The Psychometric Performance of Generic Preference-Based Measures in Informal Carers: A Systematic Review of Validation Studies.

Key Points

Background

Traditionally, economic evaluations in healthcare have considered only the costs and benefits of the person receiving care. However, international guidelines now support the inclusion of informal carer costs and benefits in economic evaluations, particularly when conducting an analysis using a societal or a public sector perspective [ 1 , 2 ]. These contemporary guidelines are also reflected in the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) Value Assessment Framework, which sought to broaden the appreciation of value in healthcare by identifying elements not traditionally considered in a cost-effectiveness analysis [ 3 ]. The inclusion of informal carer effects in a cost-effectiveness analysis is an underappreciated element, potentially undervaluing health interventions if excluded [ 3 ]. This decision to expand the scope of economic evaluations to include informal carers acknowledges that the effects of a health condition impact not only the patient, but also people around the patient. This has been described in a theoretical framework by Basu and Meltzer as spillover effects [ 4 ]. The framework suggests that the total health effect of a disease includes: (1) the direct effect of the disease on the individual’s (patient’s) utility; (2) the direct effect of the disease on the informal carer’s utility; and (3) the indirect effect of the informal carer’s utility on the individual’s utility, where the latter two effects (2 and 3) represent spillover effects [ 4 , 5 ]. Consequently, treating a disease also has a direct effect on both patients and informal carers. That is, if an intervention is provided to address a specific disease or condition of a patient, its effects may also benefit the informal carer [ 4 , 5 ]. The Spillovers in Health Economic Evaluation and Research (SHEER) task force, which was convened to provide guidance around the inclusion of family and caregiver health spillovers in economic evaluations, framed a formal definition of informal carer spillovers to ensure consistency. The definition provided by SHEER states that family and caregiver spillovers consist of the impact of the individual’s health condition(s), the intervention(s) targeting the health condition(s) or both on the family members’ and/or caregivers’ health-related quality of life (HRQoL), both positive or negative [ 2 ]. Furthermore, these health impacts may be a product of caring for or about the care recipient and/or the effect of a healthcare intervention for the condition [ 2 ].

There are several measures available to measure caregiver impacts for use in an economic evaluation. For instance, existing carer-specific preference-based measures capturing informal carer quality of life include the Adult Social Care Outcomes Toolkit for Carers (ASCOT-Carer) [ 6 ], the Care-Related Quality of Life (CarerQol) [ 7 ], and Carer Experience Scale (CES) [ 8 ]. Whilst these measures are preference based and were developed for use in economic evaluations, they were not anchored onto the 0–1 dead–full health quality-adjusted life-year scale, which is often the requirement of health technology assessment submissions. Jurisdictions such as the National Institute for Health and Care Excellence (NICE) advocate the use of the EQ-5D, a generic preference-based measure (GPBM) of HRQoL for a cost-utility analysis [ 9 ]. However, while current NICE guidelines state that “ evaluations should consider all health effects for patients, and, when relevant, carers ” [ 9 ], no minimum standards have been formulated for measuring carer HRQoL [ 10 ]. To provide some guidance, the SHEER task force recommended that economic evaluations involving informal carers use the same HRQoL measure used in patients if the aim is to aggregate both patient and informal carer outcomes. Moreover, using GPBMs as opposed to carer-specific measures, allow evaluations to be consistent and facilitates comparability across different evaluations [ 2 ].

There are several GPBMs that are available, such as the Assessment of Quality of Life (AQoL) instruments [ 11 – 16 ], the EQ-5D (-3L and -5L) [ 17 , 18 ], EuroQol Health and Wellbeing short form (EQ-HWB-9) [ 19 ], Health Utilities Index (HUI Mark 2 and 3) [ 20 ], the National Institutes of Health Patient-Reported Outcomes Measurement Information System (PROMIS) preference score (PROPr) [ 21 ], Short-Form 6-Dimension (SF-6D) versions 1 and 2 [ 22 – 24 ], Quality of Well Being Self-Administered (QWB-SA) [ 25 ] and 15D instrument [ 26 ]. Although these measures are validated generic HRQoL measures, they differ in scope and their coverage of HRQoL domains, which may lead to differences in their performance among informal carers. However, psychometric evidence on the performance of these GPBMs in informal carers has not yet been systematically evaluated. Therefore, the aim of the current study was to examine the literature around the psychometric performance of GPBMs in informal carers. The findings of this review provide a better overall understanding of the performance of GPBMs in capturing carer outcomes and highlight some potential existing gaps.

Methods

This review follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. International Prospective Register of Systematic Reviews (PROSPERO) registration for this review is CRD42023434651. Changes to the original review protocol were made that included broadening the review beyond the focus on the EQ-5D tools to include also other GPBMs. The decision to include other GPBMs was made to expand the usefulness of the findings for health technology assessment agencies that do not specifically prefer the EQ-5D in their submissions.

Literature Search

Four electronic databases (i.e. CINAHL, PsycInfo, Embase and MEDLINE) were searched for studies that evaluated the psychometric properties of GPBMs in informal carers. The electronic search was conducted in December 2023 and updated in October 2024. Details of the search strategy are available in Appendix 1 of the Electronic Supplementary Material (ESM). Search terms and MeSH headings relating to three themes were used: (1) GPBMs; (2) psychometric properties; and (3) informal carers. No filters were utilised during the electronic search. Forward and backward citation searches in Scopus were conducted based on the included studies.

Inclusion and Exclusion Criteria

Studies were included if they examined the psychometric properties of GPBMs among informal carers. Studies involving different population groups were still included if psychometric results specific to informal carers were reported and could be extracted. Any type of condition or disease of the care recipient was permitted. Studies were excluded if they were published as reviews, editorials, conference abstracts, and not in the English language. If a relevant systematic review was identified, it was utilised for cross-checking of identified studies, but was not included in this review.

Study Selection, Data Extraction and Synthesis

References identified from the electronic search were uploaded to Covidence [ 27 ], where the software automatically identified and removed duplicates. Title and abstract screening, followed by full-text screening were conducted by two independent reviewers (JF as first reviewer, VS or YBB as second reviewer). Judgement differences were resolved by a senior reviewer (LE). Prior to commencing data extraction, reviewers JF, VS, and LE met to discuss and pilot data extraction using two of the identified papers. It was agreed that data extraction was to be conducted by a single reviewer (JF), with the accuracy confirmed by a second reviewer (VS). In instances of discrepancies, these were discussed and corrected by JF and VS, with further unresolved conflicts reviewed by LE. However, there were no instances of discrepancies that warranted the senior reviewer to review data extraction. Data extracted included study characteristics (e.g. carer and care recipient characteristics and caring situation), GPBMs and other measures used, and results of the psychometric property tests. Studies were then grouped according to patient population (i.e. children, adults, older adults and general population) to supplement the analysis of evidence.

Psychometric Properties

Validity: Known-group validity assesses whether a measure can discriminate between groups that are known to differ [ 28 ]. Known-group validity results were extracted and classified as (+), meaning significant differences were observed between groups, otherwise the results were deemed (−) or not observed. In order to further aid analysis, explored groups were clustered according to: whether they were related to carer characteristics (e.g. age, gender/sex), carer quality of life (stratified by carer-specific preference-based measure scores), caring situation and burden (e.g. relationship to care recipient), carer health and well-being (e.g. self-rated health status), patient health characteristics (e.g. health condition) and other groups (e.g. compared with non-carers). Results were then reported as a ratio of the number of significant differences (+) or observed (numerator) over the total number of group analyses conducted (denominator). Not all studies used a priori hypotheses to guide known-group validity analysis as recommended by the COSMIN guidelines, as some studies adopted an exploratory approach. As part of data extraction, the number of a priori hypotheses reported by each study were also extracted to guide analysis and the risk of bias assessment.

Convergent validity assesses the degree to which a measure, in this case the GPBMs, correlates with other measures [ 28 ]. Reported correlation coefficients were judged according to the correlation strength using the following categories: ‘weak correlation’ (≤0.4), ‘moderate correlation’ (0.4–0.7) and ‘strong correlation’ (>0.7) [ 29 ]. During the analysis, correlation coefficient findings were grouped as either at least moderately correlated or weakly correlated. Measures used for convergent validity were also classified as to whether they relate to other GPBMs, carer quality-of-life measures (i.e. ASCOT-Carer, CarerQoL-7D, and CES), carer health and well-being including mental health, caregiving situation and burden, relationships, patient characteristics and other measures. Results were then reported as the number of moderate or strong correlations over the number of correlation analyses conducted. Similar to known-group validity, using a priori hypotheses is recommended by the COSMIN guidelines, which however were not always stated. The number of a priori hypotheses reported by each study were therefore extracted to guide analysis and the risk of bias assessment.

Responsiveness: This relates to the ability of an instrument to capture changes over time where a change in scores or health is expected [ 30 ]. Responsiveness to change was determined as to whether the measure showed significant gradient effects between no change, worsening, and improving of health or situation.

Reliability: Test–retest reliability is the ability of a measure to be consistent, stable and reliable over time when there is no health change [ 28 ]. Test–retest reliability was examined using intraclass correlation coefficient (ICC) and Kappa scores. Thresholds for ICC were: ≤ 0.39 ‘poor’; 0.40–0.59 ‘fair’; 0.60–0.74 ‘good’, and ≥ 0.75 ‘excellent’ [ 31 ], while thresholds for Kappa scores were: ≤ 0.20 ‘poor’; 0.21–0.40 ‘fair’; 0.41–0.60 ‘moderate’; 0.61–0.80 ‘substantial’, and ≥ 0.81 ‘almost perfect’ agreement [ 32 ].

Structural validity: This relates to the degree to which the scores of a measure adequately reflect the dimensionality of the construct to be measured [ 28 ]. This is often examined using exploratory factor analysis or confirmatory factor analysis. This was discussed narratively in the review.

Ceiling effect: The proportion of respondents reporting full health. A ceiling effect is observed when the proportion reporting full health using the GPBM is ≥15% [ 30 ].

Quality Assessment/Risk of Bias

The COSMIN risk of bias tool was used for the quality assessment [ 33 ]. Two independent reviewers (JF, VS) conducted the risk of bias assessment. Conflicts in judgement were resolved by a third reviewer (LE). Only relevant items from the COSMIN risk of bias tool were used (i.e. specific to the psychometric property assessed). Results from the risk of bias assessment were reported narratively.

Results

Study Selection and Inclusion

The literature search retrieved 3718 studies. After 1514 duplicates were removed, 2204 underwent the title and abstract screening process, with 2154 further studies being excluded. From the 50 studies that underwent full-text screening, 29 articles were excluded. Reasons for exclusion were conference abstracts, incorrect population (patient or proxy responses), caregiver measure (non-GPBM) validation, application and qualitative studies. There were 21 articles identified from the literature search. The Health-related Quality of Life Instrument with 8 Items (HINT-8) [ 34 ], a new GPBM that was not part of the original search strategy was identified during the screening process. A forward and backward citation search in Scopus did not find further studies. The percentages of agreement between screeners were 97% for title and abstract screening and 87% for a full-text review (Fig. 1 ).

Study Characteristics

Table 1 provides a detailed summary of characteristics of the included studies. There were 12 studies that tested a single GPBM (EQ-5D-3L, n = 5; EQ-5D-5L, n = 3; EQ-HWB-9, n = 2; HUI2, n = 1; QWB-SA, n = 1) [ 35 – 46 ], with the remaining studies ( n = 9) testing two or more GPBMs [ 34 , 47 – 54 ]. The EQ-5D-3L was evaluated in a total of nine studies [ 39 , 42 – 45 , 48 , 49 , 53 , 54 ], whereas the EQ-5D-5L was tested in eight studies [ 34 , 40 , 41 , 46 , 47 , 50 , 51 , 54 ]. One study compared the EQ-5D-3L and the EQ-5D-5L [ 54 ]. The EQ-HWB-9 [ 36 , 37 , 50 , 51 ] and the SF-6D version 1 (SF-6Dv1) [ 47 , 48 , 52 , 53 ] each had four studies and the HUI3 had two studies [ 49 , 52 ]. The HINT-8 [ 34 ], HUI2 [ 38 ] and QWB-SA [ 52 ] had one study each. No studies were identified for 15D or the AQoL measures.

Only limited countries were represented by the included studies. The USA was most represented by six studies [ 35 , 38 , 39 , 48 , 50 , 52 ], followed by the UK [ 41 – 44 , 47 ] and China [ 40 , 45 , 53 , 54 ] with four studies each and three studies from Australia [ 36 , 37 , 46 ]. Italy [ 51 ], Iran [ 49 ] and South Korea [ 34 ] contributed one study each. The final study included informal carers from three countries, the UK, Germany and France [ 43 ].

Population Characteristics

In terms of the care recipients, there were six studies focusing on children care recipients with mostly parent carers [ 36 , 37 , 39 , 47 , 48 , 52 ] across health conditions that included craniofacial malformations [ 52 ], meningitis [ 47 ], autism [ 39 , 48 ] and non-specific conditions [ 36 , 37 ]. There were eight adult care recipient studies [ 35 , 40 , 42 , 45 , 49 , 50 , 53 , 54 ] with various carers (i.e. spouse/partners, parents, children, relatives, friends). The health conditions also varied in the adult care recipient studies, from non-specified conditions [ 50 ] to adults using long-term or social care [ 42 ] to multiple sclerosis [ 49 ] and cancers (e.g. leukaemia, colorectal and breast) [ 35 , 40 , 45 , 53 , 54 ]. For older adults, there were six studies that primarily involved spouse/partners and children as the carers [ 34 , 38 , 41 , 43 , 44 , 46 ]. Five studies were in cognitive diseases, such as Alzheimer’s disease and dementia [ 34 , 38 , 43 , 44 , 46 ], with a further study in care recipients with dementia, stroke, mental illness or rheumatoid arthritis [ 41 ]. The last study compared informal carers and non-carers in the general population [ 51 ].

Psychometric Criteria Overview

Table 2 provides an overview of the psychometric criteria by the identified studies, and organised according to patient population (i.e. children, adults, older adults and general population). All GPBMs, except the HINT-8, had evidence of known-group validity from 17 studies [ 36 – 41 , 43 , 45 – 54 ]. The EQ-5D-3L [ 39 , 43 , 45 , 48 , 49 , 53 , 54 ], EQ-5D-5L [ 40 , 41 , 46 , 47 , 50 , 51 , 54 ], EQ-HWB-9 [ 36 , 37 , 50 , 51 ], HUI3 [ 49 , 52 ] and SF-6Dv1 [ 47 , 48 , 52 , 53 ] had at least two studies reporting known-group validity. The HUI2 [ 38 ] and the QWB-SA [ 52 ] had one study each reporting on known-group validity. For convergent validity, 19 studies reported evidence across all the GPBMs [ 35 – 50 , 52 – 54 ]. Except for the HUI2 [ 38 ] and HINT-8 [ 34 ], the EQ-5D-3L [ 39 , 42 – 45 , 48 , 49 , 53 , 54 ], EQ-5D-5L [ 40 , 41 , 46 , 47 , 50 , 54 ], EQ-HWB-9 [ 36 , 50 ], HUI3 [ 49 , 52 ], SF-6Dv1 [ 47 , 48 , 52 , 53 ] and QWB-SA [ 35 , 52 ] had at least two studies reporting on convergent validity. Only three studies evaluated responsiveness to change. Of these three studies, one was for the EQ-HWB-9 [ 36 ], another for the EQ-5D-5L [ 41 ] and the third involved both the EQ-5D-5L and the SF-6Dv1 [ 47 ]. Test–retest reliability evidence was identified for the EQ-5D-3L, EQ-5D-5L and EQ-HWB-9 across three studies, including one study that was a head-to-head comparison of the EQ-5D-3L and the EQ-5D-5L [ 36 , 40 , 54 ]. Seven studies reported ceiling effects across the EQ-5D-3L [ 39 , 44 , 45 , 53 , 54 ], EQ-5D-5L [ 46 , 50 , 54 ], EQ-HWB-9 [ 50 ] and SF-6Dv1 [ 53 ], including three studies comparing at least two GPBMS [ 50 , 53 , 54 ]. Only one study examined structural validity through an exploratory factor analysis involving the EQ-5D-3L [ 42 ].

Validity

Known-Group Validity

Table 3 shows an overview of the reported known-group validity analyses. Out of the 175 reported analyses, only 81 were based on pre-defined hypotheses across the GPBMs and studies. Both the HUI2 and the QWB-SA had no known-group validity evidence available. All other GPBMs were examined for discriminant validity across multiple sets of groups (EQ-5D-3L, 46 groups; EQ-5D-5L, 55 groups; EQ-HWB-9, 22 groups; HUI3, 15 groups; SF-6D, 34 groups). In addition, over 60% (EQ-5D-3L, 65%; EQ-5D-5L, 84%; EQ-HWB-9, 77%; HUI3, 73%; SF-6Dv1, 62%) of the known-group analyses reported that the respective GPBMs were able to discriminate between groups. There were also varied groups used across the studies, with consistency only present in head-to-head comparison studies. For example, in a head-to-head comparison of the HUI3, SF-6D and QWB-SA, both HUI3 and SF-6Dv1 were able to discriminate based on the CES-D (a measure of depression) while the QWB-SA did not [ 52 ]. In another study, comparing the EQ-5D-5L and the SF-6Dv1 over 12 groups, results were mixed with the EQ-5D-5L confirming 4 of 12 hypothesised groups and the SF-6Dv1 confirming 5 out of 12 hypotheses. The EQ-5D-5L was able to discriminate groups related to the positive and negative effects of caring on personal health and family life, whereas the SF-6Dv1 did not. However, the SF-6Dv1 was able to discriminate groups relating to providing daily care, being the main carer and work impact, whereas the EQ-5D-5L did not. Neither measure discriminated in groups related to the impact on finances, constant supervision, majority of care provision and shared household arrangement [ 47 ]. In another study comparing the EQ-5D-3L and the EQ-5D-5L, the EQ-5D-5L was able to discriminate between carer age, education and income, while the EQ-5D-3L did not show the same ability to discriminate [ 54 ]. Two other studies were head-to-head comparisons showing similar results of discrimination in identified groups, one comparing the EQ-5D-3L and the HUI3 [ 49 ], and the other comparing the EQ-5D-3L and the SF-6Dv1 [ 53 ]. A detailed list of the results across the reported groups is available in Appendix 2 of the ESM.

Convergent Validity

Table 4 provides an overview of the results for convergent validity. There were mixed results for convergent validity. The majority of the reported findings indicate that the GPBMs were at least moderately correlated with other GPBMs, carer-specific preference-based measures and carer health and well-being measures (including mental health measures). In terms of convergence with carer-specific preference-based measures, the ASCOT-Carer was at least moderately correlated with the EQ-5D-5L [ 41 ] and the EQ-HWB-9 [ 50 ]. There was a weak correlation reported between the EQ-5D-3L and the ASCOT-Carer [ 42 ]. The CarerQoL-7D was at least moderately correlated with the EQ-5D-3L [ 48 , 49 ], EQ-5D-5L [ 41 ], EQ-HWB-9 [ 50 ], HUI3 [ 49 , 52 ], SF-6Dv1 [ 48 , 52 ] and QWB-SA [ 52 ]. However, there were consistent weak correlations between the CES and EQ-5D-3L, n = 1 [ 42 ]; EQ-5D-5L, n = 2 [ 41 , 47 ]; and SF-6D, n = 1 [ 47 ]. The majority of the analyses evaluating convergence with carer-specific quality-of-life measures also showed moderate correlations, with weak correlations observed in instances where GPBMs were compared with domains of other measures. For example, the EQ-5D-3L had a weak correlation with the SF-12 mental health composite score [ 44 ], EQ-5D-5L with the WHOQOL environment and social domains [ 40 ], the EQ-HWB-9 with the PWI-A safety domain [ 36 ] and the QWB-SA with multiple SF-36 domains [ 35 ]. However, mostly weak correlations were observed between GPBMs and carer burden, with the only moderate correlations found between the CES ‘Activities outside of caring’ domain [ 42 ] and the EQ-5D-3L, and the Short Zarit Burden Interview Index and Personal domain with the EQ-5D-5L and HINT-8 [ 34 ]. Except for two regression analyses with the HUI2, relationship measures and patient health and well-being measures had weak correlations with EQ-5D-3L and EQ-5D-5L. The observations appear to be consistent across care recipient populations. Further details regarding specific measures used for convergent validity and reported correlations are presented in Appendix 3 of the ESM.

Responsiveness

Two child care recipient studies and one older adult care recipient study evaluated responsiveness to change [ 36 , 41 , 47 ]. In the child care recipient studies, the EQ-HWB-9 showed gradient effects in terms of change in carer K6 and global health status; however, it also reported non-significant gradient effects on change in carer PWI-A. In addition, no gradient effect was observed in terms of change in number of adverse life events experienced [ 36 ]. Another study in child care recipients reported that the SF-6Dv1 was not responsive to changes in recipient health status (using the EQ-5D-5L), whereas the EQ-5D-5L was responsive showing gradient observed effects. In terms of change in hours of care provision, both EQ-5D-5L and SF-6Dv1 were not responsive [ 47 ]. One older adult care recipient study evaluated the responsiveness of the EQ-5D-5L, showing that the EQ-5D-5L was not responsive to change in care recipient EQ-5D-5L health status, but responsive to change in hours of care provided [ 41 ].

Reliability

Test–retest reliability evidence was identified for the EQ-5D-3L, EQ-5D-5L and EQ-HWB-9 across three studies, reporting positive results for the three measures [ 36 , 40 , 54 ]. In the child care recipient population, an ICC of 0.87 was reported for the EQ-HWB-9 and weighted Kappa scores for domains varied between 0.38 and 0.61 over a 2-day response period [ 36 ]. The test–retest reliability of the EQ-5D-5L was evaluated in two studies among the adult care recipient population. In the first study, an ICC of 0.987 and Kappa scores between 0.79 and 0.85 over a 24-hour timeframe were reported [ 40 ], while a head-to-head comparison study showed that the reliability of the EQ-5D-5L was greater (ICC: 0.725, weighted Kappa: 0.62–0.72) than the reliability of the EQ-5D-3L (ICC: 0.657, weighted Kappa: 0.54–0.74) over a 2-day response period [ 54 ]. No study was identified exploring test–retest reliability for the older adult care recipient population.

Structural Validity

The sole study on structural validity was an exploratory factor analysis on the EQ-5D-3L that was compared to the ASCOT-Carer and the CES. This study was conducted in the adult care recipient group. The study reported that four of the five EQ-5D-3L domains (mobility, self-care, usual activities, pain and discomfort) loaded onto the same factor, while anxiety/depression did not load in any of the other factors. There were no EQ-5D-3L items that overlapped with items of the ASCOT-Carer and CES [ 42 ].

Ceiling Effect

Five of five (100%) studies observed ceiling effects for the EQ-5D-3L, where 21–33% of respondents reported full health [ 39 , 44 , 45 , 53 , 54 ]. For the EQ-5D-5L, one of two (50%) studies reported a ceiling effect, with 17% reporting full health [ 50 ]. None of the studies observed ceiling effects in the EQ-HWB-9 and the SF-6D. In three head-to-head comparison studies, the EQ-HWB-9 (3%) had less of a ceiling effect than the EQ-5D-5L (17%) [ 50 ], while the SF-6Dv1 also had less of a ceiling effect (< 15%) than the EQ-5D-3L (33%), with a third study reporting that the EQ-5D-5L (13%) had a lower ceiling effect than the EQ-5D-3L (21%) [ 54 ]. Only one of the six studies evaluating ceiling effects focused on older adults using the EQ-5D-3L, where 27% of carers, primarily partners, reported full health [ 44 ].

Risk of Bias

In total, there were 50 sets of psychometric properties assessed across the 21 studies. Rating agreement between reviewers was 88%. Based on a risk of bias assessment using the COSMIN tool, there were 36 (72%) analyses that were deemed ‘very good’, 12 (24%) were considered ‘adequate’ and 2 (4%) ‘doubtful’. None of the analyses was found to be ‘inadequate’. Two of the three test–retest analyses were deemed ‘adequate’, with the other considered ‘doubtful’. Issues common for an ‘adequate’ judgement were not reporting enough information to confirm whether carers were stable between the two time periods and not reporting of the weighting scheme used for calculation of weighted kappa scores. The ‘doubtful’ test–retest reliability resulted from the use of non-weighted kappa scores. Only one of the 19 convergent validity analyses was found to be not of ‘very good’ standard. The issue identified for the ‘doubtful’ judgement in convergent validity was non-reporting of the thresholds used to quantify the strength of the correlation. In terms of known-group validity, of the 17 analyses, nine were ‘very good’ and the other eight were ‘adequate’. The most common issues around the ‘adequate’ judgements were non-reporting of a prior hypotheses or issues with how the groups were described. No issues were identified among the responsiveness analyses. Last, the sole issue identified in the ceiling effects analysis was non-reporting of a threshold. The risk of bias assessment is available in Appendix 7 of the ESM.

Discussion

This review gathered the current psychometric evidence around the performance and suitability of GPBMs for use in the informal carer population. The review found that evidence in this area is still limited, with no evidence available for the AQoL instruments, the PROPr and the 15D. Available evidence suggests that GPBMs have acceptable validity for use in informal carers, in particular, the EQ-5D tools have the most available evidence across a wide range of care recipient and informal carer population groups. Despite limited evidence for the EQ-HWB-9 and SF-6Dv1, existing evidence supports the performance of these measures in informal carers. Overall, the evidence suggests that there are some advantages of the different instruments in different contexts, but more evidence is needed in certain areas to support where they are best used. There was inconsistency in the features and the type of psychometric analysis used across studies (e.g. different groups used in known-groups analysis, measures used in convergent validity tests, criteria in responsiveness tests, interval of responses in the test–retest reliability analysis). However, a positive observation was that, in the risk of bias assessment, the studies were of reasonable quality, with the majority of psychometric property analyses scoring ‘very good’. Most of the issues identified point to inadequate or incomplete reporting of relevant methodological inputs. Future studies can be improved by using the COSMIN risk of bias tools to guide reporting of methods and results [ 33 ].

Addressing the inconsistencies in the analyses identified across studies warrants future head-to-head comparison studies to allow the selection of the optimal measure. To summarise the current head-to-head comparison evidence, the EQ-5D-3L had a ceiling effect and was less sensitive than the EQ-5D-5L in known-group validity [ 54 ], and performed less than the SF-6Dv1 in convergent validity, known-group validity and ceiling effects [ 48 , 53 ]. The SF-6Dv1 and HUI3 performed better than the QWB-SA in terms of convergent and known-group validity [ 52 ]; however, the SF-6Dv1 was not favoured when compared with the EQ-5D-5L in the known-group and responsiveness analysis [ 47 ]. There were no studies identified examining content or face validity of GPBMs among informal carers. Moreover, one study indicated that the EQ-5D-3L domains have no overlap with carer-specific measures indicating that the EQ-5D-3L may lack content validity [ 42 ]. This means that carer-specific preference-based measures seem to measure distinct domains not captured by the EQ-5D-3L, indicating the EQ-5D-3L captures the health impacts of caring, but is not capturing the full range of spillover effects. Nonetheless, available evidence still supports the validity of GPBMs in informal carers through moderate correlations with carer-specific preference-based measures. Our findings show that the EQ-5D-3L [ 48 , 49 ], EQ-5D-5L [ 41 ], EQ-HWB-9A [ 50 ], HUI3 [ 49 , 52 ], SF-6Dv1 [ 48 , 52 ] and QWB-SA [ 52 ] have moderate correlations with the CarerQoL-7D. Furthermore, the EQ-5D-5L [ 41 ] and EQ-HWB-9 [ 50 ] also showed moderate correlations with the ASCOT-Carer. Of all GPBMs, only the EQ-HWB-9 showed a moderate correlation with the CES [ 41 ]. This shows that GPBMs may still capture some caring experience despite potentially having little overlap with the carer-specific items, as indicated by the correlation analysis results. Finally, there is evidence that the GPBMs, particularly the EQ-5D-3L, EQ-5D-5L and EQ-HWB-9, are reliable based on test–retest analysis evidence.

In a previous review, it was found that only a limited number of technology appraisals and highly specialised technology submissions to NICE included informal carer quality-adjusted life-years. To incorporate carers’ HRQoL, many submissions modelled carer HRQoL as a function of patients’ disease status, health state, treatment or death [ 55 ]. However, our review points to mixed evidence around the sensitivity of the GPBMs in discriminating carer HRQoL based on care recipient disease severity. For example, the EQ-5D-3L and SF-6Dv1 had mixed evidence around known-group validity considering various severity of disease categorical groups [ 48 ]. The EQ-5D-5L also had evidence of non-discrimination in terms of dementia severity [ 46 ]. Moreover, when modelling informal HRQoL carer effects throughout the course of a disease or condition, evidence around the GPBMs responsiveness to changes in care recipient disease severity should also be considered. Our review found limited responsiveness evidence, with only EQ-5D-5L [ 41 , 47 ] and EQ-HWB-9 [ 36 ] showing supportive evidence, while the SF-6Dv1 had no supporting evidence. Hence, simply including HRQoL in models as functions of disease status and health state might not be appropriate given the current state of psychometric evidence, as the inclusion of informal carer HRQoL in economic evaluations may impact value-for-money results as found in the study by Krol et al. [ 10 , 56 ]. To address this issue, GPBMs should be studied further in terms of responsiveness to change based on longitudinal studies.

Overall, the findings of this review support the acceptability of using GPBMs in measuring HRQoL of informal carers in economic evaluations. This supports the recommendation by the NICE ‘Task and Finish’ group as well as the SHEER task force for reporting informal carer HRQoL in health technology assessment submissions when considering carer spillover effects in a cost-utility analysis [ 2 , 9 ]. The evidence around other GPBMs also provides support for other jurisdictions, such as the Pharmaceutical Benefits Advisory Committee in Australia, which does not have any explicit requirements around which GPBM to use [ 57 ]. However, the results from the exploratory factor analysis, which showed a limited overlap between the EQ-5D-3L and carer-specific measures ASCOT-Carer and CES, must be considered [ 42 ]. Given this finding, simply relying on GPBMs presents decision makers a limited snapshot of the broader spillover effect from caregiving. However, this observation is based on a single study. Further studies must still be conducted to confirm the consistency of the exploratory factor analysis results, and through face and content validity studies with informal carers. The mixed results around the discriminatory ability of GPBMs around carer situation and burden also question the ability to capture broader spillover effects [ 41 , 46 , 47 , 50 , 53 ]. To better understand these limitations in measuring broader spillover effects, additional studies, similar to McLoughlin et al. [ 41 ], may need to be conducted that directly compared a GPBM’s ability with carer-specific preference-based measures. Given these limitations in the current literature, a possible solution is to consider inclusion of carer-specific measures as another outcome measure for exploration in a supplementary analysis. It is noteworthy that out of the GPBMs included in this review, the EQ-HWB-9 is the only measure specifically targeting informal carers in addition to patients and social care users [ 58 , 59 ]; however, the evidence we found is still limited. The evidence supporting the EQ-HWB-9 should also be considered in the context that it is based on the UK pilot value set and could potentially change once newer country-specific value sets are available. Hence, existing analyses should be revisited once value sets are made available. In addition, given that the EQ-HWB-9 is based on an experimental version of the instrument, there may be possible changes made on the final instrument.

Study Strengths and Limitations

A strength of this study was that the article screening process (i.e. screening, data extraction, analysis and quality assessment) followed PRISMA guidelines, which was performed by two reviewers independently, with disagreements resolved by a third reviewer. The literature search was supplemented with Scopus forward and backward citation searches, which provide some confidence that the identified literature was current and complete. Using the COSMIN risk of bias checklist also provided an overview of the quality of the analyses conducted by the identified studies. However, there were also limitations that need to be acknowledged. Articles may have been inadvertently not identified owing to several factors such as using limited databases and exclusion of non-English studies. The limited number of head-to-head studies, variety of measures used for convergent validity, groups for known-group validity, intervals for test–retest reliability and criteria for responsiveness also prevented a selection of the most appropriate GPBMs.

Conclusions

Despite a growing number of studies examining the psychometric performance of GPBMs among informal carers, the evidence in each psychometric property for all GPBMs is still limited. The EQ-5D tools have the most psychometric evidence in informal carers. Current findings indicate that existing GPBMs are valid tools to use in informal carers. Further research is still needed to explore the psychometric performance in carers of people with other conditions not explored in studies to date (such as in mental health) and other countries that utilise cost-effectiveness evidence in resource allocation decisions, but lack psychometric evidence. Ideally, future studies should preferably be head-to-head comparison studies examining the performance of GPBMs using longitudinal study designs. In addition, further research is needed to assess how well GPBMs capture caring-specific spillover effects and whether relying solely on GPBMs would be sufficient.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (XLSX 97 KB)

Supplementary file2 (DOCX 26 KB)

Supplementary file3 (DOCX 32 KB)

Declarations

Funding

Open Access funding enabled and organized by CAUL and its Member Institutions. Jan Faller and this work is supported by an Australian Government Research Training Program (RTP) Scholarship and the EuroQol Foundation (1481-PHD) through a grant awarded to Lidia Engel.

Conflicts of Interest

Jan Faller, Valeriia Sokolova, Yared Belete Belay, Brendan Mulhern and Lidia Engel have received previous or current funding from the EuroQol Research Foundation, who is the developer and copyright holder of the EQ-5D and EQ-HWB-9 instruments included in this study. Lidia Engel and Brendan Mulhern are members of the EuroQol Group, developers of the EQ-5D and the EQ-HWB-9. Gang Chen and Cathrine Mihalopoulos have no conflicts of interest that are directly relevant to the content of this article. Views expressed in this paper are those of the authors and are not necessarily those of the EuroQol Research Foundation.

Review Registration

The current review was registered with the International Prospective Register of Systematic Reviews (PROSPERO) under registration number CRD42023434651. The protocol was amended broadening the scope of the review to include other GPBMs.

Consent to Participate

Not applicable.

Consent for Publication

Not applicable.

Availability of Data and Material

All data and materials are available within the article and supplementary resources. Further information may be available by request to the corresponding author.

Code Availability

Not applicable.

Authors’ Contributions

Jan Faller and Lidia Engel designed the study and wrote the protocol. Lidia Engel acquired funding. Jan Faller performed the literature search. Jan Faller, Valeriia Sokolova, Yared Belete Belay and Lidia Engel contributed to article screening, extraction and quality assessment. All authors contributed to the interpretation of the data. Jan Faller and Lidia Engel produced the first draft of the manuscript. All authors read, edited and approved the final version of the manuscript.

References

1. Sanders GD, Neumann PJ, Basu A, Brock DW, Feeny D, Krahn M, et al. Recommendations for conduct, methodological practices, and reporting of cost-effectiveness analyses: Second Panel on Cost-Effectiveness in Health and Medicine. JAMA. 2016;316(10):1093–103.

2. Henry E, Al-Janabi H, Brouwer W, Cullinan J, Engel L, Griffin S, et al. Recommendations for emerging good practice and future research in relation to family and caregiver health spillovers in health economic evaluations: a report of the SHEER Task Force. Pharmacoeconomics. 2024;42(3):343–62.

3. Neumann PJ, Garrison LP, Willke RJ. The history and future of the “ISPOR Value Flower”: addressing limitations of conventional cost-effectiveness analysis. Value Health. 2022;25(4):558–65.

4. Basu A, Meltzer D. Implications of spillover effects within the family for medical cost-effectiveness analysis. J Health Econ. 2005;24(4):751–73.

5. Davidson T, Levin L-Å. Is the societal approach wide enough to include relatives? Appl Health Econ Health Policy. 2010;8(1):25–35.

6. Rand SE, Malley JN, Netten AP, Forder JE. Factor structure and construct validity of the Adult Social Care Outcomes Toolkit for Carers (ASCOT-Carer). Qual Life Res. 2015;24(11):2601–14.

7. Brouwer WB, van Exel NJ, van Gorp B, Redekop WK. The CarerQol instrument: a new instrument to measure care-related quality of life of informal caregivers for use in economic evaluations. Qual Life Res. 2006;15(6):1005–21.

8. Al-Janabi H, Flynn TN, Coast J. Estimation of a preference-based carer experience scale. Med Decis Making. 2011;31(3):458–68.

9. National Institute for Health and Care Excellence (NICE). NICE health technology evaluations: the manual. 2022. https://www.nice.org.uk/process/pmg36/chapter/introduction-to-health-technology-evaluation . Accessed 17 May 2025.

10. Dawoud D, Lamb A, Moore A, Bregman C, Rupniewska E, Paling T, et al. Capturing what matters: updating NICE methods guidance on measuring and valuing health. Qual Life Res. 2022;31(7):2167–73.

11. Maxwell A, Özmen M, Iezzi A, Richardson J. Deriving population norms for the AQoL-6D and AQoL-8D multi-attribute utility instruments from web-based data. Qual Life Res. 2016;25(12):3209–19.

12. Misajon R, Hawthorne G, Richardson J, Barton J, Peacock S, Iezzi A, et al. Vision and quality of life: the development of a utility measure. Investig Ophthalmol Vis Sci. 2005;46(11):4007–15.

13. Richardson J, Hawthorne G. The Australian Quality of Life (AQOL) instrument: psychometric properties of the descriptive system and initial validation. Aust Stud Health Serv Admin. 1998;85:343–6.

14. Richardson J, Iezzi A, Khan MA, Maxwell A. Validity and reliability of the Assessment of Quality of Life (AQoL)-8D multi-attribute utility instrument. Patient. 2014;7(1):85–96.

15. Richardson J, Iezzi A, Peacock S, Sinha K, Khan M, Misajon R, et al. Utility weights for the vision-related Assessment of Quality of Life (AQoL)-7D instrument. Ophthalmol Epidemiol. 2012;19(3):172–82.

16. Richardson JRJ, Peacock SJ, Hawthorne G, Iezzi A, Elsworth G, Day NA. Construction of the descriptive system for the assessment of quality of life AQoL-6D utility instrument. Health Qual Life Outcomes. 2012;10(1):38.

17. Devlin N, Roudijk B, Ludwig K. Value sets for EQ-5D-5L: a compendium, comparative review & user guide. 1st ed. Cham: Springer; 2022.

18. Szende A, Devlin N, Oppe M. EQ-5D value sets: inventory, comparative review and user guide. 1. Aufl. ed. Dordrecht: Springer; 2007.

19. Mukuria C, Peasgood T, McDool E, Norman R, Rowen D, Brazier J. Valuing the EQ Health and Wellbeing Short using time trade-off and a discrete choice experiment: a feasibility study. Value Health. 2023;26(7):1073–84.

20. Horsman J, Furlong W, Feeny D, Torrance G. The Health Utilities Index (HUI ® ): concepts, measurement properties and applications. Health Qual Life Outcomes. 2003;1(1):54.

21. Dewitt B, Feeny D, Fischhoff B, Cella D, Hays RD, Hess R, et al. Estimation of a preference-based summary score for the patient-reported outcomes measurement information system: the PROMIS( ® )-Preference (PROPr) scoring system. Med Decis Making. 2018;38(6):683–98.

22. Brazier J, Roberts J, Deverill M. The estimation of a preference-based measure of health from the SF-36. J Health Econ. 2002;21(2):271–92.

23. Brazier JE, Mulhern BJ, Bjorner JB, Gandek B, Rowen D, Alonso J, et al. Developing a new version of the SF-6D health state classification system from the SF-36v2: SF-6Dv2. Med Care. 2020;58(6):557–65.

24. Brazier JE, Roberts J. The estimation of a preference-based measure of health From the SF-12. Med Care. 2004;42(9):851–9.

25. Kaplan RM, Sieber WJ, Ganiats TG. The quality of well-being scale: comparison of the interviewer-administered version with a self-administered questionnaire. Psychol Health. 1997;12(6):783–91.

26. Sintonen H. The 15D instrument of health-related quality of life: properties and applications. Ann Med. 2001;33(5):328–36.

27. Covidence systematic review software. Melbourne (VIC): Veritas Health Innovation.

28. Mokkink LB, Terwee CB, Patrick DL, Alonso J, Stratford PW, Knol DL, et al. The COSMIN study reached international consensus on taxonomy, terminology, and definitions of measurement properties for health-related patient-reported outcomes. J Clin Epidemiol. 2010;63(7):737–45.

29. Akoglu H. User’s guide to correlation coefficients. Turk J Emerg Med. 2018;18(3):91–3.

30. Terwee CB, Bot SD, de Boer MR, van der Windt DA, Knol DL, Dekker J, et al. Quality criteria were proposed for measurement properties of health status questionnaires. J Clin Epidemiol. 2007;60(1):34–42.

31. Cicchetti DV. Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychol Assess. 1994;6(4):284–90.

32. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33(1):159–74.

33. Mokkink LB, de Vet HCW, Prinsen CAC, Patrick DL, Alonso J, Bouter LM, et al. COSMIN risk of bias checklist for systematic reviews of patient-reported outcome measures. Qual Life Res. 2018;27(5):1171–9.

34. Choi E, Seo HJ, Choo IH, Kim SM, Park JM, Choi YM. Health-related quality of life instrument with 8 items to measure health-related quality of life among family caregivers of people with dementia: a pilot validation study. Public Health Nurs. 2024. 10.1002/alz.083961.

35. Wagner CD, Bigatti SM. A comparison of quality of life measures in husbands of women with breast cancer. Appl Res Qual Life. 2016;11(3):955–69.

36. Bailey C, Dalziel K, Constable L, Devlin NJ, Hiscock H, Skouteris H, et al. The performance of the EQ-HWB-9 as a measure of quality-of-life of caregivers in families that have experienced adverse events. Eur J Health Econ. 2025;26(1):7–21.

37. Bailey C, Dalziel K, Jones R, Hiscock H, Devlin NJ, Peasgood T, et al. The validity of the EuroQol Health and Wellbeing Short Version (EQ-HWB-9) instrument in parents of children with and without health conditions. Pharmacoeconomics. 2024;42:163–72.

38. Bell CM, Araki SS, Neumann PJ. The association between caregiver burden and caregiver health-related quality of life in Alzheimer disease. Alzheimer Dis Assoc Disord. 2001;15(3):129–36.

39. Khanna R, Jariwala K, Bentley JP. Psychometric properties of the EuroQol Five Dimensional Questionnaire (EQ-5D-3L) in caregivers of autistic children. Qual Life Res. 2013;22(10):2909–20.

40. Li L, Liu C, Cai X, Yu H, Zeng X, Sui M, et al. Validity and reliability of the EQ-5D-5 L in family caregivers of leukemia patients. BMC Cancer. 2019;19(1):522.

41. McLoughlin C, Goranitis I, Al-Janabi H. Validity and responsiveness of preference-based quality-of-life measures in informal carers: a comparison of 5 measures across 4 conditions. Value Health. 2020;23(6):782–90.

42. Rand S, Malley J, Vadean F, Forder J. Measuring the outcomes of long-term care for unpaid carers: comparing the ASCOT-Carer, Carer Experience Scale and EQ-5D-3 L. Health Qual Life Outcomes. 2019;17(1):184.

43. Reed C, Barrett A, Lebrec J, Dodel R, Jones RW, Vellas B, et al. How useful is the EQ-5D in assessing the impact of caring for people with Alzheimer’s disease? Health Qual Life Outcomes. 2017;15(1):16.

44. Vatter S, McDonald KR, Stanmore E, McCormick SA, Clare L, Leroi I. A brief psychometric and clinimetric evaluation of self-report burden and mental health measures completed by care partners of people with Parkinson’s-related dementia. Int Psychogeriatr. 2020;32(7):875–80.

45. Sun LF, Huang S, Li YF, Yang ZQ, Yang XJ, Zou JY, et al. Health-related quality of life measured by EQ-5D-3L for the spouses of breast cancer patients. Front Oncol. 2022;12: 983704.

46. Sokolova V, Faller J, Asfia S, Engel L. Assessing the psychometric performance of the EQ-5D-5L among informal caregivers of people with dementia. Qual Life Res. 2024;33(10):2693–704.

47. Bhadhuri A, Jowett S, Jolly K, Al-Janabi H. A comparison of the validity and responsiveness of the EQ-5D-5L and SF-6D for measuring health spillovers: a study of the family impact of meningitis. Med Decis Making. 2017;37(8):882–93.

48. Brown CC, Tilford JM, Payakachat N, Williams DK, Kuhlthau KA, Pyne JM, et al. Measuring health spillover effects in caregivers of hildren with autism spectrum disorder: a comparison of the EQ-5D-3L and SF-6D. Pharmacoeconomics. 2019;37(4):609–20.

49. Faraji H, Soleymani F, Yaseri M, Sahraian MA, Abdollahiasl A, Meftah A, et al. Choosing the best instrument for measuring health spillover effect in caregivers of patients with multiple sclerosis. Value Health Reg Issues. 2024;39:49–56.

50. Kuharic M, Mulhern B, Sharp LK, Turpin RS, Pickard AS. Comparison of EQ Health and Well-Being Long and Short with other preference-based measures among United States informal caregivers. Value Health. 2024;27:967–77.

51. Kuharić M, Pickard AS, Mukuria C, Finch AP. The measurement properties of the EQ Health and Well-being and EQ Health and Well-Being Short Form in Italian population: a comparative study with EQ-5D-5L. Value Health. 2024;27:955–66.

52. Payakachat N, Tilford JM, Brouwer WB, van Exel NJ, Grosse SD. Measuring health and well-being effects in family caregivers of children with craniofacial malformations. Qual Life Res. 2011;20(9):1487–95.

53. Sun CY, Liu Y, Zhou LR, Wang MS, Zhao XM, Huang WD, et al. Comparison of EuroQol-5D-3L and Short Form-6D utility scores in family caregivers of colorectal cancer patients: a cross-sectional survey in China. Front Public Health. 2021. 10.3389/fpubh.2021.742332.

54. Zhang H, Liu B, Gu W, Cao Y, Xu J, Tan RL, et al. A head-to-head comparison of the measurement properties of EQ-5D-3L and EQ-5D-5L in Chinese family caregivers of cancer patients. Support Care Cancer. 2023;32(1):14.

55. Pennington BM. Inclusion of carer health-related quality of life in National Institute for Health and Care Excellence appraisals. Value Health. 2020;23(10):1349–57.

56. Krol M, Papenburg J, van Exel J. Does including informal care in economic evaluations matter? A systematic review of inclusion and impact of informal care in cost-effectiveness studies. Pharmacoeconomics. 2015;33(2):123–35.

57. Pharmacuical Benefits Advisory Committee (PBAC). The pharmaceutical benefits advisory committee guidelines. 2016. https://pbac.pbs.gov.au/information/about-the-guidelines.html . Accessed 17 May 2025.

58. Brazier J, Peasgood T, Mukuria C, Marten O, Kreimeier S, Luo N, et al. The EQ-HWB: overview of the development of a measure of health and wellbeing and key results. Value Health. 2022;25(4):482–91.

59. Mukuria C, Connell J, Carlton J, Peasgood T, Scope A, Clowes M, et al. Qualitative review on domains of quality of life important for patients, social care users, and informal carers to inform the development of the EQ-HWB. Value Health. 2022;25(4):492–511.

Associated Data

Supplementary Materials

Supplementary file1 (XLSX 97 KB)

Supplementary file2 (DOCX 26 KB)

Supplementary file3 (DOCX 32 KB)