Psychometric Properties of Generic Preference-Weighted Measures for Children and Adolescents: A Systematic Review

Determining the Content Validity of the EQ-5D-5L, EQ-5D-Y-3L, and CHU9D Instruments for Assessing Generic Child and Adolescent Health-Related Quality of Life: A Qualitative Study

BACKGROUND

Health technology assessment agencies typically recommend generic measures of health to generate quality-adjusted life-years. Most agencies provide recommendations on which measure to use for adults, whereas few make recommendations for children. Two widely used preference-weighted measures of child and adolescent health that have evidence of good psychometric performance are the EQ-5D-Y-3L and the Child Health Utility 9D Index (CHU9D). The EQ-5D-5L has also been used to assess adolescent health. However, evidence on their content validity-a core measurement property-is limited. The objective of this study was to explore the content validity of the EQ-5D-5L, EQ-5D-Y-3L, and CHU9D measures, including their relevance, comprehensiveness, and comprehensibility.

METHODS

We assessed the content validity of the EQ-5D-5L, EQ-5D-Y-3L, and CHU9D using online semi-structured cognitive interviews in the UK. Participants were asked to comment on the relevance, comprehensibility, and comprehensiveness of the measures, including response options, recall period, and completion instructions. Interviews were informed by a topic guide. Purposive sampling allowed for appropriate breadth in the sample, with variation in gender, and presence of health conditions, disease, or disability. Interviews were recorded and transcribed verbatim before thematic content analysis.

RESULTS

In total, we conducted 49 interviews between August 2022 and June 2023: 21 children/adolescents aged 8-17 years and 28 parents/guardians of children aged 4-17 years. The mean duration of the interviews was 45 min. Relevance was broadly supported, but issues were identified. Comprehensibility was inconsistent on some items, and participants expressed difficulty with grouped items (e.g., 'anxiety/depression'). Participants had difficulty distinguishing qualitatively between some response options (e.g., 'a little bit/a bit'). Some participants noted that instrument comprehensiveness was insufficient.

CONCLUSIONS

Although the content of the EQ-5D-5L, EQ-5D-Y-3L, and CHU9D was broadly supported, potential problems were identified in aspects of comprehensibility, relevance, and comprehensiveness. These present opportunities for future research and refinement to ultimately improve the content validity of these measures for assessing child and adolescent health.

Measurement Properties of the EQ-5D Instruments in Children and Adolescents: A Systematic Review

BACKGROUND

The EQ-5D instruments have been widely used to measure health-related quality of life (HRQoL) in child and adolescent populations, especially the EQ-5D-Y-3L and EQ-5D-Y-5L (beta version). Although not specifically designed for younger users, the adult versions (EQ-5D-3L and EQ-5D-5L) are also used in these populations. While the measurement properties of these instruments have been evaluated in children and adolescents, no systematic review to date has employed a rigorous method to assess risk of bias. Additionally, quality criteria for good measurement properties and certainty of evidence have not been thoroughly evaluated. The aim of this study was to summarize and critically appraise the evidence on the measurement properties of all EQ-5D instruments in children and adolescents.

METHODS

We conducted electronic searches on MEDLINE, EMBASE, CINAHL, EconLit, National Health Service Economic Evaluation Database (NHS-EED), and Health Technology Assessment (HTA) databases up to May 2024. We included studies measuring HRQoL using either the self-reported or proxy-reported version of the EQ-5D instruments-EQ-5D-3L, EQ-5D-5L, EQ-5D-Y-3L, and EQ-5D-Y-5L-using the descriptive system, visual analogue scale, and/or utility score in children and adolescents up to 19 years of age, and that tested at least one measurement property (e.g., reliability). The COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) methodology was followed to assess risk of bias, to score results for measurement properties, and to perform an evidence synthesis using a modified Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) approach.

RESULTS

From 3586 records identified through the search, 65 studies were included in this systematic review. We found moderate certainty of evidence of sufficient comprehensibility and comprehensiveness of the EQ-5D-Y-3L and EQ-5D-Y-5L. Furthermore, we found very low certainty of evidence of inconsistent relevance for the EQ-5D-Y-3L, whereas the EQ-5D-Y-5L had sufficient relevance. Almost all the measurement properties (reliability, hypothesis testing for construct validity, and responsiveness) considering all the EQ-5D versions ranged from moderate certainty of evidence of insufficient results to very low certainty of evidence of insufficient results.

CONCLUSION

There is moderate certainty of evidence that the EQ-5D-Y-3L and EQ-5D-Y-5L have sufficient content validity. Both instruments can be recommended to measure HRQoL in children and adolescents aged 8-15 years. However, most of the measurement properties across all EQ-5D versions showed insufficient results, with certainty of evidence ranging from moderate to very low due to inconsistency and doubtful to inadequate risk of bias. Therefore, further research is needed to improve the methodological quality of studies on EQ-5D instruments for children and adolescents.

SYSTEMATIC REVIEW REGISTRATION

International prospective register of systematic reviews (PROSPERO): CRD42020218382 and Open Science Framework: https://osf.io/r8kt9/ .

Head-to-Head Comparisons of the Distributional Characteristics and Measurement Properties of the 3-Level and 5-Level Versions of the EQ-5D-Y: A Systematic Review

OBJECTIVES

This systematic review compared the distributional characteristics and measurement properties of EuroQol's EQ-5D-Y-3L and EQ-5D-Y-5L instruments, using results from published head-to-head comparative studies.

METHODS

The review was reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses-COnsensus-based Standards for the selection of health Measurement INstruments (PRISMA-COSMIN) guideline. We searched 8 databases up to February 14, 2024 for studies on the measurement properties and distributional characteristics (missing data, ceiling/floor effects, consistency, and informativity) of both EQ-5D-Y versions. Measurement quality was assessed in terms of the proportion of tests meeting COSMIN recommended criteria for reliability, validity, and responsiveness.

RESULTS

Eighteen studies were included. The Y-5L reduced ceiling effects by 0.8% to 16.5% and had a slightly higher evenness index, indicating greater informativity. Both versions showed sufficient construct validity across patient and general population samples, with the Y-5L slightly outperforming the Y-3L in known-group validity. Both the Y-3L and Y-5L versions showed sufficient test-retest reliability for mobility, pain/discomfort, and worried, sad, or unhappy and inconsistent reliability for looking after myself and usual activities, indicating similar performance across these dimensions. The Y-5L showed better responsiveness (standardized effect size range = 0.03-2.05) than Y-3L (range = 0.13-0.94). In proxy-reported data, the Y-5L appears to have slightly lower test-retest reliability than the Y-3L, despite demonstrating better agreement with self-reported data.

CONCLUSIONS

Both EQ-5D-Y versions demonstrated varying psychometric performances across the evaluated populations, with the Y-5L slightly outperforming the Y-3L in responsiveness and proxy-child agreement. The availability of the Y-5L expands the options of health-related quality-of-life instruments for assessing pediatric populations. Further research is needed to assess its performance across diverse populations and administrative modes.

Psychometric validation of the Chinese versions of EQ-5D-Y-3L and the experimental EQ-TIPS in children and adolescents with COVID-19

OBJECTIVES

Respiratory infectious diseases like COVID-19 profoundly impacts the health of children and adolescents, but validated instruments to measure their impacts on health-related quality of life (HRQoL) are lacking. The EQ-5D-Y-3L, widely used for youth HRQoL, now features a Chinese value set. The experimental EQ-TIPS addresses HRQoL assessment for toddlers and infants. This study tested the psychometric properties of both instruments in paediatric COVID-19 patients, and compared the performance of self-complete and proxy EQ-5D-Y-3L.

METHODS

This longitudinal study recruited 861 COVID-19 patients aged 0-18 years and their parental caregivers, with 311 dyads completing the follow-up. Digital administration included the EQ-TIPS, the EQ-5D-Y-3L, and Overall Health Assessment (OHA). Controls comprised 231 healthy children. Analysis encompassed known-group validity, child-parent agreement, and responsiveness to change in disease severity and OHA.

RESULTS

COVID-19 children exhibited lower HRQoL than non-infected peers. The EQ-TIPS and the EQ-5D-Y-3L distinguished groups by disease presence, severity and symptoms, showing moderate to good known-group validity (ESs: 0.45-1.39 for EQ-TIPS, 0.44-1.91 for self-complete EQ-5D-Y-3L, and 0.32-1.67 for proxy EQ-5D-Y-3L). Child-parent agreement was moderate to good for EQ-5D-Y-3L (ICC: 0.653-0.823; Gwet's AC1: 0.470-0.738), and responsiveness was good for both EQ-TIPS Level Sum Score (LSS) (ESs: 1.21-1.39) and EQ-5D-Y-3L index scores (ESs: 1.00-1.16).

CONCLUSIONS

This study demonstrates the reliability, validity, and responsiveness of the experimental EQ-TIPS and the EQ-5D-Y-3L in paediatric COVID-19 patients. It is the first evidence of the EQ-TIPS' responsiveness, supporting its use in assessing the impact of COVID-19 on paediatric HRQoL.

Validity and responsiveness of EQ-5D-Y in children with haematological malignancies and their caregivers

The psychometric properties of the EQ-5D-Y have not been widely tested in severely ill children. The aim of this study was to assess and compare the validity and responsiveness of the EQ-5D-Y-3L and EQ-5D-Y-5L in paediatric inpatients with haematological malignancies and caregivers. Respondents completed the interviewer-administered self-complete or proxy version of the EQ-5D-Y-3L and EQ-5D-Y-5L and an overall health assessment twice on different days. Known-groups validity was assessed by comparing patients who differed in overall health and Eastern Cooperative Oncology Group (ECOG) performance. Responsiveness to worsened health was assessed using standardised effect size (SES) for patients with worsened ECOG grade, self-reported rating, or chemotherapy initiation. Ninety-six dyads completed the baseline questionnaires. A smaller proportion of patients reported "no problems" on the EQ-5D-Y-5L compared to EQ-5D-Y-3L for most of the five dimensions. Patients in poor health reported more problems in all dimensions and had higher EQ-5D-Y-5L level sum score, lower EQ VAS and EQ-5D-Y-3L index scores (Cohen's d ES: 0.32-1.38 for patients; 0.50-2.05 for caregivers). There was a mild to good responsiveness to worsened health condition based on ECOG (SES: 0.14-0.61 for patients; 0.40-0.96 for caregivers), suggesting the proxy version was slightly responsive than the self-complete version of both instruments. The results demonstrated validity and responsiveness for both the self-complete and proxy versions of the EQ-5D-Y-3L and EQ-5D-Y-5L. The proxy and 5-level versions of the instrument were more sensitive than the self-complete and 3-level versions in this patient group.

Cross-cultural adaptation and content validation of the Singapore English version of EQ-5D-Y: a qualitative study

BACKGROUND

The EQ-5D-Y is a generic preference-weighted measure for children and adolescents which was developed within Europe. Two versions exist, the EQ-5D-Y-3L (Y-3L) and EQ-5D-Y-5L (Y-5L). This study aimed to cross-culturally adapt the Y-3L and Y-5L for use in Singapore and to assess the content validity, specifically, the relevance and comprehensiveness of the EQ-5D-Y descriptive system (DS) in Asia.

METHODS

To culturally adapt the instruments, an expert panel consisting of paediatricians and primary school educators were consulted. Modifications suggested by the expert panel were tested via cognitive debriefing interviews with children aged 8-12 in Singapore. To assess the content validity of the EQ-5D-Y DS, interviews were conducted with both healthy (n = 8) and ill children (n = 6) aged 8-15. In the interviews, children discussed their experience with poor health and commented on the comprehensiveness and relevance of the EQ-5D-Y DS.

RESULTS

The cross-cultural adaptation process led to minor modifications to the UK English Y-3L and Y-5L versions, including using phrases familiar to the local children and adding examples to facilitate understanding. The five health dimensions in the EQ-5D-Y DS were spontaneously elicited when children discussed their experience with poor health. All health dimensions related to poor health elicited from the interviews fell into three broad categories: physical health (e.g. Appetite, Mobility, and Sleep), mental well-being (e.g. Annoyed/Frustrated and Scared/Worried), and social relationships (e.g. Family and Friends). The EQ-5D-Y DS was generally found to be relevant and comprehensive, although some health dimensions that may be relevant to the local population (Social relationship and Appetite) were not covered.

CONCLUSIONS

The UK English EQ-5D-Y instruments were adapted to produce the Singapore English EQ-5D-Y instrument that were comprehensible to local children as young as 8 years old. The EQ-5D-Y DS was generally relevant and comprehensive to measure poor health of local children. Future studies should ascertain the benefits of adding bolt-on items related to social relationships and appetite to the EQ-5D-Y DS.

Comparative Psychometric Performance of Common Generic Paediatric Health-Related Quality of Life Instrument Descriptive Systems: Results from the Australian Paediatric Multi-Instrument Comparison Study

OBJECTIVE

The aim of this study was to compare the psychometric performance of common generic paediatric health-related quality-of-life instrument descriptive systems (PedsQL generic core 4.0, EQ-5D-Y-3L, EQ-5D-Y-5L, Child Health Utility 9D [CHU9D], Assessment of Quality of Life 6D [AQoL-6D], and Health Utilities Index Mark 3 [HUI3]) by child age, report type, and health status.

METHODS

Data for children aged 5-18 years were from the Australian Paediatric Multi-Instrument Comparison study. Ceiling effects, test-retest reliability, known-group validity, convergent and divergent validity, and responsiveness were assessed in the total sample and by child age (5-12 years vs 13-18 years), report type (self- vs proxy report), and health status. Instruments were scored using an exploratory level sum score (LSS) approach.

RESULTS

Survey data were available for 5945 children, with follow-up data available for 2346 children. The EQ-5D-Y-3L demonstrated ceiling effects. The PedsQL, EQ-5D-Y-3L, EQ-5D-Y-5L, and CHU9D demonstrated acceptable test-retest reliability. All instruments demonstrated known-group, convergent, and divergent validity. The EQ-5D-Y-3L, EQ-5D-Y-5L, and CHU9D demonstrated responsiveness to improvements in health and the PedsQL, EQ-5D-Y-3L, EQ-5D-Y-5L, and CHU9D to worsening health. The AQoL-6D and HUI3 had inconclusive test-retest reliability and responsiveness evidence due to small sample size. Importantly, ceiling effects, test-retest reliability and responsiveness varied by subgroup.

CONCLUSION

Results reflect instrument performance using LSSs, which may differ to utility scores. In the total sample, the EQ-5D-Y-5L and CHU9D descriptive systems demonstrated evidence of good performance (i.e., meeting prespecified criteria) across all psychometric attributes tested. Performance varied by child age and report type, indicating room for considerations by population and study.

Comparing the Psychometric Performance of Generic Paediatric Health-Related Quality of Life Instruments in Children and Adolescents with ADHD, Anxiety and/or Depression

OBJECTIVE

The aim of this study was to examine the validity, reliability and responsiveness of common generic paediatric health-related quality of life (HRQoL) instruments in children and adolescents with mental health challenges.

METHODS

Participants were a subset of the Australian Paediatric Multi-Instrument Comparison (P-MIC) study and comprised 1013 children aged 4-18 years with attention-deficit/hyperactivity disorder (ADHD) (n = 533), or anxiety and/or depression (n = 480). Participants completed an online survey including a range of generic paediatric HRQoL instruments (PedsQL, EQ-5D-Y-3L, EQ-5D-Y-5L, CHU9D) and mental health symptom measures (SDQ, SWAN, RCADS-25). A subset of participants also completed the HUI3 and AQoL-6D. The psychometric performance of each HRQoL instrument was assessed regarding acceptability/feasibility; floor/ceiling effects; convergent validity; known-group validity; responsiveness and test-retest reliability.

RESULTS

The PedsQL, CHU9D, EQ-5D-Y-3L and EQ-5D-Y-5L showed similarly good performance for acceptability/feasibility, known-group validity and convergent validity. The CHU9D and PedsQL showed no floor or ceiling effects and fair-good test-retest reliability. Test-retest reliability was lower for the EQ-5D-Y-3L and EQ-5D-Y-5L. The EQ-5D-Y-3L showed the highest ceiling effects, but was the top performing instrument alongside the CHU9D on responsiveness to improvements in health status, followed by the PedsQL. The AQoL-6D and HUI3 showed good acceptability/feasibility, no floor or ceiling effects, and good convergent validity, yet poorer performance on known-group validity. Responsiveness and test-retest reliability were not able to be assessed for these two instruments. In subgroup analyses, performance was similar for all instruments for acceptability/feasibility, known-group and convergent validity, however, relative strengths and weaknesses for each instrument were noted for ceiling effects, responsiveness and test-retest reliability. In sensitivity analyses using utility scores, performance regarding known-group and convergent validity worsened slightly for the EQ-5D-Y-3L and CHU9D, though improved slightly for the HUI3 and AQoL-6D.

CONCLUSIONS

While each instrument showed strong performance in some areas, careful consideration of the choice of instrument is advised, as this may differ dependent on the intended use of the instrument, and the age, gender and type of mental health condition of the population in which the instrument is being used.

TRIAL REGISTRATION

ANZCTR-ACTRN12621000657820.

Health-related quality of life and subjective well-being among children aged 9-12 years in Shandong Province, China

PURPOSE

To investigate the health-related quality of life (HRQoL) and subjective well-being (SWB) of children aged 9-12 years in eastern China, and examine concordance within child self-reported and parent proxy-assessed.

METHODS

Data was collected from 9 to 12 years old children (including their parents) in Shandong Province in 2018. Participants self-completed a hard-copy questionnaire including Child Health Utility 9D (CHU9D), Pediatric Quality of Life Inventory (PedsQL)™ 4.0 Short Form 15 Generic Core Scales (hereafter the PedsQL™), Student's Life Satisfaction Scale (SLSS), as well as information on socio-demographic characteristics and self-report health status. Spearman's correlation coefficients and the difference between sub-groups were conducted to assess and compare the agreement on HRQoL and SWB instruments. Exploratory factor analysis (EFA) was used to ascertain the number of unique underlying latent factors that were associated with the items covered by the two generic HRQoL and the SWB instruments. The concordance of child self-reported and parent proxy-assessed was analyzed using weighted kappa coefficient and Bland-Altman plots.

RESULTS

A total of 810 children and 810 parents were invited to participate in the survey. A valid sample of 799 (98.6%) children and 643 (79.4%) parents completed the questionnaire. The child self-reported mean scores were CHU9D = 0.87, PedsQL™ = 83.47, and SLSS = 30.90, respectively. The parent proxy-assessed mean scores were PedsQL™ = 68.61 and SLSS = 31.23, respectively. The child self-reported PedsQL™ was moderately correlated with the CHU9D (r = 0.52). There was a weak correlation between CHU9D and SLSS (r = 0.27). The EFA result found 3 factors whilst seven SLSS items grouped into a standalone factor (factor 3), and the nine dimensions of CHU9D shared two common factors with the PedsQL™ (factor 1 and factor 2). A low level of concordance was observed across all comparisons and in all domains (weighted kappa < 0.20) between parents and their children. Furthermore, a high level of discordance was observed between child self-reported and father proxy-assessed.

CONCLUSIONS

CHU9D and PedsQL™ instruments have a higher agreement in measuring the HRQoL in children. CHU9D/PedsQL™ and SLSS instruments showed a low agreement and EFA result suggested that measuring SWB in children potentially may provide further information, which might be overlooked by using HRQoL instruments exclusively. Concordance of child self-reported and parent proxy-assessed was poor. Overall, mother-child concordance was higher than father-child concordance.

Systematic Review of the Psychometric Performance of Generic Childhood Multi-attribute Utility Instruments

BACKGROUND

Childhood multi-attribute utility instruments (MAUIs) can be used to measure health utilities in children (aged ≤ 18 years) for economic evaluation. Systematic review methods can generate a psychometric evidence base that informs their selection for application. Previous reviews focused on limited sets of MAUIs and psychometric properties, and only on evidence from studies that directly aimed to conduct psychometric assessments.

OBJECTIVE

This study aimed to conduct a systematic review of psychometric evidence for generic childhood MAUIs and to meet three objectives: (1) create a comprehensive catalogue of evaluated psychometric evidence; (2) identify psychometric evidence gaps; and (3) summarise the psychometric assessment methods and performance by property.

METHODS

A review protocol was registered with the Prospective Register of Systematic Reviews (PROSPERO; CRD42021295959); reporting followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guideline. The searches covered seven academic databases, and included studies that provided psychometric evidence for one or more of the following generic childhood MAUIs designed to be accompanied by a preference-based value set (any language version): 16D, 17D, AHUM, AQoL-6D, CH-6D, CHSCS-PS, CHU9D, EQ-5D-Y-3L, EQ-5D-Y-5L, HUI2, HUI3, IQI, QWB, and TANDI; used data derived from general and/or clinical childhood populations and from children and/or proxy respondents; and were published in English. The review included 'direct studies' that aimed to assess psychometric properties and 'indirect studies' that generated psychometric evidence without this explicit aim. Eighteen properties were evaluated using a four-part criteria rating developed from established standards in the literature. Data syntheses identified psychometric evidence gaps and summarised the psychometric assessment methods/results by property.

RESULTS

Overall, 372 studies were included, generating a catalogue of 2153 criteria rating outputs across 14 instruments covering all properties except predictive validity. The number of outputs varied markedly by instrument and property, ranging from 1 for IQI to 623 for HUI3, and from zero for predictive validity to 500 for known-group validity. The more recently developed instruments targeting preschool children (CHSCS-PS, IQI, TANDI) have greater evidence gaps (lack of any evidence) than longer established instruments such as EQ-5D-Y, HUI2/3, and CHU9D. The gaps were prominent for reliability (test-retest, inter-proxy-rater, inter-modal, internal consistency) and proxy-child agreement. The inclusion of indirect studies (n = 209 studies; n = 900 outputs) increased the number of properties with at least one output of acceptable performance. Common methodological issues in psychometric assessment were identified, e.g., lack of reference measures to help interpret associations and changes. No instrument consistently outperformed others across all properties.

CONCLUSION

This review provides comprehensive evidence on the psychometric performance of generic childhood MAUIs. It assists analysts involved in cost-effectiveness-based evaluation to select instruments based on the application-specific minimum standards of scientific rigour. The identified evidence gaps and methodological issues also motivate and inform future psychometric studies and their methods, particularly those assessing reliability, proxy-child agreement, and MAUIs targeting preschool children.