Some text in this chapter has been reproduced from our study protocol,1 published under the Creative Commons Attribution License (CC BY 2.0), which permits unrestricted use, distribution and reproduction in any medium, provided the original work is properly cited (https://creativecommons.org/licenses/by/2.0/). In addition, some text, boxes, figures and tables have been reproduced from Howel et al.2 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution International license (CC BY 4.0), which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/.
Socioeconomic inequalities in health among older people
A vast body of observational evidence on socioeconomic inequalities in health suggests a close relationship between access to resources and health status that is positive, progressive and has no apparent thresholds.3,4 Health inequalities are universal across time, nations, societies and the human life course. Socioeconomic differences in health persist into old age, and socioeconomic inequalities in self-reported physical and mental health widen in early old age.5,6 Individuals with mild cognitive impairment exhibit poorer financial and health literacy.7 Low household wealth is strongly associated with poorer life satisfaction and quality of life.8 The poorest older people have inadequate access to services essential to health and well-being.9 Older people, especially those in poor health, therefore often need additional income and support to maintain their independence, including payments for care, domestic help and aids and adaptations to the home.10–12
Resource-based interventions to promote health
Research into the differential provision of welfare across nations has shown that it can exert an influence on subjective well-being, morbidity and mortality. For example, older people are emotionally better off13 and at lower risk of suicide in welfare states more generous than the UK.14 Almost half of the reduction in excess winter mortality since 1999/2000 has been attributed to winter fuel payments.15 Key theories that explain how money influences health include materialist explanations (e.g. money buys health-promoting goods and the ability to engage in a social life in ways that enable people to be healthy, and enables the avoidance of health risks in the environment); psychosocial mechanisms (e.g. the stress of not having enough money may adversely affect health, and having adequate resources may relieve stress); behavioural factors (e.g. people living in disadvantaged circumstances may be more likely to have unhealthy behaviours); and being in poor health may affect education and employment opportunities in ways that affect subsequent health.16
In theory, increasing individual or group access to material, social or financial resources (so-called ‘assets-based’ approaches)17 should result in improved health,4,18,19 yet little research has directly evaluated the impact of increasing resources on health.20 Ecological studies following the reunification of Germany have suggested important impacts of increased income on mortality, although a range of other factors may have played a part.21,22 A systematic review20 of 10 North American randomised controlled trials (RCTs) of income supplementation experiments targeting a range of age groups, carried out in the late 1960s and 1970s, showed that none had reliably assessed the effects of increased income on health. The authors pointed out that, although such experiments are unlikely to be repeated, one way of assessing the health impact of increasing financial resources on health would be to evaluate the impact of assisting claimants to obtain full welfare benefit entitlements.20
Tackling health inequalities has become a major policy priority for the UK, highlighted, for example, in the White Paper on public health (Saving Lives: Our Healthier Nation).23 Following publication of the Acheson report24 and the advent of Health Action Zones in the late 1990s,25 there was an increase in welfare rights advice projects linked to primary care in the UK. Welfare rights advice provision in primary health-care settings can reduce, by an estimated 15%, the time general practitioners (GPs) spend on benefits issues, and leads to fewer repeat appointments and fewer prescriptions.26 In 1999, Reducing Health Inequalities: An Action Report27 highlighted welfare rights advice as a potentially effective intervention for reducing health inequalities. This proposal was endorsed by the UK government’s Marmot review in 2010.3
Social welfare for older people and the underclaiming of entitlement
In the UK, a range of financial and non-financial welfare benefits is available; these benefits have different purposes and different criteria for entitlement. Some are means-tested (income-related) and others are dependent on an assessment of health or other needs. Means-tested benefits include income-replacement benefits (e.g. income-related pension credit, and income-related financial support with housing costs). Non-means-tested benefits are intended to help meet the additional needs of those with health or social care problems (e.g. meeting the extra costs associated with ill-health or disability, carer’s benefits to supplement/replace earnings for people who provide support or bereavement benefits for people whose legal partner dies). In addition to financial benefits, material support can be offered to those in need (e.g. mobility aids and housing adaptations, such as rails, ramps and bathroom aids). Financial and non-financial benefits are variously administered through the Department for Work and Pensions, local authorities, housing associations and charities. Welfare rights advisors (WRAs) in local authority social services departments and third-sector organisations (e.g. Citizens Advice Bureaux, Age UK and other charities) help by undertaking financial assessments, recommending whether or not a client might be eligible for a benefit and sometimes offering active assistance with completing claims. Active assistance may also involve liaising with health or social care professionals for evidence of diagnosis or care needs.
The non-claiming of welfare entitlements among older people is a long-standing problem, which is currently increasing.28 A large proportion of eligible welfare benefits remains unclaimed,28 and many of these are health-related benefit entitlements of vulnerable groups, such as older people.28,29 Failure to claim entitlements is associated with a number of factors including the complexity of the benefits system,30 lack of knowledge about entitlements and difficulty in making claims.12,31–33 In recent British studies, the bureaucracy associated with applying and uncertainty regarding the outcome of any claim, the amount of the award, autonomy and independence were reasons cited by people who were entitled to but did not claim benefits.12,31–34 In addition to the state pension, there are a number of means-tested and non-means-tested benefits that can be awarded to older people if entitlement conditions are fulfilled. The level of underclaiming varies depending on the benefit concerned, but it is estimated to be at least 33% for Pension Credit and 40% for Council Tax Benefit.35 Entitlement to one benefit can often act as a ‘passport’ to others, as many of the benefits aimed at people over the state retirement age are linked together in a complex network of entitlements that is often difficult for people to access without expert assistance.
Welfare rights advice services and their evaluation
Our systematic review of the health, social and economic impacts of welfare rights advice services in health-care settings32 identified numerous studies that demonstrated the financial and material benefits of such services. Welfare rights advice provided by local authorities and third-sector organisations is known to increase the uptake of benefits among those eligible, particularly when this involves ‘active assistance’ with benefit claims.24,36 Studies have also shown that the receipt of benefit entitlements can be increased by providing information and advice in general practice, particularly in relation to those benefits that are health related.37–41 However, our systematic review identified only two studies that investigated the health impact of welfare rights advice,42–44 one of which found an improvement in health-related quality of life (HRQoL) in some of the subscales of the Short Form Health Survey-36 (SF-36).42,43 However, both studies demonstrated the difficulties of identifying and measuring appropriate health outcome measures when assessing the health effects of welfare rights advice in primary care. Neither study used a randomised controlled design, and both suffered from significant methodological weaknesses that rendered them inconclusive. Both studies evaluated a model of welfare rights advice delivery that was based in primary care, required referral from practice staff and did not offer a domiciliary service or active assistance with claims. Qualitative studies exploring the impact of welfare rights advice on clients in primary care have identified a range of health-related outcomes that can potentially result from the receipt of welfare rights advice, including changes in the physical, behavioural and, in particular, psycho-social domains of health.12,31–33
Since 2006, few studies have added substantively to this literature. Two papers updated analyses included in our systematic review but yielded no new conclusions.45,46 A study in a hospice setting audited referrals for welfare rights advice and found inconsistency in referral practices and a heavy workload for WRAs, but valuable potential of such referrals in terms of benefit eligibility.47 A study in a UK musculoskeletal clinic48 examined the use of the Health Assessment Questionnaire49 as a tool to identify patients’ eligibility for welfare benefits. Although the authors found relatively high levels of eligibility for new benefits among those identified, there are inconsistencies in the numbers in the paper that make it difficult to interpret. Nevertheless, it does identify a potential screening tool for use in health care that might be of value in the future. Another study from some of the authors of this report (Moffatt, Noble and White) examined the use of welfare rights advice among 1174 cancer patients, assessing the viability of a service linked to clinical teams using qualitative methods.50 The service proved feasible, with exceptionally high levels of successful claims (96%) and median benefits returned (£70/patient). Patients and staff valued the service highly, and health outcome evaluation was felt to be warranted. Krska et al.51 used a cohort study with qualitative interviews among 148 patients in primary care to evaluate a Citizens Advice Bureaux-delivered welfare rights advice service.51 They reported a decrease in prescription of hypnotic/anxiolytic medications 6 months after referral for advice, compared with 6 months before, but their analysis was uncontrolled. Practitioners felt that the service benefited patients and reduced workloads.
We conducted a pilot RCT to prepare for the definitive RCT described here, evaluating the impact of a domiciliary welfare rights advice service offered to people aged ≥ 60 years identified via primary care in disadvantaged areas.11,12,52 In this pilot trial, 58% of participants were awarded either financial (median gain £55/week), non-financial (e.g. aids and adaptations to the home) or both types of benefits,11 confirming the feasibility and success of the intervention from the point of view of accessing unclaimed benefits. It also provided vital information on the feasibility of such a trial, which has helped in planning the definitive RCT reported here. In reporting on the pilot trial we identified a number of key design and methodological issues that would need to be addressed in a definitive evaluation. These are discussed below.
Study design, level of randomisation, contamination and dilution
An appropriate trial design was required, preferably with both randomisation and concurrent controls. Individual-level randomisation was considered preferable to cluster randomisation (e.g. at general practice level), as it required a smaller sample size. The potential problem with individual-level allocation was that there could be ‘contamination’ between intervention and control participants in the same general practice. This was more likely to be the case when welfare rights advice was available from an open-access service delivered in the general practice. However, by using a WRA who only saw patients in their own homes, we found in our pilot RCT that contamination did not occur; no control participants reported independently seeking welfare rights advice during a follow-up period, albeit a period of only 6 months.11
Equipoise and the control condition
A key consideration in designing the proposed trial was whether or not there was genuine equipoise. Welfare rights advice is known to increase access to financial and material resources for eligible clients. However, our systematic review of published and grey literature indicated that there was no conclusive evidence that welfare rights advice leads to positive or negative changes in health.32 We discussed these findings with WRAs, with directors of adult social services, with a selection of GPs and with members of the public in our target age group. We found each of these groups to be in equipoise with regard to the proposed trial health outcomes. Having established this, we also carefully considered the issue of study design and the ideal and feasible control conditions.
We considered that, ideally, controls should be adults as similar as possible to intervention group participants, but should not receive welfare rights advice, nor claim or receive new benefits, during the period of outcome follow-up. In clinical trials, it is usual to withhold the intervention from the control group because the health benefits of the intervention are not proven (i.e. clinical equipoise exists). Although this is the case with regard to the health impacts of welfare rights advice, as indicated above there is adequate evidence that welfare rights advice leads to significant financial and material gains for a proportion of recipients. Thus, it was considered ethically problematic to identify that control group participants were eligible to receive additional financial benefits, but either to keep this information from them or to tell them of their eligibility but not give them advice or help with claims. To circumvent this dilemma, we proposed that control participants should not receive a welfare rights assessment until the end of the trial period (i.e. following final outcome measurement). The full intervention (i.e. a full benefit assessment and active assistance with claims until resolved) would then be offered.
We had concerns that the WRAs might feel tempted to offer benefits advice to control participants before the 2-year ‘wait period’ had elapsed. We planned to avoid this by not informing the WRAs of the names of control participants until a few weeks before their benefits assessment was due (i.e. 2 years after their baseline assessment). The study team would hold control participants’ names securely during this period.
The design would thus avoid unfairly raising expectations among control participants. It would also help to avoid the potential problem of contamination, which could have arisen if control participants independently sought welfare advice (leading to dilution of the outcome effect), although we did not propose attempting to prevent this as we found little evidence of it in our pilot RCT. The proposed control condition is, therefore, in effect a ‘wait-list’ control, whereby the control group waits to receive the intervention 24 months after the intervention group.
It is, of course, possible that some members of both the intervention and control groups will die during the proposed 24-month follow-up period, which we would expect in the course of any prospective study of this age group. In our pilot study, we recorded seven deaths (four in the intervention group and three in the control group) out of 105 after 24 months’ follow-up.11
One of the reasons that such interventions to date have not been rigorously evaluated for health impacts is in part because such research has previously been deemed unethical by researchers considering undertaking such evaluations,53 on the grounds that one cannot withhold benefits to which people are entitled. The proposed design of this RCT is fair because, at present, this kind of intervention is not routinely available to primary care patients and is generally only available to those who seek such services or are referred to them by a health or social care professional (e.g. a hospital social worker); these options remain open to patients in this trial. When targeted services are available in primary care, they tend to be short term and ad hoc. If we find any general practices in participating local authority areas with access to such services, they will be excluded from this trial. Genuine uncertainty exists about the proposed health-related outcomes because participants will not be denied any entitlement that they would otherwise receive and, at the time of the study, the health impact of the proposed intervention remains unknown.32 We have determined that the research team and key stakeholders from health and social services are in equipoise about the proposed health outcomes.
Pragmatic versus explanatory
In the proposed trial, we know that not all participants in the intervention group will be eligible for additional benefits, and that those who are will receive variable amounts of financial and non-financial benefits. Ideally, we wish to examine the health impact of receiving versus not receiving such benefits, as well as examine the potential for a gradient of effect (‘dose–response’ relationship) by the amount of benefit received. However, to do so will require a substantially larger sample size than that proposed. In practice, therefore, the receipt of welfare rights advice is the intervention we will be evaluating (rather than receipt of specific benefits), as ‘welfare rights advice’ is the service being delivered. The proposed trial is, therefore, a pragmatic (intention-to-treat) RCT of this ‘complex’ intervention.54 Nevertheless, we will also assess the potential for exploratory subgroup analyses looking at differential effects by participant characteristics (such as age and sex), and receipt/non-receipt of and levels of any benefits received. We anticipate that the trial will therefore contribute both to answering the question of whether or not the welfare rights advice intervention is effective in improving health and to providing new evidence on the more fundamental question of whether or not increasing resources leads to better health.4
Length of follow-up
To enable the accurate assessment of the health and social effects of welfare rights advice, an appropriate length of follow-up is required. Our previous work suggested that considerable time may elapse between the first advice session and the receipt of new financial or material benefits. Often this is between 3 and 6 months, but it can be longer if the case is not straightforward or if there is an appeal. For example, in our pilot RCT, 45% had received their entitlements by 3 months after their welfare assessment, 85% after 6 months, 95% after 9 months and 100% by 12 months.11 Given such delays in receipt of benefits, as well as the fact that any financial benefits may not be spent immediately, it seemed unlikely that the benefits would have substantial impacts on health within the first 12 months. In our pilot, which was not adequately powered for substantive analyses, we found no suggestion of differences in health-related outcomes between the intervention and control groups after 6, 12 or 24 months (although in the pilot trial control participants did receive the intervention after 6 months).11 Nevertheless, it seemed likely that the longer the delay between receipt of intervention and measurement of outcomes, the greater the chance of demonstrating a substantive effect on health.
To assess the acceptability of a range of delays in receiving the intervention among control group participants, we undertook an experiment in the context of a focus group discussion with a representative sample of low-income older people. To achieve this, simulations of the RCT randomisation procedures were undertaken. The first simulation concerned a typical drug trial, and participants were given sweets whose colour depended on whether they were randomised to the intervention or the control group. Then, randomisation for the proposed trial was simulated. The concept of equipoise, with regard to the health impacts of welfare rights advice, was explained. Next, each group member was given an envelope from which they found out whether they were in the control group or the intervention group. If in the intervention group, they were allocated various types of benefit (e.g. Attendance Allowance plus Council Tax Benefit plus Housing Benefit), and the monetary value of these was revealed to them. We then talked through the various possible time delays until the control group would also receive their welfare rights assessment and advice. The time delays used were 3, 6, 9, 12, 18, 24, 36 and 60 months. The initial response to the design was that it was unfair to those in the control group. However, when it was explained that (1) although such services exist, they are not routinely targeted at or delivered to all people aged ≥ 60 years but are only available on referral or demand, (2) the findings of this study could influence the development of such services, involving collaboration between health and social services, and (3) that a substantial ‘wait’ between intervention and control groups is needed to establish any differences in health outcome, the consensus of the group was that a delay of 24 months would be acceptable in the context of the proposed trial. We therefore opted for a wait-list design for the proposed RCT, with a 24-month follow-up period for the main outcome assessment, followed immediately by delivery of the full intervention to the control group.
Selection bias
In our pilot RCT, GPs wrote to random samples of people aged ≥ 60 years, inviting them to respond with an indication of their willingness to participate in the trial (i.e. to ‘opt in’).11 Using this method, 36% initially agreed to participate, 14% declined to participate and 50% failed to respond. Low levels of positive response to an ‘opt-in’ approach carries a risk of participation bias.55,56 In their work on evaluating the impact of welfare rights advice for the Department for Work and Pensions,57 Corden et al.58 routinely use an ‘opt-out’ method of recruitment for similar populations. With the approval of the NHS Research Ethics Service (NRES), we proposed using this type of recruitment method in this trial; this should have reduced the potential recruitment bias associated with ‘opt-in’ recruitment methods and increased the efficiency of trial recruitment.
Choice of outcome measures
Previously reported studies of the health effects of welfare rights advice have restricted reported health outcomes to general measures of health or psycho-social functioning (such as the SF-3642,59), together with measurement of financial gains.
In our qualitative research among recipients of welfare rights advice,12,33 we identified a range of potential benefits of advice, including:
- health (improvements in anxiety, depression, insomnia; reductions in medication or consultation)
- health-related behaviours (health-promoting changes in smoking, diet, physical activity and alcohol consumption)
- social (improvements in family or other relationships, reduction of social isolation, increased ability to work, ability to care for relatives, etc.)
- financial (debt rescheduling and receipt of new benefits, e.g. Attendance Allowance, Disability Living Allowance, Disability Living Allowance mobility component, Invalid Care Allowance, Incapacity Benefit, Housing Benefit, Income Support)
- material (e.g. access to free prescriptions, Council Tax exemption, entitlement to respite care, Meals on Wheels, rehousing or home modifications).
The qualitative findings of our pilot study12,33 summarised these perceived benefits of the intervention in terms of:
- increased affordability of necessities
- increased capacity to manage unexpected future problems
- decreased stress related to financial worries
- increased independence, including ability to travel, shop, visit the GP, etc.
- increased ability to participate in family life and society.
These findings are corroborated by a recent evidence synthesis undertaken to develop a logic model for the health outcomes of welfare rights advice services.19 The resultant logic model has been used as the basis for a realist evaluation of the health impact of Citizens Advice Bureaux services, for which a protocol has recently been published.60
In addition to qualitative work with study participants in our pilot RCT, we collected proposed outcome measures and relevant potential confounding factors.12 The pilot trial was not sufficiently powered for substantive analyses, but the feasibility of measurements was good, and well tolerated by older people. The main outcome measure we assessed was the SF-36 instrument, as used in several previous studies.42,59,61,62 Although this had demonstrated some potential, albeit in uncontrolled or non-randomised studies (e.g. positive changes in the mental health and emotional role domains),59 we were concerned that its domains did not sufficiently encompass the substantially wider range of reported impacts of welfare rights advice identified in qualitative studies, including our own (see Welfare rights advice services and their evaluation).12,33 We therefore sought an alternative HRQoL measure that might best encompass key domains, such as independence, social participation and mental health. There is no single, ideal outcome measure that captures all of these domains, but the CASP-19 (Control, Autonomy, Self-realisation and Pleasure) instrument,57,63 developed specifically with a view to measuring quality of life in older people, comes close and has been recommended by Corden et al.58 as a composite measure of the impact of welfare rights advice.57 It is a self-reported summative index, comprising 19 Likert scale items in four domains: control, autonomy, self-realisation and pleasure.63 Its performance has been examined in several prospective studies, including the English Longitudinal Study of Ageing (ELSA).64,65
Generalisability
Our pilot trial was undertaken in one social services district (Newcastle upon Tyne) and in four general practice populations.11 However, we know, from other work and discussions locally, that service delivery in welfare rights advice varies from area to area, as do general practice populations. To enhance the potential generalisability of the results, the RCT was, therefore, undertaken across a range of geographical and local authority areas (including urban and rural) and general practices. It seemed possible, indeed likely, that the present welfare regime would change during the course of the trial. The proposed intervention was not dependent on any particular set of benefits and was adaptable to any new regime. This added to its future generalisability.
Target population
Although we recognised that isolated older people who are eligible for benefits may live in all areas, in order to maximise the efficiency (and impact) of welfare rights advice services provided through primary health care, this RCT focused on practice populations in socioeconomically disadvantaged areas. Eligibility for health-related benefits (and failure to claim) increases with age, particularly post retirement; although there are other key target groups, such as single parents, non-claimants most likely to be accessed through primary care are predominantly in older age groups.33,66–68 This trial therefore focused on a predominantly post-retirement population (those aged ≥ 60 years), residing in areas of economic deprivation.
The North East has some of the poorest health outcomes in England, coupled with high levels of socioeconomic disadvantage. From the 19th century, the regional economy was based on heavy industries such as deep coal mining, shipbuilding and steel making. Since the 1970s and the demise of heavy industry, the area has suffered economic decline due, in particular, to the closure of shipbuilding, coal mines and other heavy manufacturing. The post-retirement population of interest in this trial were born in the pre-and post-Second World War years (from 1920 to 1950), and many would have worked in these industries, which are associated with industrial-related conditions such as musculoskeletal injuries and diseases of the lung.
Nature of the intervention
The intervention delivered in this trial was based on standard advice services of the type found across local authorities in England when this study was designed. Welfare rights advice is not a statutory service, and provision varies across all local authorities in England, Wales, Scotland and Northern Ireland. Conventionally, however, these services are available only on demand or by referral. Thus, for example, an older person admitted to hospital may be referred by a hospital social worker, doctor or nurse for benefits assessment prior to discharge. Only some services have undertaken the targeting of welfare rights advice at a population level.32 Those that have done so have found that there is a significant level of underclaiming in the general population and, in particular, among older people.29 The proposed intervention is, therefore, a modification of a standard welfare rights advice service to target proactively a particularly vulnerable population in which we anticipated there would be high levels of underclaiming (i.e. those aged ≥ 60 years in disadvantaged areas). The only reliable population registers in England at a local level are the primary care patient registration lists held by GPs, which were used to sample this target group selectively.
In our pilot RCT, we identified that effectiveness and efficiency (in terms of successful claims) could be maximised by making the service domiciliary, as a substantial proportion of those aged ≥ 60 years have limited mobility, and during assessments clients often need access to information they keep at home.33 Domiciliary visits also proved more popular with clients. In addition, we found that WRAs need to provide ‘active assistance’ with claims, for example completing claim forms for clients, as being unable to complete the forms is a key barrier to people claiming.12,31,32 Finally, GPs need to have appropriate awareness of welfare entitlements and, for health-related benefits in particular, an understanding of the medical criteria on which decisions are made, so that they can support reasonable claims effectively in medical assessments requested by the Benefits Agency. Good communication between GPs and WRAs is essential to facilitate this. In our pilot RCT, we delivered education and training on these issues to all GPs in participating practices,11,12 another feature that is included in the proposed definitive RCT.
Process evaluation
Process evaluation is an essential element of designing and testing complex interventions. Process evaluation is used to assess the fidelity and quality of implementation, clarify causal mechanisms and identify contextual factors associated with variation in outcomes.54 When evaluating the impact of an intervention, the emphasis of process evaluation is on providing greater confidence in conclusions by assessing the quantity and quality of what was delivered, and assessing the generalisability of its findings by understanding the role of context.69
To draw conclusions about what works, process evaluations aim to assess fidelity (whether or not the intervention was delivered as intended) and dose (the quantity of intervention implemented). Intervention fidelity relates to adherence to the content of the study protocol and the process of ensuring quality and consistency in intervention delivery to all study participants.70,71 This includes ensuring a clear intervention model, developing an intervention protocol, selecting appropriate staff to deliver the intervention, ensuring adequate training in the implementation of the intervention and providing staff with ongoing supervision.70 In complex interventions, fidelity is not a straightforward process. In pragmatic trials the standardisation of the intervention to the study protocol will promote internal validity, but limiting variation may affect generalisability.54
In addition to what is delivered, process evaluation can investigate how the intervention is delivered. This can provide vital information about how the specific intervention might be replicated, as well as generalisable knowledge on how to implement other complex interventions. Issues considered may include training and support, communication and management structures, and how these structures interact with implementers’ attitudes and circumstances to shape the intervention. Process evaluations also investigate the reach of interventions (whether or not the intended audience comes into contact with the intervention, and how).69
Exploring the mechanisms through which interventions bring about change is crucial to understanding both how the effects of the intervention occurred and how these effects might be replicated by similar future interventions. Process evaluations may test hypothesised causal pathways using quantitative data as well as using qualitative methods to better understand complex pathways or to identify unexpected mechanisms. Context includes anything external to the intervention that may act as a barrier to or facilitator of its implementation, or its effects. Implementation will often vary from one context to another, however; an intervention may have different effects in different contexts even if its implementation does not vary. Complex interventions work by introducing mechanisms that are sufficiently suited to their context to produce change, while causes of problems targeted by interventions may differ from one context to another. Understanding context is, therefore, critical in interpreting the findings of a specific evaluation and generalising beyond it. Even where an intervention itself is relatively simple, unlike welfare rights advice, its interaction with its context may still be highly complex.69 In the proposed RCT, we collected both quantitative and qualitative data to assess the process of intervention delivery, allowing us to explore the reach, uptake, fidelity and quality of the intervention, as well as provide contextual information that may prove valuable in explaining the observed outcomes.