The cost-effectiveness of dulaglutide versus liraglutide for the treatment of type 2 diabetes mellitus in Spain in patients with BMI ≥30 kg/m²

Abstract

Objective: Dulaglutide 1.5 mg once weekly is a novel glucagon-like peptide 1 (GLP-1) receptor agonist, for the treatment of type two diabetes mellitus (T2DM). The objective was to estimate the cost-effectiveness of dulaglutide once weekly vs liraglutide 1.8 mg once daily for the treatment of T2DM in Spain in patients with a BMI ≥30 kg/m².

Methods: The IMS CORE Diabetes Model (CDM) was used to estimate costs and outcomes from the perspective of Spanish National Health System, capturing relevant direct medical costs over a lifetime time horizon. Comparative safety and efficacy data were derived from direct comparison of dulaglutide 1.5 mg vs liraglutide 1.8 mg from the AWARD-6 trial in patients with a body mass index (BMI) ≥30 kg/m². All patients were assumed to remain on treatment for 2 years before switching treatment to basal insulin at a daily dose of 40 IU. One-way sensitivity analyses (OWSA) and probabilistic sensitivity analyses (PSA) were conducted to explore the sensitivity of the model to plausible variations in key parameters and uncertainty of model inputs.

Results: Under base case assumptions, dulaglutide 1.5 mg was less costly and more effective vs liraglutide 1.8 mg (total lifetime costs €108,489 vs €109,653; total QALYS 10.281 vs 10.259). OWSA demonstrated that dulaglutide 1.5 mg remained dominant given plausible variations in key input parameters. Results of the PSA were consistent with base case results.

Limitations: Primary limitations of the analysis are common to other cost-effectiveness analyses of chronic diseases like T2DM and include the extrapolation of short-term clinical data to the lifetime time horizon and uncertainty around optimum treatment durations.

Conclusion: The model found that dulaglutide 1.5 mg was more effective and less costly than liraglutide 1.8 mg for the treatment of T2DM in Spain. Findings were robust to plausible variations in inputs. Based on these results, dulaglutide may result in cost savings to the Spanish National Health System.

Keywords:

• Diabetes
• cost-effectiveness
• Spain
• dulaglutide
• liraglutide
• type 2 diabetes

Introduction

Diabetes is a metabolic disorder that is characterized by high blood glucose which can lead to serious health conditions including cardiovascular, renal, and eye diseases. The current prevalence of diabetes in the European region has been estimated at ∼60 million (9.1% of the population aged 20–79), including 23.5 million diagnosed cases¹. In 2015, the International Diabetes Federation (IDF Atlas) estimated the prevalence of type 1 and type 2 diabetes in Spain to be 10.4%, one of the highest observed rates in Western Europe¹. Other cross-sectional studies conducted in Spain have reported even higher prevalence rates (up to 13.8%), with almost half (6%) of patients being undiagnosed². The majority of diabetes patients (∼90%) suffer from type 2 diabetes mellitus¹ (T2DM).

The high prevalence and socio-economic consequences of T2DM make this disease an important public health concern worldwide. The life expectancy of patients with diabetes is reduced by up to 8 years compared to the general population, mainly due to the increased risk of cardiovascular death and stroke^3,4. In Spain, diabetes-related mortality has been estimated at 22,308 deaths in 2015¹. A Spanish study conducted in 2015 estimated the average annual direct healthcare costs for diabetic patients to be €3110.10 vs €1803.60 for non-diabetic patients⁵. Costs associated with hospitalizations and medications are a major underlying cost driver, generating ∼70% of the difference between diabetic and non-diabetic patients (cost of hospitalizations are €1306.60 and €801.60, and medication costs are €925.0 and €489.20 in diabetic and non-diabetic patients, respectively, in 2011 values)⁵.

Dulaglutide is a new GLP-1 receptor agonist administered once weekly via a disposable auto-injection pen with a fixed dosage (1.5 mg). The efficacy and safety of dulaglutide 1.5 mg has been demonstrated in the AWARD clinical trial program. Dulaglutide 1.5 mg exhibits GLP-1-mediated effects, including glucose-dependent potentiation of insulin secretion, inhibition of glucagon secretion, delay of gastric emptying, and weight loss. The combination of these effects results in a decrease in both fasting and postprandial glucose concentrations, thereby leading to improvement in overall glycemic control^6–12.

The objective of this study was to assess the cost-effectiveness of dulaglutide 1.5 mg once weekly when compared to liraglutide 1.8 mg, a once daily GLP-1 receptor agonist, for the treatment of T2DM patients with a body mass index (BMI) ≥ 30 kg/m² from the perspective of the Spanish National Health System. According to Spanish payers’ restrictions on GLP-1 receptor agonists, prescription is reimbursed only to patients with a BMI ≥30 kg/m². The analysis was based on direct head-to-head outputs from the AWARD-6 clinical trial⁸. Liraglutide was chosen as the comparator, since it is the GLP-1 receptor agonist most widely used in Spain¹³.

Methods

The IMS Core Diabetes Model (CDM), an internet-based computer simulation model, was used to project the long-term health outcomes (life expectancy, quality adjusted life expectancy, cumulative incidence of complications) and economic consequences (annual and cumulative costs per patient, cost of treatments, and complication costs), as well as the incremental cost effectiveness ratio (ICER) of the intervention of interest. The model performs real-time simulations, which take into account the baseline population characteristics, history of complications, and treatment strategies for micro-vascular, end-stage complications, and multi-factorial interventions. The disease progression is based on 17 inter-dependent Markov sub-models that simulate the complications of diabetes (i.e. cardiovascular disease, eye disease, hypoglycemia, nephropathy, neuropathy, foot ulcer, amputation, stroke, ketoacidosis, lactic acidosis, and mortality). Each sub-model incorporates time, state, and diabetes type-dependent probabilities from public sources¹⁴ (Figure 1). CDM is a robust and well established simulation model validated against results from both clinical and epidemiological studies^15–17.

Figure 1. Diagrammatic representation of the IMS CORE Diabetes Model of Palmer et al.¹⁸

The analyses were conducted from the perspective of the Spanish National Health System capturing all relevant direct medical costs (pharmacy costs plus costs of complications and management). The patient population comprised those with T2DM and reflected the AWARD-6 trial population on which the analyses were based. A targeted literature review was performed in order to identify Spanish-specific unit costs data and inflated to 2015 values, where necessary, using the Consumer Price Index (CPI) published by the Spanish National Statistics Institute¹⁹.

Given the chronic nature of T2DM, a lifetime horizon representing 40 years or death (whichever comes first) was used in the analysis. Treatment effects were based on direct comparative data from the AWARD-6 clinical trial. Costs and benefits were both discounted at 3% annually in accordance with Spanish guidelines²⁰. A series of outcomes are estimated in the CDM, although the key metric is the incremental cost per quality adjusted life year (QALY) gained.

Baseline characteristics of patients

The patient cohort with BMI ≥30 kg/m² was defined with baseline demographics, baseline complications, and physiological parameters representative of the patients enrolled in the AWARD-6 head-to-head trial where dulaglutide 1.5 mg was compared against liraglutide 1.8 mg. Other patient characteristics such as the proportion of smokers, cigarettes per day, and alcohol consumption were derived from the World Health Organization^21,22 (Table 1).

Table 1. Baseline cohort characteristics BMI ≥30 kg/m² (AWARD-6).

Characteristic^a	Mean	SD	Source
HbA1c (%)	8.05	0.8	AWARD-6 trial^8
Age (years)	55.79	9.66	AWARD-6 trial^8
Duration of diabetes (years)	6.73	5.28	AWARD-6 trial^8
Male (proportion)	0.464	NA	AWARD-6 trial^8
Ethnicity (proportion)
Caucasian	0.737		AWARD-6 trial^8
African-descent	0.051		AWARD-6 trial^8
Hispanic	0.210		AWARD-6 trial^8
Native American	0.000		AWARD-6 trial^8
Asian/Pacific Islander and other	0.002		AWARD-6 trial^8
Systolic blood pressure (mmHg)	132.22	14.85	AWARD-6 trial^8
Baseline total cholesterol (mg/dL)	181.03	41.84	AWARD-6 trial^8
Baseline HDL-C (mg/dL)	45.44	11.18	AWARD-6 trial^8
Baseline LDL-C (mg/dL)	97.78	33.46	AWARD-6 trial^8
Baseline triglycerides (mg/dL)	191.44	127.05	AWARD-6 trial^8
Body mass index (kg/m^2)	33.85	5.28	AWARD-6 trial^8
Proportion of smokers^b	0.3	NA	WHO 2013^21
Cigarettes per day^b	26	NA	WHO 2013^21
Alcohol consumption (oz/week)^b	7.56	NA	WHO 2014^22

^a Baseline cohort defined according to these characteristics in analyses comparing dulaglutide against liraglutide.

^b Values derived from World Health Organization (WHO) statistics.

Table 2 reports the modeled rates of baseline diabetes complications. The prevalence of these co-morbidities (renal, retinopathy, ophthalmological, and limb) were estimated based on the T2DM cohort defined in the NICE clinical guideline 28²³. Rates were assumed applicable due to similar profiles between cohorts, including baseline HbA1c, age, and duration of diabetes. Rates were also assumed applicable to a Spanish population (Table 2).

Table 2. Modeled rates of diabetes complications at baseline.

Complication	Mean (percentage points)	Source
Myocardial infarction	0.037	Controlled at AWARD-6^8
Angina	0.009	Controlled at AWARD-6^8
Peripheral vascular disease	0.030	Controlled at AWARD-6^8
Stroke	0.014	Controlled at AWARD-6^8
Congestive heart failure	0.005	Controlled at AWARD-6^8
Atrial fibrillation	0.023	Controlled at AWARD-6^8
Left ventricular hypertrophy	0.000	Not reported inNICE NG28^23
Microalbuminuria^a	0.006	NICE NG28^23
Gross protenuria^a	0.001	NICE NG28^23
End-stage renal disease	0.000	NICE NG28^23
Background diabetic retinopathy^b	0.159	NICE NG28^23
Proliferative diabetic retinopathy^b	0.018	NICE NG28^23
Severe vision loss	0.000	Not reported inNICE NG28
Macular edema	0.000	Not reported inNICE NG28^23
Cataract	0.000	Not reported inNICE NG28^23
Uninfected ulcer	0.000	Not reported inNICE NG28^23
Infected ulcer	0.000	Not reported inNICE NG28^23
Healed ulcer	0.000	Not reported inNICE NG28^23
Amputation	0.000	Not reported inNICE NG28^23
Neuropathy	0.065	NICE NG28^23

^a NICE NG28 reported prevalence of nephropathy of 0.7% (assumed equivalent to albuminuria); of patients with albuminuria, 9,368 of 11,723 have microalbuminuria, 2,355 of 11,723 have microalbuminuria (Parving et al.²⁴).

^b NICE NG28 reported 17.7% prevalence of retinopathy; of patients with diabetic retinopathy, 719 of 798 had non-proliferative retinopathy, 79 of 798 had proliferative diabetic retinopathy.

Treatment effects

Treatment effects associated with both treatments were drawn from the efficacy and safety population of AWARD-6 data (26-week data, background medication metformin). Treatment effects include changes in HbA1c, systolic blood pressure (SBP), total cholesterol (TC), high and low density lipoprotein cholesterol (LHD and HDL), and triglycerides (Table 3).

Table 3. Treatment effects for BMI ≥30 kg/m² (AWARD-6).

Characteristic	Mean treatment effect (Standard error, 95% CI)
	Dulaglutide 1.5 mg	Liraglutide 1.8 mg
HbA1c (%)	−1.42 (0.06, −1.53 to −1.31)	−1.37 (0.06, −1.48 to −1.25)
Systolic blood pressure (mmHg)	−2.98 (0.92)	−2.76 (0.94)
Total cholesterol (mg/dL)	−5.35 (2.37)	−0.99 (2.45)
LDL-C (mg/dL)	−4.06 (1.98)	0.07 (2.05)
HDL-C (mg/dL)	2.24 (0.5)	1.69 (0.52)
Triglycerides (mg/dL)	−15.64 (8.69)	−15.26 (9.0)
Body mass index (kg/m^2)	−1.03 (0.09, −1.22 to −0.85)	−1.36 (0.1, −1.55 to −1.17)
Nausea (event rate per 100 patient years)^a	40.00	33.00
Major hypo (event rate per 100 patient years)	0.00	0.00
Minor hypo (event rate per 100 patient years)^b	18.00	48.00

^a Input converted from provided data on proportion of patients at 12 months experiencing nausea.

^b Please note that the “minor hypo” refers to non-severe symptomatic hypo events as reported in the AWARD trials, which were based upon documented symptomatic and probable (not documented symptomatic) hypoglycemia.

In CDM, treatment effects are applied in the first year of treatment. In subsequent years, the UK Prospective Diabetes Study (UKPDS) regression approach is applied to predict the progression of model parameters over the course of the simulation. After 2 years of initial treatment, treatment groups were assumed to switch to basal insulin glargine (at a daily dose of 40 IU as per WHO daily defined dose for basal insulin). This conservative assumption of a 2-year treatment was based upon recent real world evidence from currently marketed GLP-1 receptor analogs, suggesting that most patients switch treatment after a period of ∼2 years^25,26. Impact of the switch was limited to impact on weight, and no effect was assumed on HbA1c or hypo event rate (hypo event rates were assumed to be “0” on switch, as both treatment arms receive the same therapy). In order to model impact on weight, patients switching to insulin glargine were assumed to rebound to their baseline BMI at the start of year 3 (i.e. the initial BMI effect was reversed, which is a conservative assumption). Scenario analyses explored the impact of eliminating the benefit of dulaglutide on BMI for the duration of the treatment, i.e. increasing the relative treatment benefit for liraglutide vs dulaglutide on BMI. Other parameters were assumed to follow natural progression.

Health state utilities and disutilities

Health state utilities for T2DM and its complications were derived based on a systematic literature review of reported values²⁷. Utility values express patient preferences for different health states and typically range from 0 (death) to 1 (full health without complications), although negative values for health states valued worse than death also exist²⁸. The disutilities (decrements of utility, range of values from −1 to 0) included in the model were associated with treatment-related events, e.g. nausea, hypoglycemia, and excess BMI were derived based on best available evidence (Table 4)^29,30. Nausea disutility was applied within the model based on nausea rates observed in the AWARD-6 trial (40% and 33% for patients treated with dulaglutide 1.5 mg and liraglutide 1.8 mg, respectively). The disutility for nausea (−0.04 annually) was taken from Matza et al.³⁰ and was applied only for the first 6 months of the analysis period.

Table 4. Health state and treatment related utilities.

Health state	Input value^a	Reference
T2DM without complication	0.785	Beaudet et al.^27
Myocardial infarction	−0.055	Beaudet et al.^27
Post-myocardial infarction	0.730	Assumed equal to baseline utility minus MI event^27
Ischemic heart disease	0.695	Beaudet et al.^27
Heart failure	0.677	Beaudet et al.^27
Stroke	−0.164	Beaudet et al.^27
Post-stroke	0.621	Assumed equal to baseline utility minus MI event^27
Severe vision loss	0.711	Beaudet et al.^27
Amputation	−0.280	Beaudet et al.^27
Post-amputation	0.505	Assumed equal to baseline utility minus MI event
Peripheral vascular disease	0.724	Beaudet et al.^27
Proteinuria	0.737	Beaudet et al.^27
Neuropathy	0.701	Beaudet et al.^27
Active ulcer	0.615	Beaudet et al.^27
Hemodialysis	0.621	Beaudet et al.^27
Peritoneal dialysis	0.581	Beaudet et al.^27
Renal transplant	0.762	Beaudet et al.^27
Cataract	0.769	Beaudet et al.^27
Diabetic retinopathy	0.745	Beaudet et al.^27
Macular edema	0.785	Beaudet et al.^27
Vision threatening diabetic retinopathy	0.715	Beaudet et al.^27
Excess BMI (each unit above 25 kg/m^2)^b	−0.006	Beaudet et al.^27
Major hypoglycemia event^c	−0.012	Beaudet et al.^27
Minor hypoglycemia event^c	−0.004	Beaudet et al.^27
Nausea^d	−0.04	Matza et al.^30
Injection site reaction (ISR)^e	−0.011	Boye et al.^29
Dosing frequency utility^f	0.023	Boye et al.^29

^a Within the CDM utilities associated with MI, stroke, amputation, nausea, hypoglycemia, and weight gain are input as utility decrements; all other values are input as expected utilities.

^b Disutility due to nausea and excess BMI were included in the base case. The impact of the exclusion of nausea and excess BMI disutilities was tested in two sensitivity analyses.

^c Disutility due to a hypoglycemic event was included in the base case. The impact of the exclusion of hypoglycemic event disutilities was tested in a sensitivity analysis

^d Nausea was searched for in the systematic review, but not reported, within the model, nausea is applied for 6 months only to reflect the transient nature of the condition (reported annual disutility is halved).

^e ISR disutility is applied to those patients experiencing ISR.

^f Dosing frequency utility is related to the utility gain associated with weekly vs daily/more frequent dosing.

Treatment costs

Treatment costs comprised drug cost, needle cost (where applicable), and the costs associated with self-monitoring of blood glucose (SMBG). Drug costs were estimated based on costs reported in BOTplus³¹ and calculated on the basis of the pubic price considering 7.5% discount in medications covered by the Spanish Royal Decree Law 8/2010³². Unit costs for drugs are reported in Table 5³¹ and for needle and SMBG costs in Table 6^33,34.

Table 5. Drug prices (€2014) in Spain.

Comparator	Units per pack	Pack price (€)	Dosage	Cost per day (€)	Source
Dulaglutide	1.5 mg	140.06	1.5 mg/week	5.00	BotPLUS^31
Liraglutide	3*3 ml pens, 6 mg/ml	128.20	1.8 mg/day	6.41	BotPLUS^31
Insulin Glargine (Lantus SoloStar)	5*3 ml pens, 100 U/ml	71.40	40 IU/day	1.90	BotPLUS^31

Table 6. Needle and SMBG prices (€2014) in Spain.

	Unit cost (€)	Source
Needle cost
Needle (AGO BD MICROFINE G31 5MM 100PZ)	0.065	Guisasola^33
SMBG cost
Lancet (One Touch Comfort Blood lancet)	0.11	Farmacia En-casa^34
Strips (One Touch Ultra)	0.39	Guisasola^33

Annual costs were estimated based on expected dose and SMBG requirements. The self-monitoring of blood glucose frequency used in the model followed that recommended by Owens et al.³⁵ for regimens including basal insulin, daily monitoring is recommended. Annual treatment costs are reported in Table 7. The cost of background therapies was not included in the analyses, as these therapies were assumed constant across treatment settings.

Table 7. Annual treatment costs (€) in Spain.

Drug	Drug annual cost (€)	Needles annual cost (€)	Blood glucose monitoring annual cost (€)		Total annual cost (€)
			Lancets	Strips
Dulaglutide 1.5 mg once weekly	1,826.25	0.00	0.00	0.00	1,826.25
Liraglutide 1.8 mg once daily	2,341.19	23.75	0.00	0.00	2,364.94
Rescue Treatment (Insulin glargine 40 IU once daily)	695.44	23.75	40.47	142.45	902.12

Management costs

Costs associated with the management of complications are summarized in Table 8. Complications associated with T2DM include CV disease, renal complications, acute events, ophthalmological events, and events related to neuropathy and limb complications. Costs were derived from a combination of literature and published tariffs, using diabetes-specific Spanish costs where available^36–48.

Table 8. Management costs (€annual).

Management costs	Unit cost (€2014)*	Reference
Annual statins	20.40	WHO ATC/DDD index^49; BOTPlus^31
Annual aspirin	16.96	WHO ATC/DDD index^49; BOTPlus^31
Annual angiotensin-converting-enzyme inhibitor (ACE –I)	31.34	WHO ATC/DDD index^49; BOTPlus^31
Annual eye screening	124.53	eSalud database – Hernandez-Pastor^41
Annual screening for microalbuminuria (MA)	776.99	eSalud database – Osakidetza^44
Annual screening for gross proteinuria (GRP)	204.87	eSalud database – Servicio de Salud de Castilla^39
Annual foot screening program	14.54	Ray et al.^47
Stopping ACE due to systemic embolism (SE)	55.32	Assumption based on Ray et al.^47
Myocardial infarction 1st year	8,578.70	eSalud database – Ministerios de Sanidad^43
Myocardial infarction 2nd + years	5,260.31	eSalud database – Ministerios de Sanidad^43
Angina 1st year	2,660.61	eSalud database – Schwander et al.^48
Angina 2nd + years	562.15	eSalud database –Schwander et al.^48
Congestive heart failure 1st year	4,407.30	eSalud database – Ministerios de Sanidad^43
Congestive heart failure 2nd + years	4,407.30	Assumed the same as the first year
Stroke 1st year	5,950.22	eSalud database – Pinol et al.^46
Stroke 2nd + years	2,209.20	eSalud database – Pinol et al.^46
Stroke death within 30 days	4,106.65	eSalud database – Pinol et al.^46
Peripheral vascular disease 1st year	6,903.44	eSalud database – Ministerios de Sanidad^43
Peripheral vascular disease 2nd + years	6,903.44	Assumed the same as the first year
Hemodialysis 1st year	37,903.89	eSalud database – Parra-Moncansi et al.^45
Hemodialysis 2+ years	37,903.89	Assumed the same as first year, based on Ray et al.^47
Peritoneal Dialysis 1st year	43,218.25	Ray et al.^47
Peritoneal Dialysis 2+ years	43,218.25	Assumed the same as first year, based on Ray et al.^47
Renal transplant costs 1st year	37,488.59	Ray et al.^47
Renal transplant 2+ years	11,015.33	Ray et al.^47
Major hypoglycemia (per event)	5,900.50	eSalud database – Ministerio de Sanidad^43
Minor hypoglycemia (per event)	446.639	Ray et al.^47
Ketoacidocis event	4,851.94	eSalud database – Ministerio de Sanidad^43
Lactic acid event	5,918.01	eSalud database – Ministerio de Sanidad^43
Laser treatment	130.05	eSalud database – Departament de Salut^37
Cataract operation	1,367.49	eSalud database- Consellaria de Sanidade^36
Cost following cataract operation	3,823.74	eSalud database – Ministerio de Sanidad^43
Cost blindness - year of onset	5,000.74	eSalud database – Ministerio de Sanidad^43
Cost blindness - following years	5,000.74	Assumed the same as the first year
Neuropathy 1st year	4,675.91	eSalud database – Ministerio de Sanidad^43
Neuropathy 2nd + years	4,675.91	Assumed the same as the first year
Amputation (event based)	12,535.15	eSalud database – Ministerio de Sanidad^43
Amputation prosthesis (event based)	23,27.74	eSalud database Departamento de Sanidad^38
Gangrene treatment (monthly)	11,485.81	eSalud database – Ministerio de Sanidad^43
Healed ulcer	6,736.96	eSalud database – Ministerio de Sanidad^43
Infected ulcer (monthly)	2,401.06	Assumed based on Goodall et al.^40
Standard uninfected ulcer (monthly)	1,372.20	Assumed based on Goodall et al.^40
Healed ulcer history of amputation	0.00	Assumed

* Inflated to 2014 costs using CPI indices by Instituto National de Estadistica¹⁹.

Handling uncertainty

Comprehensive sensitivity analyses were conducted to assess the robustness of model outputs to plausible changes in key model parameters. Treatment effects were varied according to the 95% confidence intervals for HbA1c and BMI for dulaglutide 1.5 mg, while the comparator inputs were kept constant. Alternate treatment durations of 1 and 3 years were assessed. Complication costs were varied by 20% along with alternate discount rates at 0% and 5%. The impact of shorter time horizons was assessed (10 and 20 years). The impact of the removal of disutilities, including BMI, nausea, and hypoglycemia, was also assessed. An alternative approach to treatment switch when HbA1c threshold exceeds 7.5% points was also applied, since exceeding this threshold indicates inadequately controlled disease²³. In addition, a sensitivity analysis was run to explore the price of liraglutide using costs equal to average daily dose costs for liraglutide 1.5 mg and 1.36 mg, based on published real world evidence of its use in practice (all other inputs including efficacy were kept constant)^25,42.

Probabilistic sensitivity analysis (PSA) was conducted to evaluate the uncertainty in model outputs associated with input parameter uncertainty. A second order Monte Carlo simulation was used, during which parameter values were sampled from fixed distributions (see Table 9 for details).

Table 9. Parameters and distributions used in PSA.

	Parameter	Distribution
Events	Probability for 1st MI, 1st stroke, angina, heart failure	For these probabilities, we sample the coefficients applied in respective risk regression equations. While the sampling distribution for the coefficients is normal (Gaussian), the resulting distribution of probabilities is beta.
Cohort characteristics	Start age, Duration of diabetes, HbA1c, SBP, Total cholesterol, HDL cholesterol, LDL cholesterol, Triglycerides, BMI	Normal distribution, gamma (triglycerides only)
Treatment effect	Change from baseline in: HbA1c, SBP, total cholesterol, HDL cholesterol, LDL cholesterol, triglycerides, BMI	Beta distribution (using standard error values, as reported in Table 3)
Costs	Statins, ACE inhibitor, aspirin, all health state costs	Log-normal
Utilities	All utility values used in the base case	Beta distribution (using standard error values, as reported in Table 4)

Results

Under base case assumptions dulaglutide 1.5 mg was found to be more effective and less costly, i.e. the dominant treatment. In comparison with liraglutide 1.8 mg, dulaglutide 1.5 mg was projected to increase quality adjusted life expectancy. Quality adjusted life expectancy was 10.281 QALYs for dulaglutide 1.5 mg and 10.259 QALYs for liraglutide 1.8 mg (an increment of 0.021 QALYs). Treatment with dulaglutide 1.5 mg resulted in a total cost reduction of €1,165, of which €993 was attributed to difference in treatment costs between dulaglutide 1.5 mg and liraglutide 1.8 mg. Amongst the diabetes-related complications tracked within the model, dulaglutide was mainly associated with lower cumulative incidence of angina, diabetes-related mortality, foot recurring ulcers, myocardial infarction (MI) events, MI related deaths, and minor hypoglycemic events. Liraglutide was mainly associated with lower cumulative incidence of congestive heart failure events (Table 10). In terms of complication incidence, the analysis did not show large differences between treatments. If a revised threshold for differences in cumulative incidence of 0.2 is used, for example, the incidence of myocardial infarction, recurring foot ulcers, angina, and diabetes related mortality favors dulaglutide 1.5 mg and neuropathy favors liraglutide 1.8 mg. Full base case results are reported in Table 10.

Table 10. Base case outputs (dulaglutide 1.5 mg – liraglutide 1.8 mg).

	Dulaglutide 1.5 mg	Liraglutide 1.8 mg	Difference
Quality-adjusted life expectancy (QALYs)	10.281	10.259	0.021
Life expectancy (years)	15.808	15.770	0.037
Lifetime direct costs (EUR)	108,489	109,653	-1,165
ICER (Incremental costs/Incremental life expectancy)	Dulaglutide is dominant
ICER (Incremental costs/Incremental QALYs)	Dulaglutide is dominant
Cumulative incidence of complications
Background retinopathy	33.653	33.655	−0.002
Proliferative retinopathy	4.634	4.732	−0.098
Macular edema	27.807	27.720	0.087
Severe vision loss	17.140	17.075	0.065
Cataract	13.934	13.808	0.126
Microalbuminuria	47.566	47.384	0.182
Gross proteinuria	19.923	19.766	0.157
End-stage renal disease	7.226	7.211	0.015
Foot ulcer, first event	47.593	47.676	−0.083
Recurring foot ulcer	94.230	94.503	−0.273
First amputation	21.013	21.020	−0.007
Amputation recurrent ulcer	8.452	8.401	0.051
Neuropathy	76.802	76.654	0.148
Congestive heart failure events	17.431	17.000	0.431
Peripheral vascular disease onset	18.052	18.132	−0.080
Angina	12.237	12.774	−0.537
Diabetes mortality	18.142	18.353	−0.211
Stroke event	7.821	7.706	0.115
Myocardial infarction death	18.520	18.743	−0.223
Myocardial infarction event	18.736	19.396	−0.660
Nausea	0.399	0.330	0.069
Minor hypoglycemic events	0.373	0.970	−0.597

Sensitivity analyses

A series of one-way sensitivity analyses indicated that dulaglutide would remain as the dominant treatment, generating both cost savings and improved quality adjusted life years, when treatment effects, economic inputs treatment duration, disutilities, time horizon, discounting, and HbA1c threshold were varied. For all of the scenarios considered, dulaglutide 1.5 mg remained the dominant strategy (Table 11).

Table 11. Sensitivity analyses (dulaglutide 1.5 mg – liraglutide 1.8 mg).

Population/analysis	Cost (Euro)			QALYs
	Dulaglutide 1.5 mg	Liraglutide 1.8 mg	Difference	Dulaglutide 1.5 mg	Liraglutide 1.8 mg	Difference	ICER (€/QALY)
Base-case	108,489	109,653	−1,165	10.281	10.259	0.021	DOMINANT
PSA	105,134	106,408	−1,273	9.948	9.918	0.030	DOMINANT
Treatment effects
BMI Upper Limit CI (−0.85)	108,353	109,653	−1,301	10.278	10.259	0.018	DOMINANT
BMI Lower Limit CI (−1.22)	108,464	109,653	−1,190	10.284	10.259	0.024	DOMINANT
HBA1c Upper Limit CI (−1.31)	109,027	109,653	−627	10.275	10.259	0.016	DOMINANT
HBA1c Lower Limit CI (−1.53)	108,082	109,653	−1,572	10.304	10.259	0.045	DOMINANT
Treatment duration
1 year	107,297	107,912	−615	10.284	10.252	0.032	DOMINANT
3 years	109,643	111,447	−1,804	10.284	10.268	0.016	DOMINANT
Economics (cost of complications)
Economics −20%	90,082	91,213	−1,130	10.281	10.259	0.021	DOMINANT
Economics +20%	126,897	128,096	−1,199	10.281	10.259	0.021	DOMINANT
Disutility exclusions
BMI disutility	108,489	109,653	−1,165	11.316	11.290	0.026	DOMINANT
Nausea disutility	108,489	109,653	−1,165	10.288	10.266	0.022	DOMINANT
Hypoglycemia disutility	108,489	109,653	−1,165	10.282	10.263	0.019	DOMINANT
Time horizon
10 years	32,739	34,114	−1,375	5.456	5.455	0.001	DOMINANT
20 years	72,578	73,991	−1,413	8.650	8.635	0.015	DOMINANT
Discounting
0% discount	183,485	184,249	−764	14.967	14.924	0.043	DOMINANT
5% discount	80,302	81,558	−1,256	8.331	8.317	0.014	DOMINANT
Treatment switch
HBA1c threshold 7.5% (inadequately controlled disease^26)	110,662	113,155	−2,493	10.283	10.264	0.019	DOMINANT
Alternate price
Alternative Lira 1.8 mg price equal to L 1.5 mg dose price	108,489	108,911	−422	10.281	10.259	0.021	DOMINANT
Alternative Lira 1.8 mg price equal to L 1.36 mg dose price	108,489	108,564	−75	10.281	10.259	0.021	DOMINANT

Probabilistic sensitivity analyses

Results of the probabilistic sensitivity analysis were in line with deterministic results, indicating that dulaglutide 1.5 mg was more effective and less costly relative to liraglutide 1.8 mg in the majority of iterations. It was observed that 86.4% of the cost-effectiveness pairs fell in the south-east quadrant of the cost-effectiveness plane. The incremental cost-effectiveness pairs for costs and QALYs gained are plotted in Figure 2. The cost-effectiveness acceptability curve (CEAC) shows that dulaglutide 1.5 mg is expected to have over 98% probability of being cost-effective at a willingness to pay threshold of €30,000 per QALY gained⁵⁰ (Figure 3).

Figure 2. Cost-effectiveness pairs for dulaglutide 1.5 mg vs liraglutide 1.8 mg.

Figure 3. Cost-effectiveness acceptability curve for dulaglutide 1.5 mg vs liraglutide 1.8 mg.

Discussion

Long-term projections suggested that dulaglutide 1.5 mg once weekly results in improvements in life-expectancy and quality adjusted life years. From a Spanish healthcare payer perspective and under base case assumptions, dulaglutide 1.5 mg was found to be dominant (more effective and less costly) when compared against liraglutide 1.8 mg for patients with BMI ≥30 kg/m² for whom GLP-1 receptor agonists are prescribed, making it a viable treatment option for patients with type 2 diabetes mellitus.

In the analysis performed against liraglutide 1.8 mg, the model inputs, and efficacy and safety data were based on robust head-to-head data from the AWARD-6 clinical trial program⁸. The results indicate that dulaglutide 1.5 mg was associated with lower treatment cost and cost of management and complications. Total QALYs generated in the dulaglutide arm were higher than those accumulated in liraglutide 1.8 mg arm (10.281 vs 10.259, respectively). These findings were robust to plausible changes in initial treatment effect (treatment effect was varied between the trial generated 95% confidence intervals in sensitivity analysis).

Exploration of uncertainty through one-way sensitivity analyses and probabilistic sensitivity analysis suggested that results were robust to plausible variations in the input parameters, with dulaglutide remaining less costly and more effective (dominant) in all cases.

There is limited evidence on studies conducted in the Spanish setting with which to compare the outcome of these analyses. Studies by Raya et al.⁵¹ and Perez et al.⁵² assessed the cost-effectiveness of liraglutide 1.2 mg and liraglutide 1.8 mg vs sitagliptin in Spain, revealing the increasing trend to assess the cost-effectiveness of various GLP-1 receptor agonists for the treatment of T2DM. While not directly comparable in terms of country setting, a previous analysis conducted by Raibouaa et al.⁵³ to estimate the cost-effectiveness of dulaglutide 1.5 mg vs liraglutide 1.8 mg in a Swedish setting resulted in the same conclusions. In this analysis, dulaglutide 1.5 mg was found to be dominant against liraglutide 1.8 mg, resulting in societal cost savings in Sweden.

This analysis was subject to limitations common to most diabetes modeling analyses. First, as with any cost-effectiveness model, in the absence of lifetime data, a number of core assumptions were applied within the CDM in order to extrapolate these short-term clinical trial data; however, these assumptions were consistent with disease progression and based on established risk equations. Emerging real world evidence on the duration of treatment and long-term treatment effects will allow for better future estimates of core model inputs⁴². However, the current analysis suggests that the dominance of dulaglutide 1.5 mg is a robust finding.

The analysis specifically focuses on patients with BMI ≥30 kg/m², which we consider to be a strength of the analysis, as physicians can only prescribe GLP-1 analogs in this specific population. Therefore, our analysis is informative to the Spanish decision-makers since the relevant patient sub-group was used to explore the cost-effectiveness.

In this analysis, patients switching to insulin glargine were assumed to rebound to their baseline BMI at the start of year 3, reversing the initial BMI effect. Scenario analyses explored the impact of eliminating dulaglutide’s treatment effect on BMI, i.e. increasing the relative treatment benefit for liraglutide vs dulaglutide on BMI. Scenario analysis using the upper and lower bounds of the BMI treatment effect were applied to the dulaglutide arm. In both scenarios, dulaglutide 1.5 mg remained the dominant treatment, in line with base case results. This suggests results are not sensitive to plausible changes in BMI treatment effect, even where the relative effect of liraglutide is increased vs dulaglutide.

Identification of precise, coherent, up-to-date, and relevant data on the many inputs required to model T2DM represents a challenge. Nevertheless, the best available data were used and tested thoroughly by means of sensitivity analyses, with a clear focus on use of local data where possible. The primary limitations of the analysis identified are common to other T2DM analyses, and computer simulation modeling remains the best option currently available to estimate the clinical and economic consequences of therapeutic interventions in the medium-to-long-term. A limitation of the analysis was that data on non-cardiovascular (CV) co-morbidities were not collected in the AWARD studies; however, non-cardiovascular events were modeled using published literature which informed the risk of these events occurring in the model⁵⁴. Associated life time cost and health benefits of dulaglutide 1.5 mg once weekly and liraglutide 1.8 mg once daily for T2D are likely to vary across regions due to variations in the prevalence between localities and population heterogeneity.

Conclusion

This analysis indicated that, from the Spanish healthcare perspective, dulaglutide 1.5 mg would be considered a dominant treatment alternative to liraglutide 1.8 mg in type 2 diabetes mellitus patients with a BMI ≥30 kg/m². The introduction of dulaglutide 1.5 mg provides additional health benefit for T2DM patients and may result in cost-savings to the Spanish payers. Although modest, the per patient cost savings and QALY gains may translate into wider cost and clinical outcome benefits for the overall Spanish healthcare system, if dulaglutide 1.5 mg is administered in clinical practice.

Transparency

Declaration of funding

The project was conducted by IMS Health and was funded by Eli Lilly and Company.

Declaration of financial/other competing interests

TD and KN are employees of Eli Lilly and Company. JL was an employee of IMS Health, who served as a paid consultant to Eli Lilly and Company during the development of this study and manuscript. DA, IC, and RA are employees of IMS Health, who served as paid consultants to Eli Lilly during the development of this study and manuscript. JME peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgments

The authors wish to acknowledge the reviewers and the editor for their detailed and helpful comments, which have contributed to the strength of this research. The authors would also like to thank Eli Lilly and Co. for supporting and funding this study.