اعتبار، امکان سنجی و پذیرش از زمان تجارت کردن و استاندارد ارزیابی گمبل در مطالعات ارزش گذاری بهداشت: مطالعه در یک جمعیت چند قومیتی آسیا در سنگاپور

Methods Through in-depth interviews performed among Chinese, Malay, and Indian Singaporeans (education ≥ 6 years), we assessed validity of SG/TTO methods for eliciting health preferences by hypothesizing that 1) SG/TTO scores for three hypothetical health states (HS) would exhibit ranked order (decreasing scores with worse HS); and 2) more subjects would rate the most severe HS as worse than dead. Subjects also evaluated feasibility and acceptability of SG/TTO using a 10-point visual analog scale (VAS) and open-ended questions. Ratings were compared using Kruskal–Wallis, Wilcoxon signed-rank tests or tests of proportions. Results Validity: In 62 subjects (90% response rate), as hypothesized, SG and TTO scores exhibited ranked order with increasing HS severity (SG: 0.85, 0.08, −19.00; TTO: 0.85, 0.00, −0.18). More subjects rated the most severe HS as worse than dead (SG: 8%, 39%, 59%; TTO: 8%, 45% and 62%). Feasibility Subjects felt SG and TTO were easy to understand (median VAS scores: 8.0 vs. 8.0, P = 0.87) and to complete (8.0 vs. 8.0, P = 0.84). Acceptability: SG and TTO were well accepted, with TTO less so than SG (median [interquartile range] offensiveness: 2.0 [0, 4.0] vs. 2.0 [0, 3.0], P = 0.045). Overall, subjects did not have a clear preference for SG/TTO (50% vs. 45%, P = 0.70). Conclusions This study suggests the validity, feasibility and acceptability of SG and TTO for population-based HS valuation studies in a multiethnic Asian population.

Preference-based health-related quality of life (HRQoL) instruments, including the EQ-5D [1], Health Utilities Index [2] and the SF-6D (derived from the SF-36) [3] summarize HRQoL in a single index score and are used in cost-utility analyses with the aim of informing clinical policy and resource allocation in health care [4]. Each preference-based HRQoL instrument typically comprises a health classification system for describing the respondent in terms of a health state (HS), and a utility function that maps each HS to a utility score. The utility function is typically derived from a population-based HS valuation study in which respondents express their preferences for individual HS. These values are then aggregated using statistical modeling techniques to derive the utility function [5,6]. A variety of valuation methods have been employed in eliciting HS preferences, with choice-based valuation methods clearly preferred [7]. Standard gamble (SG) and time trade-off (TTO) are two choice-based methods with demonstrated acceptability, reliability, and validity in Western sociocultural contexts [7,8]. Both SG and TTO have established theoretical underpinnings [9,10], with the former being frequently referred to as the “gold standard” because it is directly based on the axioms of expected utility theory [9] although the latter requires the additional assumptionthat utility in additional healthy time is linear with respect to time [11]. Nevertheless, TTO is preferred by some (though not by others [12]) for its ease of implementation [9,10]. Importantly, studies have shown that health utilities elicited by both methods are usually different, with SG typically generating higher scores [13,14]. Hence, in planning studies to use choice-based preference measures such as SG and TTO in settings in which there has been relatively little experience with these measures, it is important to obtain empiric evidence with regard to validity, feasibility, and acceptability of each of these methods. Existing studies comparing SG and TTO have largely taken a quantitative approach [7,13,15], with no studies (to the best of our knowledge) having addressed qualitative aspects to better understand individual subject’s preferences and behavior when completing these exercises. We therefore conducted such a study in a multiethnic Asian population, in which a diversity of views might be expected. In this study, which is likely to be the first head-to-head comparison of SG and TTO methods in an Asian population, we aimed to assess the validity, feasibility and acceptability of SG and TTO and to evaluate if systematic differences in SG and TTO scores observed in other studies [13,14] were also observed in this Asian population. We defined 1) acceptability as the degree to which subjects are satisfied with SG and TTO and have no objections to these methods; and 2) feasibility as the extent to which SG or TTO exercise may be done practically and successfully. We then aimed to move beyond descriptive statistics (e.g., completion rate and missing data) to gain an insight into factors influencing individuals’ preferences for SG or TTO.

In this study among Chinese, Malay, and Indian Singaporeans of various ages and educational levels, we found that both SG and TTO were valid, feasible, and acceptable for eliciting health preferences in this population. Nevertheless, there were systematic differences in utility scores elicited by both methods. We also evaluated factors influencing individuals’ preferences for SG or TTO methods. In multivariable analysis with adjustment for ethnicity, age (per 10 years), gender, and education, older age was only marginally associated with preference for TTO over SG. Hence, our results suggest that both SG and TTO may be used among subjects of genders, various ages, ethnicities, and education levels in population-based HS valuation studies to be conducted in this population. To the best of our knowledge, this is the first such study among Asians, and provides a useful framework for comparison with future studies in other Asian sociocultural contexts. Several aspects of our findings deserve comment. First, consistent with other published studies [18,19], we found that agreement between SG and TTO utility scores was generally poor, with SG scores being generally higher than TTO scores. It was interesting that the lack of agreement was consistently evident in the three HS studied (a mild, moderate, and severe state of health). Second, although the feasibility of SG in population-based study had previously been questioned [20], our results suggest that SG may be as feasible as TTO method in such a setting, given that our subjects preferred SG as much as, if not more than, TTO (except for older subjects who marginally preferred TTO to SG). Third, although subjects rated both SG and TTO instructions as easy to understand, there were 10 subjects who felt TTO instructions could be revised to improve clarity and reduce offensiveness, a finding consistent with the trend (not reaching statistical significance) that more subjects preferred SG. That nine of these subjects were Malay suggested that TTO instructions could be particularly difficult for this ethnic group (although the differences in ratings among ethnic groups were not statistically significant). As there were almost equal numbers of these subjects who completed the English (n = 4) and Malay (n = 5) language versions, the difficulty is unlikely to be due to translation. Fourth, open discussion of death has been thought to be taboo among Asians [21,22], with many avoiding this topic because of the perception of bringing “bad luck” on oneself by discussing death, even if it is just speaking the words with the same phoneme as the word for death [23]. Nevertheless, it was interesting and encouraging that our subjects were comfortable with discussing death. The results of this study have several implications, some of which would have a bearing on future research. First, to date, this is the only published study in Asia that concurrently evaluated both SG and TTO methods for eliciting health preferences. Our results suggest that both methods are valid, feasible, and acceptable and may be used in future clinical trials for direct measurement of health utility scores in this population. Second, the semiquantitative nature of this study provided glimpses of the decision making process of subjects completing such studies, which would be useful in explaining differences in preferences for the same HS. For example, we found that religious beliefs may influence an individual’s health preferences in that giving up life years is not an option for some individuals. Given that 7.9% of the Singaporean population are Indians and 13.9% are Malays [24] and that 4.8% of Indian and 10% of Malay subjects in this study felt that discussing trading off life years was offensive for religious reasons, we estimated that up to 1.8% of the Singaporean population may find that discussing trading off life years was offensive because of their religious beliefs. Thus, these individuals are unlikely to say, “I would rather die immediately than be confined to bed,” and would thus assign higher preference scores to the HS of being confined to bed than those whose religious beliefs did not inhibit trading quantity for quality of life. Nevertheless, the extent of the influence of religiosity on actual SG and TTO scores remains to be evaluated in future studies. This is because in a local study that measured health preferences for HUI3 HS among Chinese, Malays, and Indians using the SG [25], there were no ethnic differences in health preferences for these HS. Nevertheless, the study did not include any measures of religiosity. We recognize several limitations of this study. First, the study sample was not drawn at random from the Singapore population, which was not feasible because of cost and logistic issues. We therefore attempted to improve representativeness by specifying criteria to ensure equal gender and ethnic representation with a wide age range. Second, to reduce respondent burden, we asked subjects to evaluate only three HS each for SG and TTO, although the number of HS to be valued in existing valuation protocols typically exceeds three [26,27]. Hence, generalizability of our findings to HS valuation studies involving more than three HS needs confirmation. Third, we have used simple VASs to measure risk attitudes as no other suitable measures were available for this population at the time of this study. The association (or the lack of it as found in this study) between risk attitudes and preference for SG or TTO methods in this population could thus be further studied when better measures for assessing risk attitudes become available. In conclusion, this study found both SG and TTO methods to be valid, feasible, and acceptable among Chinese, Malays, and Indians of various sociodemographic backgrounds. The findings are therefore likely to be applicable for population-based HS valuation studies in this multiethnic Asian population. Nevertheless, generalizability of our study findings needs to be confirmed in larger studies surveying subjects who are representative of the population being studied. We would like to thank Mdm. Halimah Beevi and Ms. Syzawani Bte Amrun for their assistance in conducting the interviews. Conflict of Interest: It should be noted that David Feeny has a proprietary interest in Health Utilities Incorporated, Dundas, Ontario, Canada. HUInc. Distributes copyrighted Health Utilities Index (HUI) materials and provides methodological advice on the use of HUI.