepal B
      • et al.
      Adherence and persistence among patients with type 2 diabetes initiating dulaglutide compared with semaglutide and exenatide BCise: 6-month follow-up from US real-world data.
      ,
      • Johnston SS
      • Nguyen H
      • Felber E
      • et al.
      Retrospective study of adherence to glucagon-like peptide-1 receptor agonist therapy in patients with type 2 diabetes mellitus in the United States.
      ,

      Adherence: PQA Adherence Measures [Pharmacy Quality Alliance website]. Available at: https://www.pqaalliance.org/adherence-measures. Accessed March 1, 2021.

      Persistence

      Discontinuation was defined as a failure to refill the index drug within the permissible gap period of 45 days (primary analysis) or 60 days (sensitivity analysis) after the supply of the previous fill was exhausted. Persistence was determined using the number of days of therapy from the initiation of the index treatment to discontinuation or the end of the follow-up period. Persistence was defined as a lack of discontinuations during the follow-up period.

      Index Drug Dose-Related Patterns

      The mean number of index-medication fills, percentages of patients on low and high doses on the index date, percentage of patients with at least two and at least four fills, and dose patterns in dulaglutide and semaglutide initiators in the 6M and 12M cohorts were examined. Based on the dose-change patterns, patients were stratified into one of four mutually exclusive groups: (1) low dose only (started on low dose, never used high dose), (2) high dose only (started on high dose and never used low dose), (3) low dose to high dose (started on low dose, changed to high dose, and stayed on high dose); and (4) all other dose patterns.

      Other Treatment Patterns

      The use of other antihyperglycemic agents (AHAs), including insulin (any type) and oral AHAs, were evaluated in the baseline and 6M and 12M follow-up periods.

      Subgroup Analysis

      Adherence and persistence within subgroups in the 12M cohort were evaluated based on age (<65 vs ≥65 years), index dose (low vs high), baseline obesity status (yes vs no), and baseline insulin use (yes vs no). Obesity was defined as the presence of a diagnosis code for obesity or a body mass index of ≥30 kg/m2 on a medical claim.

      Statistical Analysis

      Analyses of all baseline and end point variables were conducted, summarizing continuous variables using means (SDs), and categorical variables using counts and percentages. Propensity-score matching was used to adjust for confounders between cohorts (6M and 12M separately). In each model, the probability of each patient initiating semaglutide versus dulaglutide (ie, propensity score) was estimated using logistic regression. Propensity-score models included the following baseline characteristics as covariates: age, sex, geographic region, insurance plan type, provider type, and selected medication use (insulin, sulfonylurea, dipeptidyl peptidase 4 inhibitor, or sodium-glucose co-transporter [SGLT]-2 inhibitor), and patients were matched exactly on index dose and baseline obesity status. These variables were selected based on differences seen in baseline characteristics between cohorts, as well as previously published literature indicating variables that influenced treatment selection and outcomes.
      • Alatorre C
      • Fernández Landó L
      • Yu M
      • et al.
      Treatment patterns in patients with type 2 diabetes mellitus treated with glucagon-like peptide-1 receptor agonists: higher adherence and persistence with dulaglutide compared with once-weekly exenatide and liraglutide.
      ,
      • Benedetto U
      • Head SJ
      • Angelini GD
      • et al.
      Statistical primer: propensity score matching and its alternatives.
      • Mody R
      • Yu M
      • Nepal BK
      • et al.
      Dulaglutide has higher adherence and persistence than semaglutide and exenatide QW: 6-month follow-up from US real-world data [abstract].
      Patients were matched, in a 1:1 ratio, based on their propensity scores, using the nearest-neighbor approach, with calipers of 0.2 (6M) and 0.15 (12M) of the SD of the logit of the propensity score.
      • Austin PC.
      Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies.
      ,
      • Brookhart MA
      • Schneeweiss S
      • Rothman KJ
      • et al.
      Variable selection for propensity score models.
      Prior to the assessment of end points, the validity of the propensity-score adjustment was assessed by graphical examination of the overlap in propensity scores and by comparison of the balance in all baseline characteristics across treatments and cohorts via standardized differences and variance ratios.
      • Faries D
      • Zhang X
      • Kadziola Z
      • et al.
      Real World Health Care Data Analysis: Causal Methods and Implementation Using SAS.
      ,
      • Stuart EA.
      Matching methods for causal inference: A review and a look forward.
      A standardized difference of ≤10% in absolute value and a variance ratio in the range of 0.55 to 2 were used to denote balance in baseline characteristics between the cohorts. Propensity-score matchings were finalized before the end point analyses were conducted.
      All end point measures were compared between dulaglutide and semaglutide initiators in the matched 6M and 12M cohorts. The statistical significance of differences in categorical variables was determined using the χ2 test, and in continuous variables, using the t test. Logistic regression was used to compare the likelihood of adherence (PDC ≥0.80) between dulaglutide and semaglutide initiators. Number of persistent days, calculated as days to discontinuation, was analyzed as a time-to-event end point. The probabilities of index-drug persistence throughout the 6M and 12M study periods were estimated using Kaplan-Meier curves. The log-rank test and Cox proportional hazards model were used to compare the rates of discontinuation between cohorts.
      In all analyses, a significance criterion of P < 0.05, defined a priori, was used. Statistical analysis was performed using WPS Analytics version 4.1 (World Programming, Romsey, UK) and R version 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria).

      Results

      Study Population

      Prior to propensity-score matching, 48,113 and 32,308 dulaglutide initiators met the criteria for inclusion in the 6M and 12M cohorts, respectively, and 26,284 and 13,837 semaglutide initiators met the criteria for inclusion in the 6M and 12M cohorts. After matching, the 6M cohort included 26,284 pairs and the 12M cohort included 13,837 pairs. Patients' characteristics at baseline prior to matching are presented in Supplemental Table S1. Additional, detailed baseline characteristics of the matched cohorts are presented in Table I.
      Table IBaseline demographic and clinical characteristics and antihyperglycemic medication use. Data are given as number (%) of patients unless otherwise noted.
      Characteristic6M Matched Cohort12M Matched Cohort
      Dulaglutide Initiators (n = 26,284)Semaglutide Initiators (n = 26,284)Absolute Std DiffDulaglutide Initiators (n = 13,837)Semaglutide Initiators (n = 13,837)Absolute Std Diff
      Age, mean (SD), y
      Demographic characteristics were measured on the index date.
      53.1 (9.7)52.9 (9.5)0.0153.0 (9.5)52.9 (9.3)0.00
      Male13,201 (50.2)13,076 (49.7)0.016,914 (50.0)6,960 (50.3)0.01
      Supplemental Medicare1,764 (6.7)1,670 (6.4)0.01882 (6.4)894 (6.5)0.00
      Provider type
      Provider type on the claim closest to the date of the index prescription was captured using the STDPROV field as a proxy measure, given that a substantial portion of pharmacy claims were missing the provider type in many cases. Primary care included provider codes for medical doctor, osteopathic medicine, internal medicine, family practice, geriatric medicine, preventive medicine, nurse practitioner, or physician's assistant. Endocrinology included provider codes for endocrinology and metabolism or pediatric endocrinology. Other included provider codes for all medical specialties not included in the primary care and endocrinology categories. Provider was reported as unknown or missing if a provider code was missing on office visit claims within 45 days of the index date or if there was no office visit within 45 days of index.
       Primary care14,772 (56.2)14,478 (55.1)0.027,578 (54.8)7,442 (53.8)0.02
       Endocrinology4,315 (16.4)4,510 (17.2)0.022,425 (17.5)2,495 (18.0)0.01
       Other specialty6,612 (25.2)6,649 (25.3)0.003,549 (25.6)3,585 (25.9)0.01
       Unknown/missing585 (2.2)647 (2.5)0.02285 (2.1)315 (2.3)0.01
      Index-drug dose category
       Low21,809 (83.0)21,809 (83.0)0.0011,380 (82.2)11,380 (82.2)0.00
       High4,475 (17.0)4,475 (17.0)0.002,457 (17.8)2,457 (17.8)0.00
      aDCSI score, mean (SD)0.75 (1.2)0.74 (1.2)0.010.72 (1.2)0.73 (1.2)0.01
      aDCSI component
      Neuropathy4,451 (16.9)4,449 (16.9)0.002,334 (16.9)2,359 (17.0)0.00
       Atherosclerotic

      cardiovascular

      disease
      3,450 (13.1)3,521 (13.4)0.011,714 (12.4)1,875 (13.6)0.03
       Nephropathy2,655 (10.1)2,613 (9.9)0.011,388 (10.0)1,324 (9.6)0.02
       Retinopathy2,260 (8.6)2,135 (8.1)0.021,167 (8.4)1,121 (8.1)0.01
       Peripheral

      vascular disease
      1,877 (7.1)1,826 (6.9)0.01911 (6.6)937 (6.8)0.01
       Cerebrovascular

      disease
      629 (2.4)594 (2.3)0.01305 (2.2)318 (2.3)0.01
       Metabolic disease551 (2.1)540 (2.1)0.00291 (2.1)281 (1.0)0.01
      Other clinical condition
      Clinical characteristics were measured during the 6-month preindex period (baseline).
       Hypertension16,165 (61.5)16,593 (63.1)0.038,515 (61.5)8,917 (64.4)0.06
       Dyslipidemia15,262 (58.1)15,701 (59.7)0.037,983 (57.7)8,486 (61.3)0.07
       Obesity10,266 (39.1)10,266 (39.1)0.005,417 (39.1)5,417 (39.1)0.00
       Osteoarthritis2,718 (10.3)2,886 (11.00)0.021,429 (10.3)1,505 (10.9)0.02
       Depression2,487 (9.5)2,517 (9.6)0.001,285 (9.3)1,263 (9.1)0.00
       Chronic kidney

      disease
      1,388 (5.3)1,384 (5.3)0.00696 (5.0)689 (5.0)0.00
       Heart failure737 (2.8)752 (2.9)0.00356 (2.6)368 (2.7)0.01
       Myocardial

      infarction
      360 (1.4)372 (1.4)0.0165 (1.2)198 (1.4)0.02
      Oral antihyperglycemic agent22,086 (84.0)22,162 (84.3)0.0111,665 (84.3)11,750 (84.9)0.02
       Biguanides/metformin18,929 (72.0)18,964 (72.2)0.009,981 (72.1)10,075 (72.8)0.02
       Sulfonylurea6,425 (24.4)6,103 (23.2)0.033,266 (23.6)3,191 (23.1)0.01
       SGLT-2 inhibitor5,816 (22.1)6,099 (23.3)0.033,233 (23.4)3,342 (24.2)0.02
       DPP-4 inhibitor4,035 (15.4)3,925 (14.9)0.012,163 (15.6)2,166 (15.7)0.00
       Thiazolidinedione1,478 (5.6)1,549 (5.9)0.01782 (5.7)861 (6.2)0.02
       Meglitinide203 (0.8)206 (0.8)0.00128 (0.9)100 (0.7)0.02
       α-Glucosidase

      inhibitor
      69 (0.3)59 (0.2)0.0134 (0.2)30 (0.2)0.01
       Insulin8,073 (30.7)8,189 (31.2)0.014,313 (31.2)4,392 (31.7)0.01
       Long-acting
      The intermediate-acting insulin category included the following medications: isophane insulin or NPH insulin, aspart protamine insulin, protamine lispro insulin, and insulin zinc.
      7,329 (27.9)7,328 (27.9)0.003,9133,924 (28.4)0.00
       Rapid-acting
      The rapid-acting insulin category included the following medications: insulin aspart, insulin glulisine, and insulin lispro.
      3,199 (12.2)3,275 (12.5)0.011,716 (12.4)1,812 (13.1)0.02
       Short-acting
      The short-acting insulin category included the following medications: regular insulin of human, purified bovine, and purified swine origins.
      234 (0.9)283 (1.1)0.02127 (0.9)154 (1.1)0.02
       Intermediate-

      acting
      The long-acting insulin category included the following medications: insulin glargine, protamine zinc insulin, and ultralente insulin.
      208 (0.8)224 (0.9)0.01108 (0.8)104 (0.8)0.00
      6M = 6-month follow-up; 12M = 12-month follow-up; aDCSI = adapted Diabetes Complications Severity Index; DPP = dipeptidyl peptidase; SGLT = sodium-glucose co-transporter; Std Diff = standardized difference.
      Cohorts were also well-balanced (std diff <0.1) on the following demographic and clinical characteristics not shown in table: geographic region, anxiety, asthma, cancer, dementia, gout, polycystic ovarian syndrome, and use of amylin analogues.
      a Demographic characteristics were measured on the index date.
      b Provider type on the claim closest to the date of the index prescription was captured using the STDPROV field as a proxy measure, given that a substantial portion of pharmacy claims were missing the provider type in many cases. Primary care included provider codes for medical doctor, osteopathic medicine, internal medicine, family practice, geriatric medicine, preventive medicine, nurse practitioner, or physician's assistant. Endocrinology included provider codes for endocrinology and metabolism or pediatric endocrinology. Other included provider codes for all medical specialties not included in the primary care and endocrinology categories. Provider was reported as unknown or missing if a provider code was missing on office visit claims within 45 days of the index date or if there was no office visit within 45 days of index.
      c Clinical characteristics were measured during the 6-month preindex period (baseline).
      d The intermediate-acting insulin category included the following medications: isophane insulin or NPH insulin, aspart protamine insulin, protamine lispro insulin, and insulin zinc.
      e The rapid-acting insulin category included the following medications: insulin aspart, insulin glulisine, and insulin lispro.
      f The short-acting insulin category included the following medications: regular insulin of human, purified bovine, and purified swine origins.
      g The long-acting insulin category included the following medications: insulin glargine, protamine zinc insulin, and ultralente insulin.

      Patients' Demographic and Clinical Characteristics

      Prior to propensity-score matching, most of the baseline characteristics were balanced (standardized differences ≤0.10) in both the 6M and 12M cohorts, except for index dose, geographic region, obesity status, and baseline use of a sulfonylurea (see Supplemental Table S1). A lower percentage of dulaglutide versus semaglutide initiators started on a low dose at index (6M, 68.1% vs 83.0%; 12M, 66.8% vs 82.2%) and had obesity at baseline (6M, 32.6% vs 39.1%; 12M, 32.4% vs 39.1%) (see Supplemental Table S1). A greater percentage of dulaglutide versus semaglutide initiators had baseline use of a dipeptidyl peptidase 4 (6M, 17.3% vs 14.9%; 12M, 17.6% vs 15.7%) and/or a sulfonylurea (6M, 29.2% vs 23.2%; 12M, 29.1% vs 23.1%); a lower percentage of dulaglutide versus semaglutide initiators had baseline use of an SGLT-2 inhibitor (6M, 20.0% vs 23.2%; 12M, 20.4% vs 24.2%) (see Supplemental Table S1).
      After propensity-score matching, the baseline characteristics were balanced between the dulaglutide and semaglutide initiators in both the 6M and 12M cohorts (Table I). In the matched cohorts, the mean age was 53 years, and approximately half were women. A total of 83.0% of dulaglutide and semaglutide initiators in the 6M cohort, and 82.2% in the 12M cohort, were started on low dose at index. After matching, 39.1% of dulaglutide and semaglutide initiators had obesity at baseline. Both cohorts had comparable mean (SD) adapted Diabetes Complications Severity Index scores between treatment groups (6M, 0.75 [1.2] vs 0.74 [1.2]; 12M, 0.72 [1.2] vs 0.73 [1.2]).

      End Points

      Index Drug Dose-Related Patterns

      The mean (SD) numbers of index fills of dulaglutide and semaglutide were 4.0 (2.0) versus 3.5 (1.8), respectively, in the 6M cohort and 6.6 (3.9) versus 5.9 (3.5) in the 12M cohort (both, P < 0.0001) (Table II). In the 6M cohort, 87.7% of dulaglutide versus 85.1% of semaglutide initiators received two or more fills of the index drug, as did 90.1% of dulaglutide versus 88.4% of semaglutide initiators in the 12M cohort (both, P < 0.0001). In the 6M cohort, 55.2% versus 45.0% of dulaglutide and semaglutide initiators, and 76.9% versus 72.8% in the 12M cohort, received four or more fills of the index drug (both, P < 0.0001).
      Table IIFollow-up dose-related patterns in the matched dulaglutide versus semaglutide initiators in the 6M and 12M cohorts.
      Parameter6M Matched Cohort12M Matched Cohort
      Dulaglutide Initiators (n = 26,284)Semaglutide Initiators (n = 26,284)Dulaglutide Initiators (n = 13,837)Semaglutide Initiators (n = 13,837)
      No. of index drug fills
       Mean (SD)4.0 (2.0)
      Statistically significant finding (P < 0.0001 vs semaglutide initiators).
      3.5 (1.8)6.6 (3.9)
      Statistically significant finding (P < 0.0001 vs semaglutide initiators).
      5.9 (3.5)
       ≥1 Fill, no. (%)26,284 (100)26,284 (100)13,837 (100)13,837 (100)
       ≥2 Fills, no. (%)23,038 (87.7)
      Statistically significant finding (P < 0.0001 vs semaglutide initiators).
      22,377 (85.1)12,468 (90.1)
      Statistically significant finding (P < 0.0001 vs semaglutide initiators).
      12,226 (88.4)
       ≥3 Fills, no. (%)19,335 (73.6)
      Statistically significant finding (P < 0.0001 vs semaglutide initiators).
      17,079 (65.0)11,539 (83.4)
      Statistically significant finding (P < 0.0001 vs semaglutide initiators).
      11,158 (80.6)
       ≥4 Fills, no. (%)14,520 (55.2)
      Statistically significant finding (P < 0.0001 vs semaglutide initiators).
      11,830 (45.0)10,646 (76.3)
      Statistically significant finding (P < 0.0001 vs semaglutide initiators).
      10,077 (72.8)
      Last dose of index drug before discontinuation/study end, no. (%)
      Last dose of index drug was assessed among patients with >one fill of their index drug and, therefore, does not include index dose in patients who only had one fill of the index drug.
      ,
      Statistical comparison was not performed on these variables.
       Low12,262 (55.6)13,038 (66.3)5,501 (47.3)5,895 (57.4)
       High9,780 (44.4)6,628 (33.7)6,136 (52.7)4,378 (42.6)
      Index-drug dose pattern, no. (%)
      Statistical comparison was not performed on these variables.
      ,
      The four dosing patterns reported were assessed irrespective of persistence and were defined as mutually exclusive groups.
       Started at low dose, never

      used high dose
      15,374 (58.5)18,015 (68.5)6,954 (50.3)8,451 (61.1)
       Started at high dose, never

      used low dose
      4,289 (16.3)4,086 (15.5)2,338 (16.9)2,216 (16.0)
       Started at low dose, changed

      to high dose and stayed at

      high dose
      6,062 (23.1)3,529 (13.4)4,010 (29.0)2,589 (18.7)
       All other dose patterns559 (2.1)654 (2.5)535 (3.9)581 (4.2)
      6M = 6-month follow-up; 12M = 12-month follow-up.
      low asterisk Statistically significant finding (P < 0.0001 vs semaglutide initiators).
      a Last dose of index drug was assessed among patients with >one fill of their index drug and, therefore, does not include index dose in patients who only had one fill of the index drug.
      b Statistical comparison was not performed on these variables.
      c The four dosing patterns reported were assessed irrespective of persistence and were defined as mutually exclusive groups.
      Most of the patients in the 6M and 12M cohorts received a low index dose (83.0% and 82.2%, respectively). Among patients who received a second dose of index drug in the 12M cohort, a smaller percentage of dulaglutide versus semaglutide initiators were on a low dose as the last dose during the 12M follow-up (47.3% vs 57.4%; P < 0.001).
      In the 12M cohort, dosing patterns showed that a majority received a low dose only (50.3% on dulaglutide vs 61.1% on semaglutide). The other dosing patterns were low dose to high dose (29.0% vs 18.7%), high dose only (16.9% vs 16.0%), and all other dosing patterns (3.9% vs 4.2%) (Table II).

      Adherence

      In both the 6M and 12M cohorts, the mean (SD) PDC was greater in dulaglutide versus semaglutide initiators (6M, 0.77 [0.28] vs 0.70 [0.27]; 12M, 0.70 [0.32] vs 0.64 [0.31]; both, P < 0.0001). A greater percentage of dulaglutide initiators were considered adherent (PDC ≥0.80) compared to semaglutide initiators in both the 6M and 12M cohorts (6M, 63.4% vs 47.8%; 12M, 54.4% vs 43.3%; both, P < 0.0001) (Figure 1A).
      Figure 1
      Figure 1Percentages of patients considered adherent (defined as a proportion of days covered [PDC] of 0.80) in matched dulaglutide (DULA) versus semaglutide (SEMA) initiators in the 6-month (6M) and 12-month (12M) cohorts. A, Overall adherence at 6 and 12 months. B, Adherence, by dose pattern, at 6 and 12 months. C, Rates and likelihood of adherence in subgroups stratified by age, index dose, baseline obesity status, and baseline insulin use, at 12 months. *P < 0.0001 dulaglutide versus semaglutide.
      The percentage of adherent patients was also examined by dosing pattern. In both dulaglutide and semaglutide initiators in the 6M and 12M cohorts, patients who were started on a low index dose and whose dose was escalated to and remained at a high dose were most adherent, while those who were started on a low dose and never used a high dose were least adherent (Figure 1B).
      In the subgroup analyses in the 12M cohort, the likelihood of adherence favored dulaglutide versus semaglutide initiators within each subgroup investigated (age, interaction P = 0.5787; baseline obesity status, interaction P = 0.2862; and baseline insulin use, interaction P = 0.1410). On assessment by index dose, there was a significant interaction (P = 0.0369), likely due to a difference in magnitude rather than direction of effects. The likelihood of adherence was greater with dulaglutide versus semaglutide in both subgroups of patients initiating at a low dose (odds ratio = 1.12; 95% CI, 1.11–1.14) and at high index dose (odds ratio = 1.09; 95% CI, 1.06–1.12) (Figure 1C). However, the magnitude as measured by the odds ratio was slightly greater in patients initiating at a low index dose compared to those initiating at a high dose.

      Persistence

      Using the definition of persistence of a lack of any treatment gap of ≥45 days, it was found that a greater percentage of dulaglutide versus semaglutide initiators were persistent in the 6M and 12M cohorts (6M: 71.9% vs 62.2%; 12M: 55.5% vs 45.3%; both P < 0.0001). The mean (SD) numbers of persistent days on dulaglutide versus semaglutide were 144.8 (55.7) versus 132.2 (61.6) for the 6M cohort and 254.3 (134.3) versus 220.7 (142.9) for the 12M cohort (both P < 0.0001) (Figure 2A). Of those who discontinued, a majority were on a low dose of index drug at the time of discontinuation in the 6M cohort (dulaglutide, 76.5%; semaglutide, 80.3%) and in the 12M cohort (dulaglutide, 66.9%; semaglutide, 74.5%). Cox proportional hazards models showed that dulaglutide initiators were less likely than were semaglutide initiators to discontinue therapy in the 12M cohort (hazard ratio = 0.73; 95% CI, 0.70–0.75; P < 0.0001) (Figure 2B). Results on persistence in the 6M cohort were consistent with those in the 12M cohort.
      Figure 2
      Figure 2Percentages of patients considered persistent (45-day permissible gap) among matched dulaglutide (DULA) versus semaglutide (SEMA) initiators in the 6-month (6M) and 12-month (12M) cohorts. A, Overall persistence at 6 and 12 months. B, Kaplan-Meier curves showing the risk for discontinuation (45-day permissible gap). C, Rates and risk for discontinuation by subgroups, stratified by age, index dose, baseline obesity status, and baseline insulin use, at 12 months. *P < 0.0001 dulaglutide versus semaglutide. P values on Kaplan-Meier plots represent the log-rank test comparing dulaglutide and semaglutide persistence. Hazard ratios (HR) were generated using Cox proportional hazards modeling. Patients were censored at the end of the 6M or 12M follow-up period. GLP-1 RA = glucagon-like peptide-1 receptor agonist.
      Using the definition of persistence of a lack of any 60-day gap, a greater percentage of dulaglutide versus semaglutide initiators were persistent (6M, 75.3% vs 69.5%; 12M, 59.7% vs 52.8%; both, P < 0.0001). The mean (SD) numbers of persistent days on dulaglutide and semaglutide were 147.4 (54.2) versus 139.3 (58.4) in the 6M group, and 262.6 (131.9) versus 240.7 (139.5) in the 12M group (both, P < 0.0001) (see Supplemental Figure S3).
      The discontinuation rate was significantly lower in dulaglutide versus semaglutide initiators overall and in all of the subgroups examined. Moreover, the magnitude of differences in discontinuation rates, as measured by hazard ratio, were similar across subgroups of interest in the 12M cohort. On examination by subgroup interactions in the 12M cohort, interactions did not reach statistical significance in subgroups stratified by age, index dose, baseline obesity status, or baseline insulin use (interaction P = 0.2359, 0.5204, 0.1565, and 0.3664, respectively) (Figure 2C).

      Other Treatment Patterns

      At baseline, a majority of patients in the matched cohorts were treated with at least one AHA (6M, 84.0% dulaglutide vs 84.3% semaglutide; 12M, 84.3% vs 84.9%). About a third of patients were treated with insulin in the baseline period (6M, 30.7% dulaglutide vs 31.2% semaglutide; 12M, 31.2% vs 31.7%) (Table I).
      During follow-up, in the 6M cohort, the percentages of patients using at least one AHA were 82.5% and 80.8% in dulaglutide and semaglutide initiators, respectively (P < 0.0001), which increased to 87.1% and 85.1% in the 12M cohort (P < 0.0001) . In the follow-up period, insulin use was similar between treatments in the 6M cohort, while a greater percentage of dulaglutide versus semaglutide initiators were treated with insulin at in the 12M cohort (35.7% vs 34.4%; P = 0.0233) (Table III).
      Table IIIFollow-up antihyperglycemic medication use in the matched dulaglutide versus semaglutide initiators in the 6M and 12M cohorts. Data are given as number (%) of patients.
      Parameter6M Matched Cohort12M Matched Cohort
      Dulaglutide Initiators (n = 26,284)Semaglutide Initiators (n = 26,284)Dulaglutide Initiators (n = 13,837)Semaglutide Initiators (n = 13,837)
      Oral antihyperglycemic agents21,696 (82.5)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      21,241 (80.8)12,052 (87.1)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      11,776 (85.1)
       Biguanides/metformin18,480 (70.3)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      18,063 (68.7)10,420 (75.3)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      10,167 (73.5)
       SGLT-2 inhibitor5,767 (21.9)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      6,141 (23.4)3,684 (26.6)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      3,843 (27.8)
       Sulfonylurea5,644 (21.5)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      4,776 (18.2)3,243 (23.4)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      2,720 (19.7)
       DPP-4 inhibitor2,217 (8.4)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      1,746 (6.6)1,324 (9.6)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      1,120 (8.1)
       Thiazolidinedione1,389 (5.3)1,447 (5.5)885 (6.4)951 (6.9)
       Meglitinide183 (0.7)168 (0.6)139 (1.0)118 (0.9)
       α-Glucosidase inhibitor62 (0.2)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      38 (0.1)44 (0.3)28 (0.2)
       Semaglutide, oral19 (0.1)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      58 (0.2)11 (0.1)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      51 (0.4)
      Insulin8,345 (31.7)8,150 (31.0)4,941 (35.7)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      4,761 (34.4)
       Long-acting
      The long-acting insulin category included the following medications: insulin glargine, protamine zinc insulin, and ultralente insulin.
      7,605 (28.9)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      7,344 (27.9)4,546 (32.9)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      4,332 (31.3)
       Rapid-acting
      The rapid-acting insulin category included the following medications: insulin aspart, insulin glulisine, and insulin lispro.
      3,021 (11.5)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      2,864 (10.9)1,933 (14.0)1,870 (13.5)
       Short-acting
      The short-acting insulin category included the following medications: regular insulin of human, purified bovine, or purified swine origin.
      243 (0.9)256 (1.0)152 (1.1)172 (1.2)
       Intermediate-acting
      The intermediate-acting insulin category included the following medications: isophane insulin or NPH insulin, aspart protamine insulin, protamine lispro insulin, and insulin zinc.
      170 (0.6)164 (0.6)115 (0.8)93 (0.7)
      Amylin analogues1 (<0.1)2 (<0.1)2 (<0.1)2 (<0.1)
      Injected GLP-1 RAs
      Patients in the GLP-1 RA + insulin category were not included in the injected GLP-1 RAs category unless they received a single injection of a GLP-1 RA at a different time during the postindex period.
      26,284 (100)26,284 (100)13,837 (100)13,837 (100)
       Dulaglutide26,284 (100)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      508 (1.9)13,837 (100)509 (3.7)
       Semaglutide486 (1.8)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      26,284 (100)632 (4.6)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      13,837 (100)
       Liraglutide197 (0.7)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      131 (0.5)184 (1.3)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      122 (0.9)
       Exenatide173 (0.7)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      77 (0.3)128 (0.9)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      80 (0.6)
       Albiglutide1 (<0.1)000
       Lixisenatide01 (<0.1)01 (<0.1)
      GLP-1 RA + insulin combination
      Patients in the GLP-1 RA + insulin category were not included in the injected GLP-1 RAs category unless they received a single injection of a GLP-1 RA at a different time during the postindex period.
      87 (0.3)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      45 (0.2)89 (0.6)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      60 (0.4)
       Lixisenatide + insulin60 (0.2)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      26 (0.1)59 (0.4)43 (0.3)
       Liraglutide + insulin27 (0.1)20 (0.1)32 (0.2)
      Statistically significant finding (P < 0.05 vs semaglutide initiators).
      18 (0.1)
      6M = 6-month follow-up; 12M = 12-month follow-up; DPP = dipeptidyl peptidase; GLP-1 RA = glucagon-like peptide-1 receptor agonist; SGLT = sodium-glucose co-transporter.
      low asterisk Statistically significant finding (P < 0.05 vs semaglutide initiators).
      a The long-acting insulin category included the following medications: insulin glargine, protamine zinc insulin, and ultralente insulin.
      b The rapid-acting insulin category included the following medications: insulin aspart, insulin glulisine, and insulin lispro.
      c The short-acting insulin category included the following medications: regular insulin of human, purified bovine, or purified swine origin.
      d The intermediate-acting insulin category included the following medications: isophane insulin or NPH insulin, aspart protamine insulin, protamine lispro insulin, and insulin zinc.
      e Patients in the GLP-1 RA + insulin category were not included in the injected GLP-1 RAs category unless they received a single injection of a GLP-1 RA at a different time during the postindex period.

      Discussion

      In this analysis, in which the dulaglutide and semaglutide cohorts had comparable follow-up times sufficient for assessing end points, patients initiating dulaglutide showed greater adherence and persistence compared with semaglutide initiators at 6M and 12M follow-up. The 6M adherence rate with dulaglutide in the present study was greater than those reported in prior studies (63.4% vs 54%‒61%
      • Alatorre C
      • Fernández Landó L
      • Yu M
      • et al.
      Treatment patterns in patients with type 2 diabetes mellitus treated with glucagon-like peptide-1 receptor agonists: higher adherence and persistence with dulaglutide compared with once-weekly exenatide and liraglutide.
      • Mody R
      • Grabner M
      • Yu M
      • et al.
      Real-world effectiveness, adherence and persistence among patients with type 2 diabetes mellitus initiating dulaglutide treatment.
      ,
      • Mody R
      • Yu M
      • Nepal B
      • et al.
      Adherence and persistence among patients with type 2 diabetes initiating dulaglutide compared with semaglutide and exenatide BCise: 6-month follow-up from US real-world data.
      ,
      • Mody R
      • Yu M
      • Nepal BK
      • et al.
      Dulaglutide has higher adherence and persistence than semaglutide and exenatide QW: 6-month follow-up from US real-world data [abstract].
      ,
      • Robinson S
      • Boye KS
      • Mody R
      • et al.
      Real-world effectiveness of dulaglutide in patients with type 2 diabetes mellitus: a literature review.
      ) and with other GLP-1 RAs (exenatide, 29%‒51%; liraglutide, 29%‒48%; or semaglutide, 43%
      • Alatorre C
      • Fernández Landó L
      • Yu M
      • et al.
      Treatment patterns in patients with type 2 diabetes mellitus treated with glucagon-like peptide-1 receptor agonists: higher adherence and persistence with dulaglutide compared with once-weekly exenatide and liraglutide.
      • Mody R
      • Grabner M
      • Yu M
      • et al.
      Real-world effectiveness, adherence and persistence among patients with type 2 diabetes mellitus initiating dulaglutide treatment.
      ,
      • Mody R
      • Yu M
      • Nepal B
      • et al.
      Adherence and persistence among patients with type 2 diabetes initiating dulaglutide compared with semaglutide and exenatide BCise: 6-month follow-up from US real-world data.
      ,
      • Mody R
      • Yu M
      • Nepal BK
      • et al.
      Dulaglutide has higher adherence and persistence than semaglutide and exenatide QW: 6-month follow-up from US real-world data [abstract].
      ). The dulaglutide adherence rate was greater in the present study (54.4%) than were those reported in prior 12M follow-up studies of once-weekly liraglutide (21%‒38%) or exenatide (29%).
      • Mody R
      • Huang Q
      • Yu M
      • et al.
      Adherence, persistence, glycaemic control and costs among patients with type 2 diabetes initiating dulaglutide compared with liraglutide or exenatide once weekly at 12-month follow-up in a real-world setting in the United States.
      • Yu M
      • Xie J
      • Lando LF
      • et al.
      Liraglutide versus exenatide once weekly: persistence, adherence, and early discontinuation.
      ,
      • Buysman EK
      • Liu F
      • Hammer M
      • et al.
      Impact of medication adherence and persistence on clinical and economic outcomes in patients with type 2 diabetes treated with liraglutide: a retrospective cohort study.
      ,
      • Cai J
      • Wang Y
      • Baser O
      • et al.
      Comparative persistence and adherence with newer anti-hyperglycemic agents to treat patients with type 2 diabetes in the United States.
      Persistence rates in this study were in line with prior findings showing lower rates of discontinuation, a greater percentage of persistent patients, and longer persistence periods among patients initiating dulaglutide versus other GLP-1 RAs.
      • Mody R
      • Yu M
      • Nepal B
      • et al.
      Adherence and persistence among patients with type 2 diabetes initiating dulaglutide compared with semaglutide and exenatide BCise: 6-month follow-up from US real-world data.
      ,
      • Mody R
      • Yu M
      • Nepal BK
      • et al.
      Dulaglutide has higher adherence and persistence than semaglutide and exenatide QW: 6-month follow-up from US real-world data [abstract].
      A recently published study by Uzoigwe et al
      • Uzoigwe C
      • Liang Y
      • Whitmire S
      • et al.
      Semaglutide once-weekly persistence and adherence versus other GLP-1 RAs in patients with type 2 diabetes in a real-world setting.
      reported contrasting results: estimated persistence at 360 days was greater with once-weekly injectable semaglutide compared to those with other GLP-1 RAs, including dulaglutide. However, in that study, as indicated in its section on limitations, a majority of semaglutide-treated patients were enrolled toward the end of the patient-selection period, and did not have sufficient follow-up time to experience discontinuation based on a 60-day gap, which may have overestimated persistence in semaglutide-treated patients. Furthermore, a difference in follow-up times between the study cohorts can lead to bias in the evaluate of end points,
      • Clark TG
      • Altman DG
      • De Stavola BL.
      Quantification of completeness of follow-up.
      and, thus, the conclusions may have limited robustness and validity. Additionally, Uzoigwe et al
      • Uzoigwe C
      • Liang Y
      • Whitmire S
      • et al.
      Semaglutide once-weekly persistence and adherence versus other GLP-1 RAs in patients with type 2 diabetes in a real-world setting.
      suggested that once-weekly semaglutide has adherence comparable to that of dulaglutide at 12M, yet conclusions were based on post hoc analyses unadjusted for confounders and a very small sample size of 87 patients at 12M in the semaglutide group, while the adjusted co-primary end point analyses showed significantly greater adherence at 6M follow-up, and no significant difference at 12M follow-up, in dulaglutide initiators versus semaglutide initiators.
      On analysis of adherence and persistence patterns by patient subgroup (age, index drug dose, baseline obesity status, and baseline insulin use) in the present study, greater adherence and persistence among dulaglutide versus semaglutide initiators were consistent across the subgroups studied.
      The present study did not examine the reasons for greater adherence and persistence, as these reasons are unavailable in administrative claims data. However, patients' preference of the dulaglutide-injection device over the semaglutide-injection device due to its ease of use, may be an important factor in the greater adherence and persistence seen among dulaglutide initiators.
      • Matza LS
      • Boye KS
      • Stewart KD
      • et al.
      Assessing patient preference between the dulaglutide pen and the semaglutide pen: a crossover study (PREFER).
      In the present study, the subgroup of patients started on a low dose and escalated to a high dose had the greatest adherence and persistence compared to those who stayed on a low dose of the index drug. These greater adherence and persistence rates with the high doses may have been due to the established dose-dependent efficacy of both dulaglutide and semaglutide, wherein the greater doses of both drugs have been shown to be associated with greater effects on HbA1c and weight in randomized controlled trials.
      • Aroda VR.
      A review of GLP-1 receptor agonists: evolution and advancement, through the lens of randomised controlled trials.
      Additionally, patients who were escalated from a low to a high dose may have had the opportunity to stay on the drug longer, which may also have contributed to the increased adherence and persistence.
      Adherence to treatment is crucial in the achievement of individualized treatment goals. Up to 75% of the gap in efficacy in HbA1c reduction between clinical trials and clinical practice may be attributable to poor medication adherence.
      • Carls GS
      • Tuttle E
      • Tan RD
      • et al.
      Understanding the gap between efficacy in randomized controlled trials and effectiveness in real-world use of GLP-1 RA and DPP-4 therapies in patients with type 2 diabetes.
      Although the present study of data from clinical practice did not assess HbA1c levels, the findings from a recently published study by Mody et al
      • Mody R
      • Huang Q
      • Yu M
      • et al.
      Adherence, persistence, glycaemic control and costs among patients with type 2 diabetes initiating dulaglutide compared with liraglutide or exenatide once weekly at 12-month follow-up in a real-world setting in the United States.
      suggested that greater adherence associated with dulaglutide contributed to a greater reduction in HbA1c levels and better glycemic control.
      The present study has a few limitations that may affect the generalizability of its findings to other populations of patients with T2D. A potential for bias exists due to unmeasured confounders, for example, duration of diabetes, weight, education of patients, and provider characteristics. This study did not examine costs, which can be an important consideration for patients and health care providers when choosing the appropriate treatment.
      • Almandoz JP
      • Lingvay I
      • Morales J
      • Campos C.
      Switching between glucagon-like peptide-1 receptor agonists: Rationale and practical guidance.
      Additionally, differences in insurance coverage, patient-assistance programs, manufacturer copayment cards for eligible patients, and out-of-pocket costs can vary between GLP-1 RAs and affect switching.
      • Almandoz JP
      • Lingvay I
      • Morales J
      • Campos C.
      Switching between glucagon-like peptide-1 receptor agonists: Rationale and practical guidance.
      As in other studies that have used claims data, undetected errors in coding may have existed. As in all studies of data from health claims databases, patients were assumed to have taken their medications as prescribed, but medication use could not be confirmed. Finally, patients in the present study were enrolled in commercial health insurance plans in the United States, and the findings cannot be extrapolated to patients with other types of health insurance, without insurance, or residing outside of the United States.
      The strengths of our study included similar follow-up times across dulaglutide and semaglutide initiators in both the 6M and 12M cohorts, and that all patients had data available from follow-up times sufficient for the event of interest (discontinuation) to have occurred. Both considerations are important for minimizing bias in survival analysis.
      • Clark TG
      • Altman DG
      • De Stavola BL.
      Quantification of completeness of follow-up.
      Additionally, all results were based on prespecified analyses, with appropriate adjustment for potential confounding and bias. These study design features, along with large sample sizes of both the 6M and 12M cohorts, allowed for a robust interpretation of the study findings.
      This study provides additional evidence from clinical practice to support adherence and persistence with dulaglutide over semaglutide over a longer fixed-duration follow-up period. Given that long-term treatment adherence is crucial for optimal glycemic control and better management of T2D, the use of treatments associated with greater adherence among newer injectable GLP-1 RAs may lead to improved outcomes in patients with T2D in clinical practice. Results from this analysis support the use of injectable dulaglutide over semaglutide for greater adherence and persistence, and lower discontinuation rates over 12M follow-up.

      Author Contributions

      R. Mody contributed to the conceptualization, supervision, funding acquisition, and writing (review and editing). J. Manjelievskaia contributed to the methodology, investigation, project administration, and writing (original draft). E.H. Marchlewicz contributed to the investigation, visualization, and validation. R.E. Malik contributed to the conceptualization. N.M. Zimmerman contributed to the software, formal analysis, and data curation. D.E. Irwin contributed to the methodology and supervision. M. Yu contributed to the methodology and writing (review and editing).

      DECLARATION OF INTEREST

      Supported by Eli Lilly and Company. R.M., M.Y., and R.E.M. are employees of, and may hold stock options in, Eli Lilly and were involved in the study design, interpretation of the data, writing of the manuscript, and the decision to submit the article for publication. J.M., E.H.M., N.M.Z., and D.E.I. are employees of IBM Watson Health, which received funding to complete this project from Eli Lilly. The authors have indicated that they have no other conflicts of interest with regard to the content of this article.

      Appendix

      Figure 5
      Supplemental Figure 3Persistence (60-day gap) in the matched dulaglutide versus semaglutide initiators at 6M and 12M follow-up. *An asterisk denotes the outcomes differs significantly between DULA and SEMA patients with p-values <0.0001. Abbreviations: 6M, 6-month follow-up; 12M, 12-month follow-up; N, sample size; SD, standard deviation; DULA, dulaglutide; SEMA, semaglutide.
      Supplemental Table 1 (S1)Pre-matched baseline demographic, clinical, and anti-hyperglycemic medication use
      6M Pre-Matched Cohort12M Pre-Matched Cohort
      Dulaglutide initiators

      (N=48,113)
      Semaglutide initiators

      (N=26,284)
      Absolute Std. Diff.Dulaglutide initiators

      (N= 32,308)
      Semaglutide initiators

      (N= 13,837)
      Absolute Std. Diff.
      Age (Mean, SD)
      Demographic characteristics were measured on the index date
      53.7 (9.8)52.9 (9.5)0.0853.5 (9.6)52.9 (9.3)0.07
      Male (N, %)24,591 (51.1)13,076 (49.7)0.0316,546 (51.2)6,960 (50.3)0.02
      Medicare supplemental (N, %)3,941 (8.2)1,670 (6.4)0.072,587 (8.0)894 (6.5)0.06
      Geographic Region (N, %)
       Northeast8,181 (17.0)4,138 (15.7)0.035,754 (17.8)2,306 (16.7)0.03
       North Central9,343 (19.4)4,347 (16.5)0.086,106 (18.9)2,152 (15.6)0.09
       South25,813 (53.7)15,973 (60.8)0.1417,552 (54.3)8,464 (61.20.14
       West4,689 (9.7)1,790 (6.8)0.112,840 (8.8)900 (6.5)0.09
       Unknown87 (0.2)36 (0.1)0.0156 (0.2)15 (0.1)0.02
      Provider Type (N, %)
      Provider type on the claim closest to the date of the index prescription will be captured, using the STDPROV field, as a proxy measure, given that a substantial portion of pharmacy claims are likely to be missing the provider type. Primary care includes provider codes for medical doctor, osteopathic medicine, internal medicine, family practice, geriatric medicine, preventive medicine, nurse practitioner, or physician's assistant. Endocrinology includes provider codes for: Endocrinology & Metabolism or Pediatric Endocrinology. Other includes provider codes for all medical specialties not included in the Primary Care and Endocrinology categories. Provider is reported as unknown or missing if a provider code is missing on office visit claims within 45 days of index date or if there is no office visit within 45 days of index.
       Primary care28,324 (58.9)14,478 (55.1)0.0818,649 (57.7)7,442 (53.8)0.08
       Endocrinology7,405 (15.4)4,510 (17.2)0.055,225 (16.2)2,495 (18.0)0.05
       Other specialty11,372 (23.6)6,649 (25.3)0.047,742 (24.0)3,585 (25.9)0.04
       Unknown / Missing1,012 (2.1)647 (2.5)0.02692 (2.1)315 (2.3)0.01
      Low index drug dose on index date (N, %)32,742 (68.1)21,809 (83.0)0.3521,590 (66.8)11,380 (82.2)0.36
      aDCSI (Mean, SD)0.74 (1.2)0.74 (1.2)0.000.72 (1.2)0.73 (1.2)0.01
      Components of the aDCSI

      (N, %)
      Atherosclerotic cardiovascular disease6,221 (12.9)3,521 (13.4)0.014,068 (12.6)1,875 (13.6)0.03
      Cerebrovascular disease1,170 (2.4)594 (2.3)0.01740 (2.3)318 (2.3)0.00
      Metabolic disease974 (2.0)540 (2.1)0.00625 (1.9)281 (2.0)0.01
      Nephropathy4,967 (10.3)2,613 (9.9)0.013,269 (10.1)1,324 (9.6)0.02
      Neuropathy8,065 (16.8)4,449 (16.9)0.005,348 (16.6)2,359 (17.0)0.01
      Peripheral vascular disease3,332 (6.9)1,826 (6.9)0.002,155 (6.7)937 (6.8)0.00
      Retinopathy4,139 (8.6)2,135 (8.1)0.022,664 (8.2)1,121 (8.1)0.01
      Other Clinical Conditions

      (N, %)
      Clinical characteristics were measured during the 6-month pre-index period (baseline)
      Anxiety4,302 (8.9)2,558 (9.7)0.032,768 (8.6)1,333 (9.6)0.04
      Asthma2,243 (4.7)1,311 (5.0)0.021,512 (4.7)681 (4.9)0.01
      Cancer2,050 (4.3)1,091 (4.2)0.011,342 (4.2)579 (4.2)0.00
      Chronic kidney disease (CKD)2,610 (5.4)1,384 (5.3)0.011,668 (5.2)689 (5.0)0.01
      Depression4,366 (9.1)2,517 (9.6)0.022,798 (8.7)1,263 (9.1)0.02
      Dementia110 (0.2)33 (0.1)0.0258 (0.2)15 (0.1)0.02
      Dyslipidemia27,603 (57.4)15,701 (57.9)0.0518,703 (57.9)8,486 (61.3)0.07
      Gout747 (1.6)410 (1.6)0.00485(1.5)212 (1.5)0.00
      Heart failure1,276 (2.7)752 (2.9)0.01797 (2.5)368 (2.7)0.01
      Hypertension29,273 (60.8)16,593 (63.1)0.0519,766 (61.2)8,917 (64.4)0.07
      Myocardial infarction (MI)622 (1.3)372 (1.4)0.01401 (1.2)198 (1.4)0.02
      Obesity15,701 (32.6)10,266 (39.1)0.1310,481 (32.4)5417 (39.1)0.14
      Osteoarthritis4999 (10.4)2886 (11.0)0.023323 (10.3)1505 (10.9)0.02
      Polycystic ovarian syndrome (PCOS)327 (0.7)262 (1.0)0.03206 (0.6)133 (1.0)0.04
      Oral anti-hyperglycemic agents (N, %)40,668 (84.5)22,162 (84.3)0.0127,393 (84.8)11,750 (84.9)0.00
       Alpha-glucosidase inhibitors138 (0.3)59 (0.2)0.0194 (0.3)30 (0.2)0.01
       Biguanides/Metformin34,592 (71.9)18,964 (72.2)0.0123,254 (72.0)10,075 (72.8)0.02
       DPP-4 inhibitors8,314 (17.3)3,925 (14.9)0.065,674 (17.6)2,166 (15.7)0.05
       Meglitinides404 (0.8)206 (0.8)0.01297 (0.9)100 (0.7)0.02
       Semaglutide, oral0 (0.0)0 (0.0)NA0 (0.0)0 (0.0)NA
       SGLT-2 inhibitors9,642 (20.0)6,099 (23.2)0.086,576 (20.4)3,342 (24.2)0.09
       Sulfonylureas14,039 (29.2)6,103 (23.2)0.149,417 (29.1)3,191 (23.1)0.14
       Thiazolidinediones2,871 (6.0)1,549 (5.9)0.001,944 (6.0)861 (6.2)0.01
      Insulin (N, %)14,412 (30.0)8,189 (31.2)0.039,732 (30.1)4,392 (31.7)0.04
       Long-acting
      The long-acting insulin category includes the following medications: insulin glargine, protamine zinc insulin, and ultralente insulin.
      13,130 (27.3)7,328 (27.9)0.018,847 (27.4)3,924 (28.4)0.04
       Intermediate-acting
      The intermediate-acting insulin category includes the following medications: isophane insulin or NPH inslin, aspart protamine insulin, protamine lispro insulin, and insulin zinc.
      380 (0.8)224 (0.9)0.01262 (0.8)104 (0.8)0.01
       Rapid-acting
      The rapid-acting insulin category includes the following medications: insulin aspart, insulin glulisine, and insulin lispro.
      5,426 (11.3)3,275 (12.5)0.043,694 (11.4)1,812 (13.1)0.05
       Short-acting
      The short-acting insulin category includes the following medications: regular insulin of human, purified beef and purified pork origin.
      411 (0.9)283 (1.1)0.02278 (0.9)154 (1.1)0.03
      Amylin Analogs (N, %)5 (0.0)8 (0.0)0.013 (0.0)6 (0.0)0.02
      Abbreviations: 6M, 6-month follow-up; 12M, 12-month follow-up; N, sample size; NA, not applicable; SD, standard deviation; Std Diff, standardized difference
      a Demographic characteristics were measured on the index date
      b Provider type on the claim closest to the date of the index prescription will be captured, using the STDPROV field, as a proxy measure, given that a substantial portion of pharmacy claims are likely to be missing the provider type. Primary care includes provider codes for medical doctor, osteopathic medicine, internal medicine, family practice, geriatric medicine, preventive medicine, nurse practitioner, or physician's assistant. Endocrinology includes provider codes for: Endocrinology & Metabolism or Pediatric Endocrinology. Other includes provider codes for all medical specialties not included in the Primary Care and Endocrinology categories. Provider is reported as unknown or missing if a provider code is missing on office visit claims within 45 days of index date or if there is no office visit within 45 days of index.
      ± Clinical characteristics were measured during the 6-month pre-index period (baseline)
      c The long-acting insulin category includes the following medications: insulin glargine, protamine zinc insulin, and ultralente insulin.
      d The intermediate-acting insulin category includes the following medications: isophane insulin or NPH inslin, aspart protamine insulin, protamine lispro insulin, and insulin zinc.
      e The rapid-acting insulin category includes the following medications: insulin aspart, insulin glulisine, and insulin lispro.
      f The short-acting insulin category includes the following medications: regular insulin of human, purified beef and purified pork origin.

      References

        • Pollack M
        • Chastek B
        • Williams S.
        PDB36 Impact of treatment complexity on adherence and glycemic control: an analysis of oral anti-diabetic agents.
        Value Health. 2009; 12: A103
        • Farr AM
        • Sheehan JJ
        • Curkendall SM
        • et al.
        Retrospective analysis of long-term adherence to and persistence with DPP-4 inhibitors in US adults with type 2 diabetes mellitus.
        Adv Ther. 2014; 31: 1287-1305
        • Alatorre C
        • Fernández Landó L
        • Yu M
        • et al.
        Treatment patterns in patients with type 2 diabetes mellitus treated with glucagon-like peptide-1 receptor agonists: higher adherence and persistence with dulaglutide compared with once-weekly exenatide and liraglutide.
        Diabetes Obes Metab. 2017; 19: 953-961
        • Mody R
        • Grabner M
        • Yu M
        • et al.
        Real-world effectiveness, adherence and persistence among patients with type 2 diabetes mellitus initiating dulaglutide treatment.
        Curr Med Res Opin. 2018; 34: 995-1003
        • Polonsky WH
        • Henry RR.
        Poor medication adherence in type 2 diabetes: recognizing the scope of the problem and its key contributors.
        Patient Prefer Adherence. 2016; 10: 1299
        • Cramer JA
        • Benedict A
        • Muszbek N
        • et al.
        The significance of compliance and persistence in the treatment of diabetes, hypertension and dyslipidaemia: a review.
        Int J Clin Pract. 2008; 62: 76-87
        • Mody R
        • Huang Q
        • Yu M
        • et al.
        Adherence, persistence, glycaemic control and costs among patients with type 2 diabetes initiating dulaglutide compared with liraglutide or exenatide once weekly at 12-month follow-up in a real-world setting in the United States.
        Diabetes Obes Metab. 2019; 21: 920-929
        • Yu M
        • Xie J
        • Lando LF
        • et al.
        Liraglutide versus exenatide once weekly: persistence, adherence, and early discontinuation.
        Clin Ther. 2016; 38: 149-160
        • Krass I
        • Schieback P
        • Dhippayom T.
        Adherence to diabetes medication: a systematic review.
        Diabet Med. 2015; 32: 725-737
      1. Novo Nordisk, Inc. Victoza (liraglutide) [package insert] [U.S. Food and Drug Administration website]. Revised August 2017. Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2017/022341s027lbl.pdf. Accessed May 24, 2021.

      2. Novo Nordisk, Inc. Ozempic (semaglutide) [package insert] [U.S. Food and Drug Administration website]. Revised December 2017. Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2017/209637lbl.pdf. Accessed May 24, 2021.

      3. Eli Lilly and Company. Trulicity (dulaglutide) [package insert] [U.S. Food and Drug Administration website]. Revised January 2017. Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2017/125469s007s008lbl.pdf. Accessed May 24, 2021.

      4. Amylin Pharmaceuticals, Inc. Bydureon (exenatide extended-release) [package insert] [U.S. Food and Drug Administration website]. Revised January 2012. Available at: https://www.accessdata.fda.gov/drugsatfda_docs/label/2012/022200s000lbl.pdf. Accessed May 24, 2021.

        • Ingersoll KS
        • Cohen J.
        The impact of medication regimen factors on adherence to chronic treatment: a review of literature.
        J Behav Med. 2008; 3: 213-224
        • Mody R
        • Yu M
        • Nepal B
        • et al.
        Adherence and persistence among patients with type 2 diabetes initiating dulaglutide compared with semaglutide and exenatide BCise: 6-month follow-up from US real-world data.
        Diabetes Obes Metab. 2021; 23: 106-115
        • Uzoigwe C
        • Liang Y
        • Whitmire S
        • et al.
        Semaglutide once-weekly persistence and adherence versus other GLP-1 RAs in patients with type 2 diabetes in a real-world setting.
        Diabetes Ther. 2021; 12: 1475-1489
        • Chang HY
        • Weiner JP
        • Richards TM
        • et al.
        Validating the adapted diabetes complications severity index in claims data.
        Am J Manag Care. 2012; 18: 721-726
        • Johnston SS
        • Nguyen H
        • Felber E
        • et al.
        Retrospective study of adherence to glucagon-like peptide-1 receptor agonist therapy in patients with type 2 diabetes mellitus in the United States.
        Adv Ther. 2014; 31: 1119-1133
      5. Adherence: PQA Adherence Measures [Pharmacy Quality Alliance website]. Available at: https://www.pqaalliance.org/adherence-measures. Accessed March 1, 2021.

        • Benedetto U
        • Head SJ
        • Angelini GD
        • et al.
        Statistical primer: propensity score matching and its alternatives.
        Eur J Cardiothorac Surg. 2018; 53: 1112-1117
        • Faries D
        • Zhang X
        • Kadziola Z
        • et al.
        Real World Health Care Data Analysis: Causal Methods and Implementation Using SAS.
        SAS Institute, Cary, NC2020
        • Mody R
        • Yu M
        • Nepal BK
        • et al.
        Dulaglutide has higher adherence and persistence than semaglutide and exenatide QW: 6-month follow-up from US real-world data [abstract].
        Diabetes. 2020; 69 (-P): 928
        • Austin PC.
        Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies.
        Pharm Stat. 2011; 10: 150-161
        • Brookhart MA
        • Schneeweiss S
        • Rothman KJ
        • et al.
        Variable selection for propensity score models.
        Am J Epidemiol. 2006; 163: 1149-1156
        • Stuart EA.
        Matching methods for causal inference: A review and a look forward.
        Stat Sci. 2010; 25: 1
        • Robinson S
        • Boye KS
        • Mody R
        • et al.
        Real-world effectiveness of dulaglutide in patients with type 2 diabetes mellitus: a literature review.
        Diabetes Ther. 2020; 11: 1437-1466
        • Buysman EK
        • Liu F
        • Hammer M
        • et al.
        Impact of medication adherence and persistence on clinical and economic outcomes in patients with type 2 diabetes treated with liraglutide: a retrospective cohort study.
        Adv Ther. 2015; 32: 341-355
        • Cai J
        • Wang Y
        • Baser O
        • et al.
        Comparative persistence and adherence with newer anti-hyperglycemic agents to treat patients with type 2 diabetes in the United States.
        J Med Econ. 2016; 19: 1175-1186
        • Clark TG
        • Altman DG
        • De Stavola BL.
        Quantification of completeness of follow-up.
        Lancet. 2002; 359: P1309-P1310
        • Matza LS
        • Boye KS
        • Stewart KD
        • et al.
        Assessing patient preference between the dulaglutide pen and the semaglutide pen: a crossover study (PREFER).
        Diabetes Obes Metab. 2020; 22: 355-364
        • Aroda VR.
        A review of GLP-1 receptor agonists: evolution and advancement, through the lens of randomised controlled trials.
        Diabetes Obes Metab. 2018; 20: 22-33
        • Carls GS
        • Tuttle E
        • Tan RD
        • et al.
        Understanding the gap between efficacy in randomized controlled trials and effectiveness in real-world use of GLP-1 RA and DPP-4 therapies in patients with type 2 diabetes.
        Diabetes Care. 2017; 40: 1469-1478
        • Almandoz JP
        • Lingvay I
        • Morales J
        • Campos C.
        Switching between glucagon-like peptide-1 receptor agonists: Rationale and practical guidance.
        Clin Diabetes. 2020; 38: 390-402