Real-World Clinical Effectiveness and Cost Savings of Liraglutide Versus Sitagliptin in Treating Type 2 Diabetes for 1 and 2Years

Real-World Clinical Effectiveness and Cost Savings of Liraglutide Versus Sitagliptin in Treating... Diabetes Ther (2018) 9:1279–1293 https://doi.org/10.1007/s13300-018-0432-2 ORIGINAL RESEARCH Real-World Clinical Effectiveness and Cost Savings of Liraglutide Versus Sitagliptin in Treating Type 2 Diabetes for 1 and 2 Years . . . . Qian Li Rahul Ganguly Michael L. Ganz Cory Gamble Tam Dang-Tan Received: February 23, 2018 / Published online: May 9, 2018 The Author(s) 2018 Results: A total of 3113 patients persistently ABSTRACT used liraglutide (N = 493) or sitagliptin (N = 2620) for C 1 year [mean age (standard Introduction: This study compared the clinical deviation, SD): 53 (8.5) vs. 56 (9.7) years; 48.3% and economic outcomes of long-term use of vs. 62.3% males; both p \ 0.05]; 911 (including liraglutide versus sitagliptin for the treatment of 113 liraglutide users) were persistent users type 2 diabetes (T2DM) in real-world practice in for C 2 years. During the 1st-year follow-up, the USA. liraglutide users (versus sitagliptin users, after Methods: We identified adult patients adjustment) experienced larger glycated hemo- (C 18 years old) with T2DM who initiated globin (HbA1c) reductions from baseline (rang- liraglutide or sitagliptin in 2010–2014 using a ing from 0.34%-point in quarter 1 to 0.21%- large claims database. Quarterly glycemic con- point in quarter 4); higher likelihoods of trol measures and annual healthcare costs were obtaining HbA1c reductions of C 1%-points assessed during the 1st and 2nd years of persis- or C 2%-points [odds ratios (ORs) range tent medication use. Their associations with 1.47–2.04]; and higher likelihoods of reaching medication use (liraglutide or sitagliptin) were HbA1c goals of \ 6.5% or \ 7% (ORs range estimated using multivariable regression models 1.51–2.12) (all p \ 0.05). Liraglutide users also adjusted for patient demographic and clinical experienced HbA1c reductions from baseline in characteristics. the 2nd-year follow-up (0.53–0.33%-point, all p \ 0.05). Although liraglutide users incurred Enhanced Digital Features To view enhanced digital higher healthcare costs than sitagliptin users features for this article, go to https://doi.org/10.6084/ m9.figshare.6133790. during the 1st-year follow-up, they had $2674 (per patient) lower all-cause medical costs (ad- Electronic supplementary material The online justed cost ratio: 0.67, p \ 0.05) and similar version of this article (https://doi.org/10.1007/s13300- total costs (all-cause and diabetes-related) in the 018-0432-2) contains supplementary material, which is available to authorized users. 2nd year. Conclusion: Long-term use of liraglutide for 1 Q. Li (&)  M. L. Ganz or 2 years was associated with better glycemic Evidera, 500 Totten Pond Road, 5th Floor, Waltham, control than using sitagliptin. Savings in medi- MA 02451, USA cal costs were realized for liraglutide users dur- e-mail: Qian.Li@Evidera.com ing the 2nd year of persistent treatment, which R. Ganguly  C. Gamble  T. Dang-Tan offset differences in pharmacy costs. Novo Nordisk Inc, 800 Scudders Mill Road, Funding: Novo Nordisk Inc. Plainsboro, NJ 08536, USA 1280 Diabetes Ther (2018) 9:1279–1293 Keywords: Clinical effectiveness; Costs; superior glycemic control, greater weight loss, HbA1c; Liraglutide; Long-term; Sitagliptin; and better treatment satisfaction after 52 weeks Type 2 diabetes as an add-on to metformin compared with another incretin-based therapy, sitagliptin [a dipeptidyl peptidase-4 (DPP-4) inhibitor] [9]. INTRODUCTION The findings were consistent with those observed at a shorter follow-up period [10–12], Diabetes is an emerging global epidemic—this while gastrointestinal problems were common chronic metabolic condition affected 1 in every 11 adverse reactions to liraglutide [13]. Based on adults worldwide in 2017 and 9.4% of the US data from the clinical trials, the annual mean population in 2015 [1, 2]. Diabetes can lead to cost per patient was estimated to be many micro- and macrovascular complications [3] $5103–$6523 (2012 USD) lower with liraglutide and accounted for 10.7% of global all-cause mor- than with sitagliptin to reach glycated hemo- tality among people 20–79 years old [2]. Diabetes globin (HbA1c) \ 7.0% with no hypoglycemia was estimated to cost $727 billion (US dollars; or weight gain, and liraglutide was shown to be USD) in 2017, responsible for 6–16.6% of total more cost-effective than sitagliptin after 5 years healthcare expenditures worldwide [2]. About half of use [14, 15]. of the global diabetes budget is spent in the US, Many studies have investigated the effec- where patients with diagnosed diabetes on average tiveness of liraglutide outside the controlled spend about $13700 on medical services annually, settings of clinical trials, with recent attention which is about 2.3 times higher than that spent by paid to the long-term use of the therapy [16]. A people without diabetes [1]. 2-year prospective study of routine clinical Type 2 diabetes mellitus (T2DM), characterized practice in France found liraglutide to be as by high blood glucose in the context of insulin effective as observed in clinical trials [17]. resistance and relative insulin deficiency, However, the effectiveness of liraglutide and accounts for approximately 90% of all diabetes sitagliptin has usually been compared in a cases [4]. Due to the progressive nature of diabetes, 6-month treatment period [18–21]. Effective- the initiation of pharmacologic agents is recom- ness for a longer duration has only been mended at the time of diagnosis of T2DM, along examined in a 1-year survey conducted in the with lifestyle modifications such as diet and UK from 2009 to 2011 [22]. exercise to achieve glycemic control [5–7]. With Data on the long-term effectiveness of metformin as the preferred 1st-line therapy, liraglutide are needed to provide healthcare additional anti-diabetic drugs are often consid- providers and patients with additional infor- ered and, as recommended by the American Dia- mation to support T2DM treatment and man- betes Association (ADA), patient’s history of agement decisions. This study investigated the atherosclerotic cardiovascular (CV) disease plays a clinical and cost-effectiveness of liraglutide key role in the selection of the add-on therapy [5]. compared with sitagliptin in the US in a real- Liraglutide, which belongs to the class of glu- world setting for 1 and 2 years of treatment. cagon-like peptide-1 (GLP-1) receptor agonists, is a commonly used anti-diabetic therapy. It was METHODS shown to reduce the risk of major adverse CV events in a recent clinical trial [8], and it is cur- Data Source rently the only agent in its class that has been approved to reduce the risk of major adverse cardiovascular events (cardiovascular death, This retrospective observational cohort study nonfatal MI or nonfatal stroke) in adults with utilized data from January 2010 through T2DM and established cardiovascular disease [5]. December 2014 from the Truven Health Mar- The efficacy and safety of liraglutide for the ketScan Commercial and Medicare Supple- treatment of T2DM has been extensively stud- mental Insurance Databases. The databases ied in clinical trials. Liraglutide demonstrated contain administrative claims and eligibility Diabetes Ther (2018) 9:1279–1293 1281 records for more than 30 million commercially of the competing therapy (i.e., sitagliptin for insured individuals (i.e., working-age adults and the liraglutide cohort and vice versa) during the their dependents) and 3 million enrollees in follow-up period were excluded. Medicare supplemental plans. The data repre- sent the healthcare experience of employees, Study Measures dependents, and retirees with primary or Medicare supplemental coverage through pri- Patient characteristics included demographics vately insured health plans. Data from the measured at the index date (age, gender, geo- Truven Health MarketScan Lab database, graphic region, health plan type, and year of which contains 32.6 million laboratory test treatment initiation), baseline HbA1c (defined results for approximately 1.9 million unique as the mean of values observed between 45 days privately insured patients, were linked to each prior to 7 days after the index date), and clinical patient’s healthcare claims. This article does not characteristics measured over the 6-month contain any studies with human participants or baseline period: any visit to an endocrinologist; animals performed by any of the authors. Charlson Comorbidity Index (CCI) [24]; the occurrence of common diabetes-related com- Sample Selection plications (i.e., retinopathy, nephropathy, neu- ropathy) and comorbidities (i.e., cardiovascular The study sample included patients with pre- diseases, depression, obesity, hypertension, and scription fills of either liraglutide (Victoza )or hyperlipidemia) identified via the International sitagliptin (Januvia or Janumet ) between 1 Classification of Diseases, Ninth Revision, January 2010 and 31 December 2014. The first Clinical Modification (ICD-9-CM) codes [25]; pharmacy claim for liraglutide or sitagliptin and use of anti-diabetic drugs. defined the index therapy and the index date. This study examined the following glycemic Patients were required to be at least 18 years old control endpoints: absolute reduction in at the index date and have continuous medical HbA1c from baseline; reductions in HbA1c of and pharmacy plan enrollment for at least C 1%-points and C 2%-points; and HbA1c goal 6 months prior to (the baseline period) and attainment (\ 6.5% and \ 7%). The glycemic 12 months after the index date. Patients were control endpoints were measured at the end of excluded if they (1) had a prescription fill of any each quarter during the 1st year of persistent GLP-1 receptor agonist or any DPP-4 inhibitor use of the index therapies, where the quarterly during the baseline period or (2) had a diagnosis follow-up HbA1c was the mean value in the of type 1 diabetes, secondary diabetes, gesta- 45-day window at the end of a quarter. Quar- tional diabetes, or pregnancy through the terly HbA1c was also measured during the 2nd baseline period until the end of continuous year of persistent use among the subgroup of health plan enrollment. Patients had to have at patients continuously enrolled in their health least one HbA1c measure around the index date plan and persistently used their index therapy (from 45 days prior to 7 days after the index without competing drugs for at least 2 years. date) and at least one HbA1c measure during Annual healthcare costs during the 1st and 2nd the 1-year follow-up (from 45 days post index year of persistent use of the index therapy were date to 45 days after the end of the follow-up examined. Annual all-cause costs were mea- period). sured by medical and pharmacy claims paid by The study focused on patients who contin- patients and insurers. Diabetes-related costs uously used the index therapy during the fol- were identified by ICD-9-CM diagnosis codes low-up periods (1 and 2 years, respectively) [23]. (250.xx) on medical claims and by anti-dia- Patients who stopped using their index treat- betic drug types from pharmacy claims. All ment for at least 90 consecutive days without a costs were adjusted to the 2016 value based on prescription fill were considered to have dis- the Consumer Price Index Medical Component continued treatment [19]. Patients with any use [26]. 1282 Diabetes Ther (2018) 9:1279–1293 Statistical Analysis use during the 1st year of follow-up. The ratio of costs associated with liraglutide versus sitaglip- tin was generated from the regression. The The study measures were compared between method of recycled predictions was used to treatment groups using Student’s t test for estimate differences in costs between liraglutide continuous variables and the chi-square test for and sitagliptin users. categorical variables. Generalized linear model (GLM) regressions, specific to the outcome dis- tribution, were used to assess the adjusted RESULTS association between the index therapy (liraglu- tide or sitagliptin) and the outcomes. The study sample included 3113 patients (493 Ordinary least squares regression was used to in the liraglutide cohort and 2620 in the sita- assess the relationship between the index ther- gliptin cohort) who had persistently used the apy and quarterly measures of absolute HbA1c index therapy for at least 12 months after reductions from baseline during the 1st and 2nd applying inclusion/exclusion criteria (Fig. 1). year of persistent use. Logistic regression was The characteristics of patients in the liraglutide used for the quarterly measures of HbA1c cohort differed from those in the sitagliptin reductions of C 1%- and C 2%-points and cohort in many aspects (Table 1). Liraglutide HbA1c goal attainment. A generalized estimat- users were younger [mean age (standard devia- ing equation (GEE) approach was incorporated tion, SD): 53 (8.5) years vs. 56 (9.7) years], less to account for multiple observations per indi- likely to be male (48.3% vs. 62.3%), more likely vidual. The regression adjusted for patient to live in the South, and more likely to be demographic characteristics, baseline HbA1c enrolled in health plans covering out-of-net- and other baseline clinical characteristics, and work services (all p \ 0.05). In addition, time-varying covariates (any use of other non- liraglutide users were more likely to visit insulin anti-diabetic medications and any use of endocrinologists, use insulin, and have a diag- insulin in each quarter). nosis of obesity or hypertension during the The coefficients on liraglutide versus sita- baseline period (all p \ 0.05). Baseline HbA1c gliptin were reported, including the direct was similar between the cohorts [8.1 (1.7) % vs. impact on absolute HbA1c reduction from 8.2 (1.7) %; p C 0.05]. The mean HbA1c reduc- baseline, and the odds ratio (OR) for HbA1c tion among liraglutide users during follow-up reduction of C 1%- and C 2%-points and ranged from 1.2 (%-point) in quarter 1 to 0.9 in HbA1c goal attainment. Based on the regression quarter 8; reductions ranged from 0.9 to 0.4 results, the differences in quarterly glycemic among sitagliptin users (Table 2). Baseline control endpoints associated with liraglutide pharmacy costs were higher in the liraglutide versus sitagliptin (i.e., marginal effects) were group than the sitagliptin group before starting estimated using the method of recycled predic- index therapy (Table 1). Medical costs were also tions, where the adjusted outcomes were pre- higher at baseline for the liraglutide group; dicted for each patient in the sample by however, differences were not statistically assuming liraglutide or sitagliptin as the index significant. therapy, respectively [27]. After adjustment for covariates, use of The annual healthcare costs during the per- liraglutide was associated with a higher reduc- sistent use of index therapy were analyzed by tion in HbA1c from baseline in all quarters GLM with log link and gamma distribution. The during the 1st year of follow-up, as reflected by regressions were adjusted for patient demo- the positive coefficients (all p \ 0.05; Table 3). graphic characteristics and the following base- Specifically, the larger reduction in HbA1c line covariates: HbA1c, other clinical associated with the use of liraglutide versus characteristics, drug use, and total healthcare sitagliptin was estimated to range from 0.34 (%- costs (all cause or diabetes related). The cost point) [95% confidence interval (CI) 0.23, 0.45] regressions for the 2nd year of persistent use of in quarter 1 to 0.21 (95% CI 0.07, 0.35) in the index therapy were also adjusted for drug Diabetes Ther (2018) 9:1279–1293 1283 With pharmacy claims of liraglutide or sitagliptin from 1/1/2010 to 12/31/2014: N = 869276 Adults (≥18 years old) with continuous enrollment in baseline and 12-month follow-up: N = 383032 Without diagnosis of type 1 diabetes, secondary diabetes, gestational diabetes, or pregnancy: N = 297977 Without pharmacy claims of GLP-1 receptor agonist or DPP-4 inhibitors in baseline: N = 225950 With HbA1c measure 45 days prior to, and up to 7 days after, the index date: N = 7084 With HbA1c measure 45 days after the index date to 45 days after 12-month follow-up: N = 5326 Persistent use of index drugs (without the competing index drug) for ≥12 months: N = 3113 Liraglutide cohort: N = 493; Sitagliptin cohort: N = 2620 Fig. 1 Sample selection of patients. DPP-4 dipeptidyl peptidase-4, GLP-1 glucagon-like peptide-1, HbA1c glycated hemoglobin quarter 4. The use of liraglutide was also asso- obtain HbA1c reduction C 1%-point and 4.4% ciated with a higher likelihood of obtaining an more (21.9% vs. 17.5%, p \ 0.05) would obtain HbA1c reduction of C 1%-points [ORs ranging HbA1c reduction C 2%-point in quarter 1 if from 1.47 to 2.37] and reductions of C 2%- they used liraglutide rather than sitagliptin, and points (ORs ranging from 1.67 to 2.04; all 10.4% more (34.5% vs. 24.1%, p \ 0.05) or p \ 0.05). Lastly, the use of liraglutide was 10.5% more (57.4% vs. 46.9%, p \ 0.05) associated with a higher likelihood of attaining patients would reach an HbA1c goal of \ 6.5% HbA1c goals: with ORs around 2 of achieving or \ 7.0% in quarter 1, respectively (all HbA1c \ 6.5% and around 1.6 of achieving p \ 0.05). The marginal effects were HbA1c \ 7% in all quarters (all p \ 0.05). stable across all quarters during the 1st year of Among the patients with at least 2 years of persistent medication use, while more variation persistent use of the index therapy, the use of was observed during the 2nd year of persistent liraglutide was associated with a higher reduc- medication use. tion in HbA1c in all quarters during the 2nd The unadjusted and adjusted predicted costs year of follow-up as well as with a higher chance for each cohort, along with the cost ratios, for of an HbA1c reduction of C 1%-points (except the 1st and 2nd years of persistent treatment are for quarter 6), of achieving HbA1c \ 6.5% (ex- presented in Table 4. The adjusted total and cept for quarter 6), and of achieving HbA1c \ medical costs in each cohort and year of follow- 7% (in quarter 5) (all p \ 0.05). The small up are also illustrated in Fig. 4. In the 1st year, sample size was insufficient to estimate the there was no difference in medical costs association between HbA1c reduction C 2%- between persistent users of liraglutide and sita- point and persistent use of index therapy during gliptin. The liraglutide cohort had higher total the 2nd year of follow-up. costs (both all cause and diabetes related) than Adjusted outcomes and the marginal effects the sitagliptin cohort due to the higher phar- of liraglutide versus sitagliptin were calculated macy costs (p \ 0.05). During the 2nd year of (Figs. 2, 3). The marginal effects on HbA1c persistent drug use, all-cause medical costs were reductions from baseline were the same as the 33% lower for liraglutide patients than sita- regression coefficients due to the model speci- gliptin patients (adjusted cost ratio 0.67, 95% CI fication. For the other endpoints, 13.4% more 0.50, 0.89), resulting in a savings of $2674 per patients (48.1% vs. 34.7%, p \ 0.05) would patient (adjusted costs: $5410 vs. $8084, 1284 Diabetes Ther (2018) 9:1279–1293 Table 1 Baseline characteristics of patients with at least 1-year persistent use of liraglutide or sitagliptin Baseline characteristics Liraglutide Sitagliptin Number of patients 493 2620 Demographics Age* 53 (8.5) 56 (9.7) Male* 238 (48.3%) 1631 (62.3%) Region* Northeast 80 (16.2%) 509 (19.4%) North Central 42 (8.5%) 291 (11.1%) South 287 (58.2%) 1017 (38.8%) West 84 (17%) 803 (30.6%) Plan types* Managed care plans without out-of-network coverage 151 (30.6%) 1128 (43.1%) Managed care plans with out-of-network coverage 297 (60.2%) 1253 (47.8%) Other/missing 45 (9.1%) 239 (9.1%) Index year* 2010–2011 305 (61.9%) 1699 (64.8%) 2012–2013 188 (38.1%) 921 (35.2%) Clinical characteristics HbA1c 8.1 (1.7) 8.2 (1.7) Any visit to endocrinologist* 103 (20.9%) 191 (7.3%) Charlson Comorbidity Index 2.4 (0.9) 2.5 [1] Diabetes-related complications 110 (22.3%) 593 (22.6%) Cardiovascular diseases 73 (14.8%) 480 (18.3%) Depression 29 (5.9%) 141 (5.4%) Obesity* 84 (17%) 237 (9%) Hypertension* 317 (64.3%) 1558 (59.5%) Hyperlipidemia 315 (63.9%) 1634 (62.4%) Drug use Any use of non-insulin anti-diabetic drugs 425 (86.2%) 2233 (85.2%) Metformin 365 (74%) 1963 (74.9%) Sulfonylurea 242 (49.1%) 1162 (44.4%) Thiazolidinediones 107 (21.7%) 608 (23.2%) Other non-insulin anti-diabetic drugs 15 (3%) 50 (1.9%) Any use of insulin* 96 (19.5%) 175 (6.7%) Diabetes Ther (2018) 9:1279–1293 1285 Table 1 continued Baseline characteristics Liraglutide Sitagliptin Healthcare costs (2016 $) All-cause medical costs 3316 (10792) 2803 (8608) All-cause pharmacy costs* 2037 (3047) 1563 (2217) All-cause total costs 5352 (11438) 4366 (9302) Diabetes-related medical costs 1474 (7297) 1279 (5342) Diabetes-related pharmacy costs* 643 (1065) 443 (713) Diabetes-related total costs 2117 (7511) 1722 (5457) Means (standard deviation) were reported for continuous variables, and counts (percentage) were reported for categorical variables HbA1c glycated hemoglobin *p \ 0.05 p \ 0.05). This saving offset the higher phar- The 6-month clinical effectiveness of macy costs observed in the liraglutide group in liraglutide in this study is similar to previously the 2nd year and resulted in $602 lower all- published data. Compared with sitagliptin, use cause total costs than in the sitagliptin group of liraglutide has been found to be associated (not statistically significant). No significant dif- with 0.31–0.41 (%-point) greater reduction of ferences between the cohorts were found for the HbA1c from baseline, an 8–15% greater chance diabetes-related medical costs and total costs to reach HbA1c \ 7%, and an 11% greater during the 2nd-year follow-up. chance to reach A1C B 6.5% after 6 months since treatment initiation [18, 19], and these were reflected by $994 (2013 US dollar) lower DISCUSSION diabetes-related medical costs during the 6 months of follow-up (all p \ 0.05) [19]. This study extends previous studies by provid- Trends of clinical effectiveness over 2 years ing integrated evidence on the long-term clini- of medication use continued to be favorable for cal and cost-effectiveness of treatment with liraglutide versus sitagliptin, as found in our liraglutide versus sitagliptin in real-world clini- study. Specifically, the difference in HbA1c cal practice in the US. Even after adjusting for reduction between the two treatments was baseline patient characteristics and the use of highest after 3 months of medication use and antidiabetic medications during follow-up, we then stabilized for the remainder of the year; no found that patients persistently using liraglu- obvious trend was observed in the effectiveness tide for 1 or 2 years experienced improved gly- in terms of likelihood to obtain high HbA1c cemic control (in terms of higher reduction of reduction or reach HbA1c goals in the 1st year HbA1c from baseline, higher likelihood to reach of follow-up. During the 2nd year of medication high reduction in HbA1c, and higher likelihood use, there was much more variation over time in to reach HbA1c target levels) compared with the clinical effectiveness of liraglutide versus those using sitagliptin. Patients in both cohorts sitagliptin, possibly because of the small sample incurred similar medical costs during the 1st size (liraglutide cohort: N = 113; sitagliptin year of medication use, while during the 2nd cohort: N = 798); nevertheless, the clinical year, liraglutide users had significantly lower effectiveness of liraglutide remained stable. all-cause medical costs, which offset their The improved glycemic control associated higher pharmacy costs than sitagliptin users. with liraglutide relative to sitagliptin led to cost 1286 Diabetes Ther (2018) 9:1279–1293 Table 2 Unadjusted glycemic control during persistent use of liraglutide or sitagliptin In the 1st-year persistent use of Quarter 1 Quarter 2 Quarter 3 Quarter 4 liraglutide or sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Number of patients (N) 321 1586 255 1307 204 1289 222 1234 Change in HbA1c from baseline - 1.2 (1.3) - 0.9 (1.5)* - 1.1 (1.4) - 0.8 (1.6)* - 1.1 (1.4) - 0.7 (1.5)* - 0.9 (1.3) - 0.7 (1.5) High reduction of HbA1c Reduction of HbA1c C 1%-point 155 (48.3%) 560 (35.3%)* 103 (40.4%) 463 (35.4%) 95 (46.6%) 407 (31.6%)* 85 (38.3%) 397 (32.2%) Reduction of HbA1c C 2%-point 68 (21.2%) 272 (17.2%) 51 (20%) 211 (16.1%) 40 (19.6%) 181 (14%)* 37 (16.7%) 184 (14.9%) HbA1c goal attainment HbA1c \ 6.5% 114 (35.5%) 396 (25%)* 106 (41.6%) 347 (26.5%)* 78 (38.2%) 332 (25.8%)* 83 (37.4%) 296 (24%)* HbA1c \ 7.0% 183 (57%) 780 (49.2%)* 149 (58.4%) 616 (47.1%)* 112 (54.9%) 598 (46.4%)* 112 (50.5%) 544 (44.1%) In the 2nd-year persistent use of Quarter 5 Quarter 6 Quarter 7 Quarter 8 liraglutide or sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Number of patients (N) 62 424 60 423 53 392 54 382 Change in HbA1c from baseline - 0.9 (1.2) - 0.5 (1.4)* - 1 (1.2) - 0.6 (1.4)* - 0.9 (1.2) - 0.5 (1.4) - 0.9 (1.2) - 0.4 (1.5)* High reduction of HbA1c Reduction of HbA1c C 1%-point 22 (35.5%) 119 (28.1%) 21 (35%) 124 (29.3%) 21 (39.6%) 107 (27.3%) 26 (48.1%) 100 (26.2%)* Reduction of HbA1c C 2%-point 12 (19.4%) 45 (10.6%)* 10 (16.7%) 38 (9%) 11 (20.8%) 43 (11%)* 9 (16.7%) 40 (10.5%) HbA1c goal attainment HbA1c \ 6.5% 23 (37.1%) 109 (25.7%) 21 (35%) 120 (28.4%) 16 (30.2%) 98 (25%) 16 (29.6%) 85 (22.3%) HbA1c \ 7.0% 33 (53.2%) 197 (46.5%) 32 (53.3%) 203 (48%) 23 (43.4%) 187 (47.7%) 23 (42.6%) 175 (45.8%) Means (standard deviation) were reported for continuous variables, and counts (percentage) were reported for categorical variables HbA1c glycated hemoglobin *p \ 0.05 Based on a subgroup of patients with persistent use of index medications for C 24 months (liraglutide cohort: N = 113; sitagliptin cohort: N = 798) Diabetes Ther (2018) 9:1279–1293 1287 Table 3 Multivariable regression results on glycemic control during persistent use of liraglutide or sitagliptin Association between index therapy and Reduction in HbA1c Reduction of Reduction of HbA1c goal HbA1c goal glycemic control from baseline HbA1c ‡ 1%-point HbA1c ‡ 2%-point attainment < 6.5% attainment < 7.0% Coefficient 95% CI Odds 95% CI Odds 95% CI Odds 95% CI Odds 95% CI ratio ratio ratio ratio Liraglutide vs. sitagliptin In the 1st-year persistent use of liraglutide or sitagliptin Quarter 1 0.34 0.23, 0.45* 2.37 1.79, 3.14* 1.73 1.17, 2.56* 1.82 1.40, 2.36* 1.68 1.29, 2.19* Quarter 2 0.28 0.15, 0.40* 1.47 1.07, 2.03* 2.04 1.30, 3.19* 2.06 1.58, 2.68* 1.66 1.26, 2.18* Quarter 3 0.31 0.18, 0.44* 2.01 1.43, 2.82* 1.67 1.10, 2.55* 1.99 1.49, 2.65* 1.70 1.27, 2.28* Quarter 4 0.21 0.07, 0.35* 1.56 1.10, 2.22* 1.70 1.04, 2.78* 2.12 1.60, 2.81* 1.51 1.13, 2.01* Liraglutide vs. sitagliptin In the 2nd-year persistent use of liraglutide or sitagliptin Quarter 5 0.53 0.32, 0.75* 1.99 1.09, 3.63* 2.11 1.21, 3.68* 1.87 1.01, 3.47* Quarter 6 0.36 0.14, 0.57* 1.32 0.64, 2.72 1.73 0.95, 3.15 1.61 0.93, 2.79 Quarter 7 0.32 0.09, 0.56* 1.98 1.04, 3.77* 2.13 1.22, 3.70* 1.07 0.58, 1.96 Quarter 8 0.33 0.09, 0.59* 3.02 1.65, 5.53* 1.92 1.02, 3.60* 1.40 0.77, 2.52 All regressions were adjusted for baseline characteristics (demographics, clinical characteristics, and drug use) and time-varying covariates (any use of other non- insulin anti-diabetic medications and any use of insulin in each quarter) CI confidence interval, HbA1c glycated hemoglobin *p \ 0.05 Based on a subgroup of patients with persistent use of index medications for C 24 months (liraglutide cohort: N = 113; sitagliptin cohort: N = 798) The regression could not generate consistent estimates because of the small HbA1c sample size 1288 Diabetes Ther (2018) 9:1279–1293 Fig. 2 Adjusted glycemic control outcomes during the 1st- varying covariates (any used of other non-insulin anti- year persistent use of liraglutide or sitagliptin. *p\0.05. diabetic medications, and any use of insulin in each Outcomes adjusted for baseline characteristics and time- quarter). HbA1c glycated hemoglobin savings, especially for patients treated for longer the annual all-cause healthcare costs were sig- than 1 year. The findings on the costs during nificantly lower for the liraglutide cohort. the 2nd year of persistent treatment may be The summary statistics of healthcare obscured because of the small sample size and resource utilization (see Supplementary Appen- large variation in healthcare costs. Thus, a post dix Table 5 for results) showed lower use of hoc analysis was conducted on annual costs inpatient and emergency room services when after the 1st year of persistent use of the index taking liraglutide. Specifically, patients treated therapy among patients who did not necessarily with liraglutide had a shorter length of hospital have persistent use of index therapy during the stay during the 1st year [all-cause, mean days 2nd year after treatment initiation. The results (SD): 0.2 (1) vs. 0.4 (2.3); diabetes-related, 0.2 from the post hoc analysis showed that after (0.8) vs. 0.3 (1.9); all p \ 0.05] and 2nd year [all 1 year of persistent treatment, medical costs cause: 0.1 (0.6) vs. 0.4 (2.3); diabetes related, 0.1 (both diabetes related and all cause) were sig- (0.6) vs. 0.3 (2.0); all p \ 0.01] of persistent nificantly lower in the liraglutide cohort (Sup- treatment than sitagliptin users. Liraglutide plementary Appendix Table 6). Further, no users also had fewer visits to the emergency difference between the cohorts was found in the department during the 2nd year of persistent annual costs for diabetes-related healthcare, and treatment. No differences between treatment Diabetes Ther (2018) 9:1279–1293 1289 Fig. 3 Adjusted glycemic control outcomes during the quarter). Based on a subgroup of patients with persistent 2nd-year persistent use of liraglutide or sitagliptin. use of index medications for C 24 months (liraglutide *p\0.05. Outcomes adjusted for baseline characteristics cohort; N = 113; sitagliptin cohort: N = 798). HbA1c and time-varying covariates (any used of other non-insulin glycated hemoglobin anti-diabetic medications, and any use of insulin in each groups were observed in other healthcare the liraglutide cohort were more likely to use resource utilization. These findings suggest that insulin than those in the sitagliptin cohort hospitalization and emergency room visits may during both the baseline and follow-up periods be the main drivers of cost savings associated (p \ 0.05; results not shown), but the need for with liraglutide, and further research is war- add-on insulin increased more quickly in the ranted to investigate this hypothesis. sitagliptin cohort. Specifically, the percentages The lower medical costs in the liraglutide of patients using insulin in the liraglutide cohort compared with sitagliptin during the cohort versus the sitagliptin cohort were 19.5% 2nd year of persistent treatment were statisti- vs. 6.7% in the 6-month baseline period and cally significant for all-cause healthcare services, 21.5% vs. 11% in the 1st year of follow-up. This but not for diabetes-related services. This may may indicate that liraglutide was more success- reflect the nonglycemic benefits of liraglutide ful in reaching the treatment goal and/or (including weight loss and reduced risks of CV patient satisfaction and thus was associated events), while sitagliptin had neutral effects [5]. with less need to intensify treatment with The finding also suggests that better glycemic insulin. control can improve a patient’s overall health. The results of this study should be inter- Our analyses adjusted for many potential preted with consideration of the limitations confounding factors including demographics, associated with studies based on administrative baseline clinical characteristics, baseline claims data. First, due to the nonexperimental healthcare costs, and the use of antidiabetic nature of the data and the study design, this agents during the follow-up period. During the study does not allow for causal inferences; 1st year of persistent medication use, more than unmeasured confounders such as prescription 80% of the patients used oral antidiabetic bias may influence the results. Our study con- medications (mostly metformin) in conjunc- sidered a wide range of covariates, including tion with liraglutide or sitagliptin. Patients in socio-demographic and clinical measures, to 1290 Diabetes Ther (2018) 9:1279–1293 Table 4 Annual healthcare costs during persistent use of liraglutide or sitagliptin Annual healthcare costs during the 1st-year persistent use of liraglutide or sitagliptin c c Unadjusted annual healthcare costs Adjusted cost ratio (95% CI) Adjusted annual healthcare costs Liraglutide Sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Difference All-cause healthcare costs (2016 $) Medical costs $6630 (11643) [2481] $7392 (19912) [2055] 1.04 (0.91, 1.18) 1 $7386 $7126 $260 Pharmacy costs $8418 (7983) [7175] $5979 (5024) [4971]* 1.31 (1.20, 1.44)* 1 $7858 $5995 $1863* Total costs $15048 (14619) [10736] $13371 (21101) [8033]* 1.16 (1.07, 1.25)* 1 $15196 $13144 $2052* Diabetes-related healthcare costs (2016 $) Medical costs $2884 (7342) [705] $3611 (13528) [580] 1.03 (0.88, 1.20) 1 $3626 $3532 $94 Pharmacy costs $5134 (3175) [4830] $3320 (1715) [3106]* 1.49 (1.37, 1.62)* 1 $4975 $3338 $1637* Total costs $8018 (7799) [6035] $6931 (13651) [3944]* 1.23 (1.13, 1.34)* 1 $8473 $6882 $1591* Annual healthcare costs during the 2nd-year persistent use of liraglutide or sitagliptin b b Unadjusted annual healthcare costs Adjusted cost ratio (95% CI) Adjusted annual healthcare costs Liraglutide Sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Difference All-cause healthcare costs (2016 $) Medical costs $6029 (10709) [2361] $8010 (25840) [1915] 0.67 (0.50, 0.89)* 1 $5410 $8084 - $2674* Pharmacy costs $9577 (10540) [7787] $6191 (5797) [5011]* 1.32 (1.10, 1.58)* 1 $8227 $6227 $2000* Total costs $15606 (16711) [10623] $14201 (26888) [7778] 0.96 (0.81, 1.14) 1 $13772 $14374 - $602 Diabetes-related healthcare costs (2016 $) Medical costs $2560 (8291) [664] $4226 (20850) [548] 0.75 (0.54, 1.04) 1 $2986 $4001 - $1015 Pharmacy costs $5548 (3070) [5627] $3422 (1836) [3130]* 1.49 (1.26, 1.77)* 1 $5167 $3458 $1709* Total costs $8109 (9391) [6500] $7648 (21039) [3878] 1.12 (0.94, 1.34) 1 $8373 $7463 $911 Means (standard deviation) [median] were reported for unadjusted healthcare costs *p\ 0.05 Based on a subgroup of patients with persistent use of index medications for C 24 months (liraglutide cohort: N = 113; sitagliptin cohort: N = 798) Adjusted for baseline characteristics and drug use in the 1st year of follow-up (any use of other non-insulin anti-diabetic medications, and any use of insulin) Adjusted for baseline characteristics Diabetes Ther (2018) 9:1279–1293 1291 Fig. 4 Adjusted annual healthcare costs during persistent N = 798); and outcomes adjusted for baseline character- use of liraglutide or sitagliptin. *p\0.05. Outcomes istics and drug use in the 1st year of follow-up (any use of adjusted for baseline characteristics. Based on a subgroup other non-insulin anti-diabetic medications, and any use of of patients with persistent use of index medications for C insulin) 24 months (liraglutide cohort: N = 113; sitagliptin cohort: reduce the residual confounding. Second, similar between liraglutide users and sitagliptin although clinical trials determined the efficacy users during the 1st year of persistent treatment, of liraglutide 1.2 mg separately from 1.8 mg, our but savings in medical costs were realized for study considered the overall use of liraglutide. It liraglutide users during the 2nd year of persis- is challenging to calculate the dose of liraglutide tent treatment, which offsets differences in used in clinical practice from claims data since pharmacy costs. the National Drug Codes may not completely capture the clinical practice of dosing and titration, which may also vary over the consid- ACKNOWLEDGEMENTS ered follow-up period. Third, HbA1c measure- ments were available for only a subset of T2DM patients; the availability of HbA1c is not Funding. Novo Nordisk Inc. provided the expected to be related with patient characteris- funding for the study and the article processing tics and thus should not lead to bias in our charges. All authors had full access to all of the findings. Finally, patients 65 years or older are data in this study and take complete responsi- under-represented since the MarketScan data- bility for the integrity of the data and accuracy bases used in this study contain only Medicare of the data analysis. enrollees who purchased supplemental com- mercial health insurance. However, MarketScan Editorial Assistance. Editorial support was is one of the largest commercial healthcare provided by Colleen Dumont, an employee of claims databases in the US, and our findings can Evidera, and was funded by Novo Nordisk Inc. be interpreted in a population with commercial health insurance. Authorship. All authors participated in data analysis and interpretation and contributed to the development of the manuscript. All named CONCLUSION authors meet the International Committee of Medical Journal Editors (ICMJE) criteria for In real-world clinical practice in the US, long- authorship for this manuscript, take responsi- term use of liraglutide for 1 or 2 years was bility for the integrity of the work as a whole, associated with better glycemic control com- and have given final approval to the version to pared with sitagliptin. The medical costs were be published. 1292 Diabetes Ther (2018) 9:1279–1293 Disclosures. Qian Li is employed by Evidera REFERENCES (a division of PPD), which provides consulting and other research services to pharmaceutical, 1. Centers for Disease Control and Prevention (CDC). medical device, and related organizations. In National Diabetes Statistics Report, 2017: estimates of diabetes and its burden in the United States. their salaried positions, they work with a variety https://www.cdc.gov/diabetes/pdfs/data/statistics/ of companies and organizations and are pre- national-diabetes-statistics-report.pdf. Accessed 29 cluded from receiving payment or honoraria January 2018. directly from these organizations for services 2. International Diabetes Federation (IDF). IDF Dia- rendered. Evidera received funding from Novo betes Atlas—8th Edition. http://www.diabetesatlas. Nordisk Inc. to participate in the study and for org/. Accessed 29 January 2018. the development of this manuscript. Rahul Ganguly is employed by Novo Nordisk Inc. 3. Group AC, Patel A, MacMahon S, Chalmers J, Neal B, Billot L, et al. Intensive blood glucose control Michael Ganz is employed by Evidera (a divi- and vascular outcomes in patients with type 2 dia- sion of PPD), which provides consulting and betes. N Engl J Med. 2008;358(24):2560–72. other research services to pharmaceutical, medical device, and related organizations. In 4. Centers for Disease Control and Prevention (CDC). their salaried positions, they work with a variety At a glance 2016: diabetes. https://www.cdc.gov/ chronicdisease/resources/publications/aag/pdf/ of companies and organizations and are pre- 2016/diabetes-aag.pdf. Accessed 12 April 2017. cluded from receiving payment or honoraria directly from these organizations for services 5. ADA. Pharmacologic approaches to glycemic treat- rendered. Evidera received funding from Novo ment: standards of medical care in diabetes. Dia- betes Care. 2018;41(Suppl 1):S73–85. Nordisk Inc. to participate in the study and for the development of this manuscript. Cory 6. International Diabetes Federation (IDF). Treatment Gamble is employed by Novo Nordisk Inc. Tam algorithm for people with type 2 diabetes. http:// Dang-Tan is employed by Novo Nordisk Inc. www.idf.org/treatment-algorithm-people-type-2- diabetes. Accessed 3 Feb 2016. Compliance with Ethics Guidelines. This 7. National Institute for Health and Care Excellence article does not contain any studies with (NICE). Algorithm for blood glucose lowering human participants or animals performed by therapy in adults with type 2 diabetes 2015. http://www.nice.org.uk/guidance/ng28/resources/ any of the authors. algorithmfor-blood-glucose-lowering-therapy-in- adults-withtype-2-diabetes-2185604173. Accessed 3 Data Availability. The data sets generated Feb 2016. and/or analyzed during the current study are not publicly available because of data license 8. Marso SP, Daniels GH, Brown-Frandsen K, Kris- tensen P, Mann JF, Nauck MA, et al. Liraglutide and agreement, but programming codes are avail- cardiovascular outcomes in type 2 diabetes. N Engl J able from the corresponding author upon rea- Med. 2016;375(4):311–22. sonable request. 9. Pratley R, Nauck M, Bailey T, Montanya E, Cuddihy R, Filetti S, et al. One year of liraglutide treatment Open Access. This article is distributed offers sustained and more effective glycaemic con- under the terms of the Creative Commons trol and weight reduction compared with sitaglip- Attribution-NonCommercial 4.0 International tin, both in combination with metformin, in License (http://creativecommons.org/licenses/ patients with type 2 diabetes: a randomised, paral- lel-group, open-label trial. Int J Clin Pract. by-nc/4.0/), which permits any noncommer- 2011;65(4):397–407. cial use, distribution, and reproduction in any medium, provided you give appropriate credit 10. Pratley RE, Nauck M, Bailey T, Montanya E, Cud- to the original author(s) and the source, provide dihy R, Filetti S, et al. Liraglutide versus sitagliptin for patients with type 2 diabetes who did not have a link to the Creative Commons license, and adequate glycaemic control with metformin: a indicate if changes were made. 26-week, randomised, parallel-group, open-label trial. Lancet. 2010;375(9724):1447–56. Diabetes Ther (2018) 9:1279–1293 1293 11. Charbonnel B, Steinberg H, Eymard E, Xu L, Thak- therapies in type 2 diabetes. Diabetes Obes Metab. kar P, Prabhu V, et al. Efficacy and safety over 2014;16(9):819–26. 26 weeks of an oral treatment strategy including sitagliptin compared with an injectable treatment 19. Li Q, Chitnis A, Hammer M, Langer J. Real-world strategy with liraglutide in patients with type 2 clinical and economic outcomes of liraglutide ver- diabetes mellitus inadequately controlled on met- sus sitagliptin in patients with type 2 diabetes formin: a randomised clinical trial. Diabetologia. mellitus in the United States. Diabetes Ther Res 2013;56(7):1503–11. Treat Educ Diabetes Relat Disord. 2014;5(2):579–90. 12. King AB, Montanya E, Pratley RE, Blonde L, 20. Lind M, Matsson PO, Linder R, Svenningsson I, Svendsen CB, Donsmark M, et al. Liraglutide Jorgensen L, Ploug UJ, et al. Clinical effectiveness of achieves A1C targets more often than sitagliptin or liraglutide vs sitagliptin on glycemic control and exenatide when added to metformin in patients body weight in patients with type 2 diabetes: a with type 2 diabetes and a baseline A1C \ 8.0%. retrospective assessment in Sweden. Diabetes Ther Endocr Pract. 2013;19(1):64–72. Res Treat Educ Diabetes Relat Disord. 2016;7(2):321–33. 13. Li M, Yang Y, Jiang D, et al. Efficacy and safety of liraglutide versus sitagliptin both in combination 21. Nyeland ME, Ploug UJ, Richards A, Garcia Alvarez L, with metformin in patients with type 2 diabetes: a Demuth D, Muthutantri A, et al. Evaluation of the systematic review and meta-analysis. Medicine. effectiveness of liraglutide and sitagliptin in type 2 2017;96(39):e8161. diabetes: a retrospective study in UK primary care. Int J Clin Pract. 2015;69(3):281–91. 14. Langer J, Hunt B, Valentine WJ. Evaluating the short-term cost-effectiveness of liraglutide versus 22. Evans M, McEwan P, O’Shea R, George L. A retro- sitagliptin in patients with type 2 diabetes failing spective, case-note survey of type 2 diabetes metformin monotherapy in the United States. patients prescribed incretin-based therapies in J Manag Care Pharm JMCP. 2013;19(3):237–46. clinical practice. Diabetes Ther Res Treat Educ Dia- betes Relat Disord. 2013;4(1):27–40. 15. Lee WC, Samyshkin Y, Langer J, Palmer JL. Long- term clinical and economic outcomes associated 23. Maciejewski ML, Bryson CL, Wang V, Perkins M, with liraglutide versus sitagliptin therapy when Liu CF. Potential bias in medication adherence added to metformin in the treatment of type 2 studies of prevalent users. Health Serv Res. diabetes: a CORE Diabetes Model analysis. J Med 2013;48(4):1468–86. Econ. 2012;15(Suppl 2):28–37. 24. D’Hoore W, Bouckaert A, Tilquin C. Practical con- 16. Ostawal A, Mocevic E, Kragh N, Xu W. Clinical siderations on the use of the Charlson comorbidity effectiveness of liraglutide in type 2 diabetes treat- index with administrative data bases. J Clin Epi- ment in the real-world setting: a systematic litera- demiol. 1996;49(12):1429–33. ture review. Diabetes Ther Res Treat Educ Diabetes Relat Disord. 2016;7(3):411–38. 25. Young BA, Lin E, Von Korff M, Simon G, Ciecha- nowski P, Ludman EJ, et al. Diabetes complications 17. Gautier JF, Martinez L, Penfornis A, Eschwege E, severity index and risk of mortality, hospitalization, Charpentier G, Huret B, et al. Effectiveness and and healthcare utilization. Am J Manag Care. persistence with liraglutide among patients with 2008;14(1):15–23. type 2 diabetes in routine clinical practice—EVI- DENCE: a prospective, 2-year follow-up, observa- 26. United States Department of Labor, Bureau of Labor tional, post-marketing study. Adv Ther. Statistics. Consumer Price Index 2014. http://www. 2015;32(9):838–53. bls.gov/cpi/. Accessed 12 April 2017. 18. Lee WC, Dekoven M, Bouchard J, Massoudi M, 27. Graubard BI, Korn EL. Predictive margins with sur- Langer J. Improved real-world glycaemic outcomes vey data. Biometrics. 1999;55(2):652–9. with liraglutide versus other incretin-based http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Diabetes Therapy Springer Journals

Real-World Clinical Effectiveness and Cost Savings of Liraglutide Versus Sitagliptin in Treating Type 2 Diabetes for 1 and 2Years

Free
15 pages
Loading next page...
 
/lp/springer_journal/real-world-clinical-effectiveness-and-cost-savings-of-liraglutide-DQr9Dcj8qX
Publisher
Springer Healthcare
Copyright
Copyright © 2018 by The Author(s)
Subject
Medicine & Public Health; Internal Medicine; Diabetes; Cardiology; Endocrinology
ISSN
1869-6953
eISSN
1869-6961
D.O.I.
10.1007/s13300-018-0432-2
Publisher site
See Article on Publisher Site

Abstract

Diabetes Ther (2018) 9:1279–1293 https://doi.org/10.1007/s13300-018-0432-2 ORIGINAL RESEARCH Real-World Clinical Effectiveness and Cost Savings of Liraglutide Versus Sitagliptin in Treating Type 2 Diabetes for 1 and 2 Years . . . . Qian Li Rahul Ganguly Michael L. Ganz Cory Gamble Tam Dang-Tan Received: February 23, 2018 / Published online: May 9, 2018 The Author(s) 2018 Results: A total of 3113 patients persistently ABSTRACT used liraglutide (N = 493) or sitagliptin (N = 2620) for C 1 year [mean age (standard Introduction: This study compared the clinical deviation, SD): 53 (8.5) vs. 56 (9.7) years; 48.3% and economic outcomes of long-term use of vs. 62.3% males; both p \ 0.05]; 911 (including liraglutide versus sitagliptin for the treatment of 113 liraglutide users) were persistent users type 2 diabetes (T2DM) in real-world practice in for C 2 years. During the 1st-year follow-up, the USA. liraglutide users (versus sitagliptin users, after Methods: We identified adult patients adjustment) experienced larger glycated hemo- (C 18 years old) with T2DM who initiated globin (HbA1c) reductions from baseline (rang- liraglutide or sitagliptin in 2010–2014 using a ing from 0.34%-point in quarter 1 to 0.21%- large claims database. Quarterly glycemic con- point in quarter 4); higher likelihoods of trol measures and annual healthcare costs were obtaining HbA1c reductions of C 1%-points assessed during the 1st and 2nd years of persis- or C 2%-points [odds ratios (ORs) range tent medication use. Their associations with 1.47–2.04]; and higher likelihoods of reaching medication use (liraglutide or sitagliptin) were HbA1c goals of \ 6.5% or \ 7% (ORs range estimated using multivariable regression models 1.51–2.12) (all p \ 0.05). Liraglutide users also adjusted for patient demographic and clinical experienced HbA1c reductions from baseline in characteristics. the 2nd-year follow-up (0.53–0.33%-point, all p \ 0.05). Although liraglutide users incurred Enhanced Digital Features To view enhanced digital higher healthcare costs than sitagliptin users features for this article, go to https://doi.org/10.6084/ m9.figshare.6133790. during the 1st-year follow-up, they had $2674 (per patient) lower all-cause medical costs (ad- Electronic supplementary material The online justed cost ratio: 0.67, p \ 0.05) and similar version of this article (https://doi.org/10.1007/s13300- total costs (all-cause and diabetes-related) in the 018-0432-2) contains supplementary material, which is available to authorized users. 2nd year. Conclusion: Long-term use of liraglutide for 1 Q. Li (&)  M. L. Ganz or 2 years was associated with better glycemic Evidera, 500 Totten Pond Road, 5th Floor, Waltham, control than using sitagliptin. Savings in medi- MA 02451, USA cal costs were realized for liraglutide users dur- e-mail: Qian.Li@Evidera.com ing the 2nd year of persistent treatment, which R. Ganguly  C. Gamble  T. Dang-Tan offset differences in pharmacy costs. Novo Nordisk Inc, 800 Scudders Mill Road, Funding: Novo Nordisk Inc. Plainsboro, NJ 08536, USA 1280 Diabetes Ther (2018) 9:1279–1293 Keywords: Clinical effectiveness; Costs; superior glycemic control, greater weight loss, HbA1c; Liraglutide; Long-term; Sitagliptin; and better treatment satisfaction after 52 weeks Type 2 diabetes as an add-on to metformin compared with another incretin-based therapy, sitagliptin [a dipeptidyl peptidase-4 (DPP-4) inhibitor] [9]. INTRODUCTION The findings were consistent with those observed at a shorter follow-up period [10–12], Diabetes is an emerging global epidemic—this while gastrointestinal problems were common chronic metabolic condition affected 1 in every 11 adverse reactions to liraglutide [13]. Based on adults worldwide in 2017 and 9.4% of the US data from the clinical trials, the annual mean population in 2015 [1, 2]. Diabetes can lead to cost per patient was estimated to be many micro- and macrovascular complications [3] $5103–$6523 (2012 USD) lower with liraglutide and accounted for 10.7% of global all-cause mor- than with sitagliptin to reach glycated hemo- tality among people 20–79 years old [2]. Diabetes globin (HbA1c) \ 7.0% with no hypoglycemia was estimated to cost $727 billion (US dollars; or weight gain, and liraglutide was shown to be USD) in 2017, responsible for 6–16.6% of total more cost-effective than sitagliptin after 5 years healthcare expenditures worldwide [2]. About half of use [14, 15]. of the global diabetes budget is spent in the US, Many studies have investigated the effec- where patients with diagnosed diabetes on average tiveness of liraglutide outside the controlled spend about $13700 on medical services annually, settings of clinical trials, with recent attention which is about 2.3 times higher than that spent by paid to the long-term use of the therapy [16]. A people without diabetes [1]. 2-year prospective study of routine clinical Type 2 diabetes mellitus (T2DM), characterized practice in France found liraglutide to be as by high blood glucose in the context of insulin effective as observed in clinical trials [17]. resistance and relative insulin deficiency, However, the effectiveness of liraglutide and accounts for approximately 90% of all diabetes sitagliptin has usually been compared in a cases [4]. Due to the progressive nature of diabetes, 6-month treatment period [18–21]. Effective- the initiation of pharmacologic agents is recom- ness for a longer duration has only been mended at the time of diagnosis of T2DM, along examined in a 1-year survey conducted in the with lifestyle modifications such as diet and UK from 2009 to 2011 [22]. exercise to achieve glycemic control [5–7]. With Data on the long-term effectiveness of metformin as the preferred 1st-line therapy, liraglutide are needed to provide healthcare additional anti-diabetic drugs are often consid- providers and patients with additional infor- ered and, as recommended by the American Dia- mation to support T2DM treatment and man- betes Association (ADA), patient’s history of agement decisions. This study investigated the atherosclerotic cardiovascular (CV) disease plays a clinical and cost-effectiveness of liraglutide key role in the selection of the add-on therapy [5]. compared with sitagliptin in the US in a real- Liraglutide, which belongs to the class of glu- world setting for 1 and 2 years of treatment. cagon-like peptide-1 (GLP-1) receptor agonists, is a commonly used anti-diabetic therapy. It was METHODS shown to reduce the risk of major adverse CV events in a recent clinical trial [8], and it is cur- Data Source rently the only agent in its class that has been approved to reduce the risk of major adverse cardiovascular events (cardiovascular death, This retrospective observational cohort study nonfatal MI or nonfatal stroke) in adults with utilized data from January 2010 through T2DM and established cardiovascular disease [5]. December 2014 from the Truven Health Mar- The efficacy and safety of liraglutide for the ketScan Commercial and Medicare Supple- treatment of T2DM has been extensively stud- mental Insurance Databases. The databases ied in clinical trials. Liraglutide demonstrated contain administrative claims and eligibility Diabetes Ther (2018) 9:1279–1293 1281 records for more than 30 million commercially of the competing therapy (i.e., sitagliptin for insured individuals (i.e., working-age adults and the liraglutide cohort and vice versa) during the their dependents) and 3 million enrollees in follow-up period were excluded. Medicare supplemental plans. The data repre- sent the healthcare experience of employees, Study Measures dependents, and retirees with primary or Medicare supplemental coverage through pri- Patient characteristics included demographics vately insured health plans. Data from the measured at the index date (age, gender, geo- Truven Health MarketScan Lab database, graphic region, health plan type, and year of which contains 32.6 million laboratory test treatment initiation), baseline HbA1c (defined results for approximately 1.9 million unique as the mean of values observed between 45 days privately insured patients, were linked to each prior to 7 days after the index date), and clinical patient’s healthcare claims. This article does not characteristics measured over the 6-month contain any studies with human participants or baseline period: any visit to an endocrinologist; animals performed by any of the authors. Charlson Comorbidity Index (CCI) [24]; the occurrence of common diabetes-related com- Sample Selection plications (i.e., retinopathy, nephropathy, neu- ropathy) and comorbidities (i.e., cardiovascular The study sample included patients with pre- diseases, depression, obesity, hypertension, and scription fills of either liraglutide (Victoza )or hyperlipidemia) identified via the International sitagliptin (Januvia or Janumet ) between 1 Classification of Diseases, Ninth Revision, January 2010 and 31 December 2014. The first Clinical Modification (ICD-9-CM) codes [25]; pharmacy claim for liraglutide or sitagliptin and use of anti-diabetic drugs. defined the index therapy and the index date. This study examined the following glycemic Patients were required to be at least 18 years old control endpoints: absolute reduction in at the index date and have continuous medical HbA1c from baseline; reductions in HbA1c of and pharmacy plan enrollment for at least C 1%-points and C 2%-points; and HbA1c goal 6 months prior to (the baseline period) and attainment (\ 6.5% and \ 7%). The glycemic 12 months after the index date. Patients were control endpoints were measured at the end of excluded if they (1) had a prescription fill of any each quarter during the 1st year of persistent GLP-1 receptor agonist or any DPP-4 inhibitor use of the index therapies, where the quarterly during the baseline period or (2) had a diagnosis follow-up HbA1c was the mean value in the of type 1 diabetes, secondary diabetes, gesta- 45-day window at the end of a quarter. Quar- tional diabetes, or pregnancy through the terly HbA1c was also measured during the 2nd baseline period until the end of continuous year of persistent use among the subgroup of health plan enrollment. Patients had to have at patients continuously enrolled in their health least one HbA1c measure around the index date plan and persistently used their index therapy (from 45 days prior to 7 days after the index without competing drugs for at least 2 years. date) and at least one HbA1c measure during Annual healthcare costs during the 1st and 2nd the 1-year follow-up (from 45 days post index year of persistent use of the index therapy were date to 45 days after the end of the follow-up examined. Annual all-cause costs were mea- period). sured by medical and pharmacy claims paid by The study focused on patients who contin- patients and insurers. Diabetes-related costs uously used the index therapy during the fol- were identified by ICD-9-CM diagnosis codes low-up periods (1 and 2 years, respectively) [23]. (250.xx) on medical claims and by anti-dia- Patients who stopped using their index treat- betic drug types from pharmacy claims. All ment for at least 90 consecutive days without a costs were adjusted to the 2016 value based on prescription fill were considered to have dis- the Consumer Price Index Medical Component continued treatment [19]. Patients with any use [26]. 1282 Diabetes Ther (2018) 9:1279–1293 Statistical Analysis use during the 1st year of follow-up. The ratio of costs associated with liraglutide versus sitaglip- tin was generated from the regression. The The study measures were compared between method of recycled predictions was used to treatment groups using Student’s t test for estimate differences in costs between liraglutide continuous variables and the chi-square test for and sitagliptin users. categorical variables. Generalized linear model (GLM) regressions, specific to the outcome dis- tribution, were used to assess the adjusted RESULTS association between the index therapy (liraglu- tide or sitagliptin) and the outcomes. The study sample included 3113 patients (493 Ordinary least squares regression was used to in the liraglutide cohort and 2620 in the sita- assess the relationship between the index ther- gliptin cohort) who had persistently used the apy and quarterly measures of absolute HbA1c index therapy for at least 12 months after reductions from baseline during the 1st and 2nd applying inclusion/exclusion criteria (Fig. 1). year of persistent use. Logistic regression was The characteristics of patients in the liraglutide used for the quarterly measures of HbA1c cohort differed from those in the sitagliptin reductions of C 1%- and C 2%-points and cohort in many aspects (Table 1). Liraglutide HbA1c goal attainment. A generalized estimat- users were younger [mean age (standard devia- ing equation (GEE) approach was incorporated tion, SD): 53 (8.5) years vs. 56 (9.7) years], less to account for multiple observations per indi- likely to be male (48.3% vs. 62.3%), more likely vidual. The regression adjusted for patient to live in the South, and more likely to be demographic characteristics, baseline HbA1c enrolled in health plans covering out-of-net- and other baseline clinical characteristics, and work services (all p \ 0.05). In addition, time-varying covariates (any use of other non- liraglutide users were more likely to visit insulin anti-diabetic medications and any use of endocrinologists, use insulin, and have a diag- insulin in each quarter). nosis of obesity or hypertension during the The coefficients on liraglutide versus sita- baseline period (all p \ 0.05). Baseline HbA1c gliptin were reported, including the direct was similar between the cohorts [8.1 (1.7) % vs. impact on absolute HbA1c reduction from 8.2 (1.7) %; p C 0.05]. The mean HbA1c reduc- baseline, and the odds ratio (OR) for HbA1c tion among liraglutide users during follow-up reduction of C 1%- and C 2%-points and ranged from 1.2 (%-point) in quarter 1 to 0.9 in HbA1c goal attainment. Based on the regression quarter 8; reductions ranged from 0.9 to 0.4 results, the differences in quarterly glycemic among sitagliptin users (Table 2). Baseline control endpoints associated with liraglutide pharmacy costs were higher in the liraglutide versus sitagliptin (i.e., marginal effects) were group than the sitagliptin group before starting estimated using the method of recycled predic- index therapy (Table 1). Medical costs were also tions, where the adjusted outcomes were pre- higher at baseline for the liraglutide group; dicted for each patient in the sample by however, differences were not statistically assuming liraglutide or sitagliptin as the index significant. therapy, respectively [27]. After adjustment for covariates, use of The annual healthcare costs during the per- liraglutide was associated with a higher reduc- sistent use of index therapy were analyzed by tion in HbA1c from baseline in all quarters GLM with log link and gamma distribution. The during the 1st year of follow-up, as reflected by regressions were adjusted for patient demo- the positive coefficients (all p \ 0.05; Table 3). graphic characteristics and the following base- Specifically, the larger reduction in HbA1c line covariates: HbA1c, other clinical associated with the use of liraglutide versus characteristics, drug use, and total healthcare sitagliptin was estimated to range from 0.34 (%- costs (all cause or diabetes related). The cost point) [95% confidence interval (CI) 0.23, 0.45] regressions for the 2nd year of persistent use of in quarter 1 to 0.21 (95% CI 0.07, 0.35) in the index therapy were also adjusted for drug Diabetes Ther (2018) 9:1279–1293 1283 With pharmacy claims of liraglutide or sitagliptin from 1/1/2010 to 12/31/2014: N = 869276 Adults (≥18 years old) with continuous enrollment in baseline and 12-month follow-up: N = 383032 Without diagnosis of type 1 diabetes, secondary diabetes, gestational diabetes, or pregnancy: N = 297977 Without pharmacy claims of GLP-1 receptor agonist or DPP-4 inhibitors in baseline: N = 225950 With HbA1c measure 45 days prior to, and up to 7 days after, the index date: N = 7084 With HbA1c measure 45 days after the index date to 45 days after 12-month follow-up: N = 5326 Persistent use of index drugs (without the competing index drug) for ≥12 months: N = 3113 Liraglutide cohort: N = 493; Sitagliptin cohort: N = 2620 Fig. 1 Sample selection of patients. DPP-4 dipeptidyl peptidase-4, GLP-1 glucagon-like peptide-1, HbA1c glycated hemoglobin quarter 4. The use of liraglutide was also asso- obtain HbA1c reduction C 1%-point and 4.4% ciated with a higher likelihood of obtaining an more (21.9% vs. 17.5%, p \ 0.05) would obtain HbA1c reduction of C 1%-points [ORs ranging HbA1c reduction C 2%-point in quarter 1 if from 1.47 to 2.37] and reductions of C 2%- they used liraglutide rather than sitagliptin, and points (ORs ranging from 1.67 to 2.04; all 10.4% more (34.5% vs. 24.1%, p \ 0.05) or p \ 0.05). Lastly, the use of liraglutide was 10.5% more (57.4% vs. 46.9%, p \ 0.05) associated with a higher likelihood of attaining patients would reach an HbA1c goal of \ 6.5% HbA1c goals: with ORs around 2 of achieving or \ 7.0% in quarter 1, respectively (all HbA1c \ 6.5% and around 1.6 of achieving p \ 0.05). The marginal effects were HbA1c \ 7% in all quarters (all p \ 0.05). stable across all quarters during the 1st year of Among the patients with at least 2 years of persistent medication use, while more variation persistent use of the index therapy, the use of was observed during the 2nd year of persistent liraglutide was associated with a higher reduc- medication use. tion in HbA1c in all quarters during the 2nd The unadjusted and adjusted predicted costs year of follow-up as well as with a higher chance for each cohort, along with the cost ratios, for of an HbA1c reduction of C 1%-points (except the 1st and 2nd years of persistent treatment are for quarter 6), of achieving HbA1c \ 6.5% (ex- presented in Table 4. The adjusted total and cept for quarter 6), and of achieving HbA1c \ medical costs in each cohort and year of follow- 7% (in quarter 5) (all p \ 0.05). The small up are also illustrated in Fig. 4. In the 1st year, sample size was insufficient to estimate the there was no difference in medical costs association between HbA1c reduction C 2%- between persistent users of liraglutide and sita- point and persistent use of index therapy during gliptin. The liraglutide cohort had higher total the 2nd year of follow-up. costs (both all cause and diabetes related) than Adjusted outcomes and the marginal effects the sitagliptin cohort due to the higher phar- of liraglutide versus sitagliptin were calculated macy costs (p \ 0.05). During the 2nd year of (Figs. 2, 3). The marginal effects on HbA1c persistent drug use, all-cause medical costs were reductions from baseline were the same as the 33% lower for liraglutide patients than sita- regression coefficients due to the model speci- gliptin patients (adjusted cost ratio 0.67, 95% CI fication. For the other endpoints, 13.4% more 0.50, 0.89), resulting in a savings of $2674 per patients (48.1% vs. 34.7%, p \ 0.05) would patient (adjusted costs: $5410 vs. $8084, 1284 Diabetes Ther (2018) 9:1279–1293 Table 1 Baseline characteristics of patients with at least 1-year persistent use of liraglutide or sitagliptin Baseline characteristics Liraglutide Sitagliptin Number of patients 493 2620 Demographics Age* 53 (8.5) 56 (9.7) Male* 238 (48.3%) 1631 (62.3%) Region* Northeast 80 (16.2%) 509 (19.4%) North Central 42 (8.5%) 291 (11.1%) South 287 (58.2%) 1017 (38.8%) West 84 (17%) 803 (30.6%) Plan types* Managed care plans without out-of-network coverage 151 (30.6%) 1128 (43.1%) Managed care plans with out-of-network coverage 297 (60.2%) 1253 (47.8%) Other/missing 45 (9.1%) 239 (9.1%) Index year* 2010–2011 305 (61.9%) 1699 (64.8%) 2012–2013 188 (38.1%) 921 (35.2%) Clinical characteristics HbA1c 8.1 (1.7) 8.2 (1.7) Any visit to endocrinologist* 103 (20.9%) 191 (7.3%) Charlson Comorbidity Index 2.4 (0.9) 2.5 [1] Diabetes-related complications 110 (22.3%) 593 (22.6%) Cardiovascular diseases 73 (14.8%) 480 (18.3%) Depression 29 (5.9%) 141 (5.4%) Obesity* 84 (17%) 237 (9%) Hypertension* 317 (64.3%) 1558 (59.5%) Hyperlipidemia 315 (63.9%) 1634 (62.4%) Drug use Any use of non-insulin anti-diabetic drugs 425 (86.2%) 2233 (85.2%) Metformin 365 (74%) 1963 (74.9%) Sulfonylurea 242 (49.1%) 1162 (44.4%) Thiazolidinediones 107 (21.7%) 608 (23.2%) Other non-insulin anti-diabetic drugs 15 (3%) 50 (1.9%) Any use of insulin* 96 (19.5%) 175 (6.7%) Diabetes Ther (2018) 9:1279–1293 1285 Table 1 continued Baseline characteristics Liraglutide Sitagliptin Healthcare costs (2016 $) All-cause medical costs 3316 (10792) 2803 (8608) All-cause pharmacy costs* 2037 (3047) 1563 (2217) All-cause total costs 5352 (11438) 4366 (9302) Diabetes-related medical costs 1474 (7297) 1279 (5342) Diabetes-related pharmacy costs* 643 (1065) 443 (713) Diabetes-related total costs 2117 (7511) 1722 (5457) Means (standard deviation) were reported for continuous variables, and counts (percentage) were reported for categorical variables HbA1c glycated hemoglobin *p \ 0.05 p \ 0.05). This saving offset the higher phar- The 6-month clinical effectiveness of macy costs observed in the liraglutide group in liraglutide in this study is similar to previously the 2nd year and resulted in $602 lower all- published data. Compared with sitagliptin, use cause total costs than in the sitagliptin group of liraglutide has been found to be associated (not statistically significant). No significant dif- with 0.31–0.41 (%-point) greater reduction of ferences between the cohorts were found for the HbA1c from baseline, an 8–15% greater chance diabetes-related medical costs and total costs to reach HbA1c \ 7%, and an 11% greater during the 2nd-year follow-up. chance to reach A1C B 6.5% after 6 months since treatment initiation [18, 19], and these were reflected by $994 (2013 US dollar) lower DISCUSSION diabetes-related medical costs during the 6 months of follow-up (all p \ 0.05) [19]. This study extends previous studies by provid- Trends of clinical effectiveness over 2 years ing integrated evidence on the long-term clini- of medication use continued to be favorable for cal and cost-effectiveness of treatment with liraglutide versus sitagliptin, as found in our liraglutide versus sitagliptin in real-world clini- study. Specifically, the difference in HbA1c cal practice in the US. Even after adjusting for reduction between the two treatments was baseline patient characteristics and the use of highest after 3 months of medication use and antidiabetic medications during follow-up, we then stabilized for the remainder of the year; no found that patients persistently using liraglu- obvious trend was observed in the effectiveness tide for 1 or 2 years experienced improved gly- in terms of likelihood to obtain high HbA1c cemic control (in terms of higher reduction of reduction or reach HbA1c goals in the 1st year HbA1c from baseline, higher likelihood to reach of follow-up. During the 2nd year of medication high reduction in HbA1c, and higher likelihood use, there was much more variation over time in to reach HbA1c target levels) compared with the clinical effectiveness of liraglutide versus those using sitagliptin. Patients in both cohorts sitagliptin, possibly because of the small sample incurred similar medical costs during the 1st size (liraglutide cohort: N = 113; sitagliptin year of medication use, while during the 2nd cohort: N = 798); nevertheless, the clinical year, liraglutide users had significantly lower effectiveness of liraglutide remained stable. all-cause medical costs, which offset their The improved glycemic control associated higher pharmacy costs than sitagliptin users. with liraglutide relative to sitagliptin led to cost 1286 Diabetes Ther (2018) 9:1279–1293 Table 2 Unadjusted glycemic control during persistent use of liraglutide or sitagliptin In the 1st-year persistent use of Quarter 1 Quarter 2 Quarter 3 Quarter 4 liraglutide or sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Number of patients (N) 321 1586 255 1307 204 1289 222 1234 Change in HbA1c from baseline - 1.2 (1.3) - 0.9 (1.5)* - 1.1 (1.4) - 0.8 (1.6)* - 1.1 (1.4) - 0.7 (1.5)* - 0.9 (1.3) - 0.7 (1.5) High reduction of HbA1c Reduction of HbA1c C 1%-point 155 (48.3%) 560 (35.3%)* 103 (40.4%) 463 (35.4%) 95 (46.6%) 407 (31.6%)* 85 (38.3%) 397 (32.2%) Reduction of HbA1c C 2%-point 68 (21.2%) 272 (17.2%) 51 (20%) 211 (16.1%) 40 (19.6%) 181 (14%)* 37 (16.7%) 184 (14.9%) HbA1c goal attainment HbA1c \ 6.5% 114 (35.5%) 396 (25%)* 106 (41.6%) 347 (26.5%)* 78 (38.2%) 332 (25.8%)* 83 (37.4%) 296 (24%)* HbA1c \ 7.0% 183 (57%) 780 (49.2%)* 149 (58.4%) 616 (47.1%)* 112 (54.9%) 598 (46.4%)* 112 (50.5%) 544 (44.1%) In the 2nd-year persistent use of Quarter 5 Quarter 6 Quarter 7 Quarter 8 liraglutide or sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Number of patients (N) 62 424 60 423 53 392 54 382 Change in HbA1c from baseline - 0.9 (1.2) - 0.5 (1.4)* - 1 (1.2) - 0.6 (1.4)* - 0.9 (1.2) - 0.5 (1.4) - 0.9 (1.2) - 0.4 (1.5)* High reduction of HbA1c Reduction of HbA1c C 1%-point 22 (35.5%) 119 (28.1%) 21 (35%) 124 (29.3%) 21 (39.6%) 107 (27.3%) 26 (48.1%) 100 (26.2%)* Reduction of HbA1c C 2%-point 12 (19.4%) 45 (10.6%)* 10 (16.7%) 38 (9%) 11 (20.8%) 43 (11%)* 9 (16.7%) 40 (10.5%) HbA1c goal attainment HbA1c \ 6.5% 23 (37.1%) 109 (25.7%) 21 (35%) 120 (28.4%) 16 (30.2%) 98 (25%) 16 (29.6%) 85 (22.3%) HbA1c \ 7.0% 33 (53.2%) 197 (46.5%) 32 (53.3%) 203 (48%) 23 (43.4%) 187 (47.7%) 23 (42.6%) 175 (45.8%) Means (standard deviation) were reported for continuous variables, and counts (percentage) were reported for categorical variables HbA1c glycated hemoglobin *p \ 0.05 Based on a subgroup of patients with persistent use of index medications for C 24 months (liraglutide cohort: N = 113; sitagliptin cohort: N = 798) Diabetes Ther (2018) 9:1279–1293 1287 Table 3 Multivariable regression results on glycemic control during persistent use of liraglutide or sitagliptin Association between index therapy and Reduction in HbA1c Reduction of Reduction of HbA1c goal HbA1c goal glycemic control from baseline HbA1c ‡ 1%-point HbA1c ‡ 2%-point attainment < 6.5% attainment < 7.0% Coefficient 95% CI Odds 95% CI Odds 95% CI Odds 95% CI Odds 95% CI ratio ratio ratio ratio Liraglutide vs. sitagliptin In the 1st-year persistent use of liraglutide or sitagliptin Quarter 1 0.34 0.23, 0.45* 2.37 1.79, 3.14* 1.73 1.17, 2.56* 1.82 1.40, 2.36* 1.68 1.29, 2.19* Quarter 2 0.28 0.15, 0.40* 1.47 1.07, 2.03* 2.04 1.30, 3.19* 2.06 1.58, 2.68* 1.66 1.26, 2.18* Quarter 3 0.31 0.18, 0.44* 2.01 1.43, 2.82* 1.67 1.10, 2.55* 1.99 1.49, 2.65* 1.70 1.27, 2.28* Quarter 4 0.21 0.07, 0.35* 1.56 1.10, 2.22* 1.70 1.04, 2.78* 2.12 1.60, 2.81* 1.51 1.13, 2.01* Liraglutide vs. sitagliptin In the 2nd-year persistent use of liraglutide or sitagliptin Quarter 5 0.53 0.32, 0.75* 1.99 1.09, 3.63* 2.11 1.21, 3.68* 1.87 1.01, 3.47* Quarter 6 0.36 0.14, 0.57* 1.32 0.64, 2.72 1.73 0.95, 3.15 1.61 0.93, 2.79 Quarter 7 0.32 0.09, 0.56* 1.98 1.04, 3.77* 2.13 1.22, 3.70* 1.07 0.58, 1.96 Quarter 8 0.33 0.09, 0.59* 3.02 1.65, 5.53* 1.92 1.02, 3.60* 1.40 0.77, 2.52 All regressions were adjusted for baseline characteristics (demographics, clinical characteristics, and drug use) and time-varying covariates (any use of other non- insulin anti-diabetic medications and any use of insulin in each quarter) CI confidence interval, HbA1c glycated hemoglobin *p \ 0.05 Based on a subgroup of patients with persistent use of index medications for C 24 months (liraglutide cohort: N = 113; sitagliptin cohort: N = 798) The regression could not generate consistent estimates because of the small HbA1c sample size 1288 Diabetes Ther (2018) 9:1279–1293 Fig. 2 Adjusted glycemic control outcomes during the 1st- varying covariates (any used of other non-insulin anti- year persistent use of liraglutide or sitagliptin. *p\0.05. diabetic medications, and any use of insulin in each Outcomes adjusted for baseline characteristics and time- quarter). HbA1c glycated hemoglobin savings, especially for patients treated for longer the annual all-cause healthcare costs were sig- than 1 year. The findings on the costs during nificantly lower for the liraglutide cohort. the 2nd year of persistent treatment may be The summary statistics of healthcare obscured because of the small sample size and resource utilization (see Supplementary Appen- large variation in healthcare costs. Thus, a post dix Table 5 for results) showed lower use of hoc analysis was conducted on annual costs inpatient and emergency room services when after the 1st year of persistent use of the index taking liraglutide. Specifically, patients treated therapy among patients who did not necessarily with liraglutide had a shorter length of hospital have persistent use of index therapy during the stay during the 1st year [all-cause, mean days 2nd year after treatment initiation. The results (SD): 0.2 (1) vs. 0.4 (2.3); diabetes-related, 0.2 from the post hoc analysis showed that after (0.8) vs. 0.3 (1.9); all p \ 0.05] and 2nd year [all 1 year of persistent treatment, medical costs cause: 0.1 (0.6) vs. 0.4 (2.3); diabetes related, 0.1 (both diabetes related and all cause) were sig- (0.6) vs. 0.3 (2.0); all p \ 0.01] of persistent nificantly lower in the liraglutide cohort (Sup- treatment than sitagliptin users. Liraglutide plementary Appendix Table 6). Further, no users also had fewer visits to the emergency difference between the cohorts was found in the department during the 2nd year of persistent annual costs for diabetes-related healthcare, and treatment. No differences between treatment Diabetes Ther (2018) 9:1279–1293 1289 Fig. 3 Adjusted glycemic control outcomes during the quarter). Based on a subgroup of patients with persistent 2nd-year persistent use of liraglutide or sitagliptin. use of index medications for C 24 months (liraglutide *p\0.05. Outcomes adjusted for baseline characteristics cohort; N = 113; sitagliptin cohort: N = 798). HbA1c and time-varying covariates (any used of other non-insulin glycated hemoglobin anti-diabetic medications, and any use of insulin in each groups were observed in other healthcare the liraglutide cohort were more likely to use resource utilization. These findings suggest that insulin than those in the sitagliptin cohort hospitalization and emergency room visits may during both the baseline and follow-up periods be the main drivers of cost savings associated (p \ 0.05; results not shown), but the need for with liraglutide, and further research is war- add-on insulin increased more quickly in the ranted to investigate this hypothesis. sitagliptin cohort. Specifically, the percentages The lower medical costs in the liraglutide of patients using insulin in the liraglutide cohort compared with sitagliptin during the cohort versus the sitagliptin cohort were 19.5% 2nd year of persistent treatment were statisti- vs. 6.7% in the 6-month baseline period and cally significant for all-cause healthcare services, 21.5% vs. 11% in the 1st year of follow-up. This but not for diabetes-related services. This may may indicate that liraglutide was more success- reflect the nonglycemic benefits of liraglutide ful in reaching the treatment goal and/or (including weight loss and reduced risks of CV patient satisfaction and thus was associated events), while sitagliptin had neutral effects [5]. with less need to intensify treatment with The finding also suggests that better glycemic insulin. control can improve a patient’s overall health. The results of this study should be inter- Our analyses adjusted for many potential preted with consideration of the limitations confounding factors including demographics, associated with studies based on administrative baseline clinical characteristics, baseline claims data. First, due to the nonexperimental healthcare costs, and the use of antidiabetic nature of the data and the study design, this agents during the follow-up period. During the study does not allow for causal inferences; 1st year of persistent medication use, more than unmeasured confounders such as prescription 80% of the patients used oral antidiabetic bias may influence the results. Our study con- medications (mostly metformin) in conjunc- sidered a wide range of covariates, including tion with liraglutide or sitagliptin. Patients in socio-demographic and clinical measures, to 1290 Diabetes Ther (2018) 9:1279–1293 Table 4 Annual healthcare costs during persistent use of liraglutide or sitagliptin Annual healthcare costs during the 1st-year persistent use of liraglutide or sitagliptin c c Unadjusted annual healthcare costs Adjusted cost ratio (95% CI) Adjusted annual healthcare costs Liraglutide Sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Difference All-cause healthcare costs (2016 $) Medical costs $6630 (11643) [2481] $7392 (19912) [2055] 1.04 (0.91, 1.18) 1 $7386 $7126 $260 Pharmacy costs $8418 (7983) [7175] $5979 (5024) [4971]* 1.31 (1.20, 1.44)* 1 $7858 $5995 $1863* Total costs $15048 (14619) [10736] $13371 (21101) [8033]* 1.16 (1.07, 1.25)* 1 $15196 $13144 $2052* Diabetes-related healthcare costs (2016 $) Medical costs $2884 (7342) [705] $3611 (13528) [580] 1.03 (0.88, 1.20) 1 $3626 $3532 $94 Pharmacy costs $5134 (3175) [4830] $3320 (1715) [3106]* 1.49 (1.37, 1.62)* 1 $4975 $3338 $1637* Total costs $8018 (7799) [6035] $6931 (13651) [3944]* 1.23 (1.13, 1.34)* 1 $8473 $6882 $1591* Annual healthcare costs during the 2nd-year persistent use of liraglutide or sitagliptin b b Unadjusted annual healthcare costs Adjusted cost ratio (95% CI) Adjusted annual healthcare costs Liraglutide Sitagliptin Liraglutide Sitagliptin Liraglutide Sitagliptin Difference All-cause healthcare costs (2016 $) Medical costs $6029 (10709) [2361] $8010 (25840) [1915] 0.67 (0.50, 0.89)* 1 $5410 $8084 - $2674* Pharmacy costs $9577 (10540) [7787] $6191 (5797) [5011]* 1.32 (1.10, 1.58)* 1 $8227 $6227 $2000* Total costs $15606 (16711) [10623] $14201 (26888) [7778] 0.96 (0.81, 1.14) 1 $13772 $14374 - $602 Diabetes-related healthcare costs (2016 $) Medical costs $2560 (8291) [664] $4226 (20850) [548] 0.75 (0.54, 1.04) 1 $2986 $4001 - $1015 Pharmacy costs $5548 (3070) [5627] $3422 (1836) [3130]* 1.49 (1.26, 1.77)* 1 $5167 $3458 $1709* Total costs $8109 (9391) [6500] $7648 (21039) [3878] 1.12 (0.94, 1.34) 1 $8373 $7463 $911 Means (standard deviation) [median] were reported for unadjusted healthcare costs *p\ 0.05 Based on a subgroup of patients with persistent use of index medications for C 24 months (liraglutide cohort: N = 113; sitagliptin cohort: N = 798) Adjusted for baseline characteristics and drug use in the 1st year of follow-up (any use of other non-insulin anti-diabetic medications, and any use of insulin) Adjusted for baseline characteristics Diabetes Ther (2018) 9:1279–1293 1291 Fig. 4 Adjusted annual healthcare costs during persistent N = 798); and outcomes adjusted for baseline character- use of liraglutide or sitagliptin. *p\0.05. Outcomes istics and drug use in the 1st year of follow-up (any use of adjusted for baseline characteristics. Based on a subgroup other non-insulin anti-diabetic medications, and any use of of patients with persistent use of index medications for C insulin) 24 months (liraglutide cohort: N = 113; sitagliptin cohort: reduce the residual confounding. Second, similar between liraglutide users and sitagliptin although clinical trials determined the efficacy users during the 1st year of persistent treatment, of liraglutide 1.2 mg separately from 1.8 mg, our but savings in medical costs were realized for study considered the overall use of liraglutide. It liraglutide users during the 2nd year of persis- is challenging to calculate the dose of liraglutide tent treatment, which offsets differences in used in clinical practice from claims data since pharmacy costs. the National Drug Codes may not completely capture the clinical practice of dosing and titration, which may also vary over the consid- ACKNOWLEDGEMENTS ered follow-up period. Third, HbA1c measure- ments were available for only a subset of T2DM patients; the availability of HbA1c is not Funding. Novo Nordisk Inc. provided the expected to be related with patient characteris- funding for the study and the article processing tics and thus should not lead to bias in our charges. All authors had full access to all of the findings. Finally, patients 65 years or older are data in this study and take complete responsi- under-represented since the MarketScan data- bility for the integrity of the data and accuracy bases used in this study contain only Medicare of the data analysis. enrollees who purchased supplemental com- mercial health insurance. However, MarketScan Editorial Assistance. Editorial support was is one of the largest commercial healthcare provided by Colleen Dumont, an employee of claims databases in the US, and our findings can Evidera, and was funded by Novo Nordisk Inc. be interpreted in a population with commercial health insurance. Authorship. All authors participated in data analysis and interpretation and contributed to the development of the manuscript. All named CONCLUSION authors meet the International Committee of Medical Journal Editors (ICMJE) criteria for In real-world clinical practice in the US, long- authorship for this manuscript, take responsi- term use of liraglutide for 1 or 2 years was bility for the integrity of the work as a whole, associated with better glycemic control com- and have given final approval to the version to pared with sitagliptin. The medical costs were be published. 1292 Diabetes Ther (2018) 9:1279–1293 Disclosures. Qian Li is employed by Evidera REFERENCES (a division of PPD), which provides consulting and other research services to pharmaceutical, 1. Centers for Disease Control and Prevention (CDC). medical device, and related organizations. In National Diabetes Statistics Report, 2017: estimates of diabetes and its burden in the United States. their salaried positions, they work with a variety https://www.cdc.gov/diabetes/pdfs/data/statistics/ of companies and organizations and are pre- national-diabetes-statistics-report.pdf. Accessed 29 cluded from receiving payment or honoraria January 2018. directly from these organizations for services 2. International Diabetes Federation (IDF). IDF Dia- rendered. Evidera received funding from Novo betes Atlas—8th Edition. http://www.diabetesatlas. Nordisk Inc. to participate in the study and for org/. Accessed 29 January 2018. the development of this manuscript. Rahul Ganguly is employed by Novo Nordisk Inc. 3. Group AC, Patel A, MacMahon S, Chalmers J, Neal B, Billot L, et al. Intensive blood glucose control Michael Ganz is employed by Evidera (a divi- and vascular outcomes in patients with type 2 dia- sion of PPD), which provides consulting and betes. N Engl J Med. 2008;358(24):2560–72. other research services to pharmaceutical, medical device, and related organizations. In 4. Centers for Disease Control and Prevention (CDC). their salaried positions, they work with a variety At a glance 2016: diabetes. https://www.cdc.gov/ chronicdisease/resources/publications/aag/pdf/ of companies and organizations and are pre- 2016/diabetes-aag.pdf. Accessed 12 April 2017. cluded from receiving payment or honoraria directly from these organizations for services 5. ADA. Pharmacologic approaches to glycemic treat- rendered. Evidera received funding from Novo ment: standards of medical care in diabetes. Dia- betes Care. 2018;41(Suppl 1):S73–85. Nordisk Inc. to participate in the study and for the development of this manuscript. Cory 6. International Diabetes Federation (IDF). Treatment Gamble is employed by Novo Nordisk Inc. Tam algorithm for people with type 2 diabetes. http:// Dang-Tan is employed by Novo Nordisk Inc. www.idf.org/treatment-algorithm-people-type-2- diabetes. Accessed 3 Feb 2016. Compliance with Ethics Guidelines. This 7. National Institute for Health and Care Excellence article does not contain any studies with (NICE). Algorithm for blood glucose lowering human participants or animals performed by therapy in adults with type 2 diabetes 2015. http://www.nice.org.uk/guidance/ng28/resources/ any of the authors. algorithmfor-blood-glucose-lowering-therapy-in- adults-withtype-2-diabetes-2185604173. Accessed 3 Data Availability. The data sets generated Feb 2016. and/or analyzed during the current study are not publicly available because of data license 8. Marso SP, Daniels GH, Brown-Frandsen K, Kris- tensen P, Mann JF, Nauck MA, et al. Liraglutide and agreement, but programming codes are avail- cardiovascular outcomes in type 2 diabetes. N Engl J able from the corresponding author upon rea- Med. 2016;375(4):311–22. sonable request. 9. Pratley R, Nauck M, Bailey T, Montanya E, Cuddihy R, Filetti S, et al. One year of liraglutide treatment Open Access. This article is distributed offers sustained and more effective glycaemic con- under the terms of the Creative Commons trol and weight reduction compared with sitaglip- Attribution-NonCommercial 4.0 International tin, both in combination with metformin, in License (http://creativecommons.org/licenses/ patients with type 2 diabetes: a randomised, paral- lel-group, open-label trial. Int J Clin Pract. by-nc/4.0/), which permits any noncommer- 2011;65(4):397–407. cial use, distribution, and reproduction in any medium, provided you give appropriate credit 10. Pratley RE, Nauck M, Bailey T, Montanya E, Cud- to the original author(s) and the source, provide dihy R, Filetti S, et al. Liraglutide versus sitagliptin for patients with type 2 diabetes who did not have a link to the Creative Commons license, and adequate glycaemic control with metformin: a indicate if changes were made. 26-week, randomised, parallel-group, open-label trial. Lancet. 2010;375(9724):1447–56. Diabetes Ther (2018) 9:1279–1293 1293 11. Charbonnel B, Steinberg H, Eymard E, Xu L, Thak- therapies in type 2 diabetes. Diabetes Obes Metab. kar P, Prabhu V, et al. Efficacy and safety over 2014;16(9):819–26. 26 weeks of an oral treatment strategy including sitagliptin compared with an injectable treatment 19. Li Q, Chitnis A, Hammer M, Langer J. Real-world strategy with liraglutide in patients with type 2 clinical and economic outcomes of liraglutide ver- diabetes mellitus inadequately controlled on met- sus sitagliptin in patients with type 2 diabetes formin: a randomised clinical trial. Diabetologia. mellitus in the United States. Diabetes Ther Res 2013;56(7):1503–11. Treat Educ Diabetes Relat Disord. 2014;5(2):579–90. 12. King AB, Montanya E, Pratley RE, Blonde L, 20. Lind M, Matsson PO, Linder R, Svenningsson I, Svendsen CB, Donsmark M, et al. Liraglutide Jorgensen L, Ploug UJ, et al. Clinical effectiveness of achieves A1C targets more often than sitagliptin or liraglutide vs sitagliptin on glycemic control and exenatide when added to metformin in patients body weight in patients with type 2 diabetes: a with type 2 diabetes and a baseline A1C \ 8.0%. retrospective assessment in Sweden. Diabetes Ther Endocr Pract. 2013;19(1):64–72. Res Treat Educ Diabetes Relat Disord. 2016;7(2):321–33. 13. Li M, Yang Y, Jiang D, et al. Efficacy and safety of liraglutide versus sitagliptin both in combination 21. Nyeland ME, Ploug UJ, Richards A, Garcia Alvarez L, with metformin in patients with type 2 diabetes: a Demuth D, Muthutantri A, et al. Evaluation of the systematic review and meta-analysis. Medicine. effectiveness of liraglutide and sitagliptin in type 2 2017;96(39):e8161. diabetes: a retrospective study in UK primary care. Int J Clin Pract. 2015;69(3):281–91. 14. Langer J, Hunt B, Valentine WJ. Evaluating the short-term cost-effectiveness of liraglutide versus 22. Evans M, McEwan P, O’Shea R, George L. A retro- sitagliptin in patients with type 2 diabetes failing spective, case-note survey of type 2 diabetes metformin monotherapy in the United States. patients prescribed incretin-based therapies in J Manag Care Pharm JMCP. 2013;19(3):237–46. clinical practice. Diabetes Ther Res Treat Educ Dia- betes Relat Disord. 2013;4(1):27–40. 15. Lee WC, Samyshkin Y, Langer J, Palmer JL. Long- term clinical and economic outcomes associated 23. Maciejewski ML, Bryson CL, Wang V, Perkins M, with liraglutide versus sitagliptin therapy when Liu CF. Potential bias in medication adherence added to metformin in the treatment of type 2 studies of prevalent users. Health Serv Res. diabetes: a CORE Diabetes Model analysis. J Med 2013;48(4):1468–86. Econ. 2012;15(Suppl 2):28–37. 24. D’Hoore W, Bouckaert A, Tilquin C. Practical con- 16. Ostawal A, Mocevic E, Kragh N, Xu W. Clinical siderations on the use of the Charlson comorbidity effectiveness of liraglutide in type 2 diabetes treat- index with administrative data bases. J Clin Epi- ment in the real-world setting: a systematic litera- demiol. 1996;49(12):1429–33. ture review. Diabetes Ther Res Treat Educ Diabetes Relat Disord. 2016;7(3):411–38. 25. Young BA, Lin E, Von Korff M, Simon G, Ciecha- nowski P, Ludman EJ, et al. Diabetes complications 17. Gautier JF, Martinez L, Penfornis A, Eschwege E, severity index and risk of mortality, hospitalization, Charpentier G, Huret B, et al. Effectiveness and and healthcare utilization. Am J Manag Care. persistence with liraglutide among patients with 2008;14(1):15–23. type 2 diabetes in routine clinical practice—EVI- DENCE: a prospective, 2-year follow-up, observa- 26. United States Department of Labor, Bureau of Labor tional, post-marketing study. Adv Ther. Statistics. Consumer Price Index 2014. http://www. 2015;32(9):838–53. bls.gov/cpi/. Accessed 12 April 2017. 18. Lee WC, Dekoven M, Bouchard J, Massoudi M, 27. Graubard BI, Korn EL. Predictive margins with sur- Langer J. Improved real-world glycaemic outcomes vey data. Biometrics. 1999;55(2):652–9. with liraglutide versus other incretin-based

Journal

Diabetes TherapySpringer Journals

Published: May 9, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off