Measuring Mental Health in a Cost-of-Living Crisis: a rapid review

1.1 Who is this review for

This Rapid Review was conducted as part of the ‘Health and Care Research Wales Evidence Centre’ Work Programme. The above question was suggested by the Science, Evidence and Advice (SEA) Division, Welsh Government. The intended audience are stakeholders needing to measure mental health outcomes who are seeking evidence for the appropriate methods and tools which are applicable to the UK or relating to Wales.

1.2 Background and purpose of this review

Since 2021 the UK has experienced a sharp rise in inflation. For most people, wages and welfare payments have not kept up with the rate of inflation, leading to a cost-of-living crisis. Energy, food, fuel, and housing have become less affordable, which leads to adverse short and long-term health outcomes (Roberts et al. 2022). There is evidence from past economic crises, such as the 2008/2009 global financial crisis which is also known as The Great Recession, that economic crises and increases in the cost-of-living have a detrimental effect on mental health across the population and that vulnerable populations are disproportionately affected (Guerra & Eboreime 2021). In response, national and regional public health interventions may be put into place to mitigate against the effects of a cost-of-living crisis. It is important for policymakers to be able to measure the impact of any public health measure on mental health.

This rapid review answers the following questions:

• What methods and tools are available and appropriate for monitoring the impact of the cost-of-living crisis on mental health?

• How is it best to do this for specific, vulnerable populations?

The aims of the review are to 1) identify and appraise available population-level measures for assessing the impact on mental health of any public health response to the cost-of-living crisis and 2) review the appropriateness of the measures for specific, vulnerable populations.

The purpose of this review is not to summarise the findings of included articles (i.e., impact of the economic crises on mental health or outcome of any public health interventions), or to identify how often different methods of measurement are used to evaluate the effect of economic crises on mental health, but to provide a summary of the available evaluation methods, and the comments on study design.

2.1 Overview of the Evidence Base

We searched for published articles that evaluated the impact of economic crises on population mental health, that were longitudinal in design and that used validated tools or other clear methods to measure mental health outcomes. Relevant methods were identified from secondary evidence published from 1970 to 2023 and from primary studies published from 2021 to 2023 that were not included in the reviews. We identified four systematic reviews (Frasquilho et al. 2016, Glonti et al. 2015, Saez et al. 2019, Silva et al. 2018), one scoping review (Guerra & Eboreime 2021) one clinical guideline (Martin-Carrasco et al. 2016) and one rapid review from grey literature (Preece & Bimpson 2019) where the majority of included studies met our inclusion criteria. The reviews included a total of 322 different identifiable studies, of which 52 were included in more than one review (although Martin-Carrasco reported using evidence from a total of 354 studies, but detailed data was only extracted from 69). In addition, we identified nine further peer-reviewed primary studies (Alvarez-Galvez et al. 2021, Aretz 2022, Cherrie et al. 2021, Clair & Baker 2022, Curtis et al. 2021, Kim et al. 2022, Saville 2021, Thomson et al. 2022, Wickham et al. 2020) and two reports from grey literature (Clark & Wenham 2022, Office for National Statistics 2022).

Full details of inclusion and exclusion criteria and the screening of articles can be found in section 5 A detailed summary of the methods used in each study, and an overview of identified mental health measurement tools are available in section 6 (tables 3a-3c and table 4).

2.2 Summary of the Evidence

2.3 Potential biases in study design

All reviews and studies included some discussion on strengths and limitations of study methodology (table 1).

We identified one systematic review (Saez et al. 2019) that had the aim of evaluating biases in studies assessing the effects of the Great Recession on health in Spain. This review included 53 studies and evaluated bias using an adapted tool. Four main biases were identified: problems with evaluation, time bias, failure to adjust for confounders such as seasonal effects, and inconsistencies in defining the date of onset of the Great Recession.

Problems with evaluation existed with a repeated-cross sectional design where the sample surveyed before recession onset did not consist of the same individuals as the sample surveyed after onset, and therefore were subject to confounding. Studies judged at lower risk of bias in this domain attempted to control for confounding by either matching subjects from the samples before and after the crisis; stratifying the samples by common characteristics, often into groups considered most vulnerable, or using sophisticated multilevel modelling analyses. Glonti et al (2015) also recommended other sophisticated analyses such as dynamic modelling or structural equation modelling. Other reviews also point out that it is not possible to infer causality from studies with a repeated cross-sectional design, although a strength of these studies is that they are frequently national surveys that are representative of the population. However, if a longitudinal panel survey with a repeated measures design is used, the evaluation problems do not exist, and there is stronger evidence for causality.

Time bias existed in time series studies where there were a very few data points after crisis onset, or where there was no consideration given to any potential lag between exposure and outcome. Time series studies were mainly those where the health outcome was suicide.

Other reviewers also commented that studies into suicide rates are ecological studies using aggregated data of unknown quality and that causality cannot be inferred.

Other reviewers also discussed failure or inability to adjust for confounding. A limitation of using existing datasets for secondary analysis is that data of relevance to the secondary analysis may not have been collected. For example, it was observed that UK population panel and repeated cross-sectional surveys do not always ask about both mental health and household economics in detail, and not all income- or employment-related factors of interest were included in all survey waves.

Surveys use self-reported data which may not be accurate. However, potential biases in the use of medical records were discussed in several other reviews and studies. Data in hospital and medical records were observed to be of variable quality and dependent on the structure of services and record keeping. Any reorganisation of mental health care services post-onset of an economic crisis can introduce bias by potentially altering who can access services and may disproportionate affect the most vulnerable. The use of prescribing data can introduce bias since the availability of drugs may vary, different prescribers may have different prescribing thresholds which may affect different population sub-groups differently, and the same drug may be given for conditions that are not mental-health related. One review (Silva et al. 2018) commented that people seeking medical care were not necessarily those most impacted by economic crises. Where data from medical records is aggregated, individual behaviour cannot be inferred, and the data cannot capture variation in services between population sub-groups. For example, an economic crisis may increase suicide rates for some population sub-groups but if suicide prevention strategies are provided at the same time, these may benefit different population sub-groups and the net result may be a zero increase. However, it was observed that an advantage of aggregated data was that they reflect the environmental effects of changes in the economy beyond an individual’s personal circumstances.

2.4 Vulnerable groups

All the reviews and grey literature and some primary studies identified population groups whose mental health was most likely to be affected by an economic crisis. All the mental health measurement tools reported in section 2.2 were used for this purpose, except for the WEMWBS. Tables 3a to 3c summarises the vulnerable groups identified in each review and primary study (section 6). All the reviews presented the data narratively and neither reviews nor primary studies commented on the suitability of the methods and tools when identifying vulnerable groups. However, consistently, the most vulnerable groups were identified as those who were living on lower incomes, in financial or housing insecurity or living in more deprived areas prior to the crisis. People who were unemployed or whose employment was precarious were at high risk of worsening mental health problems.

The reviews also examined data stratified by gender and age. It appeared that men’s mental health deteriorated more during economic crises, but women were more likely to experience poorer mental health at all timepoints and were more likely to access mental health services. However, men, particularly men of working age were at higher risk of suicide. Some studies linked changes in mental health in men and women to socio-economic status but not all and the evidence for age-related differences was mixed. Other groups identified as being at risk were people with pre-existing mental health problems, people living in households with dependent children, and migrants. One primary study (Office for National Statistics 2022) also identified adults with a disability or long-term illness as being at risk. No included study or review presented data from people from minority ethnic groups.

It appears that the methods and tools are suitable for use in men and women, across the socio-economic spectrum, living in different housing tenures and with different employment status. However, it is possible that some potentially vulnerable groups are not well represented when using the methods and data sources identified.

3.1 Summary of the findings

This rapid review aimed to answer two questions:

• 1. What methods and tools are available and appropriate for monitoring the impact of the cost-of-living crisis on mental health?

We identified multiple studies that examined the impact of economic crises on mental health over time using national surveys from many different countries using different methods. Data from population-level panel surveys may be most useful. The UK has many ongoing population-level surveys, providing rich data that can be used to explore the health of the population over time. These large surveys incorporate validated mental health tools and questions to determine financial security at household or individual level and may be linked to medical records and CPRD data. Several primary studies published since 2021 used data from UKHLS. Many studies are observational in design, but we also identified two relevant quasi-experimental case-control studies, that used UKHLS data to examine the impact of specific welfare policies in the UK (the “Bedroom tax” and the introduction of Universal Credit). A quasi-experimental study design, using UKHLS data, could also be suitable for measuring the impact of a specific public health initiative that has a clear roll-out date.

• 2. Are methods and mental health measurement tools suitable for specific, vulnerable populations?

All the reviews and most primary studies found people living in financial insecurity were at higher risk of poor or worsening mental health and most examined gender differences. It therefore appears that the methods and tools used were suitable for capturing data from men and women, from different socio-economic groups. Few studies presented data from other groups that are often marginalised such as migrants and people living with disabilities or long-term poor health. We also identified no studies that examined whether people from minority ethnic groups were more at risk of deteriorating mental health during economic crises. It is therefore not clear as to whether the methods identified are suitable for all marginalised populations.

3.2 Strengths and limitations of this Rapid Review

Evidence for this rapid review has been drawn from reviews that included over 300 studies, and from primary studies and grey literature that were of relevance to Wales. We have identified appropriate available population level datasets that include both validated mental health tools and collect socioeconomic outcomes and have identified appropriate study designs and provided information about their strengths and limitations.

However, our rapid review has limitations. All the included reviews summarised data from individual studies narratively and varied in the detail they provided on study designs and strengths and limitations. The identification of vulnerable groups appeared not to be systematic in most reviews and it is possible that data on vulnerable groups was not captured. There is a strong overlap in the literature with studies describing the impact of economic crises on mental health and the impact of any austerity policies that followed the Great Recession on mental health. We did not include austerity as a search term, and it is possible that relevant studies were excluded. However, the purpose of this review was not to examine the evidence that either economic crises or austerity policies affect mental health, but to describe available methods of measuring changes to population mental health in response to public health intervention in Wales. In addition, although the search strategy was comprehensive and included evidence from peer reviewed articles and from grey literature, we had to exclude much of the identified grey literature due to lack of methodological detail and lack of information on data sources and measurement tools. It is possible that there are datasets and tools used by charities and institutions that we did not identify.

3.3 Implications for policy and research

Our rapid review has identified existing methods and tools likely to be suitable for measuring the impact of public health initiatives on mental health in people from different socio-economic groups during a cost-of-living crisis. However, it is unclear as to whether the identified methods and tools adequately capture data from people from minority ethnic groups, who already experience disparities in mental health care (Ahmad et al. 2022). In the short-term, reports and studies using panel surveys such as UKHLS or that use medical records should present data on ethnicity and, where possible, plan to stratify analyses by ethnicity. In the longer-term it may be necessary to develop new methods to better capture data from people from minority ethnic groups.

3.4 Economic considerations^*

• The UK has seen sharp rises in inflation since Winter 2021. The Consumer Price Index including owner occupier housing costs (CPIH) rose from 3% in August 2021 to a peak 9.6% in October 2022. CPIH is currently 7.9% (Office for National Statistics, 2023).

• In the lead up to the cost of living crisis, Wales had the highest proportion of working age adults (21%) and pensioners (18%) in relative income poverty out of the UK nations. 28% of children in Wales were living in relative poverty (Welsh Government, 2023a). Given that over half of all mental health problems start by age 14 (and 75% by age 18) and poverty being a known risk factor for psychological illnesses, there is likely to be a long shadow of mental health continuing into future generations stemming from the cost of living crisis (Department of Health, 2015; Fell & Hewstone, 2015).

• The effects of the cost of living crisis are not felt equally. Poorer households tend to spend a greater proportion of income on items more exposed to inflationary pressures such as energy and food (Senedd Research, 2022). 90% of people in Wales felt their health was negatively affected due to increased heating costs (Royal College of Physicians, 2022). Decisions around the use of energy at home can lead to lingering anxiety that negatively impacts mental health (Marmot, 2020).

• Mental Health problems cost the Welsh economy £4.8 billion per annum. 72% of these incurred costs are attributed to productivity losses of people living with mental health conditions and costs experienced by unpaid informal carers (Mcdaid et al., 2022).

• In a survey of 2,000 Welsh participants covering the period November 2022 to January 2023, 60% of respondents agreed that rising costs of living negatively affected their quality of life (25% strongly agreed). 87% reported ‘worrying’ around the cost of living, with 38% reporting ‘worrying a lot’ (Public Health Wales, 2023).

• People with poor mental health are more likely to experience subsequent reductions in income. The negative impacts on mental health and well-being induced by the cost of living crisis could lead to future financial and health problems. Potentially creating a downward spiral even that can persist even if economic conditions improve (Thomson et al., 2022).

• Practitioners in Wales have voiced concerns over the compounding socioeconomic consequences of the COVID-19 pandemic and the subsequent cost of living crisis affecting the mental health of the Welsh population (Welsh Government, 2023b).