HEALTH ECONOMICS

Health Econ. 22: 1272–1281 (2013)

Published online 19 October 2012 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/hec.2879

HEALTH ECONOMICS LETTER

VALUING QALY GAINS BY APPLYING A SOCIETAL PERSPECTIVE

ANA BOBINAC*, N. JOB A. VAN EXEL, FRANS F. H. RUTTEN and WERNER B. F. BROUWER
Department of Health Policy and Management and Institute for Medical Technology Assessment, Erasmus University Rotterdam,
Rotterdam, The Netherlands

ABSTRACT
Interpreting the outcomes of cost utility analyses requires an appropriately defined threshold for costs per quality-adjusted
life year (QALY). A common view is that the threshold should represent the (consumption) value a society attaches to a
QALY. So far, individual valuations of personal health gains have mainly been studied rather than potentially relevant
social values. In this study, we present the first direct empirical estimates of the willingness to pay for a QALY from a
societal perspective. We used the contingent valuation approach, valuing QALYs under uncertainty and correcting for
probability weighting. The estimates obtained in a representative sample of the Dutch population (n = 1004) range from
€52,000 to €83,000, depending on the specification of the societal perspective. The scale sensitivity was weak, however.
Copyright © 2012 John Wiley & Sons, Ltd.

Received 27 October 2011; Revised 6 July 2012; Accepted 16 September 2012

KEY WORDS: WTP; QALY; social perspective

1. INTRODUCTION
Economic evaluations aim to inform decision makers on optimal resource allocation. In the healthcare sector,
results from economic evaluations are used in decisions regarding the funding or reimbursement of healthcare
interventions. In such decisions, the outcomes of an evaluation, often expressed as incremental costs per
quality-adjusted life year (QALY) gained, need to be judged against some ‘threshold’. The nature and size
of this threshold have received quite some attention (e.g., Hirth et al., 2000; Claxton et al., 2010; Bobinac
et al., 2011a). One common interpretation of the threshold is that it represents the monetary (or consumption)
value of QALY gains (Gravelle et al., 2007). When the incremental costs per QALY gained are below this
threshold, an intervention can be regarded cost-effective and good ‘value for money’. Knowledge of the
monetary value of a QALY is therefore important but still relatively scarce. Moreover, available estimates have
limited usefulness because of their diverse origins, broad ranges (e.g., Mason et al., 2008) and doubtful validity
(Pinto Prades et al., 2009; Bobinac et al., 2011b). The need for useful estimates of the value of QALY gains
therefore remains.
An important yet frequently neglected question in this context is what perspective should be taken in valuing
QALYs. So far, studies have almost exclusively taken an individual perspective (e.g., Donaldson et al., 2011),
asking respondents to value changes in their own health. Although such an approach aligns well with dominant
welfare economic tenets such as consumer sovereignty (e.g., Brouwer et al., 2008), it is questionable whether
such valuations also represent the most relevant information for societal decision making in the context of
collectively funded health care. For this purpose, the ‘social value’ of a QALY may be considered more
relevant than individual values. The individual values are defined as the amount of own consumption indivi-
duals are willing to forgo to improve their own health, whereas the social value of a QALY is defined as the

*Correspondence to: Erasmus University Rotterdam (iBMG/iMTA), P.O. Box 1738, 3000 DR Rotterdam, The Netherlands. E-mail:
bobinac@bmg.eur.nl

Copyright © 2012 John Wiley & Sons, Ltd.

 VALUING QALY GAINS 1273

amount of own consumption individuals are willing to forego to contribute to a health gain achieved in society
(which may or may not accrue to the payer). These social values are hence still calculated using averages of
obtained individual valuations.1 The crucial difference is that the ‘social value’ pertains to the value of a QALY
gained in some member of society and generated through a collectively funded healthcare system. Social valua-
tions may be lower or higher than individual valuations.2
Social values seem particularly relevant in the context of collectively funded healthcare systems, reflecting
broader objectives and principles than purely individual ones (e.g., ‘communitarianism’ or solidarity; Barry,
1989; Dolan et al., 2003). This paper reports a first attempt in the literature to move beyond the individual
perspective and estimate the social value of a QALY. It is explicitly intended as a first step but hopes to
stimulate more investigation and debate in this important area.

2. METHODS

The social value of a QALY was estimated using an online contingent valuation survey, employing a represen-
tative sample of the Dutch population in terms of age (18 – 65 years), education, income and gender (n = 1004;
Table I). A sampling agency programmed and administered the questionnaire. The introduction to the question-
naire included a graphical explanation of the concept of risk to improve respondents’ understanding of the
exercises (e.g. Corso et al., 2001; Donaldson et al., 2010; APPENDIX A).

2.1. The questionnaire

Two question formats were used to elicit two different values of a QALY: the purely other-regarding ‘social value’
(SOC; representing the value of an expected QALY gain achieved in others) and the ‘social-inclusive-individual
value’ (SII; representing the value of an expected QALY gain in others or the individual herself) (Dolan et al.,
2003). Each respondent completed one willingness to pay (WTP) question; half of the sample completed the
SOC format (n = 504) and the other half the SII format (n = 500). Applying these two ex ante perspectives enabled
inclusion of externalities and option values in the valuation process.3
Each respondent valued a health change defined by two EQ-5D health states (APPENDIX B). Respondents
were first asked to indicate which of the two health states was better and which was worse, using a visual
analog scale (VAS). Respondents were subsequently asked to imagine the entire Dutch population in the health
state they had chosen as the better one and that because of a virus half of the population was at risk of going to
the worse health state for a period of 1 year.4 In the SOC question, the respondent did not belong to the risk
group, in the SII question he or she did. Respondents were asked to indicate their WTP for a treatment that
would eliminate the risk of the health deterioration. Respondents were instructed that the risk and non-risk
groups were identical in all other respects. The treatment was described as a medicine without side effects.
The WTP was elicited in a two-step approach (APPENDIX A). Using a payment scale, respondents first
indicated the highest amount they would certainly pay for the treatment. Next, they indicated the lowest amount
they would certainly not pay on the same scale. Second, respondents indicated their exact WTP in an
‘open-ended’ (OE) format, within the range obtained in the first step. The payment vehicle was described as an

1
Note that the appropriate method for calculating and averaging values of a QALY is debatable (e.g., Gyrd-Hansen and Kjar, 2012), but this
is not the focus of our study.
2
Lower in relation to end-of-life treatments, for instance, in which case individual valuations may be (extremely) high because of diminish-
ing marginal utility of income, or higher, for instance, when, on the basis of equity considerations, society is willing to pay more for health
improvements in certain groups (e.g., severely ill children) than individual budgets of these patients would allow.
3
A health gain carries an externality value either due to altruism (subsidy or the ‘warm glow’ motive) or due to reducing own risk of con-
tagion (encompassed in both SOC and SII). The option value denotes the value of treatment availability in circumstances of uncertainty
regarding own future health needs (encompassed within SII).
4
Risk levels were 50%, 10%, 4% and 2%.

Copyright © 2012 John Wiley & Sons, Ltd. Health Econ. 22: 1272–1281 (2013)
DOI: 10.1002/hec
1274 A. BOBINAC ET AL.

Table I. Summary statistics of respondents

Variable Mean SD Min Max
Age 40.1 12.8 18 65
Sex (% male) 0.5
Marital status Married (%yes) 0.59
Divorced (%yes) 0.08
Single (% yes) 0.26
Widow (%yes) 0.014
Not stated (%yes) 0.05
Children (% yes) 0.51
Number of children (n = 3070) 2.07 1.09 1 10
Household income 2630 1561 999 10,000
Income groups Group 1 (<€1000) 0.14
Group 2 (€1000–1999) 0.34
Group 3 (€2000–3499) 0.37
Group 4 (≥€3500) 0.15
Number of people living on household income 2.49 1.08 1 20
Education status No education and elementary school 0.13
High school education 0.53
University (% yes) 0.34
Employment status Employed (% yes) 0.62
Unemployed (% yes) 0.18
Student (% yes) 0.09
Housewife/husband or retired (% yes) 0.11
Health status EQ-5D (Dutch tariff) 0.85 0.23
0.329 1
EQ-VAS 78.8 17.6 0 100
Health status category Highest (% yes) 8.05
Mid (% yes) 9.9
Lowest (% yes) 82.06
Completion time of the questionnaire (min) 18.8 60.13 9 61
WTP for an automobile 16,336 60,258 0 1,000,000
WTP for a pair of shoes 200 1168 0 23,423
Note: VAS, visual analog scale; WTP, willingness to pay.

increase in their monthly health insurance premium.5 Respondents were told that all eligible members of the
population (above 18 years) would be required to pay the additional premium. The questionnaire was pilot
tested in a random sample of 100 respondents.

2.2. The scenarios

Twenty-nine scenarios were designed by combining 42 different health states, described using the EQ-5D
system6 (Bobinac et al., 2011b, 2011c) (APPENDIX B). Larger quality-of-life decrements were combined with
smaller probabilities (vice versa) to obtain marginal expected QALY gains and to reduce the impact of income
constraints. We based QALY gain calculations on the Dutch EQ-5D tariffs (Lamers et al., 2006), with the
expected gains ranging between 0.002 and 0.066 (APPENDIX B).7 Respondents were randomly assigned to
a scenario and either the SOC or SII question, respecting scenario and question format balance.
5
Health insurance was described as social (community-based) health insurance, typical in the Dutch setting. For more discussion on the
potential differences between social insurance and taxation healthcare financing in this context, see Olsen et al. (2004).
6
The majority of these scenarios were also applied in deriving the Dutch national EQ-5D tariffs (e.g. Lamers et al., 2006) and in other
related studies (Gyrd-Hansen, 2003).
7
Both in the design of the study and in base case analyses, we used the EQ-5D tariffs (Lamers et al., 2006) rather than the VAS scores.
EQ-5D tariffs are preferred because these are derived using choice-based techniques such as the time trade-off method and commonly used
in this context (e.g., Gyrd-Hansen, 2003; Bobinac et al., 2010c). We did test whether the ‘better’ health states (given the EQ-5D tariff )
systematically received higher average valuations on the EQ-VAS than the ‘worse’ health states because otherwise respondents might
be valuing a loss. Respondents reversed the ranking in fewer than 5% of scenarios. Removing these observations from the analyses did
not change our main results.

Copyright © 2012 John Wiley & Sons, Ltd. Health Econ. 22: 1272–1281 (2013)
DOI: 10.1002/hec
 VALUING QALY GAINS 1275

2.3. The analysis

The total number of expected QALYs (TEQ) in each scenario was calculated as
TEQ ¼ ðUð HS1Þ
Uð HS2ÞÞ  p  ð16; 500; 000  0:5Þ (1)
U(HS1)
U(HS2) is the quality-of-life decrement based on the Dutch EQ-5D tariffs and p the probability of
the specific decrement occurring in half of the Dutch population (i.e., 16,500,000 * 0.5) presented to respon-
dents. Because it is known from prospect theory that people tend to treat probabilities non-linearly, overweight-
ing small probabilities and underweighting large ones (e.g., Tversky and Kahneman, 1992), probabilities used
were both uncorrected (i.e. as presented) and corrected for probability weighting in the analysis. A failure to
correct for this weighting of probabilities can lead to incorrect estimates of WTP. Two one-parameter probabil-
ity-weighting functions were used (Table II): Tversky and Kahneman (1992) function and Prelec (1998) func-
tion, using the parameters estimated (for health) by Bleichrodt and Pinto Prades (2000).
The average WTP per QALY was then calculated as
WTP per QALY ¼ ð< WTPð OEÞ= TEQ >Þ  12  13; 260; 000 (2)
<WTP(OE)/TEQ> presents the average of the ratios between the open-ended WTP estimate and TEQ for
every data row, multiplied by 128 and the number of eligible health insurance premium payers in the
Netherlands (approximately 13,260,000).
Estimates of the SOC and SII values were tested using parametric t-tests on log-transformed WTP and non-
parametric Mann–Whitney u-tests on non-transformed data. Sensitivity to scale (i.e., construct validity;
Bobinac et al., 2011b) was tested by examining the sensitivity of OE WTP estimates to the size of health gain
on offer. We tested whether larger (smaller) gains received statistically different and proportionally higher
(lower) WTP values. Theoretical validity was tested within a log-linear regression, using STATA 11.

3. RESULTS

Average WTP (OE) values for an average non-weighted QALY gain of 0.008 were €29 in SOC and €48 in SII
(p = 0.00 in parametric and nonparametric tests). SOC values were systematically lower than SII values
(ranging €52,200 to €119,600 versus €59,200 to €188,900), also when correcting for probability weighting
(Table III). Respondents who were relatively uncertain about their answers and those who reported lower
income also reported lower WTP. Ninety-three respondents indicated €0 WTP (9%).9
The elicited WTP did not vary proportionally with the size of the QALY gain on offer, in both the SII and
SOC questions, indicating poor sensitivity to scale. The insensitivity was also found when only the probability
varied. Consequently, WTP per QALY estimates varied considerably across QALY gains of different sizes.
Position on the scale (which may be seen as a proxy for severity) did not affect the valuations. Probability
weighting could not explain the insensitivity. Log-linear regressions supported the notion of insensitivity to
scale and revealed that SOC WTP estimates were not affected by the size of the gain on offer ( p = 0.28,
Table IV). The coefficient of the QALY gain in SII preformed somewhat better ( p = 0.06). These results were
similar with a correction for probability weighting.

8
This is because we asked about a monthly premium increase for the duration of 1 year.
9
Respondents were asked about their reason for opting for a zero WTP. Their responses were uniformly distributed among available expla-
nations: (i) I am unable to pay more than €0; (ii) The health gain is not worth more than €0 to me; (iii) I am not willing to pay out of ethical
considerations; (iv) Other. Zero WTP was not systematically related to income level or the size of the gain on offer. None of the zero
responses were excluded from the analysis and doing so does not alter our main conclusions. We also re-analyzed the data removing ex-
treme outliers, but the findings remained robust.

Copyright © 2012 John Wiley & Sons, Ltd. Health Econ. 22: 1272–1281 (2013)
DOI: 10.1002/hec
1276 A. BOBINAC ET AL.

Table II. Probability weighting functions and parameter estimates

Weighted probabilities (%)

Functional form (Tversky and Kahneman, 1992): Functional form (Prelec, 1998):
w( p) = pg/[2pg + (1
pg)]1/g w(p) = exp(
(
ln)a)

Non-weighted probabilities Parameter estimate (Bleichrodt and Pinto, 2000): Parameter estimate (Bleichrodt and Pinto,
(%) y = 0.674 for losses 2000): a = 0.533

50 45 44
10 17 21
4 10 16
2 6 13

4. DISCUSSION

There has been surprisingly little empirical interest in the WTP per QALY from a societal perspective. Smith’s
(2007) work is a noteworthy exception, although not directly valuing QALYs. Although the dominant interest
in individual valuations of own health gains may be related to the traditional welfare economic perspective, the
social values may be particularly informative in the context of collectively funded healthcare systems. They can
reflect equity considerations important in healthcare decision making (Bobinac et al., 2011a) and incorporate con-
cerns for the well-being of others, although caution is warranted not to double count (e.g., Bergstrom, 2006). The
ex ante SII insurance perspective may be considered most relevant in estimating WTP for health in collectively
funded systems. A purely individual valuation misses out on important notions of equity, solidarity and altruism,
whereas an SOC perspective ignores the self-regarding option value. The SII perspective combines the individual’s
self-interest with the concern for others and therefore aligns well with the context of the social insurance system.
Our study was a first attempt to derive the value of QALY gains from a societal perspective, and it had some
clear limitations (e.g., using an online questionnaire, only one WTP question per respondent). Still, the absolute
sizes of the average SOC and SII estimates were plausible (€52,000 to €83,000 per QALY, corrected for probability
weighting) and in fact close to commonly used figures in the Netherlands (e.g. Van de Wetering et al., 2011). The
fact that the SII estimates were consistently higher than the SOC estimates also was encouraging.
However, in spite of probability weighting, both estimates failed basic sensitivity to scale tests,10 an impor-
tant criterion for deciding on the usefulness of WTP estimates to policy making (Bobinac et al. 2011c). Income
constraints were an unlikely source of insensitivity given the marginality of the presented health gains and the
similarity in ratios of WTP for larger and smaller gains for respondents with an income below and above the
mean and median. Possibly, the sensitivity test failed because of relatively small differences between expected
QALY gains. Nevertheless, the largest gain presented in the experiment was three times the smallest gain,
which should be deemed sufficient.
The observed insensitivity is less surprising for SOC values, given that these might be interpreted as a
referendum about the externality value of health gains. SOC WTP estimates may reflect the worth of having
a collective healthcare system as such (e.g., Hurley, 2000) and may relate to an ‘act of giving’ that buys moral
satisfaction (Kahneman and Knetsch, 1992; Andreoni 1990; Olsen, 1997) and not indicate the monetary value
of the underlying gains. In that sense, it may be seen as encouraging that the SII valuations perform somewhat
better in terms of sensitivity (Table IV). Without qualitative data, the reasons for this result remain unclear.
Therefore, this study hopes to encourage debate and research in the important area of eliciting valuations of
health gains that are relevant for societal decision making. In this context, more research in how to arrive at
social valuations of a QALY using estimates such as the ones presented here, varying across individual respon-
dents and potentially disease areas or risk distributions, remains pivotal as well.

10
Contrary to our results, Bobinac et al. (2011c) found that correcting for probability weighting improved the validity of WTP responses in
the context of individual valuations of QALY gains.

Copyright © 2012 John Wiley & Sons, Ltd. Health Econ. 22: 1272–1281 (2013)
DOI: 10.1002/hec
 Table III. Open-ended SOC and SII WTP per QALY estimates, weighted and non-weighted, on average and by certainty level and income group
(€, rounded to hundreds)

Copyright © 2012 John Wiley & Sons, Ltd.

SOC SII

WTP per QALY OE WTP per QALY WTP per QALY WTP per QALY OE WTP per QALY WTP per QALY
n NW (SD) OE TK (SD) OE P (SD) n NW (SD) OE TK (SD) OE P (SD)
Averages 504 119,600 (307,800) 65,100 (213,300) 52,200 (198,000) 500 188,900 (481,100) 83,200 (309,100) 59,200 (284,300)
Subgroup
analysis
Certainty Surely not 22 50,000 (165,500) 15,200 (50,200) 8,000 (26,200) 18 350,723 (773,400) 118,700 (284,400) 70,100 (183,400)
level Pretty sure not 38 88,800 (183,100) 35,700 (87,800) 25,200 (77,100) 22 76,700 (134,300) 41,700 (92,800) 34,400 (89,300)
Maybe yes, 199 130,100 (291,800) 73,100 (198,500) 61,100 (184,200) 195 213,800 (528,300) 91,600 (326,700) 66,800 (302,800)
maybe no
Pretty sure yes 179 121,100 (316,324) 63,300 (242,200) 51,700 (232,500) 191 173,600 (466,200) 76,900 (337,100) 58,100 (322,300)
Totally sure 66 125,200 (411,000) 68,500 (237,700) 56,800 (216,400) 74 156,000 (342,400) 65,500 (187,500) 47,000 (165,900)
VALUING QALY GAINS

yes
Income <€1000 82 85,700 (156,400) 37,300 (75,000) 27,000 (60,800) 56 106,400 (192,000) 48,000 (135,500) 36,300 (129,800)
group €1000–1999 169 108,400 (303,800) 57,500 (191,100) 47,000 (177,500) 174 161,400 (480,900) 69,500 (325,700) 51,100 (306,100)
€2000–3499 175 126,500 (288,900) 66,000 (211,300) 53,700 (202,000) 196 192,000 (531,500) 82,000 (339,600) 60,100 (316,900)
≥€3500 78 164,300 (446,700) 99,700 (318,800) 86,200 (298,900) 74 307,700 (408,500) 129,700 (246,100) 93,400 (218,300)
Note: NW, non-weighted probabilities applied; TK, Tversky and Khaneman probability function applied (1992); P, Prelec probability function applied (1998); SOC, social value; SII,
social-inclusive-individual value; WTP, willingness to pay; QALY, quality-adjusted life year; SD, standard deviation.

DOI: 10.1002/hec
1277

Health Econ. 22: 1272–1281 (2013)

 1278

Table IV. Log-linear regressions

SII NW SOC NW SII TK SOC TK SII P SOC P

Coeff SE p Coeff SE p Coeff SE p Coeff SE p Coeff SE p Coeff SE p

Log(QALY gain) 0.18 0.09 0.06 0.08 0.28 0.05 0.11 0.65 0.08 0.18 0.06 0.16 0.06 0.33

Copyright © 2012 John Wiley & Sons, Ltd.

0.08
0.11
0.09
0.05
Own health
0.27 0.31 0.38 0.1 0.32 0.75
0.27 0.31 0.37 0.12 0.32 0.78
0.29 0.31 0.33 0.12 0.32 0.71
Age
0.03 0.04 0.43
0.04 0.04 0.27
0.03 0.04 0.46
0.04 0.04 0.29
0.04 0.04 0.38
0.04 0.04 0.28
Age (squared) 0.002 0.005 0.74 0.003 0.004 0.41 0.0 0.00 0.78 0.0 0.0 0.43 0.0 0.0 0.64 0.00 0.00 0.43
Gender (female)
0.02 0.13 0.87 0.14 0.13 0.3
0.02 0.13 0.86 0.14 0.13 0.28
0.03 0.13 0.79 0.14 0.13 0.3
Children under 18 0.18 0.18 0.31 0.01 0.18 0.96 0.18 0.18 0.32 0.01 0.18 0.95 0.19 0.18 0.28 0.02 0.18 0.91
(yes)
University education
0.08 0.14 0.58
0.02 0.14 0.88
0.1 0.14 0.48
0.03 0.14 0.84
0.10 0.14 0.47
0.03 0.14 0.84
(yes)
Employed (baseline) — — —
Unemployed (yes)
0.24 0.2 0.23
0.11 0.2 0.58
0.24 0.21 0.23
0.1 0.2 0.61
0.26 0.2 0.19
0.1 0.2 0.61
Student (yes) 0.65 0.3 0.028
0.2 0.27 0.47 0.66 0.3 0.03 -.2 0.27 0.47 0.65 0.3 0.03
0.21 0.27 0.43
Housewife/ 0.33 0.25 0.18
0.1 0.21 0.64 0.35 0.25 0.16
0.01 0.2 0.67 0.31 0.25 0.21
0.09 0.2 0.64
A. BOBINAC ET AL.

househusband (yes)
Number of people
0.09 0.05 0.08
0.06 0.06 0.27
0.09 0.05 0.08
0.06 0.06 0.28
0.09 0.05 0.08
0.06 0.06 0.25
living on household
income
Log(Household 0.68 0.14 0.0 0.21 0.13 0.099 0.69 0.14 0.0 0.21 0.13 0.1 0.68 0.14 0.00 0.21 0.13 0.10
income)
Constant
0.02 1.55 0.99 1.61 1.36 0.239
0.79 1.52 0.6 1.51 1.36 0.27
1.17 1.46 0.43 1.82 1.32 0.17
N = 465 N = 446 N = 465 N = 446 N = 465 N = 446
2 2 2 2 2 2
R adj = 0.1095 R adj = 0.092 R adj = 0.103 R adj = 0.01 R adj = 0.107 R adj = 0.09
Note: Dependent variable: log (WTP).
NW, non-weighted probabilities applied; TK, Tversky and Khaneman probability function applied (1992); P, Prelec probability function applied (1998); SOC, social value; SII,
social-inclusive-individual value; QALY, quality-adjusted life year; SE, standard error.

Health Econ. 22: 1272–1281 (2013)

DOI: 10.1002/hec
 VALUING QALY GAINS 1279

5. CONFLICT OF INTEREST

None

APPENDIX A: EXAMPLE OF THE WILLINGNESS-TO-PAY QUESTION

The introduction to the question was as follows:

Suppose that the complete Dutch population is currently in the health state you assessed as the better
one. However, due to some virus, half of the population runs the risk of moving to worse health state
tomorrow. The virus will cause a deterioration in health from the better health state to the worse
health state for a period of 1 year, after which the person affected will return to the better health state.
The other half of the population is not at risk. These groups are otherwise similar (in terms of age,
gender, lifestyle, etc.).
Please imagine that you belong to the risk group [in SII format] / Please imagine that you do not belong to
the risk group [in SOC format]. The risk is 2% (4%/10%/50%). This means that 2 (4/10/50) persons in 100
will be affected and move to the worse health state. The diagram illustrates this level of risk.
The risk of health deterioration is completely avoidable by taking a medicine (without side-effects) every
month during this year. Persons taking the medicine will certainly remain in the better health state. The drug
will be funded by an increase in the monthly health insurance premium for all eligible Dutch citizens. The
premium increase will last for a period of 12 months.

The WTP elicitation proceeded as follows:

Respondents were asked to indicate how much they would be willing to pay for the drug that would reduce
the risk of the health deterioration to zero, through an increase in their monthly health insurance premium.
• First, they were asked to indicate on a payment scale the highest amount they would certainly pay and
subsequently the lowest amount they would certainly not pay for this medicine. [Ordered low-to-high
payment scale with monthly installments of (€) 0, 1, 2, 3, 4, 5, 6, 8, 10, 12, 15, 18, 20, 22, 25, 30, 40,
50, 75, 100, 150, 250, 500].
• Next, the respondents were asked to indicate in an ‘open-ended’ format the exact WTP for this medicine,
within the range they specified on the payment scale.
• Finally, they were asked, ‘Please indicate how sure they were about your WTP answer.’ [Answer options
were as follows: (i) totally sure yes; (ii) pretty sure yes; (iii) maybe yes, maybe no; (iv) pretty sure not;
(v) surely not.]
The concept of risk was explained as follows (translated from Dutch; based on Donaldson et al., 2010):
Imagine that in this group of 100 people the probability that someone becomes ill is 40%. In
other words, 40 people will become ill, and 60 people will not become ill.

For an impression of what 40% means for the probability that you will be one of the people
becoming ill, choose a green dot and click on it.

The probability that the dot you selected changed color is 40 in 100.

Copyright © 2012 John Wiley & Sons, Ltd. Health Econ. 22: 1272–1281 (2013)
DOI: 10.1002/hec
1280 A. BOBINAC ET AL.

APPENDIX B: DESIGN OF SCENARIOS, LEVELS OF RISK AND QALY GAIN

Choice scenario Health state 1 Health state 2 Level of uncertainty (%) Expected QALY gain
1 22222 11131 10 0.021
2 33232 33323 50 0.009
3 21312 12111 2 0.011
4 22323 21312 2 0.0074
5 22323 12111 2 0.015
6 21232 32211 4 0.0092
7 11112 22121 10 0.008
8 11122 22122 10 0.012
9 21323 22233 4 0.012
10 22331 21133 4 0.0075
11 21111 12121 50 0.066
12 23232 32232 50 0.028
13 11312 11113 10 0.0144
14 12311 11211 2 0.0068
15 32311 12311 10 0.0161
16 32311 11211 2 0.01
17 21111 12211 50 0.039
18 32313 32331 50 0.002
19 11211 22211 4 0.0047
20 23313 11133 50 0.042
21 11121 22112 10 0.021
22 12223 13332 10 0.014
23 11312 11211 2 0.0077
24 11332 11312 4 0.0132
25 11332 11211 2 0.0142
26 21222 33321 2 0.0082
27 22222 13311 50 0.0415
28 11112 22112 4 0.0047
29 33212 32223 4 0.0087

ACKNOWLEDGEMENT

This study is part of a larger project investigating the broader societal benefits of health care, which was financially
supported by Astra-Zeneca, GlaxoSmithKline, Janssen-Cilag, Merck and Pfizer BV. The researchers were free in
study design, data collection, analysis and interpretation, as well as writing and submitting the manuscript for
publication. The views expressed in this paper are those of the authors.

REFERENCES

Andreoni J. 1990. Impure altruism and donations to public goods: a theory of warm-glow giving. The Economic Journal
100: 464–477.
Barry BM. A treatise on social justice. 1989. Theories of Justice, vol. I. University of California Press: USA.
Bergstrom TC. 2006. Benefit–cost in a benevolent society. American Economic Review 96: 339–351.
Bleichrodt H, Pinto Prades JL. 2000. A parameter-free elicitation of the probability weighting function in medical decision
analysis. Management Science 46: 1485–1496.
Bobinac A, van Exel NJA, Brouwer WBF, Rutten F. 2011a. Inquiry into the relationship between the equity weights and the
value of a QALY. Submitted.
Bobinac A, van Exel NJA, Brouwer WBF, Rutten F. 2011b. Get more, pay more? An elaborate test of construct validity of
willingness to pay per QALY estimates obtained through contingent valuation. Journal of Health Economics 31: 158–168.
Bobinac A, van Exel NJA, Brouwer WBF, Rutten F. 2011c. Moving towards valid estimates of individual willingness to
pay for a QALY: using uncertainty and probability weighting. Submitted.

Copyright © 2012 John Wiley & Sons, Ltd. Health Econ. 22: 1272–1281 (2013)
DOI: 10.1002/hec
 VALUING QALY GAINS 1281

Brouwer WBF, Culyer AJ, van Exel NJA, Rutten FFH. 2008. Welfarism vs. extra-welfarism. Journal of Health Economics
27(2): 325–338.
Claxton K, Walker S, Palmer S, Sculpher M. 2010. Appropriate perspectives for health care decisions. CHE Research Paper
54, Centre for Health Economics, University of York.
Corso P, Hammitt J, Graham J. 2001. Valuing mortality-risk reduction: using visual aids to improve the validity of contingent
valuation. Journal of Risk and Uncertainty 23: 165–184.
Dolan P, Olsen JA, Menzel P, et al. 2003. An inquiry into the different perspectives that can be used when eliciting preferences
in health. Health Economics 2003: 545–551.
Donaldson C, Baker R, et al. 2010. European value of a quality adjusted life year. Final Publishable Report of the EUROVAQ
project (SP5A-CT-2007-044172). Retrieved from: http://research.ncl.ac.uk/eurovaq/
Donaldson C, Baker R, Mason H, Jones-Lee M, Lancsar E, Wildman J, Bateman I, Loomes G, Robinson A, Sugden R,
Pinto Prades JL, Ryan M, Shackley P, Smith R. 2011. The social value of a QALY: raising the bar or barring the raise?
BMC Health Services Research 11: 8.
Gravelle H, Brouwer WBF, Niessen LW, Postma MJ, Rutten FFH. 2007. Discounting in economic evaluations: stepping
forward towards optimal decision rules. Health Economics 16: 307–317.
Gyrd-Hansen D. Willingness to pay for a QALY. 2003. Health Economics 12: 1049–1060.
Gyrd-Hansen D, Kjar T. 2012. Disentangling WTP per QALY data: different analytical approaches, different answers.
Health Economics 21: 222–237.
Hirth RA, Chernew ME, Miller E, Fendrick AM, Weissert WG. 2000. Willingness to pay for a quality-adjusted life year: in
search of a standard. Medical Decision Making 20: 332–342.
Hurley J. 2000. An overview of the normative economics of the health sector. In Handbook of Health Economics, vol. 1A,
Culyer AJ, Newhouse JP (eds). Elsevier: Amsterdam.
Kahneman D, Knetsch JL. 1992. Valuing public goods: the purchase of moral satisfaction. Journal of Environmental Economics
and Management 22: 57–70.
Lamers LM, McDonnell J, Stalmeier PF. 2006. The Dutch tariff: results and arguments for an effective design for national
EQ-5D valuation studies. Health Economics 15: 1121–1132.
Mason H, Baker R, Donaldson C. 2008. Willingness to pay for a QALY: past, present and future. Expert Reviews of Pharma-
coeconomics and Outcomes Research 8: 575–582.
Olsen JA. 1997. Aiding priority setting in health care: is there a role for the contingent valuation method?. Health Economics
6: 603–612.
Olsen JA, Kidholm K, Donaldson C, Shackley P. 2004. Willingness to pay for public health care: a comparison of two
approaches. Health Policy 70: 217–228.
Pinto Prades JL, Loomes G, Brey R. 2009. Trying to estimate a monetary value for the QALY. Journal of Health Economics
28: 553–562.
Prelec D. 1998. The probability weighting function. Econometrica 66: 497–528.
Smith RD. 2007. Use, option and externality values: are contingent valuation studies in healthcare misspecified? Health
Economics 16: 861–869.
Tversky A, Kahneman D. 1992. Advances in prospect theory: cumulative representation of uncertainty. Journal of Risk and
Uncertainty 5: 297–323.
Van de Wetering E, Stolk E, van Exel N, Brouwer W. 2011. Balancing equity and efficiency in the Dutch basic benefits
package using the principle of proportional shortfall. The European Journal of Health Economics, DOI: 10.1007/
s10198-011-0346-7.

Copyright © 2012 John Wiley & Sons, Ltd. Health Econ. 22: 1272–1281 (2013)
DOI: 10.1002/hec