The place of DPP-4 inhibitors in the treatment algorithm of diabetes type 2: a systematic review of cost-effectiveness studies

Transcription

1 Eur J Health Econ (2017) 18: DOI /s ORIGINAL PAPER The place of DPP-4 inhibitors in the treatment algorithm of diabetes type 2: a systematic review of cost-effectiveness studies Alexandre Baptista 1 Inês Teixeira 2 Sónia Romano 2 António Vaz Carneiro 3 Julian Perelman 4 Received: 9 October 2015 / Accepted: 30 September 2016 / Published online: 17 October 2016 Ó Springer-Verlag Berlin Heidelberg 2016 Abstract Objective To conduct a systematic review of cost-effectiveness, cost-utility, and cost-benefit studies of DPP-4 inhibitors for diabetes treatment versus other antidiabetics. Methods Three investigators searched the CRD York, Tufts CEA Registry, and MEDLINE databases through We reviewed all potentially relevant titles and abstracts, and screened full-text articles, according to inclusion criteria. We established a quality score for each study based on a 35-item list. Results A total of 295 studies were identified, of which 20 were included. The average quality score was on a 0 1 scale. All studies were performed in high- and middleincome countries, using a 3rd-party payer perspective and randomized clinical trials to measure effectiveness. Sitagliptin, saxagliptin and vildagliptin had an ICER below 25,000 /, as second-line and as add-ons to metformin, in comparison to sulfonylureas. When compared with sitagliptin, liraglutide (GLP-1 receptor agonist) had an & Alexandre Baptista alexcalaca@hotmail.com Unit of Epidemiology of the Faculty of Medicine of Lisbon, Edifício Egas Moniz, Faculdade de Medicina da Universidade de Lisboa, Av. Professor Egas Moniz, Lisbon, Portugal Centre for Health Evaluation and Research (CEFAR), National Association of Pharmacies Group, R. Marechal Saldanha, 1., Lisbon, Portugal Center for Evidence-Based Medicine (CEMBE) of the Faculty of Medicine at the University of Lisbon, Faculdade de Medicina da Universidade de Lisboa, Av. Professor Egas Moniz, Lisbon, Portugal Escola Nacional de Saúde Pública and Centro de Investigação em Saúde Pública, Universidade Nova de Lisboa, Avenida Padre Cruz, Lisbon, Portugal ICER of up to 22,724 / for the 1.2-mg dosage, and up to 32,869 / for the 1.8-mg dosage. Insulin glargine was dominant when compared with sitagliptin. Conclusions According to the WHO threshold applied to the country and year of each study, DPP-4 inhibitors were highly cost-effective as second-line, as add-ons to metformin, in comparison with sulfonylureas. More recent therapies (GLP-1 receptor agonists and insulin glargine) were highly cost-effective in comparison to DPP-4 inhibitors. These results were obtained, however, on the basis of a limited number of studies, relying on the same few clinical trials, and financed by manufacturers. Further independent research is needed to confirm these findings. Keywords Diabetes Cost-effectiveness studies DPP-4 inhibitors Systematic review JEL Classification I1 Introduction Diabetes is one of the largest global health emergencies of the twenty-first century. In addition to the 415 million adults estimated to have diabetes in 2015, there are 318 million adults with impaired glucose tolerance, which puts them at high risk of developing the disease in the future [1]. Also, according to the International Diabetes Federation (IDF), 5 million people died of diabetes in Upper middle-income and high-income countries are those where the prevalence of diabetes is highest amongst those years old [1]. Regarding the economic burden, the majority of high-income countries spend between 5 and 20 % of their total health expenditure on diabetes [1]. In France total direct costs reached 12.9 billion in In Germany the total direct cost burden

2 938 A. Baptista et al. arising from treatment has been estimated at 43.2 billion for Total direct cost in the UK has been estimated at 13.8 billion (20.2 billion using the base year rate of exchange) [2]. In 2012 the total estimated cost of diagnosed diabetes in the United States was $245 billion, including $176 billion in direct medical costs and $68 billion in indirect costs [3]. These indirect costs included increased absenteeism ($5 billion) and reduced productivity while at work ($20.8 billion) for the employed population, reduced productivity for those not in the labour force ($2.7 billion), inability to work as a result of disease-related disability ($21.6 billion), and lost productive capacity due to early mortality ($18.5 billion). On average, medical expenditures among people with diabetes are approximately 2.3 times higher than among those without diabetes [3]. As a result, diabetes has been considered to be a research priority, leading to a large increase in recent years in the number of glucose lowering medicines for treating type 2 diabetes. Dipeptidyl peptidase-4 inhibitors (DPP-4 inhibitors) are a relatively new oral hypoglycaemic drugs group. Among these, sitagliptin, vildagliptin, saxagliptin, linagliptin, and alogliptin are currently approved by the US Food and Drug Administration and the European Medicines Agency, while others are awaiting approval or are in development. In practice, the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD) clinical guideline for type 2 diabetes, suggested adding a DPP-4 inhibitor, insulin, glucagon-like peptide 1 (GLP-1) receptor agonists, sodium-glucose cotransporter 2 (SGLT2) inhibitors, thiazolidinedione, or a sulfonylurea as secondline treatment to first-line metformin [4]. Drug choice is based on patient preferences as well as various patient, disease, and drug characteristics, with the goal being to reduce glucose concentrations while minimizing side effects, especially hypoglycemia [4]. However, the guideline did not incorporate cost-effectiveness considerations regarding the newest medicines, such as DPP-4 inhibitors or SGLT2 inhibitors, which are nevertheless essential as an instrument to help resource allocation decisions. The importance is also emphasized in the current context of economic recession and pressure on tight public budgets, and considering the high epidemiological and economic burden of the disease. The objective of this study was to conduct a systematic review of cost-effectiveness, cost-utility, and cost-benefit studies of DPP-4 inhibitors versus other antidiabetics as treatment of type 2 diabetes mellitus (T2DM), and understand the implications for guidelines, policy, and further research. Methods This review followed the methodology recommended by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [5], and by the Centre for Reviews and Dissemination (CRD) of the University of York for systematic reviews of economic evaluations [6]. The literature review was based on a search for journal articles and abstracts in Medline and the CRD database from 1996 to 2015, and NHS EED, the Health Economic and Evaluations Database (HEED), and the Tufts CEA Registry to Google scholar was also searched. In addition, relevant grey literature, including models presented at recent professional meetings available solely as abstracts in conference proceedings, were also explored. Searches for economic outcomes were conducted using a variety of terms from the medical literature to describe the intervention, the comparator, the target patient population, the outcomes, and the study design. A combination of these search terms was also used for the survey. The search terms were: sitagliptin, vildagliptin, saxagliptin, linagliptin, alogliptin, DPP-4 inhibitors, cost-effectiveness, cost-utility, and cost-benefit. The search strings used for the PubMed searches are available in the Appendix. Firstly, three investigators independently reviewed all potentially relevant titles and abstracts (1st screening) and subsequently screened full-text articles (2nd screening), according to pre-established inclusion criteria. Inclusion criteria followed the PICOS approach [7]. PICOS is the acronym for population, intervention, comparator, outcomes, and study design. We followed the PRISMA flowchart in reporting study selection, as suggested by the PRISMA statement [5]. Cost-effectiveness, cost-utility studies, and cost-benefit studies should be available as a full-text publication and published in English, French, Spanish, or Portuguese languages. We excluded the incomplete economic evaluations, namely: cost consequence analyses (4 studies); patient reported outcomes (PRO) studies (1 study); the studies on sub-populations that cannot be generalized (2 studies); the health technology assessment agencies reports that were not submitted to peer-review (36 studies); studies out of scope (162), and only abstracts (66). Secondly, the three investigators used a standardized data abstraction template, as recommended by the CRD of the University of York for systematic reviews of economic evaluations [6], to independently extract data from each study, with disagreements being resolved by discussion. For each study, the information was extracted and recorded in a specific template, provided in the Appendix. Thirdly, a critical appraisal of the methodology and reporting was performed focusing on key quality issues, such as: methods of deriving the effectiveness data; measurement and valuation of resource data; measurement and valuation of health benefits (utilities); method of synthesizing the costs and effects; analysis of uncertainty; and external validity. To do so, we used the 35-item version of the BMJ checklist [7]. A score in percentage was attributed

3 The place of DPP-4 inhibitors in the treatment algorithm of diabetes type 2: a systematic 939 Flowchart of the literature search Relevant titles and abstracts identified n= st screening Relevant titles and abstracts identified after duplicated removed n= 295 Excluded n = 271 Agencies reports = 36 Sub-populations= 2 PRO studies = 1 Cost consequence study =4 Out of scope= 162 Only abstracts =66 Full copies assessed and retrieved for eligibility n= 24 2 nd screening Studies included in the review n= 20 Excluded n= 4 Out of scope = 4 Fig. 1 Flowchart of the literature search to each study, calculating the affirmative answers in the checklist. Three investigators performed all quality assessments independently, with disagreement resolved through discussion. Finally, we reported summary statistics and qualitative (descriptive) syntheses of identified cost-effectiveness, cost-utility, and cost-benefit studies in the form of summary tables. Categorical data were reported as percentages, while continuous data were reported as means with confidence intervals or standard deviations. A comparative qualitative synthesis was performed to explore relationships within and between studies. Results The literature search initially identified 295 citations (Fig. 1). Of these, 24 cost-effectiveness studies were accepted after the 1st screening, and 20 were accepted after the 2nd screening. Characteristics of the selected publications Main details of the selected studies are presented in Table 1. We organized these results according to the timing of the market introduction, that is: (1) sitagliptin versus sulfonylurea; (2) saxagliptin versus sulfonylurea; (3) saxagliptin versus insulin; (4) saxagliptin versus pioglitazone; (5) vildagliptin versus sulfonylurea; (6) liraglutide versus sitagliptin; (7) insulin glargine versus sitagliptin; and (8) GLP-1 receptor agonists versus DPP-4 inhibitors. The studies were all performed in high-income and middle-income countries (using the World Bank classification) [8]. The high-income countries were Austria, Canada, Finland, Germany, Greece, Poland, Portugal, Spain, Sweden, United Kingdom (UK), and United States (US). The middle-income countries were Argentina and Brazil. These are very recent studies with retrieval of data: one study from 2014, four studies from 2013, five studies from 2012, two studies from 2011, one study from 2010, four studies from 2009, two studies from 2008, and one

4 940 A. Baptista et al. Table 1 Characteristics of the selected publications Reference/country/authors (publication year) Year of retrieval of data Population Interventions, design and time-horizon Costs, benefits, discount rate and perspective Score [12]/6 European countries (Austria, Finland, Portugal, Spain, Scotland- UK, Sweden)/Schwarz et al. (2008) [13]/Portugal/Pereira et al. (2012) [14]/Argentina/Elgart et al. (2013) [15]/Germany/Erhardt et al. (2012) [16]/Sweden/Granstrom et al. (2012) [17]/US/Bergenheim et al. (2012) 2007 Mean age years Mean HbA1c % Sitagliptin (?Met) vs rosiglitazone or sulfonylurea (?Met) Cost-utility analysis: discrete event simulation (JADE model) Time-horizon: patient lifetime 2010 Mean HbA1c % Sitagliptin (?Met) vs sulfonylurea (?Met) Cost-utility analysis: discrete event simulation (JADE model) Time-horizon: 50 years (patient lifetime) 2009 Mean age 64 years Proportion of men 53 % Mean HbA1c 7.7 % 10.5 years 2009 Mean age years Proportion of men 52 % Mean HbA1c 7.65 % 5.4 years 2008 Mean age years Proportion of men 52 % Mean HbA1c 7.65 % 5.4 years 2009 Mean age 60 years Proportion of men 48 % Mean HbA1c NA 5.4 years Saxagliptin (?Met) vs sulfonylurea (?Met) Cost-utility analysis: discrete event simulation (Cardiff diabetes model) Time-horizon: 20 years Saxagliptin (?Met) vs sulfonylurea (?Met) Cost-utility analysis: discrete event simulation (Cardiff diabetes model) Time-horizon: 40 years (patient lifetime) Saxagliptin (?Met) vs sulfonylurea (?Met) Cost-utility analysis: discrete event simulation model Time-horizon: patient lifetime Saxagliptin (?Met) vs sulfonylurea (?Met) Cost-utility analysis: discrete event simulation (Cardiff diabetes model) Time-horizon: 5 40 years (patient lifetime) Direct cost: drugs, adverse events, (hypoglycaemia, weight), macro- and microvascular complications Benefits: Discount rate: Austria/Sweden 3 %; Spain 6 %; Finland/Portugal 5 %; Scotland 3.5 % (costs and benefits) Direct cost: drugs, macro- and microvascular complications, hypoglycaemia Benefits: Discount rate: 5 % costs and benefits (0 3 % SA) Perspective: societal (but only with direct costs) Direct cost: drugs, adverse events, macro- and microvascular complications Benefits: and LYG Discount rate: 3.5 % costs and benefits Direct cost: medicines, hypoglycaemia treatment, diabetes related complications Benefits: and LYG Discount rate: 3 % costs and benefits (0, 5, 7 and 10 % SA) Direct cost: drugs, BGSM, macro- and microvascular complications, hypoglycaemia Benefits: and LYG Discount rate: 3 % costs and benefits Direct cost: drugs, macro- and microvascular complications Benefits: Discount rate: 3 % costs and benefits

5 The place of DPP-4 inhibitors in the treatment algorithm of diabetes type 2: a systematic 941 Table 1 continued Reference/country/authors (publication year) Year of retrieval of data Population Interventions, design and time-horizon Costs, benefits, discount rate and perspective Score [18]/Portugal/Carvalho et al. (2014) [19]/Poland/Grzeszczak et al. (2012) [20]/Brazil/Nita et al. (2012) [21]/Portugal/Viriato et al. (2014) 2014 Mean age 53 years Proportion of men 64.4 % Mean HbA1c 7.0 % 0 years 2009 Mean age 52 years Proportion of men 48 % Mean HbA1c 7.9 % 1.7 years 2011 Mean age years Proportion of men 42 % Mean HbA1c 6.47 % 7.27 years 2013 Mean age 63 years Proportion of men 52 % Mean HbA1c 7.2 % 9.13 years Mean BMI kg/m 2 Saxagliptin (?Met) vs sulfonylurea (?Met) Cost-utility analysis: discrete event simulation (Cardiff diabetes model) Time-horizon: 40-year time-horizon (patient lifetime) Saxagliptin (?Met/sulfonylurea) vs NPH insulin (?Met/sulfonylurea) Cost-utility analysis: model (long-term Cardiff diabetes model) Time-horizon: 40-year (patient lifetime) Saxagliptin (?Met) vs rosiglitazone or pioglitazone (?Met) Cost-utility analysis: discrete event simulation model (UKPDS) Time-horizon: patient lifetime Vildagliptin (?Met) vs sulfonylurea (?Met) Cost-utility analysis: patient level simulation model (UKPDS) Time-horizon: 40 years (patient lifetime) Direct cost: medicines, hypoglycaemia treatment, diabetes related complications Benefits: and LYG Discount rate: 5 % costs and benefits Perspective: societal (productivity loss included for 1st year) Direct cost: drugs, macro- and microvascular complications, hypoglycaemia Benefits: Discount rate: 5 % costs and 3.5 % benefits Direct cost: hospitalization and treatment of adverse advents Benefits: and LYG Discount rate: 5 % costs Perspective: private health care system Direct cost: drugs, macro- and microvascular complications, hypoglycaemia Benefits: and LYG Discount rate: 5 % costs and benefits (0 8 % SA) [22]/US/Li et al. (2014) Mean age 54 years (liraglutide), 58 years (sitagliptin) Proportion of men 43.9 % (liraglutide), 61.8 % (sitagliptin) [23]/Sweden/Carlsson K, Persson U. (2014) 2013 Mean age 56 years Mean HbA1c 8.4 % 6 years Mean BMI 32.6 kg/m 2 Liraglutide vs sitagliptin Retrospective observational study Time-horizon: 3 years Liragutide 1.2 mg (?Met) vs sitagliptin 100 mg (?Met) Cost utility analysis: Markov model (IHE cohort model of type2 diabetes) Time-horizon: 40 years (patient lifetime) Direct cost: medicines, diabetes related complications Benefits: % points in reduction of HbA1c (liraglutide %, sitagliptin %, Difference 0.31 %) Discount rate: N/A Direct costs: medicines, diabetes related complications, hypoglycaemia Production loss: due to hypoglycaemia and diabetic complications Discount rate: 3 % costs and benefits Perspective: societal

6 942 A. Baptista et al. Table 1 continued Reference/country/authors (publication year) Year of retrieval of data Population Interventions, design and time-horizon Costs, benefits, discount rate and perspective Score [24]/Greece/Tzanetakos et al. (2014) 2013 Mean age 64.5 years Proportion of men 51.5 % Mean HbA1c 8.2 % 10.4 years Mean BMI 30.4 kg/m 2 Liragutide 1.2 mg (?Met) vs sitagliptin 100 mg (?Met) Cost-utility analysis: Markov (CORE diabetes model) Time-horizon: patient lifetime Direct cost: medicines, diabetes related complications, hypoglycaemia Benefits: Discount rate: 3.5 % costs and benefits (0 6 % SA) [25]/US/Langer et al. (2013) 2012 Mean age 55.3 years Proportion of men 52.9 % Mean HbA1c 8.4 % 6.0 years Mean BMI 32.8 kg/m 2 Liraglutide 1.2 mg and 1.8 mg (?Met) vs sitagliptin 100 mg (?Met) Cost effectiveness analysis Time-horizon: 1 year Direct cost: medicines Benefits: % patients reaching the composite endpoint (HbA1c \7.0 %, no hypoglycaemia, no weight gain) Discount rate: 0 % (1 year time-horizon) [26]/Spain/Pérez et al. (2015) [27]/Spain/Raya et al. (2013) 2012 Mean age 55.3 years Mean HbA1c 8.4 % 6.0 years Mean BMI 32.8 kg/m Mean age 55.3 years Proportion of men 52.9 % Mean HbA1c 8.4 % 6.0 years Mean BMI 32.8 kg/m 2 Liraglutide 1.8 mg (?Met) vs sitagliptin 100 mg (?Met) Cost-utility analysis: Markov (CORE diabetes model) Time-horizon: patient lifetime Liraglutide 1.2 mg (?Met) vs sitagliptin 100 mg (?Met) Cost-utility analysis: Markov (CORE diabetes model) Time-horizon: patient lifetime Direct costs: medicines, self- monitoring of blood glucose, diabetes related complications Benefits: Discount rate: 3 % costs and benefits (0 5 % SA) Direct cost: medicines, self- monitoring of blood glucose, diabetes related complications Benefits: Discount rate: 3 % costs and benefits (0 5 % SA) [28]/UK/Davies et al. (2012) 2008 Mean age 55.3 years Proportion of men 52.9 % Mean HbA1c 8.4 % 6.0 years Mean BMI 32.8 kg/m 2 Liraglutide 1.2 mg and 1.8 mg (?Met) vs sitagliptin 100 mg (?Met) Cost-utility analysis: Markov (CORE diabetes model) Time-horizon: patient lifetime Direct cost: medicines, self- monitoring of blood glucose, diabetes related complications Benefits: Discount rate: 3.5 % costs and benefits (0 6 % SA) [29]/US/Lee et al. (2012) 2011 Mean age 55.3 years Proportion of men 52.9 % Mean HbA1c 8.4 % 6.0 years Mean BMI 32.8 kg/m 2 Liraglutide 1.2 mg and 1.8 mg (?Met) vs sitagliptin 100 mg (?Met) Cost-utility analysis: Markov (CORE diabetes model) Time-horizon: 35 years Direct cost: medicines, macro- and microvascular complications, hypoglycaemia Benefits: and LYG Discount rate: 3 % costs and benefits (0 6 % SA) 0.743

7 The place of DPP-4 inhibitors in the treatment algorithm of diabetes type 2: a systematic 943 Table 1 continued Population Interventions, design and time-horizon Costs, benefits, discount rate and perspective Score Year of retrieval of data Reference/country/authors (publication year) Direct cost: medicines, diabetes related complications, hypoglycaemia Benefits: Discount rate: 5 % costs and benefits (0 % SA) Insulin glargine (?Met) vs sitagliptin (?Met) Cost utility analysis: Markov (CORE diabetes model) Time-horizon: patient lifetime 2012 Mean age 54 years Proportion of men 51 % Mean HbA1c 8.5 % 4.5 years Mean BMI 31.1 kg/m 2 [30]/Canada/Brown et al. (2014) Direct costs: healthcare costs Productivity losses, net consumption losses Benefits: Discount rate: 3 % costs and benefits (0 5 % SA) Perspective: societal GLP-1 agonists (?Met) vs DPP-4 inhibitors (?Met) vs NPH insulin (?Met) Cost-utility analysis: Markov model (IHECM-T2DM) Time-horizon: 35 years (patient lifetime) 2013 Mean age 64.7 years Proportion of men 57.5 % Mean HbA1c 7.7 % 5.6 years Mean BMI 30.9 kg/m 2 [31]/Sweden/Kiadaliri et al. (2014) N/A not applicable, Met metformin, SA sensitivity analysis, JADE Januvia diabetes economic model, UKPDS United Kingdom prospective diabetes study, IHECM-T2DM Institute for Health Economics cohort model for T2DM, BGSM blood glucose self-monitoring study from There were four studies with a societal perspective and sixteen studies with a 3rd-party payer perspective. The societal perspective is the broadest and most complex perspective, in which ideally the estimated costs reflect the true social opportunity costs, while the 3rdparty payer perspective reflects only the costs borne by that payer. Costs measurements should be fully transparent and the perspective adopted defines the scope of the analysis, with different countries having different guidelines for this issue [9]. One single study used an observational study providing real-world effectiveness data. The majority of the studies were based on one randomized controlled trial. Sitagliptin was assessed in eleven studies, saxagliptin in seven studies, vildagliptin in one study, and the three DPP-4 inhibitors (sitagliptin, saxagliptin, and vildagliptin) were also assessed jointly versus two GLP-1 receptor agonists (liraglutide and exenatide) in one study. All but one study evaluated these drugs as add-on to metformin monotherapy as second-line, that is, for those patients who did not achieve glycaemic control with metformin as first-line. One study did not specify clearly whether the drug was assessed as second-line. Finally, the majority of studies included in this review were of high quality (average score on a 0 1 scale, with a range between and 0.885). The World Health Organization (WHO) threshold for cost-effectiveness was used as Ref. [10]. This criterion defined a strategy as cost-effective when the cost per DALY averted or gained was less than 3 times the gross domestic product (GDP) per capita, and as very costeffective if it was less than the GDP per capita [11]. The threshold was calculated for each study on the basis of the year and country where the study was performed. Main results of the selected publications These findings are reported in Table 2, according to the timing of the market introduction mentioned in the Methods section. More complete details of the studies, including the characteristics of the base population, are shown in Table A.1, in the Appendix. First, we compared sitagliptin with sulfonylureas [12, 13]. The studies used the Januvia Diabetes Economic model (JADE). The JADE model is a discrete event simulation model developed to project the long-term impacts of different interventions on diabetes related outcomes [32]. It is an extension of the more well-known United Kingdom Prospective Diabetes Study (UKPDS) outcomes model, to which it adds the possibility of modelling the effect of different treatment regimes in terms of health outcomes, costs, and quality of life. The UKPDS appears as less complete than the IMS core model the most well-

8 944 A. Baptista et al. Table 2 Results of the selected publications Reference/country/authors (publication year) Intervention vs comparator Costs ( ) a Clinical outcomes/ Incremental cost-effectiveness ratio (ICER) [12]/6 European countries (Austria, Finland, Portugal, Spain, UK, Sweden)/Schwarz et al. (2008) [13]/Portugal/Pereira et al. (2012) [14]/Argentina/Elgart et al. (2013) [15]/Germany/Erhardt et al. (2012) [16]/Sweden/Granstrom et al. (2012) [17]/US/Bergenheim et al. (2012) [18]/Portugal/Carvalho et al. (2014) Sitagliptin (?Met) vs rosiglitazone or sulfonylurea (?Met) Sitagliptin (?Met) vs sulfonylurea (?Met) Saxagliptin (?Met) vs sulfonylurea (?Met) Saxagliptin (?Met) vs sulfonylurea (?Met) Saxagliptin (?Met) vs sulfonylurea (?Met) Saxagliptin (?Met) vs sulfonylurea (?Met) Saxagliptin (?Met) vs sulfonylurea (?Met) Sitagliptin vs sulfonylurea Difference 331 to 1097 across countries Sitagliptin 44,821 Sulfonylurea 44,283 Difference 538 Saxagliptin 10,883 [US$12,327.7] Sulfonylurea 9441 [US$10,694.8] Difference 1441 [US$1632.9] Saxagliptin 38,163 Sulfonylurea 36,550 Difference 1613 Saxagliptin 12,328 [116,221 SEK] Sulfonylurea 11,321 [106,727 SEK] Difference 1006 [9,484 SEK] Saxagliptin 57,503 [US$65,139] Sulfonylurea 55,056 [US$62,367] Difference 2447 [US$2772] Saxagliptin 21,959 Sulfonylurea 21,198 Difference 761 Difference across countries Sitagliptin Sulfonylurea Difference Saxagliptin 9.54 Sulfonylurea 9.32 Difference 0.22 Saxagliptin Sulfonylurea Difference 0.12 Saxagliptin Sulfonylurea Difference 0.10 Saxagliptin Sulfonylurea 8.37 Difference 2.65 Saxagliptin Sulfonylurea Difference 0.14 Austria: 20,350 / Finland: 13,737 / Portugal: 5949 / Spain: 13,440 / UK: 11,547 / Sweden: 12,219 / ,350 / across countries 11,198 / 6510 [US$7374]/ 13,931 / 9608 [91,260 SEK]/ 924 [US$1047]/ 5307 /

9 The place of DPP-4 inhibitors in the treatment algorithm of diabetes type 2: a systematic 945 Table 2 continued Reference/country/authors (publication year) Intervention vs comparator Costs ( ) a Clinical outcomes/ Incremental cost-effectiveness ratio (ICER) [19]/Poland/Grzeszczak et al. (2012) Saxagliptin (?Met/SU) vs NPH insulin (?Met/SU) [20]/Brazil/Nita et al. (2012) Saxagliptin (?Met) vs rosiglitazone or pioglitazone (?Met) [21]/Portugal/Viriato et al. (2014) Vildagliptin (?Met) vs sulfonylurea (?Met) Saxagliptin (?Met) 7765 [PLN 31,394] Insulin (?Met) 6858 [PLN 27,730] Difference 906 [PLN 3663] Saxagliptin (?SU) 7963 [PLN 32,198] Insulin (?SU) 7090 [PLN 28,668] Difference 873 [PLN 3,529] Saxagliptin 9679 [R$33,023] Pioglitazone 10,850 [R$37,019] Difference [R$3996] Vildagliptin 14,409 Sulfonylurea 13,248 Difference 1161 [22]/US/Li et al. (2014) Liraglutide vs sitagliptin Liraglutide 1403 [US$ 1589] Sitagliptin 1809 [US$ 2049] Difference -406 [US$-460] [23]/Sweden/Carlsson and Persson (2014) [24]/Greece/Tzanetakos et al. (2014) Liraglutide 1.2 mg (?Met) vs sitagliptin 100 mg (?Met) Liragutide 1.2 mg (?Met) vs sitagliptin 100 mg (?Met) Liraglutide 101, ,673 [952, ,014 SEK] Sitagliptin 95, ,563 [900, ,411 SEK] Difference [51,898-57,603 SEK] Liraglutide 39,524 Sitagliptin 36,727 Difference 2797 Saxagliptin? Met Insulin? Met Difference 0.13 Saxagliptin? SU Insulin? SU Difference 0.14 Saxagliptin Pioglitazone Difference 0.13 Vildagliptin Sulfonylurea Difference Liraglutide % Sitagliptin % Difference % %points in reduction of HbA1c Liraglutide Sitagliptin Difference Liraglutide 9.24 Sitagliptin 9.05 Difference 0.19 (?Met) 6790 [PLN 27,454]/ (?SU) 6100 [PLN 24,663]/ Dominant 9072 / NA cost per patient successfully treated to the target composite endpoint 15,780-17,060 [148, ,827 SEK]/ 15,101 /

10 946 A. Baptista et al. Table 2 continued Reference/country/authors (publication year) Intervention vs comparator Costs ( ) a Clinical outcomes/ Incremental cost-effectiveness ratio (ICER) [25]/US/Langer et al. (2013) Liraglutide 1.2 and 1.8 mg (?Met) vs sitagliptin 100 mg [26]/Spain/Pérez et al. (2015) Liraglutide 1.8 mg (?Met) vs sitagliptin 100 mg (?Met) [27]/Spain/Raya et al. (2013) Liraglutide 1.2 mg (?Met) vs sitagliptin 100 mg (?Met) [28]/UK/Davies et al. (2012) Liraglutide 1.2 and 1.8 mg (?Met) vs sitagliptin 100 mg (?Met) [29]/US/Lee et al. (2012) Liraglutide 1.2 and 1.8 mg (?Met) vs sitagliptin 100 mg (?Met) Liraglutide 1.2 mg 9 [US$10,335] Liraglutide 1.8 mg 10,377 [US$11,755] Sitagliptin 14,882 [US$16,858] Difference 1.2 mg 5759 Difference 1.8 mg 4505 Liraglutide 56,628 Sitagliptin 52,450 Difference 4177 Liraglutide 54,684 Sitagliptin 52,387 Difference 2297 Liraglutide 1.2 mg 30,222 [ 21,793] Liraglutide 1.8 mg 32,138 [ 23,175] Sitagliptin 27,667 [ 19,951] Difference (1.2 mg) 2554 [ 1842] Difference (1.8 mg) 4471 [ 3224] Liraglutide 1.2 mg 71,896 [US$81,444] Liraglutide 1.8 mg 78,127 [US$88,502] Sitagliptin 66,438 [$76,262] Difference (1.2 mg) 4575 [US$5182] Difference (1.8 mg) 11,689 [US$13,241] Liraglutide 1.2 mg 38.9 % Liraglutide 1.8 mg 49.9 % Sitagliptin 100 mg 18.6 % % of patients reaching endpoint Liraglutide 9.24 Sitagliptin 8.84 Difference 0.4 Liraglutide 9.04 Sitagliptin 8.87 Difference 0.17 Liraglutide 1.2 mg 7.52 Liraglutide 1.8 mg 7.64 Sitagliptin 100 mg 7.34 Difference (1.2 mg) 0.19 Difference (1.8 mg) 0.31 Liraglutide 1.2 mg 8.83 Liraglutide 1.8 mg 8.98 Sitagliptin 100 mg 8.62 Difference (1.2 mg) 0.20 Difference (1.8 mg) 0.36 NA, cost per patient successfully treated to the target composite endpoint 10,436 / 13,266 / (1.2 mg) 13,661 [ 9851]/ (1.8 mg) 14,513 [ 10,465]/ (1.2 mg) 22,724 [US$25,742]/ (1.8 mg) 32,869 [US$37,234]/

11 The place of DPP-4 inhibitors in the treatment algorithm of diabetes type 2: a systematic 947 Table 2 continued Reference/country/authors (publication year) Intervention vs comparator Costs ( ) a Clinical outcomes/ Incremental cost-effectiveness ratio (ICER) [30]/Canada/Brown et al. (2014) [31]/Sweden/Kiadaliri et al. (2014) Insulin glargine (?Met) vs sitagliptin (?Met) GLP-1 agonists (?Met) vs DPP-4 inhibitors (?Met) and DPP-4 inhibitors (?Met) vs NPH insulin (?Met) Insulin glargine 33,409 [$CAD 45,556] Sitagliptin 33,469 [$CAD 45,638] Difference [$CAD -1434] GLP-1 agonists 229,430 [2162,907 SEK] DPP-4 inhibitors 225,732 [2128,042 SEK] NPH Insulin 225,102 [2122,105 SEK] Difference (GLP-1 vs DPP-4 inhibitors) 3698 [34,865 SEK] Difference (DPP-4 inhibitors vs NPH Insulin 630 [5936 SEK] Insulin glargine Sitagliptin Difference GLP-1 agonists 4.75 DPP-4 inhibitors 4.65 NPH Insulin 4.50 Difference (GLP-1 vs DPP-4) 0.10 Difference (DPP-4 vs NPH Insulin) 0.15 Dominant 13,847 [$CAD - 18,882]/ GLP-1 vs DPP-4: 37,463 [353,172 SEK]/ DPP-4 vs NPH insulin 3824 [36,050 SEK]/ Met metformin, SU sulfonylurea, NA not available, PSA probabilistic sensitivity analysis, TZD thiazolidinedione Exchange rates were assessed on 15 May 2015 using the converting tool of the Portuguese National Bank ( cascambiais/pages/taxasdereferenciadiarias.aspx) a

12 948 A. Baptista et al. Table 3 Detailed descriptions of study results References Year of retrieval of data Country/Authors (pub year) Interventions Perspective Population studied Sitagliptin (?metformin) vs sulfonylurea (?metformin) Schwarz et al. (2008) [12] Pereira et al. (2012) [13] European countries (Austria, Finland, Portugal, Spain, Scotland- UK, Sweden) Schwarz et al. (2008) 2010 Portugal Pereira et al. (2012) Saxagliptin (?metformin) vs sulfonylurea (?metformin) Elgart et al Argentina (2013) [14] Elgart et al. (2013) Erhardt et al. (2012) [15] Granstrom et al. (2012) [16] Bergenheim et al. (2012) [17] Carvalho et al. (2014) [18] 2009 Germany Erhardt et al. (2012) 2008 Sweden Granstrom et al. (2012) 2009 US Bergenheim et al. (2012) 2014 Portugal Carvalho et al. (2014) Sitagliptin vs rosiglitazone or sulfonylurea all as add-ons to metformin Sitagliptin vs sulfonylurea all as add-ons to metformin Saxagliptin vs sulfonylurea all as add-ons to metformin (dose not referred) Saxagliptin vs sulfonylurea all as add-ons to metformin (dose not referred) Saxagliptin vs sulfonylurea all as add-ons to metformin Saxagliptin vs sulfonylurea all as add-ons to metformin (dose not referred) Saxagliptin vs sulfonylurea all as add-ons to metformin (dose not referred) 3rd party payer in 6 European countries Societal (but only with direct costs) 3rd party payer (Argentina social security health care system) 3rd party payer (National sick funds) Mean age from 56.7 (Finnish women) to 64.9 years (Scottish men and women) Mean HbA1c from 7.5 (Portugal and Finland) to 8.09 % (Spanish men) BMI from 26.1 (Austrian men) to 34.6 (Portuguese men) Mean HbA1c % Mean age 64 years Proportion of men 53 % Mean HbA1c 7.7 % 10.5 years Mean age years Proportion of men 52 % Mean HbA1c 7.65 % 5.4 years 3rd party payer in Sweden Mean age years Proportion of men 52 % Mean HbA1c 7.65 % 5.4 years 3rd party payer in the US Mean age 60 years Proportion of men 48 % 5.4 years Societal (Portuguese perspective) Mean age 53 years Proportion of men 64.4 % Mean HbA1c 7.0 % 0 years

13 The place of DPP-4 inhibitors in the treatment algorithm of diabetes type 2: a systematic 949 Table 3 continued References Year of retrieval of data Country/Authors (pub year) Interventions Perspective Population studied Saxagliptin (?metformin/sulfonylurea) vs insulin (?metformin/sulfonylurea) Grzeszczak et al. (2012) [19] 2009 Poland Grzeszczak et al. (2012) Saxagliptin (?metformin) vs pioglitazone (?metformin) Nita et al Brazil (2011) [20] Nita et al. (2012) Vildagliptin (?metformin) vs sulfonylurea (?metformin) Viriato et al Portugal (2014) [21] Viriato et al. (2014) Saxagliptin vs NPH insulin when used in combination with metformin or sulfonylurea Saxagliptin vs rosiglitazone or pioglitazone all as add-ons to metformin (dose not referred) Vildagliptin vs sulfonylurea all as add-ons to metformin 3rd party payer (Polish National Health Fund) Mean age 52 years Proportion of men 48 % Mean HbA1c 7.9 % 1.7 years Private health care system Mean age years Proportion of men 42 % Mean HbA1c 6.47 % 7.27 years 3rd party payer (Portuguese healthcare system perspective) Mean age 63 years Proportion of men 52 % Mean HbA1c 7.2 % 9.13 years Mean BMI kg/m 2 Liraglutide vs sitagliptin Li et al. (2014) [22] US Li et al. (2014) Liraglutide (?metformin) vs sitagliptin (?metformin) Carlsson and Persson (2014) [23] 2013 Sweden Carlsson K, Persson U. (2014) Liraglutide vs sitagliptin 3rd party payer Mean age 54 years (liraglutide), 58 years (sitagliptin) Proportion of men 43.9 % (liraglutide), 61.8 % (sitagliptin) Liraglutide 1.2 mg vs sitagliptin 100 mg all as addons to metformin Societal Mean age 56 years Mean HbA1c 8.4 % 6 years Mean BMI 32.6 kg/m 2

14 950 A. Baptista et al. Table 3 continued References Year of retrieval of data Country/Authors (pub year) Interventions Perspective Population studied Tzanetakos et al. (2014) [24] 2013 Greece Tzanetakos et al. (2014) Liraglutide 1.2 mg vs sitagliptin 100 mg all as add-ons to metformin 3rd party payer Mean age 64.5 years Proportion of men 51.5 % Mean HbA1c 8.2 % 10.4 years Mean BMI 30.4 kg/m 2 Langer et al. (2013) [25] 2012 US Langer et al. (2013) Liraglutide 1.2 and 1.8 mg vs sitagliptin 100 mg all as addons to metformin 3rd party payer Mean age 55.3 years Proportion of men 52.9 % Mean HbA1c 8.4 % 6.0 years Mean BMI 32.8 kg/ m 2 Perez et al. (2015) [26] 2012 Spain Pérez et al. (2015) Liraglutide 1.8 mg vs sitagliptin 100 mg all as addons to metformin 3rd party payer (Spanish Healthcare payer perspective) Mean age 55.3 years Mean HbA1c 8.4 % 6.0 years Mean BMI 32.8 kg/ m 2 Raya et al. (2013) [27] 2012 Spain Raya et al. (2013) Liraglutide 1.2 mg vs sitagliptin 100 mg all as addons to metformin 3rd party payer (Spanish Healthcare payer perspective) Mean age 55.3 years Proportion of men 52.9 % Mean HbA1c 8.4 % 6.0 years Mean BMI 32.8 kg/m 2 Davies et al. (2012) [28] 2008 UK Davies et al. (2012) Liraglutide 1.2 and 1.8 mg vs sitagliptin 100 mg all as addons to metformin 3rd party payer (NHS perspective) Mean age 55.3 years Proportion of men 52.9 % Mean HbA1c 8.4 % 6.0 years Mean BMI 32.8 kg/m 2 Lee et al. (2012) [29] 2011 US Lee et al. (2012) Liraglutide 1.2 and 1.8 mg vs sitagliptin 100 mg all as addons to metformin 3rd party payer in US Mean age 55.3 years Proportion of men 52.9 % Mean HbA1c 8.4 % 6.0 years Mean BMI 32.8 kg/m 2

15 The place of DPP-4 inhibitors in the treatment algorithm of diabetes type 2: a systematic 951 Table 3 continued References Year of retrieval of data Country/Authors (pub year) Interventions Perspective Population studied Insulin glargine and sitagliptin Brown et al. (2014) 2012 Canada [30] Brown et al. (2014) Insulin glargine vs sitagliptin all as add-ons to metformin 3rd party payer Mean age 54 years Proportion of men 51 % Mean HbA1c 8.5 % 4.5 years Mean BMI 31.1 kg/m 2 GLP-1 agonists (?metformin) vs DPP-4 inhibitors (?metformin) vs NPH insulin (?metformin) Kiadaliri ety al. (2014) [31] 2013 Sweden Kiadaliri et al. (2014) GLP-1 agonists vs DPP-4 inhibitors vs NPH insulin all as add-ons to metformin Societal Mean age 64.7 years Proportion of men 57.5 % Mean HbA1c 7.7 % 5.6 years Mean BMI 30.9 kg/m 2 References Data source to Effectiveness measurement Outcomes/consequences measurement/utilities source/base utilities reference Total and incremental analysis Sensitivity analysis Classification Sitagliptin (?metformin) vs sulfonylurea (?metformin) Schwarz et al. (2008) [12] Pereira et al. (2012) [13] Nauck et al. (2007) Nauck et al. (2006), Goldstein et al. (2007), Charbonnel et al. (2006) UKPDS 0.78 UKPDS UKPDS UKPDS Mean : Sitagliptin vs sulfonylurea Difference across countries Mean direct costs: sitagliptin vs sulfonylurea Difference across countries ICER (sitagliptin vs sulfonylurea) ,350 / across countries Mean : Sitagliptin Sulfonylurea Difference Mean direct costs: Sitagliptin 44,821 Sulfonylurea 44,283 Difference 538 ICER (sitagliptin vs sulfonylurea) 11,198 / Irrespective of 20 % variations in the cost and utility weights associated with diabetes-related complications, discounted ICER values remained within a narrow range ( ), as did values associated with 50 % variations in costs and utility weights associated with hypoglycemia ( ) Globally in one way sensitivity analysis and multi-way sensitivity analysis the ICERs are robust relative to the differences in utilities or in the costs of the different parameters as well as in some parameters of efficacy

16 952 A. Baptista et al. Table 3 continued References Data source to Effectiveness measurement Outcomes/consequences measurement/utilities source/base utilities reference Saxagliptin (?metformin) vs sulfonylurea (?metformin) Elgart et al. (2013) [14] RCT D1680C week trial Göke et al. (2010) and LYG Erhardt et al. (2012) [15] RCT D1680C week trial Göke et al. (2010) and LYG UKPDS Total and incremental analysis Sensitivity analysis Classification Mean discounted life expectancy: Saxagliptin years Sulfonylurea years Difference years 0.08 years Mean discounted : Saxagliptin 9.54 Sulfonylurea 9.32 Difference 0.22 Mean discounted direct costs: Saxagliptin 10,883 [$12,327] Sulfonylurea 9441 [$10,694] Difference 1441 [$1632.9] ICER (saxagliptin vs sulfonylurea) 6510 [$7374]/ Mean life expectancy: Saxagliptin years Sulfonylurea years Difference years 0.01 years Mean : Saxagliptin Sulfonylurea Difference 0.12 Mean direct costs: Saxagliptin 38,163 Sulfonylurea 36,550 Difference 1613 ICER (saxagliptin vs sulfonylurea) 13,931 / Cost-effectiveness acceptability curve illustrates a probability of less than 58 % that saxagliptin? metformin is cost-effective compared with sulfonylurea? metformin, considering a willingness to pay of 6732 [$7626]/ (GDP per capita for Argentina)/ Univariated sensitivity analyses show that a key driver of the results was the assumption that patients received combination therapy and not metformin alone from model entry. ICER of saxagliptin vs sulfonylurea fell to 2372 / (an 83.2 % reduction from base case). In a scenario where patients entered the model at age years [an increase of 25 % over the base case (57.55 years)], the ICER rose by 63.8 % to 23,175 /. Mean (HbA1c) level at baseline was also a key model driver; values both higher and lower than the base case resulted in higher ICERs: HbA1c, 7.65 % ICER 14,147 / HbA1c, 7.15 % ICER 17,840 / HbA1c, 8.15 % ICER 15,155 / ICER 10,329 in the PSA

17 The place of DPP-4 inhibitors in the treatment algorithm of diabetes type 2: a systematic 953 Table 3 continued References Data source to Effectiveness measurement Granstrom et al. (2012) [16] RCT D1680C week trial Göke et al. (2010) Bergenheim et al. (2012) [17] RCT D1680C week trial Göke et al. (2010) Carvalho et al. (2014) [18] RCT D1680C week trial Göke et al. (2010) Outcomes/consequences measurement/utilities source/base utilities reference and LYG UKPDS UKPDS and LYG 0884 Health Survey for England 2003 Total and incremental analysis Sensitivity analysis Classification Mean discounted life expectancy: Saxagliptin years Sulfonylurea years Difference years 0.0 years Mean discounted : Saxagliptin Sulfonylurea Difference 0.10 Mean discounted direct costs: Saxagliptin 12,328 [116,211 SEK] Sulfonylurea 11,321 [106,727 SEK] Difference 1006 [9484 SEK] ICER (saxagliptin vs sulfonylurea) 9608 [91,260 SEK]/ Mean : Saxagliptin Sulfonylurea 8.37 Difference 2.65 Mean direct costs: Saxagliptin 57,503 [$65,139] Sulfonylurea 55,056 [$62,367] Difference 2447 [$2772] ICER (saxagliptin vs sulfonylurea) 924 [$ 1047]/ Mean life expectancy: Saxagliptin years Sulfonylurea years Difference years 0.01 years Mean : Saxagliptin Sulfonylurea Difference 0.14 Mean costs: Saxagliptin 21,959 Sulfonylurea 21,198 Difference 761 ICER (saxagliptin vs sulfonylurea) 5307 / Weight, and its associated HRQoL decrement and impact on diabetes-related events, was an important parameter in the model. The highest cost per 29,320 [276,408 SEK] is obtained under the extreme assumption that the HRQoL decrement (for the first and subsequent years) per unit BMI gain is reduced by 75 % (0.0035). As expected, when this HRQoL decrement is reduced by half, the resulting cost per is lower, 17,652 [166,408 SEK]. Raising the HbA1c threshold when insulin is initiated to 8.0 %, increases the cost per to 18,156 [171,162 SEK]. The overall cost per in the PSA was: 11,578 [109,152 SEK] When evaluating sensitivity around the subgroup of hypoglycaemic events accruing cost, attributing a cost value only to events requiring medical assistance did not significantly change the outcome. PSA results showed a mean incremental costeffectiveness ratio (ICER) of (- 1748) [-$1980] with saxagliptin plus metformin appearing to be dominant. In one-way sensitivity analysis the value of HbA1c is a variable with a considerable impact in the results of the study. In PSA the probability of being cost-effective is 84.5 % to a WTP of 20,000 per and 87 % for a threshold of 30,000 per

18 954 A. Baptista et al. Table 3 continued References Data source to Effectiveness measurement Outcomes/consequences measurement/utilities source/base utilities reference Total and incremental analysis Sensitivity analysis Classification Saxagliptin (?metformin/sulfonylurea) vs insulin (?metformin/sulfonylurea) Grzeszczak et al. (2012) [19] RCT Jadzinsky et al. (2009) and RCT Nauck et al. (2007) UKPDS Scenario 1 (metformin? insulin vs metformin? saxagliptin) Mean life expectancy: Saxagliptin? metformin years Insulin? metformin years Difference years 0.00 years Mean : Saxagliptin? metformin Insulin? metformin Difference 0.13 Mean direct costs: Saxagliptin? metformin 7765 [PLN 31,394] Insulin? metformin 6858 [PLN 27,730] Difference 906 [PLN 3663] ICER saxagliptin vs insulin (?metformin) 6790 [PLN 27,454]/ Scenario 2 (SU?saxagliptin vs SU? insulin) Mean life expectancy: Saxagliptin? SU years Insulin? SU years Difference years 0.00 years Mean : Saxagliptin? SU Insulin? SU Difference 0.14 Mean direct costs: Saxagliptin? SU 7963 [PLN 32,198] Insulin? SU 7090 [PLN 28,668] Difference 873 [PLN 3529] ICER saxagliptin vs insulin (?SU) 6100 [PLN 24,663]/ The results were found to be sensitive to some of the basic model assumptions, although the ICER remained below 12,366 [PLN 50,000] per gained in all cases 0.686

19 The place of DPP-4 inhibitors in the treatment algorithm of diabetes type 2: a systematic 955 Table 3 continued References Data source to Effectiveness measurement Outcomes/consequences measurement/utilities source/base utilities reference Saxagliptin (?metformin) vs pioglitazone (?metformin) Nita et al. (2011) [20] DIAPS 79 Study Group (2010) and LYG UKPDS 0885 UKPDS Vildagliptin (?metformin) vs sulfonylurea (?metformin) Viriato et al. (2014) [21] Ferrannini et al. (2009) RCT and LYG UKPDS 0.78 Clarke et al. (2002) Total and incremental analysis Sensitivity analysis Classification Mean life expectancy: Saxagliptin years Pioglitazone years Difference years 0.01 Mean : Saxagliptin Pioglitazone Difference 0.13 Mean direct costs: Saxagliptin 9679 [R$33,023] Pioglitazone 10,850 [R$37,019] Difference [R$-3996] ICER (saxagliptin vs pioglitazone) Dominant In the univariate sensitivity analysis, saxagliptin remained dominant compared with TZDs after a variation of ±15 % on all selected parameters. In PSA, adding saxagliptin to the metformin therapy was dominant in 62.1 % of all scenarios versus the addition of pioglitazone. Only in 2.2 % of the simulations did saxagliptin show less effectiveness and higher costs Mean life expectancy: Vildagliptin years Sulfonylurea years Difference years years Mean : Vildagliptin Sulfonylurea Difference Mean direct costs: Vildagliptin 14,409 Sulfonylurea 13,248 Difference 1161 ICER (vildagliptin vs sulfonylurea) 9072 / Univariate analyses showed that ICER values were robust and ranged from 4195 to 16,052 per when different parameters were varied. The PSA of 100 simulated interactions suggested that for a WTP of 30,000 per treatment with metformin plus vildagliptin had a 79 % probability of being costeffective compared with metformin plus sulfonylurea 0.771

20 956 A. Baptista et al. Table 3 continued References Data source to Effectiveness measurement Outcomes/consequences measurement/utilities source/base utilities reference Total and incremental analysis Sensitivity analysis Classification Liraglutide vs sitagliptin Li et al. (2014) [22] Observational longitudinal retrospective study Liraglutide (?metformin) vs sitagliptin (?metformin) Carlsson and Persson (2014) [23] RCT Pratley et al. (2010) (ClinicalTrials.gov Identifier:NCT ) HbA1c change % points Mean life expectancy: N/A (6 months follow-up, without extrapolation) % points in reduction of HbA1c Liraglutide % Sitagliptin % Difference 0.31 % Liraglutide 1403 [$1589] Sitagliptin 1809 [$2049] Difference -406 [$-460] Liraglutide provides greater benefits at lower cost, when compared with sitagliptin (dominance) HbA1c change % points UKPDS Men, non smoker Mean : Liraglutide Sitagliptin Difference Mean direct costs: Liraglutide 101, ,673 [952, ,014 SEK] Sitagliptin 95, ,563 [900, ,411 SEK] Difference [51,898 57,603 SEK] ICER (liraglutide vs sitagliptin) 15,780 17,060 [148, ,827 SEK]/ Not applicable Ninety per cent of the predicted cost increments and increments for the comparison of liraglutide 1.2 mg vs SU were within the intervals ,701.SEK 14, ,633 SEK and , respectively. The probability that liraglutide would be considered cost-effective compared with sitagliptin was 89 % at a WTP per of 53, ,000 SEK 0.800

21 The place of DPP-4 inhibitors in the treatment algorithm of diabetes type 2: a systematic 957 Table 3 continued References Data source to Effectiveness measurement Tzanetakos et al. (2014) [24] Langer et al. (2013) [25] RCT Pratley et al. (2010) (ClinicalTrials.gov Identifier:NCT ) Outcomes/consequences measurement/utilities source/base utilities reference Total and incremental analysis Sensitivity analysis Classification UKPDS Mean life expectancy: Liraglutide years Sitagliptin years Difference 0.13 years Mean : Liraglutide 9.24 Sitagliptin 9.05 Difference 0.19 Mean direct costs: Liraglutide 39,524 Sitagliptin 36,727 Difference 2797 ICER (liraglutide vs sitagliptin) 15,101 / HbA1c % of patients reaching endpoint Liraglutide 1.2 mg 38.9% Liraglutide 1.8 mg 49.9% Sitagliptin 100 mg 18.6% Difference (liraglutide 1.2 mg) (20.3%) Difference (liraglutide 1.8 mg) (31.3%) Mean direct costs (SD): Liraglutide 1.2 mg 9 [US$10,335] Liraglutide 1.8 mg 10,377 [US$11,755] Sitagliptin 14,882 [US$16,858] Simulation results were quite sensitive to the gradual shortening of model timehorizon resulting in an increase of base case ICER for liraglutide by more than 600 % at 5 years simulation. Simulation results were quite sensitive to patients HbA1c values, underlining, as such, the importance of this biochemical parameter to health and cost outcomes of model analysis Variation in cost assumptions by ±20 % and variation in clinical inputs parameters within the range of the 95 % CI did not change the findings that the cost of reaching the composite endpoint (cost of control) with liraglutide 1.2 mg and 1.8 mg was lower than the cost with sitagliptin after 52 weeks of treatment