Typical elicitation
In the SG method, QALY weights for health states are determined by comparing a specific number of years in health state Hi with a gamble (a treatment) offering two reference outcomes, which are a probability p of full health for the same number of years and a probability 1-p of immediate death. The probability p of full health is varied until the respondent is indifferent between the two alternatives. The indifference probability is the weight to be assigned to health state Hi. On the other hand, the TTO method typically requires comparing Y years in a particular health state Hito X years in full health. The number X is varied until the respondent is indifferent between the alternatives. The QALY weight assigned to health state Hiis then set equal to X/Y. The TTO and SG are choice-based techniques that differ significantly in that TTO is riskless whereas SG is framed in terms of risk and incorporates the respondent's attitude towards risk. According to prospect theory, attitudes towards risk consist of two components, one of which reflects sensitivity to outcomes and the other sensitivity towards chance [16].
On the other hand, WTP is the only benefit measure reflecting conventional microeconomic properties and requires respondents to consider a health-wealth tradeoff, making it a cognitively more demanding method. It imposes no restrictions on which attributes of a program can be considered in its valuation [17]. Anything, over which the individual has preferences, including a particular health outcome, is considered to be an "economic good". Similarly to the QALY approach, the value of reducing a specific mortality risk in the current period depends on life expectancy, competing mortality risk and the individuals's health if he/she is to survive the risk, baseline risk and on income or wealth. Furthermore, a person's WTP is clearly limited by his/her ability to pay [18] and it has been shown that WTP is positively related to income [19].
Ethical and cultural barriers
In most facilities the review boards and clinical staff, due to perceived ethical considerations, refused to grant permission for the patients to be subjected to questions implying, even hypothetically, the age of death, i.e. the timeframe Y mentioned previously. Religious issues had to be considered also because the duration of life and time of death are perceived as "unknowns" which are controlled "from above" and not as tradable products. Furthermore, a specific timeframe, e.g. 10 or 20 years, could bias the utility scores if subjective expectations about the age of death were different [20]. In this study, the "gamble" was a kidney transplant and the outcomes were a probability p of full health and a probability 1-p of immediate death with the latter again raising concern from clinical staff. Ethical implications were foreseen in that the immediate death probability 1-p could generate false perceptions to patients about the actual risks associated with kidney transplants and this, in turn, might result in increased unwillingness from patients to undergo a transplant should they have the opportunity to do so in a real-life situation.
In the Greek health system, as well as in most others worldwide, the cost of dialysis treatments and kidney transplants is fully covered by the public sources of financing and there are no existing co-payments required from the patients. The ethical concern in the case of WTP was that patients might become severely misconceived in that they would be required to pay for a kidney transplant in a real life situation. Given the overall low educational level of the patients, many clinicians felt that the WTP questions could create a false impression about the existence of an underlying "black-market" for transplants where the wealthier and "better connected" could skip the waiting list and "purchase" a kidney transplant.
Adapting study design
These objections had to be taken into account in designing the study. The first methodology adaptation involved offering the respondents health state Hi, (their current health state) for the rest of their expected lifespan instead of a fixed number of years. The difficulty in calculating utilities, particularly in the case of TTO, came from the fact that patients' remaining life years were unknown as this is part of the contract of life. We adopted the "fair innings" argument according to which all people are entitled to a normal span of life [21] and encouraged patients to subjectively assume that dialysis would secure for them a number of life years similar to that of the general population. For calculation, we arbitrarily assumed life expectancy at 80 years of age without disclosing this information to the patients. Those already older or willing to tradeoff years which added to their current age, exceeded 80, were excluded. In the case of SG, the problem was less complicated since the fraction of remaining lifespan the individual would be willing to sacrifice to improve health does not depend on the remaining lifespan, a condition known as "constant proportional tradeoff longevity for health" [22].
In both SG and TTO, the health state to be assessed is compared to an alternative that can be framed either as a gain or a loss and it is well known, from many studies, that the type of framing affects behavior, especially in the case of the SG method [23]. The reason for this diversion may be loss aversion, meaning that if a change is perceived as a loss compared to a reference level, it results in a greater change in utility than if the same change is perceived as a gain. Since dialysis patients are already in a less than ideal health state it is easier for them to specify the maximum probability 1-p of the unfavorable outcome rather than the minimum probability p of the favorable one. This particular adaptation of the SG elicitation methodology helped to overcome one more problem. Specifically, when outcomes are not certain but occur with known probabilities, people transform these probabilities into decision weights and in particular, they overestimate small probabilities and underestimate large probabilities [24].
Although the WTP methods used in this study are comparable to those undertaken in other countries, the actual elicitation techniques were chosen according to particularities of the sample and concerns from the physicians. Considering, once again, the low educational level of the respondents and that WTP studies are, to date, practically unknown in Greece, it was important that the elicitation format not create suspicions about the intent of the questions. For example, the bidding game, in which an auction process is simulated often resembling actual market situations, could make the patients uncomfortable and unwilling to participate or even make them question if health care is truly free. In any case, bidding games require interviewers or interactive computer programs and are therefore more costly than other methods, which can be carried out via self-administration.
In this study, WTP was measured using the dichotomous format followed by an open-ended question, a technique that has been shown to increase the statistical efficiency of the responses [25]. Patients were initially asked if they were willing to pay, out-of-pocket, €15,000 for a kidney transplant. This bid is close to the actual cost of a kidney transplant in Greece. It is unarguably a heavy economic burden for the average Greek, however it is realistic and, in most cases, could be raised with the help of the greater family or friends. The objective was for respondents to value the transplant in comparison to other personal and/or family needs and produce realistic answers. In view of this, they were advised to consider that paying this amount would imply that other needs might not be satisfied. It was emphasized once more that the situation was strictly hypothetical and no payments would be requested in a real situation.
In a subsequent open-ended question, patients responding "yes" to the dichotomous question were asked if they would be willing to pay more than €15,000 and, if so, to specify the exact amount. Those answering "no" were asked if they would be willing to pay a smaller amount and again to specify. In cases where only the dichotomous question was answered, the final WTP was taken as €15,000 for the "yes" respondents whereas the "no" respondents were taken as true zero bidders and included in the analysis. The actual questions asked in this study have been translated and are shown in Fig. 1.
Sample and data collection
We randomly selected 25% of the dialysis facilities currently operating in Greece (32 out of 128) and requested permission to conduct the study using the standard elicitation methods. Each review board examined the survey and only three granted immediate permission without any methodological or ethical objections. The other twenty-nine perceived at least one of the problems mentioned previously. We reapplied with the adapted versions of elicitation, to overcome initial denial. This resulted in twenty-one more facilities agreeing to participate (24 in total – 75%), with the other eight remaining unconvinced and eventually excluded. The adapted versions of elicitation were used in all dialysis centers.
To facilitate a large population study without the need for costly interviewers, the questions were paper-based for self-administration, a technique that has been shown to be a reliable substitute to typical utility elicitation methods in the case of SG [26, 27]. As for WTP, self-administration is once again a logical substitute to the otherwise preferred method of face-to-face interviews. However, the latter represents the most costly way of collecting data and, not surprisingly, in a review of 71 WTP surveys of health and health care published in English during the period of 1985-1998, only 27 (38%) employed face-to-face data interviews compared to 34 (47.9%) employing self-administration or post [28].
Each facility appointed a dialysis nurse who was trained to distribute the questionnaire, explain the purpose of the study and provide assistance when needed. Adult patients (aged 18+) were eligible for the study and were chosen by the clinical and nursing staff in each facility on the basis of their mental and physical ability to read, comprehend and complete the self-administered survey, with the least possible assistance. Others not fulfilling this criteria, e.g. minority groups or illiterates, were deemed unable to participate. In order to ensure informed consent, the patients were asked to read an accompanying letter emphasizing that participation was voluntary and anonymous and that only aggregate results would be reported. The survey consisted of common socio-demographic and clinical questions, TTO and SG utility questions and two WTP questions. On aggregate, 606 dialysis patients from the 24 participating facilities were deemed suitable candidates. The response rate was 78.5% with 504 patients eventually completing the survey over the period April 2004 – December 2004. Monetary values are reported in Euros for 2004 (1€ = 1.3 USD).
Analysis
Spearman's correlation coefficients were used to analyze the direction and the strength of the relationship between the health-state utilities and willingness to pay. Parametric tests (independent samples t-test and ANOVA) were performed to examine differences in scores, for each preference-based utility measure and WTP, as a result of various socio-demographic and clinical factors such as age, sex, education, familial status, employment, comorbidities and previous unsuccessful kidney transplant and multiple linear regression analysis was performed to determine the most significant predictors. All analyses were performed using SPSS software, version 12.0 (SPSS Inc., Chicago IL).