NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Raine R, Fitzpatrick R, Barratt H, et al. Challenges, solutions and future directions in the evaluation of service innovations in health care and public health. Southampton (UK): NIHR Journals Library; 2016 May. (Health Services and Delivery Research, No. 4.16.) doi: 10.3310/hsdr04160-69

Cover of Challenges, solutions and future directions in the evaluation of service innovations in health care and public health

Challenges, solutions and future directions in the evaluation of service innovations in health care and public health.

Show details

Essay 5Evaluating health-care equity

, , , and .

Author Information

Abstract

The evaluation of health-care equity necessitates measuring both horizontal and vertical equity components to establish whether or not patients receive the health care across levels of need. Examining interactions between social factors of concern and need, or stratifying analyses according to different levels of need, can be used to identify horizontal and vertical equity. Increased data linkage across sectors and settings is vital for identification of sources of inequity and, crucially, to ascertain whether or not any identified variation has clinical relevance. Macro-, meso- and micro-level determinants of equity should always, ideally, be considered. However, examination of the ‘gap’ or the ‘gradient’ will depend on the intervention studied. Emerging techniques should be harnessed, for example machine learning to more completely exploit data sources on need, case mix and outcomes; and interactive multimedia techniques to examine social variations in clinical decision-making.

Scientific summary

Universal health-care systems aim to provide care for all, solely according to clinical need. Despite this, the ‘inverse care law’ has been demonstrated to operate in these systems. In this essay we discuss key methodological challenges in the measurement of health-care equity and propose ways forward.

The ascertainment of equity in health care requires the evaluation of both horizontal and vertical components: horizontal equity refers to the equal treatment of those with equal needs and vertical equity recognises that people with greater clinical needs should have more intervention. Methods to examine varying levels of need include stratifying analyses according to different levels of need or examining interactions between the social factor of concern and a measure of need.

Data on the population in need of care and case-mix data may be poorly recorded or available only as free text. Emerging innovations in machine learning enable extraction of valuable information from both existing and hitherto unused data sources. Data linkage of data from different settings is also required to identify source(s) of inequity and whether or not it matters in terms of outcomes.

Careful consideration should be given on whether it is most appropriate to examine the gap or the gradient in sociodemographic differences in the uptake of health-care/public health interventions. Statistical techniques are now available to calculate the sample size needed for gradient analyses.

In addition to examining equity at the intervention level, innovative methods are now being developed to examine the impact of national policies. Evidence from North America is that dismantling universal systems will not only create greater inequities in access but also mean increases in total costs of health care. Researching inequalities at the individual level and in clinical encounter are also beginning to be addressed. Recent approaches include the use of web-based interactive multimedia vignettes with actor ‘patients’ to simulate key features of health-care consultations. Qualitative, longitudinal study designs which allow the exploration of decision-making at different points of a patient’s health-care journey could also yield valuable insights.

The development of innovative techniques and application of emerging methods for collecting and analysing data will enable equity to be measured with greater accuracy, precision, relevance and comprehensiveness. This will, in turn, better inform interventions for their remediation.

Introduction

Universal health-care systems, whether publicly funded or insurance based, aim to provide health care for all, according to clinical need, and undistorted by social or economic factors, geographical location or ability to pay. Yet observers have long recognised the presence of the so-called ‘inverse care law’ operating in these systems.1 This term was first coined in 1971 by Dr Julian Tudor Hart, a general practitioner (GP) who worked in socially deprived mining communities in the Welsh valleys. His observation that ‘the availability of good medical care tends to vary inversely with the need for it in the population served’ was largely based on his personal experiences rather than on empirical research. However, a wealth of confirmatory data has since been published internationally. In England, this led the Chief Medical Officer to observe in 2005 that, in the publicly funded NHS:

. . . healthcare [. . .] is to some extent inequitable at present: the preference of clinicians, the socioeconomic status and empowerment of patients, and decisions regarding specific local resource allocation may influence clinical practice as much as the actual health needs of patients, the behaviour of any pathological process or the scientific evidence base.

p. 22 (reproduced under the terms of the Open Government Licence for Public Sector Information)2

For those health-care systems which aim to provide universal coverage based solely on clinical need, health-care inequity matters because it undermines the capacity of the system to remain true to its core values or its constitution. The robust measurement of equity is, therefore, crucial.

In this essay we discuss key methodological challenges in the measurement of health-care equity and propose some ways forward. We also briefly review evidence on policies to achieve equity using the method of the natural experiment.

Equality, equity and its horizontal and vertical components

Equity is about fairness and justice and implies that everyone should have an equal opportunity to attain their full potential for health or for the use of health care.3 It should be distinguished from the related concept of equality. Equality is about the equal distribution of shares (of health or health care) so that each individual receives the same amount. The notion of equity transcends equality. Some inequalities may be unavoidable and therefore are not generally considered unjust. Others, for example associated with one’s area of residency, ethnic group, sex, age, socioeconomic circumstances (SEC) or disability, might be avoided and so are considered inequitable. They may be unfair or unjust as well as unequal and, because they are not solely determined by need, they may lead to differences in outcomes.

We recognise that definitions of fairness vary according to libertarian, liberal and collectivist perspectives. However, universal health-care systems define fairness in terms of needs. Mooney4 pointed out that equitable distribution according to clinical need might refer to the distribution of expenditure, access, use or health. With respect to expenditure distribution, in the UK, the NHS uses formulae to promote equitable allocation of funding for care. The Resource Allocation Working Party (RAWP) and its successors have developed different methods of formula funding for the four countries of the UK (England, Scotland, Northern Ireland and Wales) which recommend that money should be distributed on the basis of population size, weighted for relative need and accounting for variations in unavoidable costs of providers (e.g. higher living costs in London, excess costs for delivering services in remote areas in Scotland).5 The components of the RAWP formula have been reviewed and revised for over 30 years and they continue to be subject to reanalysis. For example, although the RAWP recommendation to account for relative need by weighting for age and sex using national average rates of utilisation is relatively uncontroversial, their use of the standardised mortality ratio to account for additional need has raised many questions, which are discussed by Bevan.5 In 1999, a new objective for the allocation of resources in the English NHS was introduced to contribute to the reduction in avoidable health inequalities.6 As a consequence, a health inequalities component was introduced into the allocation formula and increases in allocations have since favoured more deprived areas. Local NHS commissioners were free to use these additional funds to purchase primary or secondary health care or public health services, to better meet their population’s needs and improve the quality of care received. The underlying rationale is that additional health-care expenditure translates into improved population health outcomes. A recent analysis of this policy found that geographical inequalities in mortality from causes amenable to health care declined in absolute terms during the 10-year period in which this allocation policy was applied.7 Moreover, the association between additional NHS funds and reduced mortality was stronger in deprived areas than in more affluent areas. Analyses such as these, using aggregated, routinely available data, cannot exclude the possibility that the associations observed were attributable to unmeasured confounders, such as smoking, or to other, concurrent non-NHS policies implemented to tackle social exclusion in disadvantaged areas. Moreover, they do not tell us about the types and content of services received and their relative contributions to outcomes.

Indicators of access include the availability of resources, waiting times, user charges and others barriers to care. Thus, equity of access is a purely supply-side phenomenon, in the sense that equal services are made available to patients in equal need. The difficulty with measuring equity according to this definition is that individuals may not use services to which they have access. Their reasons for non-use are socially patterned and are, at least in part, because of structural and environmental barriers which are proportionately greater for the socioeconomically disadvantaged, older people, people with disabilities or those for whom English is not their first language. Routine data such as medical records may not, therefore, strictly give a measure of access, because they can supply information only about services which have been taken up.

The definition of equity which takes this into account is equal use for equal need. This definition recognises the influence of both demand and supply factors on the pursuit of equity. These factors are influenced by the preferences, perceptions and prejudices of both patient and health-care provider. Most studies on equity and access analyse utilisation rates (often adjusted by sociodemographic and clinical indicators of need) as a proxy measure of access.

The final definition is equal health for equal need. This definition addresses the fact that there are inequalities in health arising from the level of resources, housing conditions, exposure to environmental hazards and different lifestyles and behaviours. It is an ideal rather than an operational definition, as both avoidable and unavoidable influences affect our health. Neighbourhood renewal8 and Sure Start9 are examples of initiatives in the UK that move towards this ideal.

There is another issue surrounding distribution according to need which should be considered. This is that distribution on the basis of need comes in two versions: a horizontal version (people with equal needs should be treated the same) and a vertical version [people with greater clinical needs should have more intervention (provided it is effective) than those with lesser needs (unequal use for unequal need)]. From a macroeconomic perspective, Mooney and Jan10 defined vertical equity in terms of positive discrimination, arguing that achievement of horizontal equity is rarely enough. They propose differential weighting of the level of need by socioeconomic group in order to more thoroughly investigate the distribution of health and health care and to facilitate the movement of disadvantaged people up to the level of the advantaged.10 Mooney and Jan10 do not provide the differential weights that would allow examination of the presence of vertical equity. Sutton11 takes the approach forward by specifying a target level of use for people at different levels of need and then comparing whether or not actual use equals the target.

An alternative way of operationalising vertical equity is generated by the understanding that the demonstration of equal use for equal need does not necessarily indicate unequal use for unequal need.12 For example, although male and female patients with a mild form of a disease may be treated equally (horizontal equity), it cannot be assumed that the likelihood of treatment varies according to the degree of abnormality in both men and women. Men with severe disease may be more likely to receive treatment than men with mild disease (vertical equity), but the likelihood of treatment for women may not differ with disease severity (vertical inequity). The vertical component is often overlooked by researchers and policy-makers, and this prevents the comprehensive measurement of equity, the likely consequence of which is overestimation of fair use of care. Unless both components of equity are measured, it cannot be concluded that patients are receiving the health care that they need. Studies that use multivariable analysis alone to adjust for need assume that social differences in use are the same at every level of need, which may not be the case. Other analysis strategies should be employed to examine varying levels of need, such as stratifying analyses according to different levels of need or examining interactions between the social factor of concern and a measure of need.12

Identifying the population in need of care

To examine the equitable distribution of health care and public health interventions, we need to understand how interventions are distributed across the social factors of concern. This requires identification of both the complete population in need and their sociodemographic characteristics.

This is not always straightforward. Many of the social factors of interest, such as ethnicity, language skill or religion, are poorly recorded in national data sets or available only as free text (which is unstandardised). These data may not always be missing at random, potentially leading to selection bias. An emerging innovation is to apply machine learning techniques to ‘code’ free-text data in medical records. Thoughtful application of these methods has the potential to increase our ability to extract valuable information from both existing and hitherto unused or underused data sources. Machine learning could revolutionise systematic data collection in terms of comprehensiveness, completeness, timeliness and accuracy.

There are many sources of data that can be used to characterise socioeconomically disadvantaged groups. Area-based measures of SEC are often freely available and readily linked with postcode data; for example, in England the Index of Multiple Deprivation (IMD) scores and ranks areas containing approximately 650 households across several domains of deprivation.13 Relying on area-based measures to ascertain SEC for individuals, however, can result in ecological fallacy, where lack of precision or incorrect inference stems from within-area pockets of relative advantage and disadvantage masked by the ‘average’ area score. In the absence of individual-level data, such effects can potentially be reduced by adding in data from commercially available data sets at smaller geographic units which provide information on consumer habits. Such data may distinguish the advantaged from the disadvantaged within an area and provide greater resolution on SEC when used in combination with the IMD.14 The method is not bias free, however, as relatively high levels of consumerism may be the result of high debt, not affluence. Health and health-care researchers are perhaps less familiar with using sources such as local and national taxation information and education data, but forging links with data providers to explore such sources has the potential to reap benefits for future equity research.

Some relationships between health and measures of SEC are best characterised by J-shaped curves, meaning that the most disadvantaged group may not be where you expect to find them. For example, in the USA, health insurance coverage may not be linear with income; the wealthy may opt not to have health insurance, and benefit recipients may have state-provided coverage. Policy-related changes leading to changes in insurance patterns for different social groups over time complicate longitudinal analyses. The relationship between coverage and health for working poor is likely to be highly variable and associated at the micro level with heterogeneous contextual factors such as size of employer and family composition. In England, benefit recipients (and those with long-term chronic conditions) are exempt from prescription charges, so in this context it may be the working poor paying for prescriptions who suffer the highest financial burden of ill health. Examining and characterising the group most at risk by considering the context of the population setting is a crucial step in measuring inequalities. Quadratic terms can be added to regression models to allow for non-linear relationships.15

Accurate ascertainment of the population in need of health care may be hampered by several factors. First, incidence and prevalence from health-care data are likely to be inaccurate if there are a high number of undiagnosed cases of the condition in the community. Diagnosis (and diagnostic delays) are influenced by help-seeking behaviour, which is itself socially patterned. Second, disease presentation can also vary by social group. For example, women with ischaemic heart disease may present with symptoms that differ from the typical chest pain presentation by men.16 This could lead to underascertainment in women if these differences are not reflected in study diagnosis criteria. Third, there may be a lack of consensus about the definition of need for intervention, for example in total hip replacement. Where the clinical rationale for intervention is not clear-cut, investigation and treatment may be influenced as much by the availability of doctors and diagnostic equipment, or by financial factors (such as a fee-for-service system), as by clinical need.17 This highlights the problem of inequity owing to the overuse of medical interventions, which puts patients at risk of complications unnecessarily and drives up the cost of health care. Finally, health-care systems which are not universal are limited by their access to medical record databases which relate only to the insured population, excluding those without coverage, and there may be also unknown missing data bias for out-of-area medical visits.

Examining whether or not inequalities matter

We have already described why it is necessary to examine whether or not health-care inequalities matter in terms of their impact on outcomes. For example, in a study of the effect of secondary prevention in 30-day stroke survivors it was found that people aged 80–89 years were only half as likely to receive a lipid-lowering drug as those aged 50–59 years.18 This treatment inequality was important because the receipt of secondary drug prevention was associated with a halving of the mortality risk. Crucially, there was little evidence that the effect of treatment differed by age. Therefore, the undertreatment of older people cannot be justified, unless it is explained by informed patient choice. If patient preferences explain some of the differences observed, then it is important to unravel their origin. This is a theme that we discuss later, in Social variations in the clinical encounter.

The gap versus the gradient

Policy-makers distinguish between the gap (the relative difference between advantaged and disadvantaged groups) in service provision, and the gradient (the continuum along which increasingly worse health is associated with a unit drop in the social factor of interest). An example of an approach to target the most disadvantaged subgroups only is the Nurse Family Partnership in the USA,19 named the Family Nurse Partnership (FNP) in the UK20 and VoorZorg in the Netherlands,21 which provides intensive support to multiply disadvantaged first-time mothers. Three evaluations of these programmes in the USA report positive results for mothers and their babies, including improvements in birth outcomes, children’s cognitive development and uptake of preventative health care.19,22,23 A trial of the intervention in the Netherlands reported lower smoking rates in nurse-visited pregnant women and increased breastfeeding duration. However, in England, where FNP was added to the usually provided health and social care, no additional benefits were achieved in the primary outcomes, which included smoking in pregnancy, birthweight, rates of second pregnancies and emergency hospital visits for the child.24 There were important differences in both the trial design and the health-care context between the US and English studies: the English trial was a large, pragmatic, independently led evaluation, in contrast to the US evaluations, which were single centre and led by the intervention developers with a greater emphasis on efficacy. Women in England have access to more statutory public health, health care and social services than US women. The lack of any additional benefit from the English FNP suggests that macroeconomic policies and structural changes are required to complement intensive support services when tackling complex, multifactorial and enduring problems.

The costs of inequalities are borne not only by those at the bottom of the socioeconomic hierarchy, but by those at every level. Policies that target the most disadvantaged subgroups only, or which aim to narrow the gap between the most and least disadvantaged, underestimate the pervasive effect across the socioeconomic hierarchy and exclude those in need in the intermediate socioeconomic groups. Even for targeted interventions which are found to be effective, the population-level impact maybe smaller for these than for universal interventions. Such arguments augur in favour of tackling the gradient to address inequalities. Although it is rarely done, the gradient can be measured to examine the effect of universal interventions which are expected to reach the whole population, for example in Wardle et al.25 Where gradients are measured, the time frame over which effects are evaluated should be carefully chosen. This is because universal interventions can increase health inequalities in the short term, as the more advantaged are usually the first to take up new services, but this effect can level over time as the advantaged groups plateau in terms of their ability to benefit.26

Calculating the sample size needed for gradient analyses can present challenges. One strategy was developed for a universal screening programme intervention in which uptake needed to increase more in disadvantaged groups than in advantaged groups.25 Here, the solution was to use the weighted averages of the association between the deprivation quintile and the previously observed response to screening instead of the usual proportions in the formula, with the response rate held a constant across quintiles.27 Efficiency can be optimised by post-stratification, where treatment groups are stratified with a pre-treatment variable, treatment effects within the strata are estimated, and the weighted average of these estimates is used to calculate the overall average treatment effect estimate.28 In studies that aim to estimate average effects, reporting by disadvantaged subgroup, even with insufficient within-study power, increases the pool of health inequality studies available for potential synthesis. There are also additional approaches that can be taken alongside subgroup analyses to estimate within-study differential effects. One method is to conduct a latent class analysis across the whole sample where key response patterns are identified across different social groups of participants.29 This method has the advantage of not being sample size dependent, although very small samples with many multicategory variables may run into estimation problems.

Identifying the sources of inequitable provision of care

Most research undertaken in this area examines sociodemographic variations in health-care use for a defined intervention (or package of interventions) at one point in the management pathway. Inequalities have been demonstrated at each stage of the pathway: in participation in population-based screening programmes in the community, in the management of health problems in primary care, in the access to and use of diagnostic and therapeutic procedures within secondary care, and in rehabilitation and end-of-life care.

For example, the UK breast, cervical and bowel cancer screening programmes are run by the NHS without financial cost to participants. Nonetheless, the uptake of all programmes shows a gradient by SEC.30,31 The strongest gradient is for bowel cancer screening. This involves offering a guaiac faecal occult blood testing kit for use at home. In the first 2.6 million invitations in 2006–9, uptake was 61% in the least deprived quintile of residential areas and only 35% in the most deprived quintile.32 Bowel cancer screening is currently being extended to include one-off flexible sigmoidoscopy of the lower bowel to identify polyps that may develop into bowel cancer. Overall uptake in the first six pilot centres was 33% in the most deprived areas, rising to 53% in the most affluent areas.33

Within primary care, Lyratzopoulos et al.34 demonstrated that, in England, younger patients, ethnic minorities and women are more likely to have visited their GP a minimum of three times prior to referral to hospital for cancer diagnosis, suggesting potential avoidable delay in their management.

Sociodemographic variations in the likelihood of referral by GPs into secondary care has also been found to vary depending on the presence of explicit national guidelines on referral.35 This study used The Health Improvement Network, a widely used primary care database, to examine sociodemographic variations in referral for potentially life-threatening conditions where national guidance on referral has been published [post-menopausal bleeding (PMB) and dyspepsia in people > 55 years of age] and for symptoms where there is clinical uncertainty regarding the decision to refer (hip pain and dyspepsia in people < 55 years of age). For the three conditions examined, older patients were less likely to be referred, after adjusting for comorbidity. Women were less likely than men to be referred for hip pain [hazard ratio (HR) 0.90, 95% confidence interval (CI) 0.84 to 0.96]. More deprived patients with hip pain and dyspepsia (if < 55 years old) were less likely to be referred. Adjusted HRs for those in the most deprived quintile compared with the least deprived were 0.72 (95% CI 0.62 to 0.82) and 0.76 (95% CI 0.68 to 0.85), respectively. There was no socioeconomic gradient in referral for PMB. These findings are important given the widespread prevalence of non-specific symptoms for which explicit referral guidance does not exist, but which could, nonetheless, be indicative of serious underlying pathology (e.g. lung, colorectal and ovarian cancers).

Access to timely secondary care has also been demonstrated to be inequitable: in England, patients from deprived areas, older people and women are more likely to be admitted to hospital as emergencies than electives for colorectal, breast and lung cancer. This was inequitable (rather than simply unequal) because people from deprived areas and older people were also less likely to receive preferred surgical procedures such as breast-conserving surgery and lung cancer resection.36 The research was limited by the data available from routinely collected Hospital Episode Statistics. The findings suggest that social variations in both timely presentation and pathways to care need to be prospectively examined. Prospective data collection would allow the impact of potential confounders, such as tumour characteristics including stage, as well as case mix and patient preferences, to be examined.

Once in the secondary care system, women from less affluent neighbourhoods in France have lower probability of receiving ‘best practice’ treatments (Or Z, Rococo E, Bonastre J, Institute for Research and Information in Health Economics, France, 2016), and in England, patients referred from deprived practices have reduced levels of diagnostic angiography and higher waiting times.37 There is also significant variation for women with breast cancer who live in deprived areas in England: they are more likely to be diagnosed at end stage (Stage IV)38 and less likely to receive surgery and radiotherapy,39 and area-based disparity in receipt of treatment may be the major driver of variation in lung cancer survival in England.40 Inequalities in the effective provision of rehabilitation care are also evident.41

When inequalities such as these are uncovered it is not possible to conclude whether the disparity occurs at the level of the intervention under consideration, or as a consequence of inequalities in the provision of preceding interventions, or in direct (and appropriate) response to the results of previous investigations. For example, a review of sex bias in use of cardiac care found that high-quality prospective studies reported sex differences in favour of men in the use of angiography.12 However, there was consistent evidence of no sex difference in those patients in whom the results of previous investigations had been taken into account. This indicates that the sex inequalities identified were not inequitable, in that they were fair and made on the basis of clinical need (identified by the earlier investigation). Thus, the entire management pathway needs to be examined to establish the reasons for the differences found. The challenge is to identify data that will permit unbiased ascertainment of need and use across two or more settings (primary, secondary and community care). It is here that carefully specified individual record linkage between survey data to databanks of routinely collected data, and linking different sources of routinely collected data, is of high value.

Furthermore, while attention to the meso (organisational) level is appropriate, a comprehensive understanding of the sources of inequity also requires consideration of the impact of national policies (i.e. macro-level factors).42 For example, although the European Union provides universal coverage, the comprehensiveness of care varies from country to country. This is in part explained by the density and position of generalist clinicians, the presence of out-of-pocket payments and the referral system (gatekeeping or not), all of which have an impact on the use of preventative, primary and specialist care.43 In addition, in social insurance-based systems, for example in France, complex rules for reimbursement for different services appear to act as a disincentive for service use among some population groups.44

In the UK, Cookson et al.45 evaluated these policies by examining trends in primary care access, quality and outcomes by SEC between 2004 and 2012. They did this using routinely available whole population data including health data from four national administrative databases. The IMD (IMD 2010)46 was used to assign socioeconomic status to neighbourhoods of approximately 1500 people each. Slope indices of inequality were measured in four indicators: patients per family doctor, primary care quality, preventable hospitalisation and amenable mortality. They found that, during this period, the NHS succeeded in substantially reducing socioeconomic inequalities in primary care access and quality but made only modest reductions in health-care outcome inequalities.

Such analyses cannot assess the extent to which observed trends in preventable hospitalisation and amenable mortality are attributable to, for example, trends in multimorbidity outside the control of the NHS or in social variations in illness behaviour. However, they both provide much-needed evidence of the influence of national policies and highlight the need to address health-care inequity in every sector and setting.

Finally, synthesising equity effects of health-care interventions can be challenging. One approach that may be useful when comparing several different interventions is Qualitative Comparative Analysis (QCA),47 which can be used in tandem with quantitative analyses to work out what works for whom and under what conditions,48 very much under a realist perspective.49 Blackman and Dunstan50 illustrate QCA’s utility in health inequalities research by applying the method to survey data in order to understand the influence of place-based contextual factors related to variation in narrowing mortality gaps. Individual participant data meta-analyses (e.g. Virtanen et al.51) or other synthesis techniques such as QCA (e.g. Thomas et al.52) could be applied to studies in the health-care setting to explore variation in outcomes with regard to contextual effects such as the policy environment, and differential effects by disadvantage. The role of evidence synthesis in health services research is further discussed in Essay 1 of this volume.

Social variations in the clinical encounter

A potential contributing cause of demonstrated health inequalities are social variations in individual behaviour and interactions between patients and health-care professionals. However, there are significant methodological challenges to researching the clinical encounter. Direct observation of doctors and patients offers no opportunity to control patients’ clinical and sociodemographic characteristics, and would require observation of prohibitively vast numbers of consultations to obtain the necessary numbers in specific risk or demographic categories. Use of ‘standardised’ patients (i.e. consultations with doctors by trained actors) is considered a gold-standard method because it enables more control over patient characteristics, but it is costly. The use of fictional patient profiles (vignettes) can provide a valid, generalisable and efficient approach to studying variations in decision-making by health-care professionals. Most studies, however, use text-based vignettes, and omit many features of real consultations such as real-time responses to clinicians’ questions or nuanced presentation of patient characteristics. This risks bias by offering clinicians a limited selection of response options, which primes them to consider certain actions. Many studies are small, with limited generalisability.

In one novel study, a website was constructed using interactive multimedia vignettes with actor ‘patients’ to simulate key features of consultations. GPs undertook consultations from each of six clinical profiles which varied according to the ‘patients’ sociodemographic characteristics and lung cancer risk. No GP saw the same actor twice. Within this constraint, allocation of GPs to vignettes was random. This achieved balance by sex, ethnicity and SEC, and thus GPs’ decisions to initiate lung cancer investigation could be studied in a factorial design across different combinations of clinical and sociodemographic characteristics (Sheringham J, Sequeira R, Myles J, Hamilton W, McDonnell J, Offman J, et al., University College London, 2016). This research demonstrated that, regardless of clinical risk, GPs were less likely to investigate older and black ‘patients’.

Patient vignettes can provide insight into clinical decision-making but do not contribute towards our understanding of the patient perspective or of factors influencing the interaction between patients and health-care professionals. It is often suggested that patient choice underlies inequalities in uptake of health care and public health interventions. Even if decisions to forego some aspects of care reflect ‘informed’ choice, we still need to understand the origins of and influences on these choices. Choices are influenced by individual-level factors (motivation, perceived consequences of different actions and values placed on those consequences), social networks (families, peer groups), local environments (e.g. school ethos) and the national context (taxation, regulation, advertising, subsidies). There appear to be systematic differences in expectations for good health and in perceptions of risk and benefits of treatment between advantaged and less advantaged groups.53 Patients’ ‘choice’ for less intensive treatment may reflect inaccurate perceptions about the availability, effectiveness or risk of treatment, or be influenced by accurate observations that outcomes are worse in their community.53 Furthermore, communication in clinical encounters is socially patterned, with clinicians providing less information and adopting a less participatory consulting style when consulting with less articulate, socially disadvantaged patients.54 This is likely to directly impact on the decisions that patients make.55

Innovative approaches are therefore needed to explore and explain patient ‘choice’. In-depth, observational, qualitative, longitudinal study designs are likely to be required which allow the exploration of decision-making at different points of a patient’s health-care journey. For example, interviews before and after consultations about beliefs and expectations, and how they change, together with non-participant observation of clinical encounters, could yield valuable insights.

Designing interventions to improve equity

Given (and in spite of) the current pre-eminence of the concept of individual choice and responsibility, these observations suggest three levels of intervention. At the level of the patient–clinician interaction it is important to understand patients’ and health professionals’ assumptions about risks and benefits of medical interventions and their accuracy. Health-care delivery preferences may be shaped by quality of service, so we need to ensure equity in the quality of services and focus on quality improvement where certain communities are poorly served. At the societal level, the observation that preferences/perceptions of opportunities are often defined by social, economic and cultural factors underscores the importance of addressing fundamental structural inequities and sociocultural norms.

One area emerging in response to the problem of complex policy and population settings is to design intervention strategies on multiple levels. A host of interventions can often be theorised; for example, simultaneous economic, organisational and behavioural interventions may be needed to tackle diabetes. The intervention framework needs consider factors at macro, meso and micro levels and their interactions, even though much of the dynamism may be focused at only one of these levels. Hawe56 suggests ways to move towards interventions characterised not by a programme but by relationships, routines, power structures and sets of values. A number of essays in this volume take forward in a number of different ways the issue of complexity and how it can be appropriately addressed in evaluative research (see particularly Essays 1, 6 and 7).

Can austerity threaten equity?

Cutler57 argued that consequences of governments aiming for equity by increasing coverage were increases in costs, which led to top-down policies to control costs, which, in turn, resulted in a deterioration in quality (such as long waiting times), and the interest in market-type reforms to remedy that problem. Tuohy58 has argued that governments’ search for a system that achieves equity, controls costs with high-quality care and results in policy cycling as governments emphasise policies that tackle the most serious failure of these three objectives and implicitly neglect the other two. So, following the global financial crisis, fiscal pressures mean that governments now focus on cost control rather than on equity and quality. An obvious question is whether or not there is evidence that systems of universal coverage that are mainly free at the point of delivery, on grounds of promoting equity of access, are poorly designed to control costs. This essay has demonstrated that even with these structural characteristics, there is evidence that the inverse care law1 still applies. Changing to partial coverage and high user charges would mean that that law would apply with even greater force. But the question is whether or not austerity would justify cycling to such policies as a way of controlling costs.

In trying to assess the impacts of government policies for the twin objectives of equity and cost control, we typically lack controlled experiments. Although there is the famous landmark study of the RAND Health Insurance Experiment,59 which randomly allocated people to different insurance packages (with differing co-payments and co-insurance) and also included a Health Maintenance Organisation, this still falls short of assessing the macro questions of whether or not a financial system with universal coverage and services free at the point of delivery will have more serious problems of cost control than one with partial coverage and high user charges. Evans et al.6062 used the ‘natural experiment’ between the USA and Canada to investigate the impact of policies to improve equity of access to health care in terms of control of total costs. Prior to 1970, both countries were structurally similar in how health care was financed (with multiple insurers, partial coverage and high user charges) and in the delivery of care (with hospitals and physicians being independent of government and paid charges and fees for services delivered). After 1970, in Canada, only the financial system changed with the introduction of universal coverage free at the point of access for hospital care and physicians’ services. Evans60 emphasises that an unintended outcome of policy cycling by the Canadian government was that these policies that were directed at equity were later found to be highly effective in controlling total costs of health care in Canada as compared with the USA.

Evidence from ‘natural experiments’ is always open to challenge from potential influence from other factors and other outcomes that have not been measured. Their value is hence even more strongly dependent on developing a sound theoretical explanation, which is just what Evans60 does in explaining what is a paradoxical outcome. Given the characteristics of health care, it is folly to regard patients as well-informed consumers making cost-conscious choices when confronted with high user charges, which are a poor way of trying to control costs when doctors typically make decisions. Indeed, Brook et al.63 summarised the results of the RAND study as showing that ‘. . . cost sharing can be a blunt tool. It reduced both needed and unneeded health services’ (p. 4)63 and that ‘. . . subsequent RAND work on appropriateness of care found that economic incentives by themselves do not improve appropriateness of care or lead to clinically sensible reductions in service use’ (p. 4).63 Evans60 goes on to argue that the key to cost control is targeting not patients but suppliers, and that is most effectively done by empowering government as the single payer with monopsony power in negotiations with physicians and hospitals. Thus, he concludes: ‘The standard theoretical analyses of health insurance, focussing on (an incomplete specification of) the incentives faced by patients, counts the peanuts but ignores the elephants’.60

Spend on health care as a percentage of Gross Domestic Product in 2013 was 16.4% for the USA, 10.2% for Canada and 8.4% for the UK.64 Thus, although Canada and the USA have similar delivery systems, what appears to matter in controlling costs is that Canada and the UK have similar institutional financial systems. So, although austerity strains universal systems which are largely free at the point of delivery, the evidence from North America is that dismantling that system will not only create greater inequities in access but also mean increases in total costs of health care. Such universal systems need to be developed to manage insurance to relate to differences in patients’ preferences (as revealed, e.g., through shared decision-making) but that is beyond the scope of this essay.

Conclusions

In this essay we have highlighted strategies to address some of the methodological challenges faced when evaluating health-care equity. These include the measurement of both horizontal and vertical equity, the use of machine learning to enhance the completeness and quality of data collection and considerations on the appropriateness of gap or gradient analyses. Linkages between routinely collected data from primary, hospital and social care; disease cohort and audit data, deprivation indices, mortality registers; and judiciously chosen other sources (e.g. from education and consumer surveys) increase our ability to precisely identify sources of inequity along the patient pathway and to disentangle the influences of case mix, social and organisational context. Together with innovative methods to capture expectations and patient and health professional decision-making in real time, these strategies will inform the design and evaluation of national-, organisational- and individual-level interventions to improve health-care equity.

Acknowledgements

The authors thank Professor David Byrne for his helpful comments.

Contributions of authors

Rosalind Raine (Professor, Applied Health Research) wrote the first draft of the essay.

Zeynep Or (Research Director, Health Economist) and Gwyn Bevan (Policy Analyst) provided additional material and along with Stephanie Prady (Research Fellow, Social Epidemiologist) commented on the draft.

References

1.
Hart JT. The inverse care law. Lancet 1971;1:405–12. 10.1016/S0140-6736(71)92410-X. [PubMed: 4100731] [CrossRef]
2.
Department of Health. The Chief Medical Officer on the State of Public Health: Annual Report 2005. London: Department of Health; 2006. URL: http://webarchive​.nationalarchives​.gov.uk​/20130107105354/http://www​.dh.gov.uk/prod_consum_dh​/groups/dh_digitalassets​/@dh​/@en/documents/digitalasset/dh_4137367​.pdf (accessed 10 April 2016).
3.
Roberts T. What is the difference between equity and equality? J Health Serv Res 1997;2:129.
4.
Mooney GH. Equity in health care: confronting the confusion. Eff Health Care 1983;1:179–85. [PubMed: 10310519]
5.
Bevan G. The search for a proportionate care law by formula funding in the English NHS. Finance Account Manag 2009;25:391–410. 10.1111/j.1468-0408.2009.00484.x. [CrossRef]
6.
Department of Health. The NHS Plan: A Plan for Investment, A Plan for Reform. London: The Stationery Office; 2000.
7.
Barr B, Bambra C, Whitehead M. The impact of NHS resource allocation policy on health inequalities in England 2001–11: longitudinal ecological study. BMJ 2014;348:g3231. 10.1136/bmj.g3231. [PMC free article: PMC4035504] [PubMed: 24865459] [CrossRef]
8.
Social Exclusion Unit, Cabinet Office. A New Commitment to Neighbourhood Renewal – National Strategy Action Plan. London: Social Exclusion Unit, Cabinet Office; 2001.
9.
Belsky J, Barnes J, Melhuish EC. The National Evaluation of Sure Start: Does Area-Based Early Intervention Work? Bristol: The Policy Press; 2007.
10.
Mooney G, Jan S. Vertical equity: weighting outcomes? Or establishing procedures? Health Policy 1997;39:79–87. 10.1016/S0168-8510(96)00851-2. [PubMed: 10164908] [CrossRef]
11.
Sutton M. Vertical and horizontal aspects of socio-economic inequity in general practitioner contacts in Scotland. Health Econ 2002;11:537–49. 10.1002/hec.752. [PubMed: 12203756] [CrossRef]
12.
Raine R. Bias measuring bias. J Health Serv Res Policy 2002;7:65–7. 10.1258/1355819021927584. [PubMed: 11822263] [CrossRef]
13.
Noble M, McLennan D, Wilkinson K, Withworth A, Barnes H. The English Indices of Deprivation 2007. London: Communities and Local Government; 2008.
14.
Sheringham J, Sowden S, Stafford M, Simms I, Raine R. Monitoring inequalities in the National Chlamydia Screening Programme in England: added value of ACORN, a commercial geodemographic classification tool. Sex Health 2009;6:57–62. 10.1071/SH08036. [PubMed: 19254493] [CrossRef]
15.
Gray L, Leyland AH. A multilevel analysis of diet and socio-economic status in Scotland: investigating the ‘Glasgow effect’. Public Health Nutr 2009;12:1351–8. 10.1017/S1368980008004047. [PubMed: 19026094] [CrossRef]
16.
Brewer LC, Svatikova A, Mulvagh SL. The challenges of prevention, diagnosis and treatment of ischemic heart disease in women. Cardiovasc Drugs Ther 2015;29:355–68. 10.1007/s10557-015-6607-4. [PMC free article: PMC5516639] [PubMed: 26210899] [CrossRef]
17.
Welch WP, Miller ME, Welch HG, Fisher ES, Wennberg JE. Geographic variation in expenditures for physicians’ services in the United States. N Engl J Med 1993;328:621–7. 10.1056/NEJM199303043280906. [PubMed: 8429854] [CrossRef]
18.
Raine R, Wong W, Ambler G, Hardoon S, Petersen I, Morris R, et al. Sociodemographic variations in the contribution of secondary drug prevention to stroke survival at middle and older ages: cohort study. BMJ 2009;338:b1279. 10.1136/bmj.b1279. [PMC free article: PMC2669853] [PubMed: 19372118] [CrossRef]
19.
Olds DL. Prenatal and infancy home visiting by nurses: from randomized trials to community replication. Prev Sci 2002;3:153–72. 10.1023/A:1019990432161. [PubMed: 12387552] [CrossRef]
20.
FNP. Family Nurse Partnership. 2015. URL: http://fnp​.nhs.uk/ (accessed 10 April 2016).
21.
Mejdoubi J, van den Heijkant SC, van Leerdam FJ, Crone M, Crijnen A, HiraSing RA. Effects of nurse home visitation on cigarette smoking, pregnancy outcomes and breastfeeding: a randomized controlled trial. Midwifery 2014;30:688–95. 10.1016/j.midw.2013.08.006. [PubMed: 24041564] [CrossRef]
22.
Olds DL, Henderson CR Jr, Chamberlin R, Tatelbaum R. Preventing child abuse and neglect: a randomized trial of nurse home visitation. Pediatrics 1986;78:65–78. [PubMed: 2425334]
23.
Kitzman H, Olds DL, Henderson CR Jr, Hanks C, Cole R, Tatelbaum R, et al. Effect of prenatal and infancy home visitation by nurses on pregnancy outcomes, childhood injuries, and repeated childbearing. A randomized controlled trial. JAMA 1997;278:644–52. 10.1001/jama.1997.03550080054039. [PubMed: 9272896] [CrossRef]
24.
Robling M, Bekkers MJ, Bell K, Butler CC, Cannings-John R, Channon S, et al. Effectiveness of a nurse-led intensive home-visitation programme for first-time teenage mothers (building blocks): a pragmatic randomised controlled trial. Lancet 2016;387:146–55. 10.1016/S0140-6736(15)00392-X. [PMC free article: PMC4707160] [PubMed: 26474809] [CrossRef]
25.
Wardle J, von Wagner C, Kralj-Hans I, Halloran SP, Smith SG, McGregor LM, et al. Effects of evidence-based strategies to reduce the socioeconomic gradient of uptake in the English NHS Bowel Cancer Screening Programme (ASCEND): four cluster-randomised controlled trials. Lancet 2016;387:751–9. 10.1016/S0140-6736(15)01154-X. [PMC free article: PMC4761689] [PubMed: 26680217] [CrossRef]
26.
Victora CG, Vaughan JP, Barros FC, Silva AC, Tomasi E. Explaining trends in inequities: evidence from Brazilian child health studies. Lancet 2000;356:1093–8. 10.1016/S0140-6736(00)02741-0. [PubMed: 11009159] [CrossRef]
27.
Brentnall AR, Duffy SW, Baio G, Raine R. Strategy for power calculation for interactions: application to a trial of interventions to improve uptake of bowel cancer screening. Contemp Clin Trials 2012;33:213–17. 10.1016/j.cct.2011.09.021. [PubMed: 22008246] [CrossRef]
28.
Miratrix LW, Sekhon JS, Yu B. Adjusting treatment effect estimates by post-stratification in randomized experiments. J Roy Stat Soc B Met 2013;75:369–96. 10.1111/j.1467-9868.2012.01048.x. [CrossRef]
29.
Lanza ST, Rhoades BL. Latent class analysis: an alternative perspective on subgroup analysis in prevention and treatment. Prevent Sci 2013;14:157–68. 10.1007/s11121-011-0201-1. [PMC free article: PMC3173585] [PubMed: 21318625] [CrossRef]
30.
Maheswaran R, Pearson T, Jordan H, Black D. Socioeconomic deprivation, travel distance, location of service, and uptake of breast cancer screening in North Derbyshire, UK. J Epidemiol Community Health 2006;60:208–12. 10.1136/jech.200X.038398. [PMC free article: PMC2465550] [PubMed: 16476749] [CrossRef]
31.
Bang JY, Yadegarfar G, Soljak M, Majeed A. Primary care factors associated with cervical screening coverage in England. J Public Health (Oxf) 2012;34:532–8. 10.1093/pubmed/fds019. [PubMed: 22518028] [CrossRef]
32.
Logan RF, Patnick J, Nickerson C, Coleman L, Rutter MD, von Wagner C. Outcomes of the Bowel Cancer Screening Programme (BCSP) in England after the first 1 million tests. Gut 2012;61:1439–46. 10.1136/gutjnl-2011-300843. [PMC free article: PMC3437782] [PubMed: 22156981] [CrossRef]
33.
McGregor LM, Bonello B, Kerrison RS, Nickerson C, Baio G, Berkman L, et al. Uptake of bowel scope (flexible sigmoidoscopy) screening in the English national programme: the first 14 months [published online ahead of print 20 September 2015]. J Med Screen 2015. 10.1177/0969141315604659. [PubMed: 26387824] [CrossRef]
34.
Lyratzopoulos G, Neal RD, Barbiere JM, Rubin GP, Abel GA. Variation in number of general practitioner consultations before hospital referral for cancer: findings from the 2010 National Cancer Patient Experience Survey in England. Lancet Oncol 2012;13:353–65. 10.1016/S1470-2045(12)70041-4. [PubMed: 22365494] [CrossRef]
35.
McBride D, Hardoon S, Walters K, Gilmour S, Raine R. Explaining variation in referral from primary to secondary care: cohort study. BMJ 2010;341:c6267. 10.1136/bmj.c6267. [PMC free article: PMC2995017] [PubMed: 21118873] [CrossRef]
36.
Raine R, Wong W, Scholes S, Ashton C, Obichere A, Ambler G. Social variations in access to hospital care for patients with colorectal, breast, and lung cancer between 1999 and 2006: retrospective analysis of hospital episode statistics. BMJ 2010;340:b5479. 10.1136/bmj.b5479. [PMC free article: PMC2806941] [PubMed: 20075152] [CrossRef]
37.
Hippisley-Cox J, Pringle M. Inequalities in access to coronary angiography and revascularisation: the association of deprivation and location of primary care services. Br J Gen Pract 2000;50:449–54. [PMC free article: PMC1313721] [PubMed: 10962781]
38.
Cuthbertson SA, Goyder EC, Poole J. Inequalities in breast cancer stage at diagnosis in the Trent region, and implications for the NHS Breast Screening Programme. J Public Health (Oxf) 2009;31:398–405. 10.1093/pubmed/fdp042. [PubMed: 19423544] [CrossRef]
39.
Downing A, Prakash K, Gilthorpe MS, Mikeljevic JS, Forman D. Socioeconomic background in relation to stage at diagnosis, treatment and survival in women with breast cancer. Br J Cancer 2007;96:836–40. 10.1038/sj.bjc.6603622. [PMC free article: PMC2360059] [PubMed: 17311024] [CrossRef]
40.
Forrest LF, Adams J, Rubin G, White M. The role of receipt and timeliness of treatment in socioeconomic inequalities in lung cancer survival: population-based, data-linkage study. Thorax 2015;70:138–45. 10.1136/thoraxjnl-2014-205517. [PubMed: 24923873] [CrossRef]
41.
Taylor E, Jones F. Lost in translation: exploring therapists’ experiences of providing stroke rehabilitation across a language barrier. Disabil Rehabil 2014;36:2127–35. 10.3109/09638288.2014.892636. [PubMed: 24564356] [CrossRef]
42.
Hunter DJ. Role of politics in understanding complex, messy health systems: an essay by David J Hunter. BMJ 2015;350:h1214. 10.1136/bmj.h1214. [PubMed: 25753054] [CrossRef]
43.
Jusot F, Or Z, Sirven N. Variations in preventive care utilisation in Europe. Eur J Ageing 2011;9:15–25. 10.1007/s10433-011-0201-9. [PMC free article: PMC5547316] [PubMed: 28804404] [CrossRef]
44.
Or Z, Jusot F, Yilmaz E. Impact of Health Care System on Socioeconomic Inequalities in Doctor Use. Paris: IRDES; 2008.
45.
Cookson R, Asaria M, Ali S, Ferguson B, Fleetcroft R, Goddard M, et al. A framework for monitoring NHS equity performance – small area analysis of national administrative data from 2004/5 to 2011/12. J Epidemiol Community Health 2015;69:A28. 10.1136/jech-2015-206256.43. [CrossRef]
46.
The English Indices of Deprivation 2010. Department of Communities and Local Government.London, 2011. URL: www​.gov.uk/government​/statistics/english-indices-of-deprivation-2010(accessed 11 April 2016).
47.
Ragin CC. The Comparative Method: Moving Beyond Qualitative and Quantitative Strategies. Berkeley, CA: University of California Press; 1987.
48.
Rihoux B, Lobe B. The Case for Qualitative Comparative Analysis (QCA): Adding Leverage for Thick Cross-Case Comparison. In Byrne D, Ragin CC, editors. The SAGE Handbook of Case-Based Methods. London: Sage Publications; 2009. pp. 222–43. 10.4135/9781446249413.n13. [CrossRef]
49.
Pawson R, Tilley N. Realistic Evaluation. London: Sage Publications; 1997.
50.
Blackman T, Dunstan K. Qualitative comparative analysis and health inequalities: investigating reasons for differential progress with narrowing local gaps in mortality. J Soc Policy 2010;39:359–73. 10.1017/S0047279409990675. [CrossRef]
51.
Virtanen M, Nyberg ST, Batty GD, Jokela M, Heikkila K, Fransson EI, et al. Perceived job insecurity as a risk factor for incident coronary heart disease: systematic review and meta-analysis. BMJ 2013;347:f4746. 10.1136/bmj.f4746. [PMC free article: PMC3738256] [PubMed: 23929894] [CrossRef]
52.
Thomas J, O’Mara-Eves A, Brunton G. Using qualitative comparative analysis (QCA) in systematic reviews of complex interventions: a worked example. Syst Rev 2014;3:67. 10.1186/2046-4053-3-67. [PMC free article: PMC4079172] [PubMed: 24950727] [CrossRef]
53.
Katz JN. Patient preferences and health disparities. JAMA 2001;286:1506–9. 10.1001/jama.286.12.1506. [PubMed: 11572745] [CrossRef]
54.
Willems S, De Maesschalck S, Deveugele M, Derese A, De Maeseneer J. Socio-economic status of the patient and doctor-patient communication: does it make a difference? Patient Educ Couns 2005;56:139–46. 10.1016/j.pec.2004.02.011. [PubMed: 15653242] [CrossRef]
55.
Karnieli-Miller O, Eisikovits Z. Physician as partner or salesman? Shared decision-making in real-time encounters. Soc Sci Med 2009;69:1–8. 10.1016/j.socscimed.2009.04.030. [PubMed: 19464097] [CrossRef]
56.
Hawe P. Lessons from complex interventions to improve health. Ann Rev Public Health 2015;36:307–23. 10.1146/annurev-publhealth-031912-114421. [PubMed: 25581153] [CrossRef]
57.
Cutler DM. Equality, efficiency, and market fundamentals: the dynamics of international medical-care reform. J Econ Lit 2002;40:881–906. 10.1257/jel.40.3.881. [CrossRef]
58.
Tuohy CH. Reform and the politics of hybridization in mature health care states. J Health Polit Policy Law 2012;37:611–32. 10.1215/03616878-1597448. [PubMed: 22466051] [CrossRef]
59.
Newhouse JP, Insurance Experiment Group. Free for All? Lessons from the RAND Health Insurance Experiment. Cambridge, MA: Harvard University Press; 1993.
60.
Evans RG. Public health insurance: the collective purchase of individual care. Health Policy 1987;7:115–34. 10.1016/0168-8510(87)90026-1. [PubMed: 10312072] [CrossRef]
61.
Evans RG, Lomas J, Barer ML, Labelle RJ, Fooks C, Stoddart GL, et al. Controlling health expenditures – the Canadian reality. N Engl J Med 1989;320:571–7. 10.1056/NEJM198903023200906. [PubMed: 2492637] [CrossRef]
62.
Evans RG, Barer ML, Hertzman C. The 20-year experiment: accounting for, explaining, and evaluating health care cost containment in Canada and the United States. Ann Rev Public Health 1991;12:481–518. 10.1146/annurev.pu.12.050191.002405. [PubMed: 1904728] [CrossRef]
63.
Brook RH, Keeler EB, Lohr KN, Newhouse JP, Ware JE, Rogers WH, et al. The Health Insurance Experiment: A Classic RAND Study Speaks to the Current Health Care Reform Debate. Santa Monica, CA: RAND Corporation; 2006.
64.
Organisation for Economic Co-operation and Development (OECD). OECD Health Statistics 2015. Paris: OECD; 2015.

List of abbreviations

CI

confidence interval

FNP

Family Nurse Partnership

GP

general practitioner

HR

hazard ratio

IMD

Index of Multiple Deprivation

PMB

post-menopausal bleeding

QCA

Qualitative Comparative Analysis

RAWP

Resource Allocation Working Party

SEC

socioeconomic circumstances

Declared competing interests of authors: None

This essay should be referenced as follows: Raine R, Or Z, Prady S, Bevan G. Evaluating health-care equity. In Raine R, Fitzpatrick R, Barratt H, Bevan G, Black N, Boaden R, et al. Challenges, solutions and future directions in the evaluation of service innovations in health care and public health. Health Serv Deliv Res 2016;4(16). pp. 69–84.

Copyright © Queen’s Printer and Controller of HMSO 2016. This work was produced by Raine et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Included under terms of UK Non-commercial Government License.

Bookshelf ID: NBK361257DOI: 10.3310/hsdr04160-69

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (2.2M)

Other titles in this collection

Related information

  • PMC
    PubMed Central citations
  • PubMed
    Links to PubMed

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...