EQ-5D(-Y) Valuation from Adult and Child Perspectives: Where Does the Empirical Evidence Leave Us and How Should We Proceed?

Abstract

Valuing pediatric health-related quality of life (HRQOL) is essential for economic evaluations in child healthcare. Instruments like EQ-5D-Y were developed for this purpose. A key methodological innovation—though controversial—has been the use of the child perspective for valuation of EQ-5D-Y health states, where adults value health states imagining a 10-year-old child. This paper critically reviews empirical findings on this approach, examines potential biases, assesses alignment with stakeholder views, and explores alternatives. We relied on a targeted review of empirical literature, including studies comparing adults valuing their own health (adult perspective) and using child perspectives, as well as stakeholder opinion studies. Findings were synthesized into ten key learnings: (1) Child-perspective valuations are typically higher than adult ones for the same health states. (2) Adults prioritize pain/discomfort and being sad/unhappy differently for children. (3) Child age has minimal impact. Mechanisms contributing to differences between adult and child perspectives include (4) discomfort with child death, (5) different valuations of life duration, (6) psychological distance, (7) emotional difficulty deciding for others, and (8) external goals influencing results. Stakeholder engagement shows that (9) the effects of using child perspectives do not align well with societal preferences, and (10) stakeholders express a preference for approaches that directly involve children and adolescents in valuation tasks. We conclude that relying on child perspectives may introduce systematic biases, potentially undermining the validity of pediatric health utilities. A re-evaluation of current valuation methods for EQ-5D-Y may be warranted, with greater consideration for direct child involvement, mapping techniques, and group-based deliberative approaches.

Key Points for Decision Makers

Economic evaluations in pediatric populations rely on instruments like EQ-5D-Y, for which adults provide ‘child perspective’ valuations to balance ethical and practical concerns.

Evidence shows child perspective utilities are usually higher than adult perspective valuations for the same health states. These differences are not only due to perceived severity, but are also related to factors like unwillingness to trade-off life duration or consider death for children, emotional discomfort, psychological distance, and external goals, which may introduce systematic bias.

There is widespread stakeholder opposition to relying only on child perspective valuations. Current EQ-5D-Y valuation methods may need to incorporate direct child involvement, mapping approaches, and deliberative techniques.

Introduction

In the past decades, the number of economic evaluations of medical innovations in pediatric populations has grown substantially [1–3]. In particular, an increase in the number of cost-utility analyses can be observed [1], which is closely interlinked with the recent developments made in the methods for measurement and valuation of children and adolescents’ health-related quality of life (HRQOL). These methods enable the calculation of quality-adjusted life years (QALYs), which express the utility associated with health gains or losses realized through medical innovation. QALYs are calculated by multiplying the length of relevant changes in health by a utility weight representing the HRQOL experienced [4, 5]. In pediatric economic evaluation, these utility weights are often obtained with preference-accompanied instruments for measuring and valuing children and adolescents’ HRQOL, such as EQ-5D-Y and Child Health Utility 9D (CHU-9D) [6].

Instruments such as EQ-5D-Y and CHU-9D have been designed to enable self-report of HRQOL by children as young as 8 years old, with proxy reports by parents possible at even younger ages [7, 8]. The instruments include different dimensions of HRQOL, which are scored individually and described in language amenable to children. These scores on relevant dimensions of HRQOL (e.g., mobility or schoolwork) can be translated to utility weights using country-specific ‘value sets’ (e.g., for the Netherlands, see [9, 10]). To construct value sets for EQ-5D-Y-3L, the version of EQ-5D-Y with three response levels, the standard valuation methodology used for adult EQ-5D instruments was taken as a starting point [11], i.e., a combination of time trade-off (TTO) and discrete choice experiments (DCEs) completed by a representative sample of the adult general public [12]. In adult EQ-5D valuation, the sample is asked to imagine they themselves live in a set of EQ-5D health states (henceforth, the ‘adult perspective’); in the EQ-5D-Y-3L protocol [12], instead, adults are asked to make these trade-offs considering their views about a 10-year-old child (henceforth, the ‘child perspective’).

The use of a child perspective in EQ-5D-Y-3L valuation, in our view, is a pragmatic compromise. Given that TTO and DCE are seen as complicated and involve considerations of death, involving children themselves is often seen as not feasible and potentially ethically problematic [12–14]. Furthermore, adults’ preferences are traditionally seen as essential as they fit the taxpayer perspective considered relevant for economic evaluation, meaning that QALYs should be calculated based on utilities obtained from those who collectively finance the medical innovation being evaluated [12, 15]. As such, given the (perceived) need to rely on adults to complete valuation methods, the introduction of a child perspective aligns the context in which valuation tasks are completed with the intended end-users of the instrument (8–15-year-old children). Yet, the shift from an adult to a child perspective in EQ-5D(-Y) valuation has spurred discussions on its rationale and implications [15–17].

Many studies have explored how shifting from adult to child perspectives affects valuation outcomes (for an up-to-date overview, see also [18]). Roudijk et al. [19] compared EQ-5D-5L value sets (elicited with an adult perspective) with EQ-5D-Y-3L value sets (elicited with a child perspective) for the countries where value sets for both instruments were available and concluded that (1) the five dimensions are valued differently across instruments, and (2) the EQ-5D-Y-3L typically has a smaller value range, meaning fewer states are seen as worse than dead, and the worst state is rated higher than in EQ-5D-5L. Devlin et al. [20] pointed out that these differences between EQ-5D-5L and EQ-5D-Y-3L value sets could have multiple causes. Some of these causes Devlin et al. [20] consider legitimate, i.e., reflecting of the severity of health impairments described by the instrument, and we will refer to such legitimate causes as severity-related, whereas if they are unrelated to actual health state disutility, we refer to them as non-severity-related. Devlin et al. [20] argue that at least the following two severity-related causes may explain differences between EQ-5D-Y-3L and EQ-5D-5L value sets: (1) the EQ-5D-Y-3L instrument uses milder severity labels than the EQ-5D-5L instrument (e.g., the worst level for mobility are a lot of problems walking about for EQ-5D-Y-3L versus unable to walk about for EQ-5D-5L), and (2) adults may view children’s suffering as less severe. Yet, Devlin et al [20] also speculate that differences between EQ-5D-5L and EQ-5D-Y-3L may have non-severity-related causes—which can only be identified from empirical work directly comparing utilities elicited for the same instrument in both perspectives.

The aim of our contribution is to summarize this empirical work (including qualitative and quantitative studies capturing stakeholder opinions), review potential severity- and non-severity-related causes of differences between adult and child perspectives, and provide suggestions for the way forward. We rely on a targeted narrative review, which focused on the empirical literature on adult and child perspectives1 in valuing pediatric health states, which mostly involves studies on EQ-5D-Y. The review synthesizes evidence from key quantitative and qualitative studies to explore differences in valuation outcomes, underlying mechanisms, and alignment with stakeholder preferences. Evidence was drawn from a broad range of sources, including but not limited to work on EQ-5D instruments, to ensure conceptual breadth. Reference lists and citations of foundational studies (e.g., [23–25]) were screened to identify additional relevant contributions. Findings were subjected to interpretative synthesis to derive ten key learnings on the implications of involving adults in valuing child and adolescent health (also summarized in Table 1).

Table 1.

Summary of the ten key learnings from empirical and stakeholder evidence on valuing child health states using child perspectives

#	Main learning	Potentially non-severity-related?
1	Child perspective utilities are generally higher than adult perspective utilities	X
2	Different dimensions may matter more in child perspectives than in adult perspectives
3	The age of the child considered in the child perspective may matter (depending on country and dimensions)
4	In child perspectives immediate death is avoided	X
5	Little to no evidence for differences in the value of life years between adult and child perspectives	X
6	Utilities in child perspectives differ because of psychological distance	X
7	Child perspectives increase complexity and evoke strong emotions	X
8	The external goals that affect child perspective valuation differ from those that affect adult perspective valuation	X
9	Differences between utilities in adult and child perspectives do not align with priority-setting preferences
10	The use of a child perspective for EQ-5D-Y is misaligned with taxpayers and other stakeholders’ preferences

Comparing Adult and Child Perspective Health Valuation: Ten Lessons Learned

Learning 1: Child Perspective Utilities are Generally Higher than Adult Perspective Utilities

Empirical work aligns with Roudijk et al. [19] in finding that utilities elicited from a child perspective are generally higher than those from an adult perspective. Kreimeier et al. [23] compared adults’ TTO valuation of EQ-5D-3L and EQ-5D-Y from adult and child perspectives, noting differences between the instruments as well as between perspectives. Shah et al. [26] further demonstrated that the worst health state on EQ-5D-Y-3L (33333) was valued higher when assessed from a child perspective using TTO. Subsequent research provided mixed findings. Most studies replicated higher utilities for child perspectives using various methods [27–32], while a minority reported no significant differences [24, 33–35]. A meta-regression conducted on 25 studies comparing adult and child perspectives suggests that differences between perspectives range between 0.02 and 0.06 (on the QALY scale) depending on the health state severity and valuation methods [18]. Other work also suggests the difference could be associated with country context [33]. Interestingly, to our knowledge, only one study found lower valuations from a child perspective: Kind et al. [25] reported lower EQ-5D-Y-3L values using visual analog scales (VAS), a result not replicated in subsequent VAS-based research [24, 26].

Learning 2: Different Dimensions May Matter More in Child Perspectives than in Adult Perspectives

Several studies have explored differences in the perceived importance of EQ-5D dimensions between adult and child perspectives. In quantitative work, Hitch et al. [31] found that in child perspectives, pain/discomfort are prioritized more than in adult perspectives, where instead self-care is emphasized. These dimension, it appears, may receive special consideration in child perspectives, as indicated in two qualitative studies: Dewilde et al. [30] suggests adults perceive children as better at coping with pain, potentially explaining these differences. Reckers-Droog et al. [36] also found that adults expect children to be better at coping with impaired health in general. They also show that in adult perspectives, being independent is given high priority, whereas adults using child perspectives consider 10-year-old children less affected by self-care limitations than adults. However, Åström et al. [37], in a qualitative study, observed no clear link between perspective and dimensional importance, a finding echoed in quantitative research [34]. Synthesizing evidence, de Silva et al. [18] concluded that pain/discomfort and self-care were consistently ranked as the most and least important dimensions, respectively, across both perspectives, though the ranking of other dimensions varied. The operationalization of the child perspective also affects these rankings. For example, the EQ-5D-Y valuation protocol instructs adults to consider health states from a 10-year-old child’s perspective, but it does not clarify whether adults should let their preferences reflect their own judgment of what is best or what the child would likely choose. This subtle distinction influences results: adults reflecting what they believe is best for the child tend to assign lower utilities to health states involving severe pain or anxiety, whereas adults considering what the child would want tend to assign higher utilities [38].

Learning 3: The Age of the Child Considered in the Child Perspective May Matter (Depending on Country and Dimensions)

The EQ-5D-Y-3L valuation protocol for the child perspective asks adults to consider a 10-year-old, raising questions about the applicability of resulting value sets across the instrument’s full age range [15]. Studies investigating whether a child’s age affects utilities yield mixed results. Quantitative findings suggest age-related consistency depends on social and cultural context. For example, Essers et al. [39] found minimal differences in utilities for 10- and 15-year-olds in the Netherlands, while significant differences emerged in China. Similarly, age-related differences were reported in the UK but not the USA, across age groups (i.e., 5–7-year-olds, 8–10-year-olds, 11–13-year-olds, and 14–15-year-olds, see [40]). It has also been found that age significantly influenced utilities for certain health states using VAS, with disparities between ages 4, 10, and 16 [41]. The age of the child may affect the disutility of some dimensions more than others. Jumamyradov et al. [42] showed that the relevance of self-care issues varies with a child’s age, while Reckers-Droog et al. [36] noted adults view self-care problems as less severe for 10-year-olds, assuming they receive help, but more serious for 15-year-olds and adults. These results suggest the child’s age is a critical factor in determining utilities.

Learning 4: In Child Perspectives Immediate Death is Avoided

The child perspective is often employed in valuation studies as it is considered undesirable or infeasible to involve children in hypothetical decisions involving death [14, 43, 44]. In tasks like TTO and some DCE variants, adults must decide if certain states are worse than death for a hypothetical child, a moral and ethical challenge [37]. For example, in the composite TTO task, adults are asked whether it is better for a child to live 10 years in a health state or die immediately. Studies suggest adults avoid immediate death options for themselves [45, 46], and this tendency may be stronger in child perspectives. Zero utilities, indicative of death being avoided, are more frequent in child perspectives [28], while tasks not requiring such judgments yield lower utilities [47]. The better-than-dead (BTD) method, where choices are made between living in an EQ-5D(-Y) state and immediate death, similarly shows less preference for death in child perspectives [23]. Importantly, Lang et al. [48] show that the tendency to consider states better or worse than dead depends strongly on the presence or absence of information on duration: in ranking tasks without duration, far fewer states are considered worse than dead, and no differences between perspectives are observed—whereas when duration is present, child perspectives show fewer preferences for immediate death. Collectively, these studies suggest that, as Devlin et al. [20] wrote, ‘Dead (…) invokes special considerations regarding children’s survival, and it cannot be ruled out this exerts an upwards bias on child perspective utilities.’

Learning 5: Little to No Evidence for Differences in the Value of Life Years Between Adult and Child Perspectives

Many methods for eliciting utilities in pediatric populations rely on trade-offs between length and quality of life, as seen in some DCE and TTO approaches, often modeled using a linear QALY framework [5, 49]. In this framework, each additional year is equally valuable, irrespective of timing or recipient. While normatively appealing, this model poorly reflects actual trade-offs involving life duration [50–53]. Empirical evidence shows individuals often value future life years less—a phenomenon known as discounting [54–57]. Ignoring this bias can distort utility estimates [58], prompting calls to incorporate discounting in TTO [59] and DCE methods [60]. Some research has explored discounting differences between adult and child perspectives [27, 28]. Adults may devalue their own future years more strongly, whereas child years are perceived as more precious [30]. To our knowledge, three studies exist that studied differences in discounting between perspectives. Lang et al. [27] noted that correcting for discounting eliminated differences in utilities between adult and child perspectives, but this result is not observed in Lipman et al. [28]. Yu et al. [61], on the other hand, did find slightly stronger discounting in adult perspectives than in child perspectives. Additionally, Hoogenboom and Lipman [29] found no evidence of differences in loss aversion—a related form of non-linearity (i.e., individuals prefer avoiding losses versus realizing equivalent gains)—between adult and child perspectives. These findings suggest discounting may be a mechanism that subtly affects valuations in different perspectives, but the current evidence does not point to it consistently varying by perspective.

Learning 6: Utilities in Child Perspectives Differ Because of Psychological Distance

When adopting a child perspective instead of an adult perspective, preference elicitation tasks shift fundamentally, as preferences are elicited for another person, which a large literature suggests involves distinct considerations compared to deciding for oneself [62–66]. Therefore, child and adult perspectives may yield different outcomes because the former requires the assessment of preferences for another person, whose needs and preferences are unknown [36]. Studies examining this ‘deciding for others’ effect suggest it influences valuations [24, 67]. Drawing on construal level theory [68], Lang et al. [32] find that child perspectives generate higher utilities due to increased psychological distance. Trzebiński et al. [69] extended this by comparing parents and non-parents, showing that parents perceive children in child perspectives as psychologically closer. Interestingly, if increased psychological distance decreases willingness to trade off life years (i.e., higher utilities), parents should have lower utilities using child perspectives due to lower experienced distance. However, some studies report the opposite—parents value child health states higher—likely reflecting moral aversion to contemplating a child’s death [37, 70, 71]. Trzebiński et al. [69] study this apparent contradiction, and find no clear association between measures of psychological distance and EQ-5D-Y utilities (both in parents and non-parents) in child perspectives. Overall, it is not entirely clear if and how psychological distance contributes to utilities in child perspectives.

Learning 7: Child Perspectives Increase Complexity and Evoke Strong Emotions

Child perspectives add complexity to already challenging health state valuation tasks [47]. Qualitative research highlights this difficulty. Powell et al. [67] summarized that adults are generally less willing to trade life years for children than for themselves, and deciding for others (especially children) intensifies the challenge. Some participants simplify tasks by imagining health states for themselves instead. Other studies report that valuing health for children is more mentally and emotionally taxing than for adults, with respondents feeling burdened by the responsibility of stating preferences on behalf of children [36]. Åström et al [37] similarly noted that participants preferred valuing their own health state over someone else’s, describing their experience when valuing health states using child perspectives as ‘horrible’ and ‘grotesque,’ particularly when required to trade life years for a 10-year-old. Powell et al. [72] further observed that adults often find it difficult to imagine health states from a child’s perspective and feel uncomfortable deciding for others. These findings suggest that while the child perspective aims to address ethical concerns about directly involving children, it introduces emotional and cognitive strain, potentially generating a mechanism that affects the reliability of the results.

Learning 8: The External Goals that Affect Child Perspective Valuation Differ from Those that Affect Adult Perspective Valuation

In adult perspectives, health state valuation with TTO is affected by external age-related goals, such as the desire to reach a specific age or life milestone (e.g., to see children grow up, see [73–76]). This tendency to consider external goals extends to child perspective valuation: empirical studies suggest that external goals, like completing school or reaching adulthood, influence child perspective valuations. For example, Reckers-Droog et al. [36] observed that adults considered (developmental) milestones, like children finishing school or experiencing a first romantic love, when valuing child perspectives [36]. Powell et al. [67] also find that adults consider what children would be doing at different ages explicitly when trading off life in child perspectives. Think-aloud interviews revealed that respondents used external goals to structure trade-offs and manage the emotional and cognitive difficulty of valuation tasks [36], but it may also introduce biases [30]. For instance, Reckers-Droog et al. [36] found that in child perspectives, reaching the age of 18 was often seen as a common extrinsic goal—symbolizing the point at which responsibility for making trade-offs could be ‘handed over.’ In the way TTO is typically operationalized [11], this tends to yield high(er) utilities and may bias child perspective valuations because some adults resist making further trade-offs once doing so feels ethically inappropriate, leading them to disengage from the task or stop at the point where they feel the child should decide for themselves. A second set of studies focused on external goals has explored the role of subjective life expectancy (SLE), which in adults has been shown to bias TTO utilities because respondents anchor their willingness to give up life years on how long they personally expect to live, leading those with shorter or longer perceived lifespans to trade differently for reasons unrelated to the health state being valued [75–77]. The role of SLE has also been studied in child perspectives with mixed results. While Lipman [78] found no evidence that SLE affects child perspective utilities, a null result was also observed in two subsequent studies [27, 33]. Ongoing work by Lang et al. [32], however, suggested EQ-5D-Y utilities are positively associated with child SLE. As such, the current evidence only supports the conclusion that adults consider different types of external factors in child perspectives compared with adult perspective valuations; however, there is no clear evidence regarding the magnitude or effect of these considerations on health state utilities.

Learning 9: Differences Between Utilities in Adult and Child Perspectives Do Not Align with Priority-Setting Preferences

While TTO and DCE tasks do not directly compare the lives of children and adults, differences in health state utilities between these perspectives may align with preferences for resource allocation. In that sense, the typically higher utilities in child perspectives compared to adult perspective for the same health state are sometimes interpreted as reflecting the inherent value adults assign to children’s lives [30]. Indeed, one implication of higher utilities in child perspectives is that, ceteris paribus, more QALYs are gained by extending life in children over adults for the same health state, meaning that the same life extension should be preferred in children (assuming QALY maximization is the goal). This pattern is consistent with studies on priority setting [79–82]. However, this line of reasoning also implies that, while life-extending interventions might generate more QALYs and could thus be prioritized for children (given their higher utilities), ceteris paribus, pure HRQOL improvements would be prioritized in adults, since the relative QALY gains from improving lower adult utilities would be greater [15, 33]. If both these implications of differences in utilities on resource allocation decisions completely align with stakeholder preferences for priority setting, concerns about EQ-5D-Y-3L value sets derived from different perspectives might be partially mitigated (e.g., such as those raised in [20]). This alignment, theoretically, could even replace the need for equity weights in resource allocation decisions [15]. However, Attema et al. [33] found little difference in utilities across perspectives, yet their sample still strongly prioritized HRQOL improvements in children over adults, implying that equity considerations remain necessary for resource allocation decisions. Similarly, Peasgood et al. [82] reviewed evidence favoring prioritization of HRQOL improvements in children rather than in adults. These findings suggest utility differences between perspectives may not reliably reflect the relative value of children’s versus adults’ lives.

Learning 10: The Use of a Child Perspective for EQ-5D-Y is Misaligned with Taxpayers and Other Stakeholders’ Preferences

Recent work has engaged different stakeholders [83], including experts in economic evaluation [84], the general public [36, 67, 72, 85, 86], health technology assessment decision bodies [83, 86, 87], and younger populations themselves [85]. These efforts sought reflections on the EQ-5D-Y-3L valuation protocol [12]. A near consensus emerged against exclusively relying on adults valuing EQ-5D-Y from a child perspective. International experts [84] and diverse stakeholders, particularly in the USA and Canada, recommended including adolescents valuing health from their own perspective [83, 86, 87]. Qualitative studies also echoed these suggestions, with adults questioning the legitimacy of their preferences for valuation of child HRQOL and advocating for direct involvement of younger respondents [36, 37]. Focus group participants in the UK also opposed the exclusion of individuals with relevant experience from the valuation process [72, 88]. This shows that taxpayers, whose perspectives are central due to their role in financing healthcare, prefer approaches that involve children and adolescents directly. This preference for direct involvement is supported not only empirically but also ethically: Article 12 of the UN Convention on the Rights of the Child affirms that children capable of forming their own views have the right to express them freely in all matters affecting them, and that their views should be given due weight according to age and maturity [89]. Collectively, these studies provide minimal support for exclusively relying on adult perspectives in EQ-5D-Y valuation. Instead, they underscore the importance of inclusive methods that engage adolescents, aligning valuation approaches more closely with stakeholder preferences.

Adult and Child Perspective Valuations are Different: Where Does that Leave Us?

The empirical evidence synthesized above suggests that adult and child perspective valuations for the same instrument are different. That is, (1) utilities are generally higher in child perspectives, (2) different dimensions appear to matter, and (3) the age of the child described could play a role. These differences can, in principle, be completely due to severity-related causes, i.e., adults genuinely consider children to have better or different HRQOL when they live in some EQ-5D health state compared to the HRQOL the adults would consider themselves to have when living in the same health state (although the extent to which this is true may depend on how young the child is and which health problems the child experiences exactly).

One way to determine if the differences between adult and child perspectives we synthesized have severity-related or non-severity-related causes is to explore how differences depend on the method chosen for valuation. If differences are only present in specific methods, they cannot be considered to be genuinely severity-related, as it is at least partially an artifact of the method selected. In that sense, it seems problematic that many of the learnings summarized above seem specific to use of child perspectives in methods that invoke trade-offs between length of life and quality of life, such as TTO and DCE with duration. For example, for learning 1, de Silva et al. [18], in a meta-analysis, provide evidence that the difference between adult and child perspectives is larger in TTO methods than in methods that do not involve trading off length of life (i.e., VAS). Differences in the dimensions that matter more for children compared to adults (learning 2) and between children imagined at different ages (learning 3), on the other hand, are observed across methods [40, 41], suggesting that they are driven by severity-related causes.

Our synthesis also sheds light on several mechanisms underlying valuation in adults and children. If these mechanisms are causing differences between adult and child perspectives, value sets relying on either perspective will partially reflect these potentially non-severity-related causes, which limits comparability of utilities between perspectives [20]. Indeed, avoidance of immediate death (learning 4) and differences in the value of life duration (learning 5) were considered by Devlin et al. [20] as potential sources of bias in valuation of health for children, i.e., as non-severity-related causes. Our evidence synthesis identifies three additional mechanisms that may be a source of bias: (1) psychological distance, (2) the emotional response to increased complexity, and (3) the presence of different external goals.

First, in learning 6, we synthesize works studying the role of psychological distance [32, 69], which tends to increase when child perspectives are adopted. Although the evidence on its exact effects is mixed, effects related to the increase in psychological distance appear to stem not from the severity of the health states themselves, but from the task instructions for eliciting utilities or respondents’ difficulty imagining children in poor health. Clearly, it seems undesirable if value sets used in economic evaluation are systematically influenced by psychological distance, as this effect reflects limits on adult respondents’ ability to imagine HRQOL in children, rather than the disutility of health impairments in children. Hence, effects of psychological distance reflect a non-severity-related cause of differences between child and adult perspective utilities.

Second, in learning 7, we discuss how the use of child perspectives increases complexity, which potentially affects utilities. Whether or not such emotional responses to the difficulty in child perspectives are severity-related is far from straightforward: the emotional responses seems at least partially related to the severity of health impairments and therefore potentially relevant for inclusion in value sets for child health. That is, considering children in impaired HRQOL invokes strong emotions (particularly in parents [36, 90, 91], and the extent to which this happens may reflect how good or bad the health state is. Yet, respondents in qualitative work expressed particularly strong emotions when considering shortening or even ending children’s lifespan [36, 37]. Such considerations would, by extension, be less relevant and as a consequence less likely to influence child perspective valuation in methods that do not involve trading off life duration for HRQOL improvement, suggesting that this mechanism may also be (partially) non-severity-related.

Third, in learning 8, we discuss how child perspectives may introduce the presence of (child-specific) external goals. Somewhat similar to adult perspectives, in child perspectives, the attainment of symbolic transitions like reaching age 18 are emphasized [36, 47]. In contrast to adult perspectives, SLE appears to have limited impact on child perspective utilities in most studies [78]. Although the evidence is difficult to interpret, whether or not this mechanism is severity-related is more straightforward. External goals seem especially likely to be considered when adults imagine children attaining different ages, i.e., in methods where children’s life duration is traded off. In methods without duration, there is typically no reduction in lifespan implied, making age-related external goals less relevant or focal. Hence, if future work provides more evidence that (age-related) external goals affect child perspective valuation [32], this would suggest that another non-severity-related cause drives differences between adult and child perspectives.

Although the learnings above raise some concerns that using child perspectives introduces differences in utilities due to non-severity-related causes that are not present when adult perspectives are used (see also Table 1 for an overview), this may be entirely acceptable if the methods employed are fully aligned with societal preferences regarding the prioritization between adults and children, or with how relevant stakeholders believe children’s HRQOL should be or is best valued. Yet, this is clearly not the case (learning 9 and 10). Exclusively involving adult respondents in valuation of child HRQOL is not supported by stakeholders, taxpayers, experts, or children themselves. Furthermore, we cannot rely on value sets with adult and child perspectives alone to align resource allocation with the generally prevailing societal preference for prioritizing children [82].

Which Alternatives to a Child Perspective in EQ-5D-Y Valuation Can Be Considered?

Collectively, the ten lessons reported here suggest that while the child perspective was introduced as a pragmatic compromise—balancing the ethical concerns of involving children in valuation with the practical need to obtain utilities for pediatric health states reflecting a taxpayer perspective—this approach may have limitations that warrant closer consideration. The observed systematic differences appear to stem from non-severity-related causes, and the limited stakeholder support for the current methodology suggests that this compromise may not fully align with broader societal expectations or preferences, indicating that alternative approaches should be explored. This section briefly outlines three such alternatives (partially overlapping with Devlin et al. [20]), while noting that the most promising option—practically, ethically, and in terms of stakeholder alignment—remains an open question for future research.

Alternative 1: Using an Age-Invariant Perspective for Valuation of Adult and Child HRQOL

A straightforward approach to avoid non-severity-related influences of child perspectives is to use value sets based on adult perspectives for both adult and child HRQOL instruments. Valuation with an age-invariant perspective can be implemented in two distinct ways: (1) mapping and (2) blinded health valuation. First, instead of creating unique value sets for children, mapping algorithms may be developed that translate child HRQOL measurements to value sets used in adults that rely on adult perspectives. This approach has been used for the Health Utilities Preschool Scale (HuPS) [92], which mirrors the Health Utility Index Mark 3 (HUI3) dimension structure; instead of valuing HuPS states directly, parents indicated which HuPS levels corresponded to HUI3 levels, enabling the estimation of mapping to adult perspective HUI3 value sets. This approach may be extended to develop utility values for EQ-5D-Y instruments by finding the most similar response levels in EQ-5D-5L (used for adults). The approach used for HuPS relies on respondents’ pairings of levels and dimensions of two instruments, whereas mapping algorithms between different EQ-5D instruments typically rely on response mapping. This entails collecting responses on both instruments and predicting scores from one instrument on the other without respondents explicitly being asked about their similarity [93]. Beyond avoiding potential biases related to shifting from adult to child perspectives, a key advantage of mapping is that its reliance on a single adult-derived value set enhances consistency and comparability across economic evaluations of adult and pediatric populations [13]. However, a major disadvantage is that mapping provides no insight into how society values specific impairments in children, despite clear evidence that the adult general public weighs some dimensions differently for adults and children [18, 19]. For example, applying adult-derived weights may overvalue outcomes in domains like self-care that are less important in child perspectives, potentially leading to the disproportionate prioritization of interventions targeting such areas. Second, adult respondents may complete EQ-5D-Y valuation tasks from their own perspective without knowing the health states originate from a pediatric instrument, as in CHU9D valuation [10]. In this approach, adults are essentially ‘blinded’ to the instrument’s intended use, which allows for the direct generation of EQ-5D-Y values and avoids potential noise or estimation error from mapping algorithms. It shares the key advantages and disadvantages of mapping, namely, greater consistency and comparability across populations but limited ability to reflect what society considers important for children. However, stakeholder feedback—including input from experts and focus groups—indicates limited support for the notion that adults complete valuation tasks while being unaware that their preferences are being applied to child HRQOL valuation [72, 84].

Alternative 2: Involving Children

Another approach to reduce the reliance on potentially biased child perspective valuation is greater involvement of children themselves in valuation. Ample evidence indicates that standard valuation tasks, such as DCEs, can be completed meaningfully by adolescents [21, 22, 61, 94]. Related methods such as best-worse scaling have been used successfully in children as young as 11 years old [95–97], and some studies have experimented with involving children with similar ages in TTO and/or standard gamble (SG) tasks [95, 98–100]. Beyond avoiding potential bias related to the use of child perspectives, advantages of involving children as participants in standard valuation tasks are that (1) it aligns with stakeholder calls for greater involvement of children [72, 83–85, 87], (2) such involvement is in line with several normative principles (e.g., the right for children to be represented in decisions affecting them [89]), and (3) it may enhance the validity of pediatric health valuation by capturing children’s unique insights into what matters to them [101]. Key disadvantages of involving children in standard valuation tasks may include the complex or sensitive nature of these tasks harming the (ethical) feasibility, reliability, or validity of valuation [12]. These concerns are typically considered separately for tasks involving comparisons of health states, e.g., DCE or best-worst scaling, and tasks involving trading off life duration and/or immediate death, e.g., TTO and SG [87], where generally involving children in the former is considered feasible or desirable but involving children in the latter is not [84, 87]. Yet, Crump et al. [102], in a systematic review of studies directly eliciting preferences in children, suggested that although the number of studies was low, even TTO and SG methods may be considered feasible, reliable, and valid for children, suggesting that exploration of the use of these methods in children need not be a non-starter. Another consideration around involving children in standard valuation tasks is that stakeholders may consider their preferences necessary but not sufficient for guiding economic evaluation [83, 86, 87], as consensus also exists around the need to involve adults in child health valuation [84]. Consequently, different strategies may be considered to include preferences from both adults and children, which may include (1) separate value sets for both groups [21, 31], (2) combining preferences from separate representative samples of adults and children into a single value set [22], and (3) lowering the age of inclusion for samples of the general population such that children are included proportionately in line with their representation in society. Nazari et al. [22] show that the actual influence child preferences have on value sets and subsequently economic evaluation may depend on what strategy of combining preferences is selected. Hoogenboom et al. [101], instead of only considering children completing standard valuation tasks, argue that greater involvement of children could take different forms regarding their involvement in standard valuation tasks. Indeed, we believe a promising avenue for future research, which may avoid some of the disadvantages involving children has, is (re)designing (standard) valuation tasks to ensure greater involvement of children is feasible. For instance, children might consult or advise adult respondents completing valuations from a child perspective (e.g., by telling adults how they would experience the health state), or provide other forms of validation of child perspective preferences (e.g., providing ‘confirmation’ values elicited in adults, using methods such as those reported in [103]). Ideally, valuation tasks may be conceived that enable children to express their preferences meaningfully, considering that the approaches outlined in Hoogenboom et al. [101] may reduce but not fully eliminate adult-related biases, as adults valuing on behalf of children would remain part of the process.

Alternative 3: Deliberate Health State Valuation

A third option, suggested by Hausman [104], is group-based deliberative valuation, in which a diverse group of participants discuss and reflect on the value of different health states collectively. While the use of such small panels deliberating (e.g., in focus groups) to determine valuations is common practice in the estimation of disability weights for calculation of disability-adjusted life years [105], it has been only very infrequently experimented with in the valuation of generic HRQOL instruments in adults [106–108], and to our knowledge no such work exists for pediatric instruments for measuring HRQOL. Several major challenges and opportunities exist when considering the use of deliberate health state valuation for younger populations. Key among them is group composition; prior work often used convenience samples such as university students [106, 107], but pediatric valuation may require input from experts like developmental psychologists and pediatricians, alongside children and adolescents [104]. Practical uncertainties remain, including ideal group size and deliberation length—though smaller groups with longer sessions have been recommended [107]. When considering deliberate methods involving both adults and children, study designers may draw from experience developed in designed deliberate methods in related fields such as pediatric ethics and priority setting [109–111], in which children and adolescents are often involved. However, critically, it is still unclear how deliberative methods can produce valid value sets. Existing studies used either multi-criteria decision analysis [107] or limited TTO tasks [106, 108], neither of which offer sufficient health state variation or anchor on the QALY scale. One promising avenue may be the use of personal utility function approaches [112], which encourage deliberation and, in principle, can generate value sets from a single group process. A key disadvantage of the use of deliberate methods, however, is that it is difficult to ensure sample representativeness (and consequently reproducibility), which may be of crucial importance to stakeholders.

Concluding Remarks

While the use of child perspectives in valuing EQ-5D-Y health states has offered a pragmatic interim solution, accumulating empirical evidence indicates that this approach introduces systematic biases and does not adequately reflect societal or stakeholder preferences. The consistent divergence between adult and child valuations—coupled with limited stakeholder support—highlights a need for methodological innovation in pediatric health valuation. Future research should prioritize the development and testing of alternative approaches, such as direct child involvement in valuation tasks, robust mapping techniques, and deliberative frameworks. Advancing these methods is essential to ensure that pediatric HRQOL value sets are both empirically valid and ethically and socially acceptable for use in health technology assessment and policy decision-making.

Declarations

Conflict of interest

Dr. Lipman is a member of the EuroQol group. The views expressed by the author in the publication do not necessarily reflect the views of the EuroQol group.

Ethics approval, Informed consent, and Data availability

No research data were generated or re-used as part of this Current Opinion article. The lack of data also precludes ethics approval or informed consent.

Author contributions

SL conceptualized the paper, co-developed the first draft of the manuscript, provided significant edits, and supervised the paper. ZG provided significant feedback, co-developed the first draft, and edited the manuscript.