Exploring the causal and effect nature of EQ-5D dimensions: an application of confirmatory tetrad analysis and confirmatory factor analysis

Comparing Generic Paediatric Health-Related Quality-of-Life Instruments: A Dimensionality Assessment Using Factor Analysis

BACKGROUND

Widely used generic instruments to measure paediatric health-related quality of life (HRQoL) include the EQ-5D-Y-5L, Child Health Utility 9 Dimension (CHU-9D), Paediatric Quality of Life Inventory (PedsQL) and Health Utilities Index (HUI). There are similarities and differences in the content of these instruments, but there is little empirical evidence on how the items they contain relate to each other, and to an overarching model of HRQoL derived from their content.

OBJECTIVE

This study aimed to explore the dimensionality of the instruments using exploratory factor analysis (EFA).

METHODS

Data from the Australian Paediatric Multi-Instrument Comparison (P-MIC) Study were used. EQ-5D-Y-5L, CHU-9D, PedsQL and HUI data were collected via proxy or child self-report data. EFA was used to investigate the underlying domain structure and measurement relationship. Items from the four instruments were pooled and domain models were identified for self- and proxy-reported data. The number of factors was determined based on eigenvalues greater than 1. A correlation cut-off of 0.32 was used to determine item loading on a given factor, with cross-loading also considered. Oblique rotation was used.

RESULTS

Results suggest a six-factor structure for the proxy-reported data, including emotional functioning, pain, daily activities, physical functioning, school functioning, and senses, while the self-report data revealed a similar seven-factor structure, with social functioning emerging as an additional factor.

CONCLUSION

We provide evidence of differences and similarities between paediatric HRQoL instruments and the aspects of health being measured by these instruments. The results identified slight differences between self- and proxy-reported data in the relationships among items within the resulting domains.

Development of the Young Disability Questionnaire (spine) for children with spinal pain: field testing in Danish school children

OBJECTIVE

The objective of this study was to finalise the development of the Young Disability Questionnaire (YDQ-spine) to measure the consequences of neck, midback and low back pain, relevant for schoolchildren aged 9-12 years.

DESIGN

A cross-sectional field test of the YDQ-spine was carried out.

SETTING

Danish primary schools.

PARTICIPANTS

Children aged 9-12 years from all Danish schools were invited to complete the questionnaire.

METHODS

Eight hundred and seventy-three schools were invited to participate. Consenting schools received information material, instructions and a link to an electronic version of the prefinal YDQ-spine. Local teachers distributed the electronic YDQ-spine to children aged 9-12 years. Descriptive statistics and item characteristics were carried out. Item reduction was performed using partial interitem correlations (scrutinising correlations>0.3) and factor analyses (items loading>0.3 were retained) to eliminate redundant items and to obtain insight into the structure of the questionnaire.

RESULTS

A total of 768 children from 20 schools answered of the questionnaire and 280 fulfilled the inclusion criteria of having back and/or neck pain (36%). Multisite pain was reported by 38%. Partial interitem correlations and factor analyses resulted in elimination of four items which were considered redundant leaving 24 items in the final YDQ-spine with an optional section on what matters most to the child. The factor analyses showed a two-factor structure with a physical component (13 items) and a psychosocial component (10 items) in addition to one standalone item (sleep).

CONCLUSION

The YDQ-spine is a novel questionnaire with satisfactory content validity measuring physical and psychosocial components (including sleep disturbances) of spinal pain in children aged 9-12 years. It also offers an optional section on what matters most to the child allowing targeted care in clinical practice.

Validation of the National Aeronautics and Space Administration Task Load Index (NASA-TLX) adapted for the whole day repeated measures context

Our objective was to investigate the validity of four-item and six-item versions of the National Aeronautics and Space Administration Task Load Index (NASA-TLX, or TLX for short) for measuring workload over a whole day in the repeated measures context. We analysed data on 51 people with type 1 diabetes from whom we collected ecological momentary assessment and daily diary data over 14 days. The TLX was administered at the last survey of every day. Confirmatory factor analysis fit statistics indicated that neither the TLX-6 nor TLX-4 were a unidimensional representation of whole day workload. In exploratory analyses, another set of TLX items we refer to as TLX-4v2 was sufficiently unidimensional. Raw sum scores from the TLX-6 and TLX-4v2 had plausible relationships with other measures, as evidenced by intra-person correlations and mixed-effects models. TLX-6 appears to capture multiple factors contributing to workload, while TLX-4v2 assesses the single factor of 'mental strain'. Practitioner Summary: Using within-person longitudinal data, we found evidence supporting the validity of a measure evaluating whole-day workload (i.e. workload derived from all sources, not only paid employment) derived from the NASA-TLX. This measure may be useful to assess how day-to-day variations in workload impact quality of life among adults.Abbreviations: NASA-TLX or TLX: National Aeronautics and Space Administration Task Load Index; TLX-6: six item version of the NASA-TLX; TLX-4: four item version of the NASA-TLX, TLX-4v2: four item NASA-TLX version two; NIOSH: National Institute for Occupational Safety and Health; CFA: confirmatory factor analysis; T1D: type 1 diabetes; EMA: ecological momentary assessment; BG: blood glucose; SD: standard deviation; CV: coefficient of variation; RMSEA: root mean square error of approximation; CFI: comparative fit index; TLI: Tucker-Lewis Index; SRMR: standardized root mean square residual; AIC: Akaike information criterion; BIC: Bayesian information criterion; χ2: Chi-square statistic.

Longitudinal study of patients' health-related quality of life using EQ-5D-3L in 11 Swedish National Quality Registers

OBJECTIVE

To compare problems reported in the five EQ-5D-3L dimensions and EQ VAS scores at baseline and at 1-year follow-up among different patient groups and specific diagnoses in 11 National Quality Registers (NQRs) and to compare these with the general population.

DESIGN

Longitudinal, descriptive study.

PARTICIPANTS

2 66 241 patients from 11 NQRs and 49 169 participants from the general population were included in the study.

PRIMARY AND SECONDARY OUTCOME MEASURES

Proportions of problems reported in the five EQ-5D-3L dimensions, EQ VAS scores of participants' own health and proportions of participants and mean/median EQ VAS score in the Paretian Classification of Health Change (PCHC) categories.

RESULTS

In most of the included registers, and the general population, problems with pain/discomfort were the most frequently reported at baseline and at 1-year follow-up. Mean EQ VAS score (SD) ranged from 45.2 (22.4) among disc hernia patients to 88.1 (15.3) in wrist and hand fracture patients at baseline. They ranged from 48.9 (20.9) in pulmonary fibrosis patients to 83.3 (17.4) in wrist and hand fracture patients at follow-up. The improved category of PCHC, improvement in at least one dimension without deterioration in any other, accounted for the highest proportion in several diagnoses, corresponding with highest improvement in mean EQ VAS score.

CONCLUSIONS

The study documented self-reported health of several different patient groups using the EQ-5D-3L in comparing with the general population. This demonstrated the important role of patient-reported outcomes in routine clinical care, to assess and follow-up health status and progress within different groups of patients. The EQ-5D-3L descriptive system and EQ VAS have an important role in providing a 'common denominator', allowing comparisons across NQRs and specific diagnoses.

TRIAL REGISTRATION NUMBER

ClinicalTrials.gov (NCT04359628).

Combining EQ-5D-5L items into a level summary score: demonstrating feasibility using non-parametric item response theory using an international dataset

Background

The EQ-5D-5L is a well-established health questionnaire that estimates health utilities by applying preference-based weights. Limited work has been done to examine alternative scoring approaches when utility weights are unavailable or inapplicable. We examined whether the Mokken scaling approach can elucidate 1) if the level summary score is appropriate for the EQ-5D-5L and 2) an interpretation of such a score.

Methods

The R package “mokken” was used to assess monotonicity (scaling coefficients H, automated item selection procedure) and manifest invariant item ordering (MIIO: paired item response functions [IRF], H^T). We used a rich dataset (the Multiple Instrument Comparison, MIC) which includes EQ-5D-5L data from six Western countries.

Results

While all EQ-5D-5L items demonstrated monotonicity, the anxiety/depression (AD) item had weak scalability (H_i = 0.377). Without AD, scalability improved from H_s = 0.559 to H_s = 0.714. MIIO revealed that the 5 items can be ordered, and the ordering is moderately accurate in the MIC data (H^T = 0.463). Excluding AD, H^T improves to 0.743. Results were largely consistent across disease and country subgroups.

Discussion

The 5 items of the EQ-5D-5L form a moderate to strong Mokken scale, enabling persons to be ordered using the level summary score. Item ordering suggests that the lower range of the score represents mainly problems with pain and anxiety/depression, the mid-range indicates additional problems with mobility and usual activities, and middle to higher range of scores reveals additional limitations with self-care. Scalability and item ordering are even stronger when the anxiety/depression item is not included in the scale.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11136-021-02922-1.

The Young Disability Questionnaire-Spine: item development, pilot testing and conceptualisation of a questionnaire to measure consequences of spinal pain in children

OBJECTIVE

The objective of the present study was to develop a questionnaire to measure the consequences of neck, midback and low back pain, relevant for schoolchildren aged 9-12 years.

DESIGN

The development of the questionnaire was carried out in three phases: (1) generation of items, (2) pilot testing and (3) conceptualisation.

SETTING

Danish primary schools.

PARTICIPANTS

Children aged 9-12 years from a local school were invited for completion of questionnaires and subsequent interviews.

METHODS

In phase 1 an extensive literature search identified items from existing questionnaires measuring musculoskeletal disability in children. These were added to items from a previously conducted qualitative study and constituted the basis for the new questionnaire. In phase 2 two consecutive pilot tests were performed to test for comprehension and feasibility of the questionnaire. Phase 3 consisted of a categorisation of the newly developed items according to the WHO's International Classification of Functioning, Disability and Health (ICF).

RESULTS

The combination of the previously conducted qualitative study and the literature review resulted in an item pool of 35 items. Through the process of item formulation and pilot testing, these were reduced to 28 items in the final questionnaire, which represented all categories in the ICF model. The qualitative study identified codes giving relative weight to four important domains. These were not included in any of the existing questionnaires but were added to the new questionnaire.

CONCLUSIONS

We developed the first version of a questionnaire to measure the consequences of back or neck pain in children. The process showed the importance of combining research methods, each adding important contributions to the final product. Subsequent work will finalise the questionnaire, allowing various options for use of the questionnaire.

Validity of EQ-5D-5L health-related quality of life questionnaire in self-reported diabetes: evidence from a general population survey

Background

This study aimed to assess the validity of the EQ-5D-5L in respondents with self-reported diabetes coming from a representative general population survey.

Methods

2974 respondents from the general adult population of Poland, chosen with multi-stage random sampling, were surveyed with HRQoL instruments (EQ-5D-5L, EQ VAS, SF-12, EQ-5D-3L) and a screening question about diabetes. To obtain EQ-5D index values, we used country-specific Polish value sets. We compared the instruments in terms of the ceiling effect, discriminatory power and frequency of individual health states. We evaluated construct validity in terms of known-groups validity and convergent validity of EQ-5D-5L dimensions and index values with other HRQoL measures.

Results

In respondents with diabetes (n = 247), the percentage reporting 'no problems' with EQ-5D-3L was reduced by 34.5% with the use of EQ-5D-5L (from 14.2% to 9.3%, respectively). A significant improvement in informativity was noticed in mobility and pain/discomfort dimensions (a relative increase of 23.1% and 22.7%, respectively). Known-groups construct validity analysis confirmed prior hypotheses—index scores were higher in the following groups: younger respondents, males, those taking no medication or oral antidiabetic drugs, and respondents with higher levels of education. The convergence between related EQ-5D-5L and EQ-5D-3L or SF-6D dimensions was stronger than between unrelated dimensions. The Bland–Altman analysis showed a mean difference between EQ-5D-5L and EQ-5D-3L, SF-6D, EQ VAS/100 index scores of 0.047, 0.165 and 0.231 respectively.

Conclusions

Our results support the validity of the EQ-5D-5L descriptive system and EQ-5D-5L index, based on the directly measured value set in respondents with self-reported diabetes coming from the general population.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12955-021-01780-2.

Psychometric properties of Short Form-36 Health Survey, EuroQol 5-dimensions, and Hospital Anxiety and Depression Scale in patients with chronic pain

Supplemental Digital Content is Available in the Text.

This large-sample item response theory-based evaluation assessed the measurement properties of SF-36, EQ-5D, and hospital anxiety and depression scale for chronic pain patients in clinical settings.

Recent research has highlighted a need for the psychometric evaluation of instruments targeting core domains of the pain experience in chronic pain populations. In this study, the measurement properties of Short Form-36 Health Survey (SF-36),EuroQol 5-dimensions (EQ-5D) and Hospital Anxiety and Depression Scale (HADS) were analyzed within the item response-theory framework based on data from 35,908 patients. To assess the structural validity of these instruments, the empirical representations of several conceptually substantiated latent structures were compared in a cross-validation procedure. The most structurally sound representations were selected from each questionnaire and their internal consistency reliability computed as a summary of their precision. Finally, questionnaire scores were correlated with each other to evaluate their convergent and discriminant validity. Our results supported that SF-36 is an acceptable measure of 2 independent constructs of physical and mental health. By contrast, although the approach to summarize the health-related quality of life construct of EQ-5D as a unidimensional score was valid, its low reliability rendered practical model implementation of doubtful utility. Finally, rather than being separated into 2 subscales of anxiety and depression, HADS was a valid and reliable measure of overall emotional distress. In support of convergent and discriminant validity, correlations between questionnaires showed that theoretically similar traits were highly associated, whereas unrelated traits were not. Our models can be applied to score SF-36 and HADS in chronic pain patients, but we recommend against using the EQ-5D model due to its low reliability. These results are useful for researchers and clinicians involved in chronic pain populations because questionnaires' properties determine their discriminating ability in patient status assessment.

Psychometric properties of the EQ-5D-5L: a systematic review of the literature

PURPOSE

Although the EQ-5D has a long history of use in a wide range of populations, the newer five-level version (EQ-5D-5L) has not yet had such extensive experience. This systematic review summarizes the available published scientific evidence on the psychometric properties of the EQ-5D-5L.

METHODS

Pre-determined key words and exclusion criteria were used to systematically search publications from 2011 to 2019. Information on study characteristics and psychometric properties were extracted: specifically, EQ-5D-5L distribution (including ceiling and floor), missing values, reliability (test-retest), validity (convergent, known-groups, discriminate) and responsiveness (distribution, anchor-based). EQ-5D-5L index value means, ceiling and correlation coefficients (convergent validity) were pooled across the studies using random-effects models.

RESULTS

Of the 889 identified publications, 99 were included for review, representing 32 countries. Musculoskeletal/orthopedic problems and cancer (n = 8 each) were most often studied. Most papers found missing values (17 of 17 papers) and floor effects (43 of 48 papers) to be unproblematic. While the index was found to be reliable (9 of 9 papers), individual dimensions exhibited instability over time. Index values and dimensions demonstrated moderate to strong correlations with global health measures, other multi-attribute utility instruments, physical/functional health, pain, activities of daily living, and clinical/biological measures. The instrument was not correlated with life satisfaction and cognition/communication measures. Responsiveness was addressed by 15 studies, finding moderate effect sizes when confined to studied subgroups with improvements in health.

CONCLUSIONS

The EQ-5D-5L exhibits excellent psychometric properties across a broad range of populations, conditions and settings. Rigorous exploration of its responsiveness is needed.

Understanding the association between osteoarthritis and social participation: a CLSA population-based study

OBJECTIVE

The focus on disability in osteoarthritis (OA) has largely been on performing specific activities, neglecting wider implications for social participation (SP). We investigated the association between OA and SP considering activity limitations (AL) and instrumental supports (IS) as intervening variables in the association.

METHODS

Data were from 21,214 respondents aged 45-85 from cycle 1, Canadian Longitudinal Study on Aging. The questionnaire elicited: self-reported doctor-diagnosed OA; difficulty with 14 activities; perceived availability and receipt of IS; and 17 SP activities. Structural equation modeling was used.

PRIMARY OUTCOME

SP; Primary predictor: OA; Intervening variables: AL, received IS, perceived IS. Latent variables were developed for intervening and SP variables. Covariates: age, sex, body mass index, income, education, smoking, comorbidity count.

RESULTS

Mean age was 63 years, 51% were female, and 26.5% reported OA. Two distinct SP indicators were identified, SP-Diversity and SP-Intensity. Without considering intervening variables, minimal/no association was found between OA and SP. When considered, unique pathways linking OA and SP were found. The overall negative association between AL and SP was, in part, direct and, in part, buffered by both receipt and perceived availability of IS. Absent AL, OA was associated with greater SP.

CONCLUSIONS

Enhanced SP in people with OA without AL may reflect proactive steps taken by those with mild OA to maintain activity and social engagement. For those with AL, findings highlight the need for interventions to mitigate limitations and draw particular attention to the importance of both provision and awareness of available IS in maintaining SP.

Does the EQ-5D usual activities dimension measure what it intends to measure? The relative importance of work, study, housework, family or leisure activities

Background

The EQ-5D is the most widely used generic preference-based health-related quality of life measure. It comprises five dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. The usual activities dimension asks respondents to evaluate the severity of problems in their usual activities, such as work, study, housework, family or leisure activities. The primary aim of this study is to investigate whether the EQ-5D (five-level) usual activities dimension captures those activities that it intends to capture. We further assess the relative importance of each of these activities for the usual activities dimension.

Methods

Data include 7933 respondents from six countries: Australia, Canada, Germany, Norway, the UK, and the US. Logistic regression and ordinary least square regression models investigate the relationship between the usual activities dimension and its main predictors (work/study, housework, family, and leisure activities). A Shapley value decomposition method was applied to measure the relative importance of each predictor.

Results

Work/study, housework, family, and leisure activities were all significant (p < 0.001) determinants of usual activities dimension. The respective marginal contribution (in %) of housework, leisure, work/study and family to UA dimension (as a share of goodness-of-fit) is 28.0, 26.2, 20.8 and 14.7 in the logistic regression model. This finding is consistent when linear regression is used as an alternative model.

Conclusions

The usual activities dimension in EQ-5D reflects the specific activities described to respondents. Therefore, the usual activities dimension measures what it really intends to measure.

Electronic supplementary material

The online version of this article (10.1007/s11136-020-02501-w) contains supplementary material, which is available to authorized users.

A conceptual map of health-related quality of life dimensions: key lessons for a new instrument

Purpose

Quality-adjusted life years (QALYs) represent a critical metric in economic evaluations impacting key healthcare decisions in many countries. However, there is widespread disagreement as to which is the best of the health state utility (HSU) instruments that are designed to measure the Q in the QALY. Instruments differ in their descriptive systems as well as their valuation methodologies; that is, they simply measure different things. We propose a visual framework that can be utilized to make meaningful comparisons across HSU instruments.

Methods

The framework expands on existing HRQoL models, by incorporating four distinctive continua, and by putting HRQoL within the broader notion of subjective well-being (SWB). Using this conceptual map, we locate the five most widely used HSU-instruments (EQ-5D, SF-6D, HUI, 15D, AQoL).

Results

By individually mapping dimensions onto this visual framework, we provide a clear picture of the significant conceptual and operational differences between instruments. Moreover, the conceptual map demonstrates the varying extent to which each instrument moves outside the traditional biomedical focus of physical health, to also incorporate indicators of mental health and social well-being.

Conclusion

Our visual comparison provides useful insights to assess the suitability of different instruments for particular purposes. Following on from this comparative analyses, we extract some important lessons for a new instrument that cover the domains of physical, mental and social aspects of health, i.e. it is in alignment with the seminal 1948 WHO definition of health.

Exploring the Internal Structure of the EQ-5D Using Non-Preference-Based Methods

BACKGROUND

When the EuroQol EQ-5D is applied in settings other than resource allocation, a non-preference-based score may be more appropriate than societal, preference-weighted utility. To develop a psychometric score for the EQ-5D, its structural relationship, ie, how the 5 items/dimensions interrelate, must be understood to inform appropriate methods of summarizing the instrument.

OBJECTIVES

To explore psychometrically derived approaches of elucidating the 5-level EQ-5D (EQ-5D-5L) item structure.

METHODS

Three measurement models were assessed. All 5 items were modeled as reflective indicators using confirmatory factor analysis. EQ-5D-5L items were conceptualized as formative indicators, and other health scales (eg, the short form 36 health survey) were conceptualized as reflective indicators in Multiple Indicators Multiple Causes models (external MIMIC). The EQ-5D-5L items were modeled as a combination of formative and reflective indicators in internal MIMIC models. Results across 9 data sets from various countries and patient groups were examined to determine their robustness.

RESULTS

All items loaded well (0.63-0.96) in the confirmatory factor analysis except for anxiety/depression (0.20-0.66, excluding 1 outlier). The model fit statistics of the external MIMIC models were poor, and the coefficients of the Self-Care dimension were small. The internal MIMIC model with Mobility, Pain/Discomfort, and Anxiety/Depression as formative indicators and Self-Care and Usual Activities as reflective indicators fit best. The model results of the Spanish valuation data set were outliers.

CONCLUSIONS

Although there were some variations in results across subgroups, the relationship between the items remained robust. The evidence calls for testing of formative/reflective combination approaches to summarize the EQ-5D-5L.