Feasibility study for survey of incomes and assets of adults with social care needs: summary report
Published 7 November 2024
Applies to England
This research was commissioned under the previous government (11 May 2010 to 5 July 2024) and therefore does not reflect the policies of the current government. The views expressed are the authors’ and do not necessarily reflect those of the government.
Background, aims and methodology
This document summarises the findings from a research project exploring the feasibility of collecting data about the care needs, income and assets of people with care and support needs in England.
There is currently a lack of robust data about the income and assets of the population using adult social care services, as well as other information such as demographics and spending on care. The collection of this data can be complex and sensitive with important ethical implications. In December 2021 the Department of Health and Social Care (DHSC) therefore commissioned Ipsos jointly with the Care Policy and Evaluation Centre (CPEC) at the London School of Economics and Political Science (LSE) to conduct a study to explore the feasibility and acceptability of collecting data on the characteristics of people with care needs, including their income and assets. The overall project explored different options for potential data collection and their benefits and drawbacks. From the different options explored throughout this project, DHSC sought to understand the feasibility of answering the following questions from any data:
- how the assets of people with care and support needs are impacted as they move through the care system
- how funding arrangements play out in the real world, and the issues faced by those in different funding scenarios
- what the demographic characteristics of people with care and support needs are
The project was divided into 5 workstreams which were conducted iteratively between December 2021 and October 2023. All the evidence from workstreams 1 to 5 have been brought together in this summary report. Workstream specific reports are also available.
Workstream 1
Workstream 1 consisted of a rapid evidence review and was led by CPEC. It identified and reviewed past studies on the income and assets of people with care needs. It also looked at the following existing large scale national population data sets and bespoke surveys of people receiving care and support and their carers that collect financial data, with the aim of finding out the extent to which these existing surveys could provide the data DHSC needs:
- English Longitudinal Study of Ageing (ELSA)
- Family Resources Survey (FRS)
- Health Survey for England (HSE)
- UK Household Longitudinal Study (UKHLS)
Workstream 2
Workstream 2 involved in-depth interviews with 21 care providers, local authorities and other key informants, followed by a meeting with an expert reference group. This workstream explored different ways that data on the characteristics, income and assets of people with care and support needs are currently recorded or collected and participants’ views on additional data collection activities.
Workstream 3
Workstream 3 consisted of secondary data analysis (phase 1) and an options appraisal (phase 2). For the secondary analysis, CPEC investigated the response rates to questions about the financial circumstances of respondents in 2 major national longitudinal surveys, ELSA and UKHLS. The objective was to identify how well these questions in ELSA and UKHLS capture the corresponding financial information they are designed to capture.
In the options appraisals, Ipsos looked at the different data collection options available. Three groups of options were considered and their benefits and drawbacks identified. Each option was appraised on a range of topics including:
- coverage of the target population
- information that the option would provide and whether this would meet DHSC needs
- impact on people with care needs
- impact on carers and families
- required involvement of organisations
- technical and practical considerations and data analysis and use
Findings from workstreams 1 and 2 and from CPEC’s secondary data analysis fed into the options appraisal. This options appraisal was conducted prior to the 2022 autumn statement announcement that the implementation of charging reform would be delayed.
Workstream 4
Workstream 4 involved 20 in-depth interviews with people with care needs, unpaid carers and people with lasting power of attorney for the financial affairs of a family member with care needs. These interviews explored participants’ views on taking part in a survey asking about the care needs, assets and income of people with care needs. These were followed by 20 cognitive testing interviews, to test potential survey questions. A favourable ethical opinion was sought from a research ethics committee at the Health Research Authority at the start of this workstream. A second expert reference group meeting was convened to discuss the overall project findings and their implications.
Workstream 5
As part of workstream 5, CPEC conducted further secondary analysis of ELSA data to explore how potential high level proxy measures of financial circumstance were related to more detailed financial evidence of the type needed for modelling.
Findings
The research identified a range of options for collecting data on the care needs, assets and incomes of people with care and support needs in England. These options were divided into 3 groups. While conducted iteratively, the 5 workstreams all explored aspects of these options. Below we provide a thematic summary of each group of options explored.
Options group 1: setting up a new survey of people with care needs, administered by local authorities, care providers or a third party on behalf of DHSC
A survey conducted annually could potentially provide up to date and relevant data about the care needs and financial situation of people with care and support needs. However, a number of challenges and considerations would need to be resolved, from the perspective of people with care and support needs and their families, and from the perspective of the organisations that would have to be involved. There would also be costs associated with setting up and running a survey.
There are challenges in collecting financial information through a survey. These challenges include:
- trust and sensitivity about providing financial data
- burden for participants in providing detailed financial information, which those who have had a care and financial assessment with their local authority will have already provided
The interviews with 20 people with care needs, unpaid carers and people with lasting power of attorney for the financial affairs of a family member showed there was considerable reluctance to provide information about income, savings, assets, housing tenure and property value. Greatest concern was expressed about sharing information on savings and assets. These were considered private matters, and participants were concerned about scams, being defrauded, why they were being asked about these, and how the information would be stored and used (to inform government policy and/or make decisions about their care).
Despite providing an information sheet and a consent form ahead of the interview, and explaining these to participants at the start of the interview, there was still some confusion about the aims of the survey among participants. Some participants believed the survey was a local authority assessment or similar, and that taking part and/or answering all the questions would help them access care and support, or was a requirement if they wanted to access care and support. When the aims and use were correctly understood, participants were keen to know what difference taking part would make in terms of informing government policy and making a difference to others in the future, and why this information is not already held by the government in a form which can be used for research.
In-depth and cognitive interviews with people with care needs and carers showed how willingness to provide information varied.
Participants were most willing and able to provide accurate data about care needs, types of care received, attitudes to and experience of care and paying for care, perceived financial status, benefit receipt and (for some) house value or location. Given the challenges involved in collecting detailed information, it would be more realistic to collect simpler proxy financial measures, rather than the more detailed financial data required for modelling. However, this would rely on these measures being acceptable proxies from an analytical perspective. Analysis of ELSA data which includes both detailed, high quality financial measures and proxy measures tested in cognitive interviews showed that proxy demographic and financial characteristics among the older population are statistically significantly associated with benefit unit non-housing assets and with individual income. However, the analyses suggest that the potential proxies examined are not sufficiently robust for modelling purposes. The proxy measures overestimated the incomes (non-housing assets) of those with low incomes (assets). Likewise, they under-estimated the incomes of those with high incomes. This limits the value of proxy financial measures for modelling, even in combination with other options. Nonetheless, measures of attitudes and experience of care and paying for care collected in a survey could be of value in combination with other options for gathering the financial data.
Setting-up a new survey would also require consideration of how data about people lacking mental capacity could be collected. These considerations include the need for family input and the burden this would place on them. Participants thought that it would be acceptable to invite a family member of a person lacking mental capacity to complete the survey instead, as long as the consultee took into account the wishes and feelings of the person with care needs when deciding about their participation and answering questions. The ethical implications of this would need to be more fully considered.
There are advantages and disadvantages to different ways of administering the survey. Conducting the survey through care providers or local authorities (that is, with them sending out the survey invitations on behalf of DHSC) could encourage local ownership and improve response rates, as the target audience would be familiar with the sender and therefore less suspicious. This approach would minimise issues around data sharing as care providers and local authorities would not need to share the name and contact details of people with care needs for the survey to be administered. However, capacity and workload issues among local authorities’ and care providers’ staff would be an important barrier to this approach.
Conducting the survey centrally (that is, with a third party sending out the survey invitations) would greatly reduce the burden on local authorities and care providers but this would involve overcoming issues about sharing contact details for the sample, which might be possible for local authorities but not for care providers. Importantly, while local authorities have the names and contact details of people eligible for local authority funded care, currently they do not have comprehensive records of self-funders - these would need to be accessed through care providers or through other means. Achieving comprehensive coverage of self-funders is crucial as their care needs and financial profile are different from the profile of people whose care is funded by their local authority and this is the group about which DHSC currently knows least.
Options group 2: using existing surveys to collect the required data
ELSA, FRS, HSE and UKHLS provide high quality financial information and detail about care needs and receipt of care. They are commonly used for analysis of social care receipt because they survey nationally representative samples and include a significant number of questions on the need for care and support, the sources of care and support, and in some cases, how the care and support received by individual respondents is paid for. Table 1 summarises the key characteristics of these surveys, in terms of sample size, coverage, sample frame and availability of information on social care, assets and income.
Table 1: characteristics of ELSA, FRS, HSE and UKHLS
ELSA | FRS | HSE | UKHLS | |
---|---|---|---|---|
Type | Longitudinal | Cross-sectional | Cross-sectional | Longitudinal |
Sample size | 8,445 individuals (2016 to 2017) | 19,000 households (2018 to 2019) | 10,250 individuals (2018 to 2019) | 49,685 individuals (2017 to 2019) |
Sample frame | Follow-up from HSE - private households but follows into care homes | Private households | Private households except for care home sample in 2000 | Private households with some proxy information on people in care homes |
Survey frequency | Bi-yearly | Annual | Annual | Annual |
Social care information | 50-plus social care module | All adults but limited | 65-plus social care module | 65-plus social care module |
Income | Detailed | Detailed | Detailed | Detailed |
Assets | Detailed | Only financial, not property | Only financial, not property | Detailed |
Paying for care | Detailed | Not included | Detailed | Detailed |
Additional features | End of life module (introduced in 2012) | Focus on financial resources | No additional features | Large immigrant and ethnic boost sample |
Each of the data sets offers the potential for boost samples, follow-ups or linking with administrative data.
As ELSA and UKHLS collect more comprehensive data on the assets of respondents, (compared to the other 2 surveys), an analysis of missing values on the financial questions asked in these surveys was undertaken, to determine their feasibility as approaches to collecting representative data on the income and assets of individuals receiving social care, whether in their current form or with adaptations. Both of these are longitudinal surveys which means respondents are familiar with them and are unlikely to suspect scams or fraud when they are asked questions on their finances (unlike a new survey as described under option 1).
For UKHLS, the analysis showed that across all the financial variables, up to 12% of respondents refused to provide an answer. However, low outlier values on some financial variables may indicate hidden missing values. For example, 12% of non-missing values reported the last pension credit payment was £0. Similarly, 17% of non-missing values reported the last private pension payment was £10 or less. Rates of refusal to answer financial questions were only slightly higher among individuals aged 65 or over than among other age groups, and the same pattern was observed among people aged 65 or over with one or more activity of daily living limitation.
On ELSA (which only covers people aged 50 and over), response rates to the financial variables varied from 76% to 100%: housing tenure (99.9%) and house value (95%) were the variables with the highest response rates, whereas interest earned from current or savings account (76%) and amount held in the current or savings account (82%) were the variables with the lowest response rates. Of the 13 variables investigated, 8 had a rate of valid response of 90% or higher. When considering rates of missingness due to refusal to answer a question, this was considerably higher for the variable capturing savings in current or saving accounts than for other variables, accounting for 10% of those asked the question. When focusing the analysis on people aged 65 and over only, or people with 1 or more activity of daily living limitation, the overall rate of missingness was slightly higher for many of the financial variables.
These 4 surveys are all limited in their sample frame, as they do not routinely include people in residential care, and no social care information is available for people with care needs under the age of 50 or 65 in 3 of these surveys. Sample sizes, albeit large, are not necessarily large enough for DHSC’s analysis purposes when the analysis is focused on specific groups of interest, such as people of working age with care needs, especially when rates of non-response to financial questions are taken into account. However, participants in these general population samples include those who will be care service users in the future. Compromises would need to be made in terms of coverage, and modelling and imputation would be needed, or the sampling approach would need to involve sample boosts for particular groups of interest. While sample boosts would incur costs and be subject to the agreement of the organisations commissioning these surveys, existing data is available at no extra costs - an advantage compared with the option of a new survey (option 1).
Options group 3: using administrative data to assemble the required information
This option had 3 different approaches.
1. Using existing administrative data held by local authorities
High quality data on finances and care needs is already collected by local authorities during care needs and financial assessments. Using this data would minimise the burden on people with care needs and carers as it would make use of data they have already provided. However, it would only cover people who make contact with their local authority and are eligible for a full assessment. At present this excludes most self-funders (unless the local authority commissions care on their behalf), though this will change when the charging reforms are introduced. Local authorities will collect information on self-funders who choose to meter towards the cap once they have eligible care needs and self-funders who request the option for local authorities to commission care home places for them. Significant data protection issues would still need to be resolved. In-depth consideration would also need to be given to the feasibility of fully anonymising data before it is shared and the level of detail that would be required, and could be provided, without risking identification of those with less common demographic characteristics or assets. Examples include:
- the first half of the postcode as opposed to local authority name only
- detailed or aggregate ethnic groups
- type of assets or their total value
- income or assets brackets or value
2. Using existing administrative data held by care providers
Theoretically this would enable coverage of self-funders who are not on local authorities’ records - however, care providers hold very limited information about the finances of people who pay for their services. While they may do some initial checks when taking on a new client to assess their ability to pay, they do not usually store the information, or they store the minimal amount of information they need, and it does not get updated over time. This means that this option is unlikely to provide the financial data that DHSC needs. Information about the care needs of the people they support is likely to be held in inconsistent formats which cannot be easily aggregated, or care providers may be unwilling or unable to share it. Significant data completeness, data quality and data protection issues would need to be resolved.
3. Linking centrally held data (such as data held by HM Revenue and Customs, HM Land Registry and the Department for Work and Pensions) with surveys or local authorities’ fnancial and care assessment data
Bringing together data across government bodies and linking this with local authority administrative or survey data would create a comprehensive and detailed data set on the incomes and assets of people with care needs. This data could possibly be updated over time and could therefore provide longitudinal data. There are a number of uncertainties that would need further exploration to determine the viability of this option. This option would require buy-in from a number of government organisations. The data governance arrangements of each would need to be investigated and this would inform the level of data that could be shared and linked together. Expert reference group members commented that the recently set up client level data (CLD) could potentially be used to complement these sources: it includes information about unit costs and amount paid for care.
Conclusion: a combined approach
To conclude, this project showed that collecting the information DHSC needs on the care needs, assets and income of people with care and support needs will most likely require a combination of options, as none of the options examined could in isolation provide all the information required. The most promising combination of options will be feasible when the charging reforms are introduced, as local authorities will need to consistently collect names and contact details of self-funders who choose to start metering toward the cap, or when they request local authorities to commission care on their behalf.
A potential combination of approaches could include the following.
1. Ad-hoc surveys with people in receipt of care and proxy respondents
This approach would involve short, simple ad hoc surveys of people in receipt of or entering care collecting proxy information which gives an indication of care needs, care use and financial status, as well as attitudes and experience of planning for and paying for care and demographic information. Individual contact details would be provided to a third party organisation to enable contact to be made with people with care and support needs who have been assessed by the local authority (needs assessment for local authority and self-funders). The approach would need to enable proxy respondents (family or friends) to respond on behalf of the person who draws on care and support with their consent or through a consultee process if the person lacks capacity. The Confidentiality Advisory Group (CAG) and the research ethics committee would need to advise on what is permissible. Ideally this would be done with CAG approval which means the data could be provided without opt-in consent, though people interacting with the local authority would need to be told their data could be used in this way.
2. Using existing surveys to model the link between proxy measures and financial information to create derived variables
Under this approach, existing surveys with detailed high quality financial and social care data could be analysed to model the link between proxy measures and detailed financial information which could allow creation of derived variables on surveys with proxy measures (although the relationship between proxy and detailed measures is limited). Although this data has limitations in terms of coverage and sample size, by combining waves or surveys and modelling future care needs, this data is the best option for providing DHSC with information at pace. This could be supplemented by other work carried out by the Office for National Statistics to explore information about self-funders using Care Quality Commission regulated providers and census information with future plans to gain insight into the individual characteristics of those who self-fund their care. This approach, which relies on information provided by care providers, supplemented by information about the local population, may be able to address some of the data needs of DHSC.
3. Linking data from surveys to data held by local authorities or government departments
This approach involves linking data from surveys to administrative data at local authority or national level (with permission) to enable inclusion of additional information which cannot easily be asked in surveys. This permission is already available for the long running national surveys. With the development of client level data for local authorities with adult social care responsibility, it may offer further potential for linkage with ad hoc surveys, if the data governance is set up appropriately. However, it is important to note that it would take time to resolve information governance challenges and organise permissions for data linkage across government departments. Furthermore, at present the data would not cover self-funders, but this will change once charging reform is implemented as it will increase the number of self-funders being assessed by their local authority, which in turn would increase the benefits of this approach.
Longer term, there is potential to use boost samples for the existing general population surveys to improve coverage of the key groups of interest to DHSC, with a sub-option to extend the social care module to all adults, not just those aged 65 and over. However, it would be challenging to achieve large numbers of respondents in residential care or people of working age.
As DHSC develops new approaches to managing administrative data, implements charging reform and works on the digitisation of social care, consideration could be given to how these changes could be set up in a way which facilitates future data sharing and research to address their information requirements, while maintaining robust data governance and information security controls.