Economic evaluations of interventions that target or include children require health state utilities (HSUs). Despite the availability of preference-weighted measures for children, methods for valuing child health states and estimating child utilities are not as well established as those for adult HSUs. The objective of this task force was to develop emerging good practice recommendations for valuing child and adolescent health to generate HSUs for use in economic evaluation. This task force identified and described the interrelated methodological choices regarding the valuation of child health to generate HSUs. The task force considered available evidence related to 4 key issues: (1) whose preferences should be sought, (2) whose health is imagined, (3) which method should be used, and (4) the comparability between adult and child utilities. Best practices may vary depending on the modeling context, characteristics of the health states, and the health technology assessment setting in which the HSUs will be used. For any individual study, methods will be informed by empirical evidence, value judgments, and recommendations from healthcare decision makers. Rather than recommending an approach that would apply to every study, this task force presents options to consider when determining the preference elicitation approach to generate utilities for child health states, along with the strengths and limitations of each. Given that child HSUs can affect the outcomes of a cost-utility analysis and subsequent decisions about healthcare resource allocation, this task force recommends that researchers be transparent about methodological choices and their impact on HSUs.
Measuring health-related quality of life (HRQoL) in the very young (i.e., infants and toddlers, aged 0–47 months) presents unique conceptual and methodological challenges. As infants and toddlers cannot reliably self-report their HRQoL, observer inference is necessary. This raises questions about which concepts should be included; how to manage proxy reporting; and how to capture genuine variations in HRQoL, not changes in development or assessment. This commentary outlines six key challenges in measuring infant and toddler HRQoL and details eight proposed recommendations for HRQoL researchers. The piece draws on insights from the EuroQol Toddler and Infant Populations (EQ-TIPS) project, aimed to develop a generic preference-weighted measure (PWM) of HRQoL for infants and children, but has broader applicability to HRQoL measurement in the very young. Key issues addressed include: (i) what concepts to measure; (ii) how to identify which concepts matter to very young children; (iii) managing proxy reporting and reducing bias; (iv) accounting for rapid developmental changes; (v) managing continuity across life-course instruments; and (vi) separating child HRQoL from family spillover effects. Proposed recommendations include greater consensus on a core HRQoL model; prioritising primary caregiver perspectives; justifying and operationalising observable aspects of subjective HRQoL; careful design to capture genuine HRQoL, not developmental change or caregiver spillover; and developing measurement systems to prioritise age-based sensitivity or comparability across time. These recommendations offer a foundation for future consensus-building and research to refine and harmonise best practices in infant and toddler HRQoL measurement, particularly for use in health technology assessment.
This study aimed to evaluate the impact of adding four bolt-ons (skin irritation (SI), self-confidence (SE), social relationships (SR) and sleep (SL)) on psychometric performance of EQ-5D-5L and compare it to PROMIS-29 and Skindex-16 in patients with psoriasis (Ps) and atopic dermatitis (AD). Paper-based questionnaires were completed by 125 Ps and 111 AD patients at dermatology outpatient clinics in Egypt, most had moderate-to-severe disease. Psychometric evaluation included ceiling effect, informativity, convergent validity, explanatory power for variances in EQ VAS, and the discriminatory ability across severity subgroups defined by clinical and patient reported indices. Both dimension-level scores and Level Sum Scores (LSS), standardized to a 5–25 scale, were used to assess the psychometric performance of EQ-5D-5L with bolt-ons. The EQ-5D-5L showed a low ceiling effect (3.4
OBJECTIVES:The scaling factor model (SFM) uses parameters of existing EQ-5D value sets to estimate value sets for new EQ-5D descriptive systems expanded with bolt-ons. This study aimed to compare the performance of SFM and the standard modeling approach (ie, "standalone" model) for modeling the general public's preferences for EQ-5D-5L health states expanded with bolt-on. METHODS:In a composite time trade-off (cTTO) valuation study, we selected EQ-5D-5L and bolt-on health states using an orthogonal array design. We randomized 597 respondents to valuing EQ-5D-5L health states (arm 1), EQ-5D-5L states with vision bolt-on (arm 2), or EQ-5D-5L states with cognition bolt-on (arm 3). We modeled the cTTO data derived from arm 1 and used the estimated coefficients to model the cTTO data derived from arm 2 and arm 3 separately using SFM. Both additive and cross-attribute level effects model specifications were used in the analysis. Using a cross-validation method, we evaluated the predictive accuracy of both SFM and standalone models. RESULTS:SFM showed better predictive accuracy than standalone model in the cross-validation analysis. For EQ-5D states with vision bolt-on, the mean absolute errors were 0.049 and 0.064 for SFM and 0.183 and 0.085 for standalone model; for cognition, the mean absolute errors were 0.051 and 0.047 for SFM and 0.152 and 0.063 for standalone model. CONCLUSIONS:This study suggests that the SFM could be a viable approach to utilizing existing EQ-5D value sets to estimate value sets for new, bolt-on enhanced EQ-5D health-state descriptive systems.
