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

Unpaid Caring and Health-Related Quality of Life: Longitudinal Analysis of Understanding Society (the UK Household Longitudinal Survey)

OBJECTIVES

Decision models for economic evaluation are increasingly including health-related quality of life (HRQoL) for informal/unpaid carers, but these estimates often come from poor quality data and typically rely on cross-sectional analysis. We aimed to identify within-person effects using longitudinal analysis of 13 waves of Understanding Society (the UK Household Longitudinal Survey).

METHODS

We analyzed data for coresident carer and care-recipient dyads, where the carer reported "looking after or giving special help to" the care recipient in any of the 13 waves. We used fixed-effects models to study the effects of caring for the care recipient (the "caregiving" effect) using volume of care (hours per week) and continuous duration of caregiving (years) and caring about the care recipient (the "family" effect) using the care recipient's HRQoL on the carer's HRQoL. HRQoL was measured using the Short Form 6 Dimension, calculated from the Short Form 12.

RESULTS

We found consistent evidence for the family effect: improving care recipient's HRQoL by 0.1 would improve carer's HRQoL by approximately 0.012. We also consistently found evidence of a small but statistically significant decrement to carer's HRQoL for each additional year of caring. These findings were robust to scenario analyses. Evidence for the relationship between volume of care and carer's HRQoL was less clear.

CONCLUSIONS

We propose that our estimates can be used to populate economic models to predict changes in carers' HRQoL over time and allow disutilities to be estimated separately for the family and caregiving effect.

How Does Bereavement Affect the Health-Related Quality of Life of Household Members Who Do and Do Not Provide Unpaid Care? Difference-in-Differences Analyses Using the UK Household Longitudinal Survey

BACKGROUND

Guidelines for modelling in economic evaluation recommend that it may be necessary to consider costs and outcomes until all modelled patients have died. Some guidelines also recommend that carers' health-related quality of life (HRQoL) outcomes should be included. However, it is unclear whether economic evaluations should continue to include carers' HRQoL after patients have died, and whether there is any evidence to support an additional bereavement effect for carers.

METHODS

We used the UK Household Longitudinal Study waves 1-12. We used Difference-in-Differences to estimate the short- and long-term bereavement effects on the SF-6D for people who reported that they did and did not provide care to a household member who then died. We assumed parallel trends conditional on age, sex, long-term health conditions, education, and household income.

RESULTS

Carers and non-carers experienced a significant loss in HRQoL in the year immediately following bereavement. Carers potentially experienced a loss in HRQoL in the year before bereavement, whereas the bereavement effect may have lasted longer for non-carers. For both groups, HRQoL became comparable to the non-bereaved population around 3 years after bereavement.

CONCLUSIONS

Bereavement has a statistically significant negative impact on HRQoL in the short-term, for both carers and non-carers. However, the effect size is small and is not sustained, suggesting that including bereavement in economic evaluation would make little difference to results.

Maternal Mental Health Spillovers From Child Illness and Disability: A Dynamic Panel Analysis

OBJECTIVES

There is growing recognition of the consequences of a person's health and illness experience for the health and wider welfare of those close to them. However, estimation of these health spillovers is challenging. This study adopts a longitudinal approach to examine maternal mental health spillovers associated with various forms of child illness and disability.

METHODS

Dynamic panel models are used in estimating maternal mental health spillovers related to 7 subcategories of chronic child illness and disability. In particular, we use longitudinal data from the Growing Up in Ireland study and a system generalized method of moments approach. We also consider heterogeneity in these spillovers by the severity of the child's illness/disability and by household deprivation.

RESULTS

We find that a child's experience of chronic nervous system conditions and chronic mental and behavioral disorders are associated with 10.8 and 5.1 percentage point increases in the probability of maternal depression, respectively. Similar associations were not observed for other health conditions. Spillover magnitude is also found to be strongly related to illness/disability severity. Finally, subsample analyses reveal a larger association between severe child illness and maternal depression among deprived households.

CONCLUSIONS

This analysis, in observing health spillovers related to certain disease categories but not others, draws further attention to their context specificity. Our findings also further corroborate calls for inclusion of caregiver and family member outcomes in the economic evaluation of child health services and support consensus guidelines for collection of these outcomes alongside patient outcomes in clinical trials.

Including carer health-related quality of life in NICE health technology assessments in the United Kingdom

The impact of health technologies may extend beyond the patient and affect the health of people in their network, like their informal carers. The National Institute for Health and Care Excellence (NICE) methods guide explicitly allows the inclusion of health-related quality of life (HRQoL) effects on carers in economic evaluations when these effects are substantial, but the proportion of NICE appraisals that includes carer HRQoL remains small. This paper discusses when inclusion of carer HRQoL is justified, how inclusion can be substantiated, and how carer HRQoL can be measured and included in health economic models. Inclusion of HRQoL in economic evaluations can best be substantiated by data collected in (carers for) patients eligible for receiving the intervention. To facilitate combining patient and carer utilities on the benefit side of economic evaluations, using EQ-5D to measure impacts on carers seems the most successful strategy in the UK context. Alternatives to primary data collection of EQ-5D include vignette studies, using existing values, and mapping algorithms. Carer HRQoL was most often incorporated in economic models in NICE appraisals by employing (dis)utilities as a function of the patient's health state or disease severity. For consistency and comparability, economic evaluations including carer HRQoL should present analyses with and without carer HRQoL.

Development of stakeholder-informed recommendations for inclusion of family spillover effects in health technology assessment

BACKGROUND

The impacts of disease and treatment on a patient's family members and informal caregivers are known as "family spillover effects." Although many formal value frameworks call for the consideration of these effects, they are often not included in health technology assessments (HTAs) and cost-effectiveness analyses (CEAs). A formal evaluation of stakeholder perspectives may help address the disconnect for inclusion of family spillover effects observed in practice.

OBJECTIVE

To develop stakeholder-driven recommendations for the measurement and use of family spillover effects in the United States and to identify research opportunities.

METHODS

We first conducted a targeted literature review of US-based CEAs and HTA reports from the past 10 years to assess the current use of family spillover effects. We then used a purposeful sampling technique to conduct 25 qualitative interviews with outcomes researchers, patient advocates, health economists, and health policy and payer experts to gather perspectives on when and how family spillover effects should be considered in HTA processes. We conducted a thematic analysis of the interview transcripts to identify key themes and develop preliminary recommendations. Finally, we conducted an online workshop with 8 stakeholders to discuss, rate, and refine preliminary recommendations to develop final recommendations.

RESULTS

A key theme identified in the stakeholder interviews was the role that data availability, analyst preferences, and prior precedence play in limiting the inclusion of spillover effects in HTAs. Additional themes included support for the inclusion of both qualitative and quantitative spillover effects and the need to capture broad and diverse impacts across populations. We developed 15 recommendations from the consensus building workshop addressing measurement, CEA modeling, and HTA processes. Key recommendations included (1) a transparent process for deciding when family spillover effects should be included, (2) measurement of direct and indirect costs with priority based on the magnitude of impact, (3) the use of validated measures, (4) the use of proxy information and expert elicitation when quality data are unavailable, and (5) the use of a modified impact inventory table for transparency of included effects. Research opportunities included patient involvement in family spillover effect research and HTAs, mapping algorithms and non-preference-based caregiver measures to generate utilities, and consensus best practices for modeling.

CONCLUSIONS

The inconsistent inclusion of family spillover effects in HTAs and CEAs remains a persistent challenge. The stakeholder-driven recommendations and research opportunities identified in this study may help improve the transparency, measurement, and use of family spillover effects in assessing the clinical and economic value of novel medical technologies.

Socioeconomic burden of cystic fibrosis in Canada

BACKGROUND

Cost of illness studies are important tools to summarise the burden of disease for individuals, the healthcare system and society. The lack of standardised methods for reporting costs for cystic fibrosis (CF) makes it difficult to quantify the total socioeconomic burden. In this study, we aimed to comprehensively report the socioeconomic burden of CF in Canada.

METHODS

The total cost of CF in Canada was calculated by triangulating information from three sources (Canadian CF Registry, customised Burden of Disease survey and publicly available information). A prevalence-based, bottom-up, human capital approach was applied, and costs were categorised into four perspectives (ie, healthcare system, individual/caregiver, variable (ie, medicines) and society) and three domains (ie, direct, indirect and intangible). All costs were converted into 2021 Canadian dollars (CAD) and adjusted for inflation. The cost of cystic fibrosis transmembrane conductance regulator (CFTR) modulator therapies was excluded.

RESULTS

The total socioeconomic burden of CF in Canada in 2021 across the four perspectives was $C414 million. Direct costs accounted for two-thirds of the total costs, with medications comprising half of all direct costs. Out-of-pocket costs to individuals and caregivers represented 18.7% of all direct costs. Indirect costs representing absenteeism accounted for one-third of the total cost.

CONCLUSION

This comprehensive cost of illness study for CF represents a community-oriented approach describing the socioeconomic burden of living with CF and serves as a benchmark for future studies.

A Systematic Review of Methods and Practice for Integrating Maternal, Fetal, and Child Health Outcomes, and Family Spillover Effects into Cost-Utility Analyses

BACKGROUND

Maternal-perinatal interventions delivered during pregnancy or childbirth have unique characteristics that impact the health-related quality of life (HRQoL) of the mother, fetus, and newborn child. However, maternal-perinatal cost-utility analyses (CUAs) often only consider either maternal or child health outcomes. Challenges include, but are not limited to, measuring fetal, newborn, and infant health outcomes, and assessing their impact on maternal HRQoL. It is also important to recognize the impact of maternal-perinatal health on family members' HRQoL (i.e., family spillover effects) and to incorporate these effects in maternal-perinatal CUAs.

OBJECTIVE

The aim was to systematically review the methods used to include health outcomes of pregnant women, fetuses, and children and to incorporate family spillover effects in maternal-perinatal CUAs.

METHODS

A literature search was conducted in Medline, Embase, EconLit, Cochrane Collection, Cumulative Index to Nursing and Allied Health Literature (CINAHL), International Network of Agencies for Health Technology Assessment (INAHTA), and the Pediatric Economic Database Evaluation (PEDE) databases from inception to 2020 to identify maternal-perinatal CUAs that included health outcomes for pregnant women, fetuses, and/or children. The search was updated to December 2022 using PEDE. Data describing how the health outcomes of mothers, fetuses, and children were measured, incorporated, and reported along with the data on family spillover effects were extracted.

RESULTS

Out of 174 maternal-perinatal CUAs identified, 62 considered the health outcomes of pregnant women, and children. Among the 54 quality-adjusted life year (QALY)-based CUAs, 12 included fetal health outcomes, the impact of fetal loss on mothers' HRQoL, and the impact of neonatal demise on mothers' HRQoL. Four studies considered fetal health outcomes and the effects of fetal loss on mothers' HRQoL. One study included fetal health outcomes and the impact of neonatal demise on maternal HRQoL. Furthermore, six studies considered the impact of neonatal demise on maternal HRQoL, while four included fetal health outcomes. One study included the impact of fetal loss on maternal HRQoL. The remaining 26 only included the health outcomes of pregnant women and children. Among the eight disability-adjusted life year (DALY)-based CUAs, two measured fetal health outcomes. Out of 174 studies, only one study included family spillover effects. The most common measurement approach was to measure the health outcomes of pregnant women and children separately. Various approaches were used to assess fetal losses in terms of QALYs or DALYs and their impact on HRQoL of mothers. The most common integration approach was to sum the QALYs or DALYs for pregnant women and children. Most studies reported combined QALYs and incremental QALYs, or DALYs and incremental DALYs, at the family level for pregnant women and children.

CONCLUSIONS

Approximately one-third of maternal-perinatal CUAs included the health outcomes of pregnant women, fetuses, and/or children. Future CUAs of maternal-perinatal interventions, conducted from a societal perspective, should aim to incorporate health outcomes for mothers, fetuses, and children when appropriate. The various approaches used within these CUAs highlight the need for standardized measurement and integration methods, potentially leading to rigorous and standardized inclusion practices, providing higher-quality evidence to better inform decision-makers about the costs and benefits of maternal-perinatal interventions. Health Technology Assessment agencies may consider providing guidance for interventions affecting future lives in future updates.

Examining how well economic evaluations capture the value of mental health

Health economics evidence informs health-care decision making, but the field has historically paid insufficient attention to mental health. Economic evaluations in health should define an appropriate scope for benefits and costs and how to value them. This Health Policy provides an overview of these processes and considers to what extent they capture the value of mental health. We suggest that although current practices are both transparent and justifiable, they have distinct limitations from the perspective of mental health. Most social value judgements, such as the exclusion of interindividual outcomes and intersectoral costs, diminish the value of improving mental health, and this reduction in value might be disproportionate compared with other types of health. Economic analyses might have disadvantaged interventions that improve mental health compared with physical health, but research is required to test the size of such differential effects and any subsequent effect on decision-making systems such as health technology assessment systems. Collaboration between health economics and the mental health sciences is crucial for achieving mental-physical health parity in evaluative frameworks and, ultimately, improving population mental health.