Understanding Canadian stakeholders’ views on measuring and valuing health for children and adolescents: a qualitative study

There has been a growing interest in measuring and valuing health for children and adolescents in recent years [1]. Since the development of the EQ-5D-Y, a child-friendly version of the EQ-5D in 2010 [2, 3], a series of methodological research and empirical applications using the EQ-5D-Y have been conducted worldwide [4‐13]. For adults, methods for the measurement and valuation of health have been well established. However, there are unique challenges when measuring and valuing child health. It has been demonstrated that health state preferences of children and adolescents differ from those of adults and therefore led to the argument supporting the development of a separate value set for children and adolescents [4, 5, 14]. Further research was then conducted to explore the impact of the perspective used in the valuation tasks [8, 15]. In the meantime, the EQ-5D-Y has been applied in observational research and routine outcome assessments [11‐13].

In 2020, an international valuation protocol for the 3-level of the EQ-5D-Y (EQ-5D-Y-3L) was published [16]. A few countries have developed a value set for the instrument using the protocol [17‐24]. In the meantime, the EQ-5D-Y-5L, a revised version of EQ-5D-Y with expanded response levels, has been developed to measure health for children and adolescents with improved sensitivity [25]. Despite the publication of the valuation protocol, key issues remain, including the choice between the EQ-5D-Y-3L and Y-5L (the difference between these two health state descriptive systems and the acceptability of increasing response levels), source of health state preferences (who should be asked to provide health state preferences for child health valuation?), and perspective (adult’s or child’s own perspective?) [13, 26].

The choice of valuation methods used to elicit preferences are also critical for valuing health for children and adolescents. Traditional methods such as time trade-off (TTO) have been established in adult populations. However, empirical evidence about these methods in child health valuation is lacking. Discrete choice experiment (DCE) is potentially promising with its lower cognitive burden and the feasibility through online survey. The latent utility generated by DCE needs to be anchored using additional tasks, for example, with immediate death presented as an additional option or duration presented as an additional attribute [27‐29]. Whether these variations of DCE work for child health valuation is less clear [13, 26].

Besides, child health valuation study designed according to such a protocol would fail to incorporate considerations deemed important in local contexts. A utility value set will eventually be used to support health technology assessment (HTA) and local reimbursement decision making in which the priority is to efficiently allocate resources in the local setting rather than to make an international comparison. Engagement of stakeholders is therefore recommended to identify and incorporate context-specific factors into child health valuation [30]. While there is a lacking such evidence of taking views of stakeholders into account for valuing health for children and adolescents.

Research in child health has been a top priority in Canadian health research. A prominent example is the Ontario Child Health Study (OCHS) that has been conducted since 1983 [31, 32]. Canadian researchers developed one of the first child health instruments, the Health Utility Index (HUI) in 1992 [33]. Although no specific instrument is recommended by the HTA agency in Canada, most of the research related to child health valuation has been limited to HUIs. To date only a few studies have used the EQ-5D-Y to measure child health in Canada [6, 34, 35]. The aim of this study was, therefore, to understand the views of Canadians stakeholders on the issues related to measuring and valuing child health using the EQ-5D-Y, including the preference of using the EQ-5D-Y-3L or the Y-5L, the source of health state preferences, the perspective of the valuation task, the preference of using TTO or DCE approaches, and potential ethical issues when directly asking children to answer the valuation tasks.

We approached 17 stakeholders and 15 agreed to participate. The mean (SD) time duration of the interview was 51.3 (9.5) mins. As shown in Table 1, 11 (73.3%) participants were female. Five (33.3%) participants were aged 31–40 years, three (20.0%) 41–50 years, and seven (46.7%) 51–65 years. Ten (66.7%) participants had PhD degree and 13 (86.7%) were parents. Two (13.3%) participants were from the pharmaceutical industry, four (26.7%) the HTA agencies, four (26.7%) health care providers, and five (33.3%) universities (including two child health researchers and a bioethicist). Most participants had previous experiences of economic evaluation (n = 10, 66.7%), were familiar with the concept of health utility (n = 13, 86.7%) or involved in measuring health utilities (n = 9, 60.0%). Table 2 provides a summary of responses to the key interview questions.

Measuring and valuing child health poses unique methodological and normative questions. Stakeholder engagement can provide a useful insight to tackle these issues. Canadian stakeholders suggested that we elicit health state preferences from children and adolescents directly from their own perspective. Furthermore, most participants considered DCE with duration an easier preference elicitation method compared with other types of DCE or TTO. This stakeholder engagement provides important feedback that will inform the design of child health valuation in Canada.

Assessing innovative health technologies for pediatric populations requires patient-reported outcome measures suitable for children. There are 15 preference-based instruments that have already been developed and validated to measure health for children of various ages [38, 39]. These instruments were designed to generate health utilities. Empirical studies have consistently found that children’s health state preferences were different from those of adults [4, 5]. This was a key rationale for a separate value set for the EQ-5D-Y stated in the international protocol [16]. However, when it comes to the source of preferences, the protocol recommends using general adult populations based primarily on the legal age threshold for voting and paying taxes (18 years in most countries) [16]. Most of our stakeholders disputed the legal age justification for child health valuation. They provided convincing arguments that the voices of cognitively able children should be considered when allocating healthcare resources. Only two participants in our interviews who supported the use of adults (both happened to be health economists) stated that the use of adult preferences would conform to the current HTA guidelines. However, it is important to note that current guidelines are for adults and in fact no HTA agencies have provided any recommendations on child health valuation [38]. If a separate value set for a child preference-based instrument is needed because health state preferences differ between children and adults, why are the preferences of children not used in developing the value set?

Ethical concerns were another factor for the recommendation in the protocol. It is true that talking about death could be sensitive and have potential impact on some children. However, it might not be possible to completely shield children from being exposed to these subjects in the internet era. Some participants raised the concern, but they nonetheless considered that children should be asked for health state preferences if we approach this carefully and with the consent from parents. Indeed, there are alternative designs to avoid talking about death in the valuation task, for instance, DCE with duration.

A complication of recommending the use of preferences from adults as proxy for child health valuation is from which perspective the tasks should be framed. Adults are asked to state their preferences from the perspective of a 10-year-old child according to the protocol [16]. In general, many adults (parents in particular) would agree that it is difficult to know what a child thinks or prefers. Moreover, the 10-year-old child perspective was an arbitrary choice initially [4], but subsequently has been used by other studies for comparability [5, 10, 14]. This perspective, or any perspective other than the respondent’s own, creates new challenges. Evidence has shown that the image of a hypothetical 10-year-old child is different among adults, which caused variations in health state preferences [8, 40]. Furthermore, adults’ preferences change with the age of the hypothetical child [41]. Using the child’s own perspective will avoid this complication.

Given the challenges in child health valuation, an international protocol can provide methodological guidance and allow for comparability across studies. However, addressing some of these challenges requires empirical evidence, for others they impose normative questions that go beyond science. Therefore, stakeholder engagement can be important to inform the design of child health valuation to align with the expectations of local stakeholders [30]. Unfortunately, it is rare that stakeholders are engaged prior to a child health valuation study for various reasons. The recently published roundtable discussion in the US was one of the few [42]. Most of US stakeholders expressed that adolescents could relate to a 10-year-old’s perspective better than adults and were capable of self-completing valuation tasks, and thus should be directly included in the valuation study. They also raised concerns that adults would be inconsistent in their views about a 10-year-old [42]. Similar to what we found in our study, US stakeholders who were more familiar with HTA were also more inclined to favor the use of adult preferences based on the legal age for voting and paying taxes [42].

There are a few strengths in our stakeholder engagement. First, interviews were chosen to allow us to have sufficient time to engage with each participant and obtain their insights into this rather complex topic. We preferred interview to other qualitative study designs (e.g., focus group) so we can capture the views from individual stakeholders with diverse background and expertise, instead of reaching a consensus among them. Second, our stakeholders include pediatricians, child health researchers, and a bioethicist who can share their direct experience with children. Another strength is that all interviews were conducted by one author, which ensured consistency across the interviews.

One limitation of this qualitative study was that we only recruited those who were highly educated health care professionals or experts in health-related research. We were concerned about the complexity of this topic for laypersons in the public. The two participants who were not familiar with the concept of health utility felt that some parts of the interview (e.g., preference elicitation methods) were rather technical. During the interview, we used examples of TTO and DCE tasks to help participants understand the main process of preference elicitation. Providing introductive material in advance may be also helpful to facilitate understanding. Another limitation is that we designed the interview by focusing on the issues around child health valuation methods. Their potential impact on health policy making was only briefly discussed due to the time constraint. We felt this issue is nonetheless important, especially for decision makers, to be aware of important implications related to child health valuation.

Most Canadian stakeholders supported eliciting health state preferences directly from adolescents using their own perspective for child health valuation. DCE was considered a more appropriate preference elicitation method for this age group. Future studies should assess the feasibility of directly eliciting preferences for the EQ-5D-Y from adolescents in Canada.