DIABETICMedicine

DOI: 10.1111/dme.12772

Research: Health Economics

The cost-effectiveness of dapagliflozin versus
sulfonylurea as an add-on to metformin in the treatment
of Type 2 diabetes mellitus

M. Charokopou1, P. McEwan2,3, S. Lister4, L. Callan5, K. Bergenheim6, K. Tolley7, R. Postema8,

R. Townsend9 and M. Roudaut8
1
Pharmerit International, Rotterdam, The Netherlands, 2Centre for Health Economics, Swansea University, 3HEOR, Monmouth, 4Bristol-Myers Squibb
Pharmaceuticals, Uxbridge, 5AstraZeneca UK, Luton, UK, 6AstraZeneca, Mo €lndal, Sweden, 7Tolley Health Economics, Buxton, UK, 8Bristol-Myers Squibb, Rueil-
Malmaison, France and 9AstraZeneca, Brussels, Belgium

Accepted 25 March 2015

Abstract
Aims To assess the cost-effectiveness of dapagliflozin, a sodium–glucose co-transporter–2 (SGLT–2) inhibitor,
compared with a sulfonylurea, when added to metformin for treatment of UK people with Type 2 diabetes mellitus
inadequately controlled on metformin alone.
Methods Clinical inputs sourced from a head-to-head randomized controlled trial (RCT) informed the Cardiff diabetes
decision model. Risk equations developed from the United Kingdom Prospective Diabetes Study (UKPDS) were used in
conjunction with the clinical inputs to predict disease progression and the incidence of micro- and macrovascular
complications over a lifetime horizon. Cost and utility data were generated to present the incremental cost-effectiveness
ratio (ICER) for both treatment arms, and sensitivity and scenario analyses were conducted to assess the impact of
uncertainty on the final model results.
Results The dapagliflozin treatment arm was associated with a mean incremental benefit of 0.467 quality-adjusted life
years (QALYs) [95% confidence interval (CI): 0.420; 0.665], with an incremental cost of £1246 (95% CI: £613; £1637).
This resulted in an ICER point estimate of £2671 per QALY gained. Incremental costs were shown to be insensitive to
parameter variation, with only treatment-related weight change having a significant impact on the incremental QALYs.
Probabilistic sensitivity analysis determined that dapagliflozin had a 100% probability of being cost-effective at a
willingness-to-pay threshold of £20 000 per QALY.
Conclusions Dapagliflozin in combination with metformin was shown to be a cost-effective treatment option compared
with sulfonylurea from a UK healthcare perspective for people with Type 2 diabetes mellitus who are inadequately
controlled on metformin monotherapy.
Diabet. Med. 32, 890–898 (2015)

patient’s lifetime, and provide an important source of

Introduction
Decision modelling is becoming an increasingly important
factor in the economic analyses of healthcare interventions
because it allows for the costs and benefits of treatments to
be compared over a significant period [1]. Through the
extrapolation of clinical data to estimated ‘real-world’
outcomes, economic models can offer insights into the
impact that safety and efficacy parameters can have over a
Correspondence to: Klas Bergenheim.
E-mail: klas.bergenheim@astrazeneca.com
information to healthcare decision makers [2].
Type 2 diabetes mellitus is a chronic progressive condition
estimated to account for 7–12% of the United Kingdom (UK)
National Health Service (NHS) expenditure [3]. However,
around 80% of these costs arise from the treatment of
avoidable micro- and macrovascular complications, which
can potentially be reduced through effective management of
the disease [4]. Economic analysis assessing the available
Type 2 diabetes mellitus therapeutics is critical to ensuring
that limited resources are used effectively.
The primary treatment goal of Type 2 diabetes mellitus is
to reduce HbA1c levels to below 48 mmol/mol (6.5%) (or

ª 2015 The Authors.

890 Diabetic Medicine ª 2015 Diabetes UK
 Research article
What’s new?
• Dapagliflozin is a selective sodium–glucose co–trans-
porter–2 (SGLT-2) inhibitor licensed in the European
Union (EU) for use in people with Type 2 diabetes
mellitus when diet and exercise plus metformin fail to
achieve glycaemic control.
• Sulfonylureas are commonly prescribed as second-line
agents and therefore are relevant comparators for
dapagliflozin.
• This is the first health economic model to assess the
cost-effectiveness of dapagliflozin versus sulfonylureas
as an add-on to metformin in the treatment of Type 2
diabetes mellitus.
• Dapagliflozin in combination with metformin was
shown to be a cost-effective treatment option compared
with sulfonylurea added on to metformin from a United
Kingdom (UK) healthcare perspective.
below 59 mmol/mol, 7.5%, when patients are at an
increased risk of hypoglycaemia) [3]. Type 2 diabetes mell-
itus represents an unmet medical need because 30–40% of
patients fail to reach this target [5,6]. Furthermore, addi-
tional complications are often associated with over three-
quarters of patients who are either overweight or obese [6].
Currently, metformin in conjunction with lifestyle modifica-
tion is recommended as a first-line treatment to reduce
HbA1c levels [7]. However, due to the progressive nature of
the disease, many patients will require additional therapy as
an add-on to metformin to maintain glycaemic control.
Sulfonylureas are commonly prescribed as second-line
agents, although these are associated with inherent short-
comings such as weight gain and an increased risk of
hypoglycaemia [8].
Dapagliflozin was the first selective sodium–glucose co-
transporter–2 (SGLT–2) inhibitor licensed in the European
Union (EU) for use in people with Type 2 diabetes mellitus
when diet and exercise plus metformin failed to achieve
glycaemic control. The treatment effect of SGLT–2 inhibitors
is achieved through a novel insulin-independent mechanism
of action, associated with favourable outcomes in terms of
weight and hypoglycaemic risk [9]. In clinical trials, dapa-
gliflozin has demonstrated non-inferior HbA1c control, and
significantly reduced weight and hypoglycaemic incidences
when compared with a sulfonylurea, glipizide, when both
were added to metformin [10]. By using a decision-modelling
approach, the available short-term clinical data can be
extrapolated to provide meaningful economic inputs in order
to fully assess the cost-effectiveness of each treatment option.
This study aimed to assess the long-term health economic
consequences of dapagliflozin versus sulfonylurea, as an add-
on to metformin, from the perspective of the UK NHS.
ª 2015 The Authors.
Diabetic Medicine ª 2015 Diabetes UK
DIABETICMedicine
Assessments of the cost-effectiveness of dapagliflozin versus
other antidiabetes agents used as an add-on to metformin, and
in indications other than an add-on to metformin and settings
other than the UK have been presented elsewhere [11,12].
Patients and methods
Patient population
Baseline characteristics of the model population were
sourced from a head-to-head Phase III clinical trial of
dapagliflozin plus metformin versus a sulfonylurea plus
metformin (Table 1) [10]. The model cohort was considered
to be representative of UK patients who would be eligible to
receive dapagliflozin as part of a UK treatment strategy.
Model structure
The previously published and validated Cardiff Diabetes
Model was used as the basis for this economic analysis
[13,14]. Risk equations, developed from the UK Prospective
Diabetes Study (UKPDS) 68 were used to estimate long-term
micro- and macrovascular complications including ischaemic
heart disease, myocardial infarction, congestive heart failure,
stroke, amputation, blindness and end stage renal disease
(ESRD), as well as diabetes- and non-diabetes-related mor-
tality [15]. The probability of drug-related hypoglycaemic
events and additional adverse events such as urinary tract
and genital infections (which have been shown to be
associated with the mechanism of action of an SGLT–2
inhibitor) were also included in the model. Disease-related
complications were based on the time- and treatment-
dependent evolution of modifiable risk factors, including
HbA1c, BMI, the ratio of total and HDL cholesterol, and
systolic blood pressure. All-cause mortality events were
estimated using gender specific life tables for the UK [16],
which were converted to six-monthly probabilities and
evaluated at each modelled cycle. Because mortality related
to cardiovascular events and diabetes are accounted for in
Table 1 Summary of baseline variables applied in the health economic
model [10]
Characteristic Baseline value
Age; years 58.40
Proportion female; % 44.90
Duration of diabetes; years 6.32
Height; m 1.67
Proportion Afro-Caribbean; % 6.20
Proportion smokers; % 17.60
HbA1c; mmol/mol (%) 61 (7.72)
Weight; kg 88.02
Systolic blood pressure; mmHg 133.30
Total cholesterol; mg/dl 182.54
HDL cholesterol; mg/dl 45.87
891
DIABETICMedicine Cost-effectiveness of dapagliflozin vs. sulfonylurea  M. Charokopou et al.

the UKPDS risk equations, the probability of death from
cardiovascular and diabetes-related mortality (as predicted
using the UKPDS 68 event, event fatality and diabetes
mortality equations) is subtracted from the all-cause mortal-
ity defined in the life tables to avoid double counting.
Patients in the intervention and comparator groups were
simulated through six-month time intervals over a total
period of 40 years, indicative of a lifetime horizon for
Type 2 diabetes mellitus. The six-month cycle was chosen
because it is a standard duration of trial follow-up and
treatment decisions [7,17]. At the end of each cycle, fatal and
non-fatal complications are calculated. Once a fatal event
occurred, life years and quality-adjusted life years (QALYs)
were updated and the simulation ended for the patient. In the
base case analysis, 100 cohorts of 30 000 individual patients
were modelled to ensure stability in the simulation results.
The primary outcome used in this cost-effectiveness
analysis was the incremental cost-effectiveness ratio (ICER),
measured as cost per QALY gained. A discount rate of 3.5%
was applied to both costs and health effects as recommended
by the National Institute for Health and Care Excellence
(NICE) in the UK [7].
Treatment sequence
The first modelled therapies were dapagliflozin + metformin
and sulfonylurea + metformin for the intervention and
comparator groups, respectively. On losing glycaemic con-
trol, patients switched therapy, first to insulin + metformin,
and subsequently to intensified insulin (simulated by increas-
ing the dose by 50%) where they remained for the duration
of the time-horizon. The switching threshold therefore
played an integral role in determining the treatment duration
of each therapy. In this study, the HbA1c threshold for
therapy change was defined as 61 mmol/mol (7.7%), because
it was the average baseline HbA1c value of patients entering
the Phase III clinical trial before progressing to dual oral
therapy [10]. Patients would also have a separate risk of
discontinuing therapy during the first cycle, which was
applied based on the clinical trial data [10].
Treatment effects
The introduction of a new treatment resulted in a one-year
reduction in HbA1c according to the relative efficacy of the
interventions based on a Phase III randomized controlled
trial (Table 2) [10]. The timeframe of the treatment effect
was reflective of the availability of clinical data [10] and was
followed by a continued rise in HbA1c due to disease
progression, which was derived from a regression analysis of
the UKPDS dataset [15]. Similar assumptions were used for
systolic blood pressure and cholesterol, where treatment
effects were applied during the first year, before progressing
in accordance with the UKPDS 68 regression equations [15].
The annual probability of treatment discontinuation due to
adverse events was sourced from the Phase III randomized
controlled trial and applied during the first model cycle only
(Table 2) [10].
Because weight has been shown to impact on the health-
related quality of life (HRQoL) in people with Type 2 diabetes
mellitus, and increase the risk of cardiovascular complica-
tions, it was included as a parameter in the model. Initially,
the progression of weight was determined by the relative
treatment effects observed in the 52-week clinical trial [10]. In
the dapagliflozin arm, patient weight was then assumed to be
maintained in the second year based on open-label extension

Table 2 Overview of treatment efficacy and tolerability estimates for each antidiabetes treatment included in the health economic model

Dapagliflozin + Sulfonylurea + Insulin + Intensified

Variable metformin‡‡ metformin‡‡ metformin† insulin‡

ΔHbA1c*; % 0.52 0.52 1.10 1.11

ΔWeight* (kg) 3.22 1.44 1.08 1.90††
ΔSBP*; mmHg 4.30 0.80 0** 0**
ΔTC*; mg/dl 0.071 0.028 0** 0**
ΔHDL-C*; mg/dl 0.069 0.002 0** 0**
Probability of discontinuation§ 0.091 0.059 0** 0**
Probability. of hypoglycaemic 0.035 0.408 0.011 0.616
events (symptomatic)¶
Probability of hypoglycaemia (severe)¶ 0.000 0.007 0.037 0.022
Probability of urinary tract infection¶ 0.108 0.064 0** 0**
Probability of genital infection¶ 0.123 0.027 0** 0**

*Effects apply to the first year after treatment initiation.

†
Monami et al. [39].
‡
NICE HTA report Chapter 4 [32].
§
Probability of discontinuation was applied during the first model cycle.
¶
Probabilities of adverse events were applied during every model cycle.
**No estimate available and/or zero value assumed.
††
Weight change from Monta~ nana et al. [40] chosen as most recent study reporting weight effect included in the NICE HTA report.
‡‡
Nauck et al. [10].
SBP, systolic blood pressure; TC, total cholesterol; HDL-C, high-density lipoprotein cholesterol; Δ, absolute change from baseline.

ª 2015 The Authors.

892 Diabetic Medicine ª 2015 Diabetes UK
 Research article DIABETICMedicine

96 Costs
INS intensified
94 MET+INS weight effect
weight effect Economic data used in the model were sourced from a
92
systematic literature review and included the direct acquisi-
90
tion costs of the individual treatments, as well as the costs
Weight (kg)

88 Treatment
effect
Natural weight incurred for individual treatment paths and the management
progression
86
Years to loss of of associated adverse events (Table 3). A UKPDS study [19]
weight effect. A
84 linear loss is was utilized as a key source for model inputs because it
Years of assumed
82 maintained presented high-quality data relevant to modelling diabetes;
weight loss

80 all costs were indexed to 2011 using the Hospital and

0 2 4 6 8 10
Years Community Health Services Pay and Price index. Acquisition
Treatment Control costs were sourced from England and Wales Drug Tariff
costs [20].
FIGURE 1 Illustration of the dynamic weight profile implemented in
the model. MET, metformin; INS, insulin.
Health-Related Quality of Life (HRQoL)

data from the Phase III clinical trial [18]. Because there was no
further long-term data on patient weight available at the time
the analysis was performed, it was assumed that all weight loss
would be fully regained, in a linear manner, by the time the
patient switches to the next treatment line. An illustration of
how the model simulates patient weight for both treatment
and comparator arms is provided in Fig. 1.
A systematic review was carried out to identify studies for
HRQoL outcomes and the impact of drug-related adverse
events in Type 2 diabetes mellitus. The UKPDS 62 study [21]
identified in the review was used extensively as a source of
EuroQol–5 dimension (EQ–5D) utility outcomes in the
model because the utilities were derived from the same
patient cohort that informed the risk equations. Utility data

Table 3 Overview of cost inputs as applied in the model

Drug acquisition cost Price per tablet* Dose per tablet/pen Daily dose Annual cost

Dapagliflozin £1.31 10 mg 10 mg £476.92

Sulfonylurea (Gliclazide) £0.04 80 mg 160 mg £27.90
Metformin £0.02 500 mg 2000 mg £23.46
Insulin† (insuman basal) £0.47/day 300 IU 40 IU £170.23
Intensified insulin £0.70/day 300 IU 60 IU £256.96

Diabetes related complication cost‡ Fatal Non-fatal Maintenance Reference

Ischaemic heart disease – £3 479 £1 149 Clarke et al. [19]

Myocardial infarction £2 244 £6 709 £1 105 Clarke et al. [19]
Congestive heart failure £3 880 £3 880 £1 360 Clarke et al. [19]
Stroke £5 658 £4 103 £776 Clarke et al. [19]
Amputation £13 359 £13 359 £771 Clarke et al. [19]
Blindness – £1 752 £742 Clarke et al. [19]
End stage renal disease – £34 806 £34 806 Baboolal et al. [41]

Adverse event, renal monitoring

and discontinuation costs Cost input Source

Severe hypoglycaemic event§ £390 Hammer et al. [42]

Renal monitoring £38.67 Assumed to incur one GP visit cost and a 24-hour creatine clearance
determination [43]
Urinary tract infection, genital infection £36 Assumed to incur one GP visit cost [43]
Discontinuation £36 Assumed to incur one GP visit cost [43]

*The daily costs are based on pack costs and have been rounded. The source of the unit costs is the England and Wales Drug Tariff, February
2012.
†
The cost of insulin was based on a patient baseline weight of 88 kg, which if it remained stable would equate to an annual cost of £170.23
(and £256.96 for intensified insulin). However, in the model weight changed over time, hence the actual annual cost of insulin (with dosage
according to weight) in the economic analysis varies according to the simulated change in weight. Insulin daily cost per kg = £0.0053,
intensified insulin daily cost per kg = £0.008.
‡
Prices were indexed to 2011 using the Hospital and Community Health Services Pay and Prices index.
§
Costs based on 65% hospital treatment versus 35% non-hospital treatment.
GP, General practitioner; IU, international units.

ª 2015 The Authors.

Diabetic Medicine ª 2015 Diabetes UK 893
DIABETICMedicine Cost-effectiveness of dapagliflozin vs. sulfonylurea  M. Charokopou et al.

Table 4 Overview of utility decrements for each health state applied in

the model  10%, and total non-drug costs were varied by  25%.
Scenario analyses were conducted to explore the impact of
Utility how the treatment effect on patient weight was extrapolated
Event decrement Reference on treatment progression. In addition, the HbA1c switching
threshold was investigated as a scenario analysis because it
Diabetes-related complications
Ischaemic heart disease 0.090 Clarke et al. [21]
was an important factor in determining the treatment
Myocardial infarction 0.055 Clarke et al. [21] duration, and therefore the treatment effect, for each
Congestive heart failure 0.108 Clarke et al. [21] intervention in the model. Because of the expected effect of
Stroke 0.164 Clarke et al. [21]
Amputation 0.280 Clarke et al. [21]
weight on patients’ quality of life, the associated utility was
Blindness 0.074 Clarke et al. [21] also tested in scenario analyses where alternative utility
End stage renal disease 0.263 Currie et al. [22] sources were used.
Hypoglycaemia
Symptomatic 0.042 Currie et al. [25]
Nocturnal 0.008 Currie et al. [25]
Severe 0.047 Currie et al. [25]
Results
Adverse events
Urinary tract infection 0.00283 Barry et al. [23]
In the deterministic base-case analysis, dapagliflozin + met-
Genital infection 0.00283 Assumed to be the formin was associated with a mean incremental cost of
same as urinary £1246 (95% CI: £613; £1637) (Table 5). Analysis of the
tract infection
BMI changes Utility change
disaggregated costs calculated for dapagliflozin + metformin
Per unit increase 0.0472 Lane et al. [24] and sulfonylurea + metformin confirmed that this was
Per unit decrease +0.0171 Lane et al. [24] mainly driven by dapagliflozin’s higher acquisition cost
(Table 6). The mean incremental benefit of 0.467 QALYs
(95% CI: 0.420; 0.665) calculated for dapagliflozin + met-
for end-stage renal disease and BMI were sourced from formin was associated with the significantly different treat-
additional identified studies [22,23]. The parameters for BMI ment effects on weight between the comparators. An ICER
utility were derived from a bespoke study that made a point estimate of £2671 per QALY gained was calculated,
distinction in terms of utility change for BMI increase and well within the acceptable UK willingness to pay threshold of
decrease, elicited from Canadian people with Type 2 diabe- £20 000 per QALY gained [28]. Similarly, the probabilistic
tes mellitus [24]. The fear of hypoglycaemia following base-case analysis resulted in an ICER of £2187. Further
symptomatic and severe episodes has also been shown to details of the number of predicted cumulative events
have an effect on HRQoL [25] and has been included in observed per patient are presented in Table 7. The mean
other economic evaluations of Type 2 diabetes mellitus treatment duration for both treatment arms are presented in
treatments [11,26,27]; therefore, this was also included in the Supporting Information (Table S1).
the quality of life analysis (Table 4). The tornado graphs (Fig. 2) highlight the variation range of
both the incremental cost and incremental QALY for the
parameters that had the greatest effect on each outcome
Approach to sensitivity analysis
during the sensitivity analysis. As can be seen from these
Deterministic and probabilistic sensitivity analyses were graphs, the point estimate for incremental costs was relatively
carried out to assess the impact of uncertainty on model insensitive to the variation applied to the model parameters
results. Parameters including BMI utilities, weight and (Fig. 2). The point estimate for incremental QALYs was most
HbA1c were varied around their 95% confidence/credible sensitive to variation of the treatment effect of sulfonylurea
intervals in the deterministic sensitivity analysis. Where data and dapagliflozin on body weight, and the utility values
were unavailable, the standard error was assumed to be a assigned to BMI change. The largest impact was seen when
percentage of the mean in line with the magnitude of other varying the weight effect of sulfonylurea between the outer
standard errors for similar variables. As such, disutilities for limits of its 95% CI (from 0.382 to 0.552 QALYs),
Type 2 diabetes mellitus complications were altered by representing an ICER range of £3290 to £2241. The differ-

Table 5 Discounted base-case results (deterministic)

Technologies Total Incremental ICER

Add-on to metformin Costs Life years QALYs Costs Life years QALYs (Incremental cost per QALY gained)

Sulfonylurea £11 658 14.71 11.28 – – – –

Dapagliflozin £12 904 14.76 11.74 £1 246 +0.050 +0.467 £2 671

Abbreviations: QALYs, Quality adjusted life years; ICER, Incremental cost-effectiveness ratio.

ª 2015 The Authors.

894 Diabetic Medicine ª 2015 Diabetes UK
 Research article DIABETICMedicine

Table 6 Summary of disaggregated costs (£) calculated for cases, dapagliflozin remained cost-effective with reported
dapagliflozin + metformin and sulfonylurea + metformin treatment ICERs of £6487 and £5441 when weight profiles converged
sequences
prior to switching to second-line and third-line treatments,
respectively. Additional details of the scenarios including
Dapagliflozin Sulfonylurea
Parameter + metformin + metformin Incremental illustrations of how weight was modelled over time are
provided in the Supporting Information (Table S2; Figs S1
Treatment related and S2).
Drug treatment £4 502 £2 977 £1 525
(total) The impact of varying the HbA1c threshold was investi-
Hypoglycaemia 115 123 8 gated through scenario analyses by inputting alternative
Other adverse 67 14 53 higher and lower parameter values. NICE clinical guidelines
event related
costs (including specify a threshold of 59 mmol/mol (7.5%); therefore, this
renal value was included in the analysis [7]. However, because
monitoring)
studies have shown that many patients will exceed this target
Event related
Ischaemic heart £1 168 £1 194 26 in practice, switching thresholds of 64 mmol/mol (8.0%) and
disease 69 mmol/mol (8.5%) were also investigated [29,30]. As the
Myocardial £2 311 £2 355 43
infarction
switching threshold increased, the incremental costs ranged
Congestive heart 583 616 33 from £863 to £3282 when considering both scenarios. An
failure increase in incremental QALYs was also observed when
Stroke 582 590 8
Amputation 399 396 3 increasing the switching threshold, resulting in ICERs rang-
Blindness £2 682 £2 878 196 ing from £1830 to £5633 per QALY. Another scenario
End stage renal 497 516 19 analysis demonstrated that by using alternative sources for
disease
Total £12 904 £11 658 £1 246 BMI utility effects, [31,32], the ICER increased from £2671
to £7555 and £8373 per QALY, respectively. The distribu-
tion of the ICER estimates from the probabilistic sensitivity
analysis indicated that dapagliflozin + metformin is both
ence of only £1049 between the upper and lower limits of this more effective and costly than sulfonylurea + metformin
ICER demonstrates the robustness of the results produced by (Fig. 3). However, at a willingness-to-pay threshold of
the model and highlights the likelihood of dapagliflo- £20 000 per QALY gained, dapagliflozin + metformin had
zin + metformin being cost-effective. Two scenario analyses a 100% probability of being cost-effective compared with a
were conducted to explore the impact of converging weight sulfonylurea + metformin treatment strategy. The parame-
profiles immediately prior to switching treatment. In both ters and distributions used in the probabilistic sensitivity

Table 7 Lifetime predicted cumulative number of events per patient; dapagliflozin + metformin vs sulfonylurea + metformin

Dapagliflozin + metformin Sulfonylurea + metformin Incremental

Event Non-fatal Fatal Non-fatal Fatal Non-fatal Fatal

Macrovascular
Ischaemic heart 0.129 0.000 0.132 0.000 0.003 0.000
disease
Myocardial 0.202 0.116 0.205 0.121 0.003 0.005
infarction
Stroke 0.087 0.015 0.092 0.017 0.004 0.002
Congestive heart 0.058 0.039 0.059 0.017 0.001 0.023
failure
Microvascular
Blindness 0.078 0.000 0.078 0.000 0.000 0.000
Nephropathy 0.014 0.015 0.015 0.016 0.001 0.001
Amputation 0.026 0.026 0.027 0.028 0.001 0.001
Adverse events
Urinary tract 0.440 0.100 0.340 0.000
infection
Genital infection 8.803 10.058 1.254 0.000
Symptomatic 0.448 0.469 0.022 0.000
hypoglycaemia
Severe 0.129 0.132 0.003 0.000
hypoglycaemia

ª 2015 The Authors.

Diabetic Medicine ª 2015 Diabetes UK 895
DIABETICMedicine Cost-effectiveness of dapagliflozin vs. sulfonylurea  M. Charokopou et al.
FIGURE 2 Univariate sensitivity analyses: tornado graphs of
incremental costs (upper) and incremental QALYs (lower). Variations
in selected parameters are displayed as a range from the base-case value
(y–axis). ΔHbA1c, change in HbA1c from baseline; ΔWeight, change in
weight from baseline; Comp., Comparator (sulfonylurea + metfor-
min); QALY, quality-adjusted life year; dapagliflozin + metformin,
dapagliflozin added on to metformin; T2DM, Type 2 diabetes mellitus.
It can be observed from the tornado graph for incremental costs that
assuming either a larger or smaller effect of dapagliflozin + metformin
on HbA1c reduction would result in lower incremental costs than in the
base case. This can be explained by the model structure: in the case of a
smaller HbA1c reduction patients will sooner switch to the next
treatment line, leading to lower drug costs.
analyses are described in the Supporting Information
(Table S3).
Discussion
The economic analysis of dapagliflozin + metformin versus
sulfonylurea + metformin has shown dapagliflozin to be a
valuable treatment option for patients who are inadequately
controlled by metformin monotherapy. By utilizing risk
equations to simulate the long-term effects of both interven-
tions, the economic model can generate evidence to help
inform effective healthcare decisions and maximize the use of
NHS resources. In a clinical setting, dapagliflozin has
demonstrated sustained HbA1c control, as well as superior
weight reduction and systolic blood pressure outcomes when
compared with sulfonylurea [10]. These outcomes, in addi-
tion to a lower risk of hypoglycaemic events, were key
drivers in the associated gain in QALYs observed.
The cost-effectiveness model has been published, validated
and used successfully with HTA authorities in several
896
s
Quadrant s
North East 100.0%
South East 0.0%
North West 0.0%
South West 0.0%
Acceptability curve
FIGURE 3 Scatterplot of the ICER estimates of the probabilistic
sensitivity analysis (upper) and cost-effectiveness acceptability curve
(lower) for dapagliflozin + metformin vs sulfonylurea + metformin.
ICER, incremental cost-effectiveness ratio; WTP, willingness to pay;
QALY, quality-adjusted life year.
jurisdictions [14,27,33–36]. For this analysis, the baseline
characteristics used were all sourced from a non-inferiority
randomized controlled trial [10] because it represented a
direct comparative study between dapagliflozin + metformin
and sulfonylurea + metformin. This provided a source of
high-quality data inputs for the simulation and allowed the
model to take into account factors such as the effect of
cholesterol levels and systolic blood pressure on patient
outcomes. The primary limitation of the model was a lack of
high-quality, long-term data regarding the development of
diabetes related micro- and macrovascular complications
associated with dapagliflozin or sulfonylurea as an add-on to
metformin. Although the UKPDS risk equations used to
predict lifetime events are effective for determining cardio-
vascular-related Type 2 diabetes mellitus complications,
these also have associated limitations [37]. Moreover,
updated risk equations have been reported, but were not
available at the time of the current analysis. However,
because the UKPDS 68 equations were derived from a study
of a large UK population (N > 5000), and are commonly
used in other Type 2 diabetes mellitus models, they were
deemed to be suitable for this analysis.
The respective effect on weight for each treatment arm
was shown to be an important risk factor in determining
QALYs during the sensitivity analysis. However, the
extrapolation approach utilized in the model can be
considered conservative in light of a recent extension study,
which confirms that the weight effect is maintained up to
ª 2015 The Authors.
Diabetic Medicine ª 2015 Diabetes UK
 Research article
four years [38]. Therefore, any beneficial effect associated
with weight loss was likely to be underestimated in the
model. In addition, given the modelling approach used, the
differential established at therapy initiation for systolic
blood pressure is maintained throughout the patient’s
lifetime. This could potentially be to the benefit of dapa-
gliflozin, which has a more favourable initial efficacy
profile.
There was some variation between literature sources
describing the impact of increasing/decreasing BMI on utility
values. The source for weight utilities used in the base-case
analysis was derived from a bespoke investigation into the
impact of weight gain (and loss) on the health state of people
with Type 2 diabetes mellitus, which was why it was selected
as the most appropriate [24]. However, the ICER remained
below £10 000 per QALY when the other sources were
employed [31,32]. The value of the HbA1c switching
threshold was also shown to have an effect on the dapagli-
flozin + metformin versus sulfonylurea + metformin ICER;
however, the ICER still remained below £20 000 per QALY
in all cases.
The results of this analysis can be considered to be
generalizable to patients in a UK setting because the
populations defined from the international trial [10] were
representative of the treatment indication in the UK. In
addition, the multicentre study included 12 sites within the
UK, and the average baseline demographics were similar to
the UK Type 2 diabetes mellitus population recruited for the
UKPDS studies [15]. Cost and resource inputs were also
derived from a UK perspective and utility data were largely
sourced from the UKPDS studies.
Conclusion
This analysis confirmed that dapagliflozin is cost-effective in
the UK in comparison with sulfonylurea, when both are used
as a dual therapy added-on to metformin. The favourable
weight effect associated with dapagliflozin, along with the
reduced risk of hypoglycaemic events, resulted in an increase
in patients’ quality of life, which mitigated the increased
costs associated with drug acquisition.
Funding sources
This study was funded by Bristol-Myers Squibb, Rueil-
Malmaison, France and AstraZeneca, Brussels, Belgium.
Employees of BMS and AstraZeneca had input into model
design, results interpretation and manuscript development.
Competing interests
MC, PM and KT are paid consultants of Bristol-Myers
Squibb. SL, MR and RP are employees of Bristol-Myers
Squibb. LC and KB are employees of AstraZeneca. RT was
an employee of AstraZeneca at the time of the research.
ª 2015 The Authors.
Diabetic Medicine ª 2015 Diabetes UK
DIABETICMedicine
Acknowledgements
Editorial support was provided for this manuscript by
Costello Medical Consulting Ltd.
References
1 Coyle D, Lee KM, O’Brien BJ. The role of models within economic
analysis: focus on type 2 diabetes mellitus. PharmacoEconomics
2002; 20(Suppl 1): 11–19.
2 Haddix AC, Teutsch SM, Corso PS. Prevention Effectiveness: a
guide to decision analysis and economic evaluation: Oxford:
Oxford University Press, 2003.
3 The National Collaborating Centre for Chronic Conditions. Type 2
Diabetes: national clinical guideline for management in primary
and secondary care (update). Royal College of Physicians, 2008.
Available at http://www.nice.org.uk/guidance/cg66/resources/cg66-
type-2-diabetes-full-guideline2 Last accessed 1 November 2014.
4 House of Commons Committee of Public Accounts. The manage-
ment of adult diabetes services in the NHS (October 2012).
Available at http://www.publications.parliament.uk/pa/cm201213/
cmselect/cmpubacc/289/289.pdf Last accessed 3 April 2015.
5 Inzucchi SE, Bergenstal RM, Buse JB, Diamant M, Ferrannini E,
Nauck M et al. Management of hyperglycemia in type 2 diabetes: a
patient-centered approach. Position statement of the American
Diabetes Association (ADA) and the European Association for the
Study of Diabetes (EASD). Diabetes Care 2012; 35: 1364–1379.
6 National Diabetes Audit. National Diabetes Audit Executive
Summary. Available at http://www.hscic.gov.uk/catalogue/
PUB12421/nati-diab-audi-11-12-care-proc-rep.pdf Last accessed
1 November 2014.
7 National Collaborating Centre for Chronic Conditions, Centre for
Clinical Practice at NICE. NICE Clinical Guideline 87. Type 2
Diabetes: the management of type 2 diabetes. London: National
Institute for Health and Clinical Excellence, 2009.
8 Phung OJ, Scholle JM, Talwar M, Coleman CI. Effect of noninsulin
antidiabetic drugs added to metformin therapy on glycemic control,
weight gain, and hypoglycemia in type 2 diabetes. JAMA 2010;
303: 1410–1418.
9 List JF, Whaley JM. Glucose dynamics and mechanistic implica-
tions of SGLT2 inhibitors in animals and humans. Kidney Int Suppl
2011; 120: S20–S27.
10 Nauck MA, Del Prato S, Meier JJ, Dur an-Garcıa S, Rohwedder K,
Elze M et al. Dapagliflozin versus glipizide as add-on therapy in
patients with type 2 diabetes who have inadequate glycemic control
with metformin: a randomized, 52-week, double-blind, active-
controlled noninferiority trial. Diabetes Care 2011; 34: 2015–2022.
11 Charokopou M, McEwan P, Lister S, Callan L, Bergenheim K,
Tolley K et al. The cost-effectiveness of dapagliflozin (Forxiga)
versus a DPP–4 inhibitor in the treatment of type 2 diabetes
mellitus (T2DM) in England and Wales. Value Health 2013; 16:
A440.
12 van Haalen HG, Pompen M, Bergenheim K, McEwan P, Townsend
R, Roudaut M. Cost effectiveness of adding dapagliflozin to insulin
for the treatment of type 2 diabetes mellitus in the Netherlands.
Clin Drug Invest 2014; 34: 135–146.
13 McEwan P, Evans M, Bergenheim K. A population model
evaluating the costs and benefits associated with different oral
treatment strategies in people with type 2 diabetes. Diabetes Obes
Metab 2010; 12: 623–630.
14 The Mount Hood 4 Modeling Group. Computer modeling of
diabetes and its complications. Diabetes Care 2007; 30: 1638–1646.
15 Clarke P, Gray A, Briggs A, Farmer AJ, Fenn P, Stevens AJ et al. A
model to estimate the lifetime health outcomes of patients with type
897
DIABETICMedicine Cost-effectiveness of dapagliflozin vs. sulfonylurea  M. Charokopou et al.
2 diabetes: the United Kingdom Prospective Diabetes Study
(UKPDS) Outcomes Model (UKPDS no. 68). Diabetologia 2004;
47: 1747–1759.
16 Office of National Statistics. UK Life Expectancy Tables. Available
at www.ons.gov.uk/ons/taxonomy/index.html?nscl=Life+Tables
Last accessed 1 November 2014.
17 Barlow J, Krassas G. Improving management of type 2 diabetes:
findings of the Type2Care clinical audit. Aust Fam Physician 2013;
2: 57.
18 Del Prato S, Nauck M, Rohwedder K et al., editors. Long-term
efficacy and safety of dapagliflozin vs add-on glipizide in patients
with type 2 diabetes inadequately controlled with metformin:
2 year results. Presented at the 47th Annual Meeting of the
European Association for the Study of Diabetes, 2011.
19 Clarke P, Gray A, Legood R, Briggs A, Holman R. The impact of
diabetes-related complications on healthcare costs: results from the
United Kingdom Prospective Diabetes Study (UKPDS Study No.
65). Diabet Med 2003; 20: 442–450.
20 National Health Service (NHS) UK. England and Wales Drug
Tariff (Feb 2012). Available at www.nhsbsa.nhs.uk/prescriptions
Last accessed 1 November 2014.
21 Clarke P, Gray A, Holman R. Estimating utility values for health
states of type 2 diabetic patients using the EQ-5D (UKPDS 62).
Med Decis Making 2002; 22: 340–349.
22 Currie C, McEwan P, Peters J, Patel TC, Dixon S. The routine
collation of health outcomes data from hospital treated subjects in the
Health Outcomes Data Repository (HODaR): descriptive analysis
from the first 20,000 subjects. Value Health 2005; 8: 581–590.
23 Barry HC, Ebell MH, Hickner J. Evaluation of suspected urinary
tract infection in ambulatory women: a cost-utility analysis of
office-based strategies. J Fam Pract 1997; 44: 49–60.
24 Lane S, Levy AR, Sambrook J, Mukherjee J, Leahy M, Tildesley H.
PDB72 The impact on utilities of weight loss and weight gain
among canadian patients with type 2 diabetes. Value Health 2012;
15(4): A183.
25 Currie CJ, Morgan CL, Poole CD, Sharplin P, Lammert M,
McEwan P. Multivariate models of health-related utility and the
fear of hypoglycaemia in people with diabetes. Curr Med Res Opin
2006; 22: 1523–1534.
26 Erhardt W, Bergenheim K, Duprat-Lomon I, McEwan P. Cost
effectiveness of saxagliptin and metformin versus sulfonylurea and
metformin in the treatment of type 2 diabetes mellitus in Germany.
Clin Drug Invest 2012; 32: 189–202.
27 Granstrom O, Bergenheim K, McEwan P, Sennf€alt K, Henriksson
M. Cost-effectiveness of saxagliptin (Onglyza) in type 2 diabetes
in Sweden. Primary Care Diabetes 2012; 6: 127–136.
28 National Institute for Health and Care Excellence. Guide to the
Methods of Technology Appraisal. 2013.
29 Fox KM, Gerber RA, Bolinder B, Chen J, Kumar S. Prevalence of
inadequate glycemic control among patients with type 2 diabetes in
the United Kingdom general practice research database: a series of
retrospective analyses of data from 1998 through 2002. Clin Ther
2006; 28: 388–395.
30 Calvert MJ, McManus RJ, Freemantle N. Management of type 2
diabetes with multiple oral hypoglycaemic agents or insulin in
primary care: retrospective cohort study. Br J Gen Pract 2007; 57
(539): 455–460.
31 Bagust A. Modelling EuroQol health-related utility values for
diabetic complications from CODE-2 data. Health Econ 2005; 14:
217–230.
32 Waugh N, Cummins E, Royle P, Clar C, Marien M, Richter B et al.
Newer agents for blood glucose control in type 2 diabetes:
898
systematic review and economic evaluation. Health Technol Assess
2010; 14: 1–248.
33 Cummins E, Scott N, Rothnie K, Waugh N, Fraser C, Philip S et al.
Evidence Review Group report; Dapagliflozin for the treatment of
type 2 diabetes. 2013. Available at http://www.nice.org.uk/guidance/
ta288/documents/diabetes-type-2-dapagliflozin-evidence-review-
group-report3 Last accessed 3 April 2015.
34 Palmer AJ. Computer modeling of diabetes and its complications: a
report on the fifth Mount Hood Challenge meeting. Value Health
2013; 16: 670–685.
35 Charokopou M, Sabater F, Townsend R, Roudaut M, Verheggen
BG. Overview of health economic models in type 2 diabetes mellitus:
a systematic literature review. Value Health 2013; 16: A591.
36 Scottish Medicines Consortium. Dapagliflozin for the Treatment of
Type 2 Diabetes. 2012. Available at https://www.scottishmedicines.
org.uk/SMC_Advice/Advice/799_12_dapagliflozin_Forxiga/dapagli
flozin_Forxiga Last accessed 3 April 2015.
37 Yang F, Ye J, Pomerantz K, Stewart M. Potential modification of
the UKPDS risk engine and evaluation of macrovascular event rates
in controlled clinical trials. Diabetes Metab Syndr Obes 2013; 6:
247–256.
38 Del Prato A, Nauck M, Duran-Garcia S, Rohwedder K, Theuerkauf
A, Langkilde AM et al. Durability of dapagliflozin vs glipizide as
add-on therapies in T2DM indadquately controlled on metformin:
4 year data. Presented at the 73rd American Diabetes Association
Meeting, Chicago IL, June 21–25 2013.
39 Monami M, Marchionni N, Mannucci E. Long-acting insulin
analogues versus NPH human insulin in type 2 diabetes: a meta-
analysis. Diabetes Res Clin Pract 2008; 81: 184–189.
40 Monta~ nana CF, Herrero CH, Fern andez MR. Less weight gain and
hypoglycaemia with once-daily insulin detemir than NPH insulin in
intensification of insulin therapy in overweight Type 2 diabetes
patients—The PREDICTIVETM BMI clinical trial. Diabet Med
2008; 25: 916–923.
41 Baboolal K, McEwan P, Sondhi S, Spiewanowski P, Wechowski J,
Wilson K. The cost of renal dialysis in a UK setting—a multicentre
study. Nephrol Dial Transplant 2008; 23: 1982–1989.
42 Hammer M, Lammert M, Mejias SM, Kern W, Frier BW. Costs of
managing severe hypoglycaemia in three European countries. J Med
Econ 2009; 12: 281–290.
43 Personal Social Services Research Unit. Unit Costs of Health and
Social Care. Available at http://www.pssru.ac.uk/project-pages/
unit-costs/2012/index.php Last accessed 1 November 2014.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Figure S1. Scenario 1. Convergence of weight curve imme-
diately prior to switch to second line treatment.
Figure S2. Scenario 2. Convergence of weight curve imme-
diately prior to switch to third line treatment.
Table S1. Duration in years on each treatment line.
Table S2. Results of scenario analyses exploring convergence
of weight profiles.
Table S3. Parameters and distributions – probabilistic sen-
sitivity analysis.
ª 2015 The Authors.
Diabetic Medicine ª 2015 Diabetes UK