Setting
Kermanshah city, the capital of Kermanshah province, is located in western Iran. Based on the 2016 population census of Iran, the city had a total population of about two million people. The socio-economic status of the people is low, and the city's contribution to the national gross domestic product (GDP) is only about 1.7% to 2%.
Study design, study period, and sample size
A cross-sectional study was conducted on a total sample of 943 adults aged 18 years and above, from the general population of Kermanshah city, to elicit their WTP for one additional QALY gained from a hypothetical life-saving treatment during September to December 2019. The Mitchell and Carson (20) was used to determine the appropriate sample size.
Where n represents the calculated sample size at =10%, V=2.5, and (the difference between the true and estimated WTP values). Thus, the computed sample size was at 786. However, considering the attrition rate of 20% and generalizability of the findings, the final sample for the study was 943. We used a multistage sampling technique to select the study participants by dividing the city into the western, eastern, central, northern, and southern geographic areas. Then, we equally divided the final sample into five (n=189) and selected the study participants using a systematic random sampling technique.
Data collection and variables
We used a self-administrated questionnaire to obtain data for eliciting the participants’ WTP for one additional QALY gained by oneself and a family member using a hypothetical life-saving treatment (4). The self-administered questionnaire focused on the participants’ current health state, WTP for one additional QALY gained from life-saving treatment, and sociodemographic characteristics (Appendix 1). Before the final data collection, five health economists checked the questionnaire for its content validity, a revision was made based on their opinions, and a pilot test was conducted on 30 participants to ensure the understandability of the questions and the hypothetical scenarios.
We used the Iranian version of the EQ-5D-3L as well as a visual analogue scale (VAS) measure to obtain the respondents’ health utility values (4). The EQ-5D-3L had five dimensions consisting of mobility, self-care, usual activities, pain/discomfort, and anxiety/depression dimensions with; three-level responses of: no problems, some problems, and extreme problems. A participant had to use one of these responses in each dimension to indicate his/her current health state (21). Again, we allowed the respondents to identify their current health state on a 100-unit thermometer analogue scale extending from 0 (almost dead) to 100 (perfect health) for the VAS valuation(4, 22).
As in a previous study (4), we used two hypothetical life-threatening condition for the individual participants and their family members to estimate their maximum WTP value for one additional QALY gained. The assumption used for the participant in the first scenario was as follows: “Suppose you had a life-threatening disease for the past year. There is a cure (treatment) for the disease., and if you do not get the treatment now, you will die today. If you get treated, you will be back to your original health state and live only for one more year”. In the second scenario, we presented the respondents with a similar to the first one but asked them to imagine the situation for his/her family members (a different perspective).
We elicited the maximum WTP value for one more QALY gained by a family member from the hypothetical intervention using the contingent valuation method (CVM) (23). The CVM is one of the most commonly employed methods to elicit the WTP of individuals for one additional QALY gained using an intervention. For example, a systematic review that determined the WTP per QALY reported that 92.85% of the studies applied the CVM, and only one study used a discrete choice experiment method to estimate the WTP of participants (16). Furthermore, we used the payment card (PC) method, one of the CVM methods, accompanied by a follow-up of open-ended questions to identify the participants' WTP. The PC applies a visual scale consisting of a range of potential bid values presented to the respondents to indicate their best WTP value (24). The PC was comprised of 15 bid values ranging from the lowest US$ 78 to the highest US$ 19,381, and we presented the bid to those that showed a positive attitude toward the WTP. However, we included the values below US$ 78 and above US$ 19,381 on the PC scale to avoid limiting the participants' chance of not responding.
The follow-up questions elicited the respondents' exact WTP values. We used values ranging from zero to more than US$ 19,380 from a pilot study conducted in 2019. In this study, we utilized the PC with closed-ended questions because it covered a wide range of bids and helps avoid fatigue and confusion of the respondents in the valuations. These limitations are likely to occur when using other CVM methods such as the dichotomous choice, bidding game formats, and the multiple bounded discrete choice methods, where the respondents had to bargain to show the WTP values. The value of US$ 1 at the time of the study was equivalent to 128,986 Iranian Rials (IRRs) (25). Sociodemographic related variables included in the analysis were age, sex, marital status, individual monthly income, education status, health insurance coverage, birthplace, and having a chronic disease.
Data analysis
We calculated the utility scores from the additional QALY gained using the EQ-5D-3L and VAS valuation methods. The data were initially scaled on the VAS from the best to the worst imaginable health state and then rescaling the scores of the respondents from 100 to 0 using the following formula:
Where VASrchs and VASraw respectively represent the scores of the rescaled current health state and current health state. Deathraw and ‘11111’raw are the scores of almost dead and perfect health states, respectively.
In this study, the additional QALY gained by each respondent was the difference between the utility measure of the current health state using the EQ-5D-3L or VAS and the almost dead health state, and calculated using the following mathematical equation:
Where Udeath is the utility from the dead health state which is equal to 0.000 and Uchs is the utility from the current health state. The WTP for the additional QALY gained is the amount of WTP per an additional QALY gained by oneself or a family member.
We used the Mann-Whitney and chi-square tests to explore the association between the continuous and categorical explanatory variables and the WTP for the life-saving treatment of the respondents, respectively. The data on WTP for the additional QALY gained from the life-saving treatment using the EQ-5D-3L and VAS methods were positively skewed. Similar to previous studies (26-29), we applied the Tobit regressi
Where Udeath is the utility from the dead health state which is equal to 0.000 and Uchs is the utility from the current health state. The WTP for the additional QALY gained is the amount of WTP per an additional QALY gained by oneself or a family member.
We used the Mann-Whitney and chi-square tests to explore the association between the continuous and categorical explanatory variables and the WTP for the life-saving treatment of the respondents, respectively. The data on WTP for the additional QALY gained from the life-saving treatment using the EQ-5D-3L and VAS methods were positively skewed. Similar to previous studies (26-29), we applied the Tobit regression model to explore the relationship between the WTP for the additional QALY gained and the explanatory variables, and to handle the possible limitations that may arise when using other models. Furthermore, we estimated the marginal effect of the and , where is the explained marginal effect for the probability of being uncensored and is the explained marginal effect for the expected WTP value conditional on being uncensored: E (WTP | WTP>0). The age, gender, educational level, health insurance coverage, marital status, birthplace, chronic disease status on oneself, chronic disease status in a family member, death of a family member in the past year, and monthly household income were the dependent variables. The Stata statistical software package version 14.2 performed all the analyses, and we considered the findings as statistically significant at the p-value of less than 0.05.
on model to explore the relationship between the WTP for the additional QALY gained and the explanatory variables, and to handle the possible limitations that may arise when using other models. Furthermore, we estimated the marginal effect of the and , where is the explained marginal effect for the probability of being uncensored and is the explained marginal effect for the expected WTP value conditional on being uncensored: E (WTP | WTP>0). The age, gender, educational level, health insurance coverage, marital status, birthplace, chronic disease status on oneself, chronic disease status in a family member, death of a family member in the past year, and monthly household income were the dependent variables. The Stata statistical software package version 14.2 performed all the analyses, and we considered the findings as statistically significant at the p-value of less than 0.05.