Reducing health inequalities through general practice: a realist review and action framework

Article history

The research reported in this issue of the journal was funded by the HSDR programme or one of its preceding programmes as project number NIHR130694. The contractual start date was in January 2021. The final report began editorial review in September 2022 and was accepted for publication in February 2023. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HSDR editors and production house have tried to ensure the accuracy of the authors’ manuscript and would like to thank the reviewers for their constructive comments on the final manuscript document. However, they do not accept liability for damages or losses arising from material published in this manuscript.

Copyright statement

Copyright © 2024 Gkiouleka et al. This work was produced by Gkiouleka et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2024 Gkiouleka et al.

Research context: interventions aiming to reduce health inequalities in general practice

The last systematic review of the evidence on health service interventions that can reduce inequalities was published more than two decades ago. 38 The study concluded that health services either alone or through sharing initiatives with other agencies can implement interventions to reduce health inequalities. Characteristics of successful interventions included a systematic, intensive and multidisciplinary approach, and enhanced access and service utilisation, addressing the needs of the target populations and community involvement. Since then, the evidence base has expanded, and the healthcare landscape has changed substantially.

A significant milestone was the development of the PROGRESS-Plus conceptual framework for the understanding of inequality based on multiple criteria. 39 Proposed by Campbell and Cochrane Collaborations, the framework encouraged researchers to adopt a more systematic approach to equity by considering multiple social categories that stratify social and health opportunities and outcomes. The criteria include place of residence; race/ethnicity/culture/language; occupation; gender/sex; religion; education; socio-economic status; social capital as well as social categories associated with discrimination (e.g. disability and age); features of relationships (e.g. smoking parents); and time-dependent relationships (e.g. being a carer). Attwood and colleagues explored equity in primary-care-based physical activity interventions across PROGRESS-Plus criteria. 40 They found that the majority of randomised controlled trials (RCTs) of physical activity interventions in primary care do record information on selected PROGRESS-Plus criteria. However, only a few of them finally report details of relevant analyses to determine which groups may benefit or be further disadvantaged by such interventions. 40 Moreover, Terens and colleagues41 reviewed trials of quality-improvement interventions aimed to reduce health inequities among people with diabetes in primary care and explored the extent to which experimental studies addressed and reported equity issues. Among the reviewed studies almost all reported the age, gender/sex and race distribution of participants. Most of them also reported on at least one additional PROGRESS-Plus criterion, mostly education and income. Finally, although only a few studies examined differential intervention effects by PROGRESS-plus criteria, they showed that a quality-improvement strategy improves care for older and less educated people and African Americans. 41

Another review of the evidence on the impact of CVD primary prevention on health inequalities42 showed that adopting a high-risk approach (i.e. screening and treating high-risk individuals only) increases socio-economic inequalities in screening, healthy diet advice, smoking cessation, statin and anti-hypertensive prescribing, and adherence. The authors stressed that interventions targeted at high-risk patients usually focus on individual action either of patients or healthcare professionals. Instead, they suggested a shift towards an approach that integrates policy and population exposure to risk factors. These findings point to what we call IGIs. 25 Such inequalities emerge when public health interventions, although benefiting health overall, increase health inequalities by disproportionately benefitting socio-economically privileged groups or unintentionally harming disadvantaged groups. 25 Interventions with such unintended aggravating impact on health inequalities tend to be those that require considerable individual material or psychosocial resources without accounting for pre-existing inequalities in such resources across socio-economic groups. 25,26,40 In contrast, a review study on the impact of shared decision-making in primary and secondary care showed that relevant interventions reduce socio-economic inequalities by particularly benefitting disadvantaged groups through increased knowledge, informed choice, participation and reduced decisional conflict. 43

Despite the significance of the findings, the evidence regarding interventions that can effectively reduce health inequalities in general practice is still limited and disparate. Most of the available evidence reviews focus on research trials with often small samples and variant study quality. Further, they often exclude interventions that target the SDH and/or IGIs. Acknowledging these gaps, our study provides up-to-date, extended and integrated evidence on aspects of care and interventions in general practice that decrease inequalities in health and healthcare. Also, it identifies which groups are more likely to benefit and, finally, suggests a framework for the design and delivery of equitable general practice.

Social context: general practice in the times of the COVID-19 pandemic

General practice in the UK consists of 6000–7000 small- to medium-sized businesses that are contracted by the NHS to provide a programme of core services. 44 They are primarily funded through a mixture of capitation via a contractual reimbursement formula (commonly known as the Carr-Hill Formula), nationally directed financial incentives and provision of national or local enhanced services (e.g. immunisation, minor surgery). In addition, practices can receive income from a range of other activities, such as dispensing medication, participating in research and undertaking teaching and training. 45,46 In 2019, Primary Care Networks (PCNs) were established and incentivised practices to form around local populations of 30,000 to 50,000 patients. 47 The workforce within general practice has become increasingly multidisciplinary, with the incentivisation of additional roles provided through PCNs; these include social prescribers, physician associates, paramedics and pharmacists. 48 Most practices are run and owned by a group of two or more GPs called a partnership. However, several practices are owned by single-handed GPs or large limited multi-national companies. Decisions are generally made by the partnership usually in collaboration with a practice manager, who is sometimes part of the partnership. 44 The partnership model is being increasingly debated because of financial and time commitments coupled with a workforce and workload crisis. 49

The pandemic occurred when general practice as a sector was already affected by shortages in workforce, substantial workloads and funding pressures. 50–52 Evidence shows that these problems have been affecting practices in deprived areas more, with fewer GPs, direct patient care (DPC) staff and paramedics per 10,000 patients employed in disadvantaged areas31 and funding which does not sufficiently account for deprivation levels. 53 In the aftermath of the COVID-19 pandemic, general practice has been faced with an immense backlog of routine long-term condition care due to the redirection of workforce and resources to the vaccination programmes. 54,55 Falling numbers of qualified full-time GPs and record high numbers of appointment bookings have resulted in an increase of 300 patients per practitioner since 2015. 56,57 In parallel, the systematic demoralisation of the profession in the public discourse during the pandemic58 coupled with remote working conditions59 and increased pressure has left GPs struggling with feelings of frustration and loss in terms of autonomy and control over their working lives. 60 As expected, these challenges are even harder for practices in socio-economically disadvantaged areas with pre-existing shortages in workforce and resources. 31

During these challenging times, there have been also some changes in response to the pandemic which do not immediately look negative or positive but rather warrant a thorough evaluation in terms of their impact on inequalities. One of them has been the increase in remote consultations to limit the transmission of COVID-19. 2 Evidence shows that within a few weeks, GPs achieved at least an 80% increase in remote consultations compared with fewer than 10% before the pandemic. 61 The use of technology enabled this shift and has certainly made GPs and patients more confident to use available telephone services and online tools for the purpose of triage and consultation. 61 However, careful evaluation is needed to assess the extent to which remote services and the use of technology increase access to general practice services and if so for whom. 62 Data show that more than 10 million people in the UK have limited digital literacy and large groups of the population live without consistent access to the internet. 63 At the same time, many minors, women and migrants live in contexts where privacy and safety are not guaranteed and this makes a remote consultation with a healthcare professional often impossible. 1 So, it is likely that a change which was introduced out of necessity and is understood as positive, at least to some extent, could have a worsening impact on some already disadvantaged groups if adopted as a general undifferentiated strategy. A qualitative study on primary care staff working in the most deprived areas of the country during the COVID-19 pandemic stresses how digital poverty and limited IT literacy should be seriously considered in future NHS plans to expand the use of digital care. 64

Finally, during the same period there have also been some examples of positive and promising practice. These concerned mostly an increased sense of community that emerged in several places through the implementation of vaccination programmes and public health initiatives which brought together general practices, local councils, faith leaders, community groups and charity organisations. 65 The ‘Everyone In’ initiative for rough sleepers is such an example: GPs worked together with local councils and other actors to protect those experiencing or being at risk of rough sleeping from the spread of COVID-19. 66 Similarly, vaccination programmes across Crawley, east Surrey and mid-Sussex run by the GP Federation, Alliance for Better Care (ABC), included an explicit orientation towards equity in vaccination uptake, local partnerships with small voluntary sector organisations and charities, and a mentality of working with people to reach populations who have been historically marginalised by healthcare services (e.g. refugees, people in poverty, Gypsy Roma Travellers). 67 The success of such initiatives demonstrates that given the mandate and funding, general practice can play a vital role in achieving public health goals and be a cornerstone of intersectoral collective initiatives for the equitable distribution of SDH like housing, safety and care. 66,67

Together with the negative impact of the COVID-19 pandemic on services’ capacity and the well-being of the healthcare workforce, there are important lessons to be learned from this period in terms of the role of general practice in reducing health inequalities and mitigating their impact on people’s health. Although it seems that we have reached a post-peak pandemic period, the health, social and economic consequences of the pandemic are still experienced by large shares of the population and especially by disadvantaged individuals and groups. 1 Moreover, the latest cost of living crisis aggravates even further the deterioration of life conditions for increasing numbers of people. 68 In this context and against a backdrop of alarming unmet need,55 it is urgent that general practice be supported to play its important role in promoting health for all and protecting the most vulnerable. Currently, there are voices which talk about general practice as a ‘sinking ship’69 but at the same time, others talk about the things we can do to secure general practice’s future as healthier, more equitable and effective for its people and its patients. 60,65 Our work aims to strengthen these latter voices by providing evidence on aspects of routine care and interventions in general practice that can reduce health inequalities in chronic conditions specifically and also inform general practice more broadly in terms of providing equitable care.

Conceptual framework: understanding health inequalities in their complexity

Health inequalities is not a one-dimension concept, and neither is it necessarily understood in the same way by healthcare professionals, researchers and policy-makers. The same applies for the drivers and causes of health inequalities. A recent study by Olivera and colleagues70 suggests that local healthcare systems in England lack a clear conceptual framework for addressing health inequalities and rather use a series of value judgements which are not linked to concrete action. However, consistency and transparency regarding the way we understand the problem and its causes are necessary for suggesting solutions. Thus, this section briefly discusses the different conceptualisations of health inequalities used among public health researchers and professionals, including GPs. Also, we clarify which of these conceptualisations have informed the current study.

Review questions and objectives

Our study aims to answer the following research questions:

What types of interventions and aspects of routine care in general practice decrease or increase inequalities in healthcare and outcomes among people with or at risk of CVD, cancer, diabetes and/or chronic obstructive pulmonary disease (COPD)?

For whom do these interventions and aspects of care work best, why, and in what circumstances?

Our main objective is to synthesise this evidence to produce specific guidance and a framework for healthcare professionals and decision-makers at a local and national level about how best to tackle health inequalities in general practice services.

Expert panel

An expert panel was recruited for the EQUALISE (Reducing Health Inequalities through General Practice: A Realist Review) study to provide content expertise for the development and refinement of the programme theory. The panel consisted of 12 experts including academics, clinicians (recruited mainly through regional clinical research networks) and patient representatives and met twice: once in March 2021 at the stage of developing the initial programme theory (IPT), and once in May 2022 for the refinement of the results and programme theory. Additional communications with some of the expert panel members took place via e-mails throughout the study for the exchange of relevant information or material. The two meetings were held virtually on the Microsoft Teams platform due to COVID-19-related restrictions and lasted for 2 hours each. More details about the meetings are available in Table 1.

Ahead of the meetings, we provided the participants with an agenda and preparatory material (e.g. the aim and objectives of the study, progress update). The meetings started with a brief slide-presentation by our team to introduce the participants to the topic(s) of the discussions. The discussions focused on the relevance and effectiveness of the programme theory and were facilitated by members of our team (AG, JF or GW) to ensure that all attendants would have the time and space to share their input. Notes from the meetings were used to inform the progress of the review and refine the programme theory. The expert panel members who required a reimbursement for their organisations were offered the appropriate payment according to the University’s guidance.

Patient and public involvement

Patient and public involvement (PPI) was strong and consistent across all the steps of the study. The project team included three regular patient representatives, AM, SM and RH. All of them were recruited through pre-existing connections with the research team, they contributed to research design and the writing of the research funding proposal, and participated (all or in rotation) in every project meeting. Additionally, patient representatives recruited through the Healthwatch and the Patients Association participated in the expert panel (n = 2) and the deliberative workshop (n = 2). All the patient representatives are women involved in patient and carer advocacy coming from the East of England. They are affected by social disadvantage that cuts across gender, ethnic minority status, age, religion, neurodiversity, socio-economic position and being a carer. AG led the PPI component of the review. At the beginning of the project, she invited the three regular PPI members to attend an initial meeting during which they discussed the scope and terms of their involvement. At the end of that initial meeting the PPI members signed a letter of engagement to be used as a point of reference and mutual commitment. At least one of the regular PPI members attended the monthly project meetings. Before each project meeting, AG contacted PPI members separately to address any potential questions or issues related to the meetings’ agenda.

Two additional separate meetings for the regular PPI members took place in November 2021 (virtual) and in March 2022 (face to face). In the meeting held in November, AG and the PPI members discussed the process of data extraction and coding with some examples from the included literature. The meeting lasted for almost an hour and at the end the PPI members agreed to contribute to the coding and data-extraction process. The PPI members were sent five papers each, which they coded on their own time and then sent their codes and thoughts on the process to AG via e-mail. This enabled AG to expand and enhance her scope during the coding and data-extraction process and also provided the PPI members with an insight into the data-extraction process, which was necessary for a better understanding of the included literature, the review process and the evidence synthesis. The meeting held in March 2022 was a day workshop that took place at the Cambridge Biomedical Campus. Two of the regular PPI members attended together with AG. The focus of the day was the relevance of the findings and the resulting suggestions for practice. The PPI members had been sent the relevant material in advance and had the opportunity to share their thoughts and feedback during their meeting. The meeting took place in an informal setting and functioned as an opportunity for AG to express her appreciation of the PPI contribution to the project. AG used the notes of the meeting to refine the findings by integrating the PPI members’ suggestions. All the PPI members (regular or not) were reimbursed for their contribution to the project, including meeting attendance and preparatory work, on a set hourly rate (£25) according to the Cambridge University guidance and were reimbursed for any other costs related to their involvement to the project (e.g. travel expenses).

Deliberative workshop

The deliberative workshop took place in May 2022 at the Pitt building in Cambridge. It included health professionals recruited mainly through regional clinical research networks and specifically GPs (n = 4), social prescribers and care coordinators (n = 2), patient representatives (one member of the regular PPI group and two additional PPI representatives), one well-being coach, and academics (n = 3) together with three members of the research team (AG, JF and GW). The aim of the workshop was the discussion of the study findings and the suggested framework and practice qualities for equitable general practice. The meeting was facilitated by a professional facilitator. It involved a presentation of the study and findings by members of the research team (AG and JF) at the beginning, and three subsequent sessions focusing on the relevance and effectiveness of findings, missing elements from the results, and suggested ways to translate the findings into practice. During every session, the participants worked in small groups and added their collective reflections and comments in flipchart sheets provided by the facilitator. At the end of every session each working group shared their input and there was an open discussion among all the participants. At the end of the day, all the sheets with participants’ notes were collected by the facilitator and given to AG, who integrated them into a summary document. The document informed a discussion on the refinement of results held by AG and JF.

Locating existing theories

Our first step was to identify the key theories that explain how, for whom, why and in what circumstances interventions or care delivered in general practice may increase or decrease health inequalities. We achieved this through three inter-connected processes. (1) We conducted an exploratory background search to get familiarised with the relevant literature (including literature on IGIs). This search was deliberately conducted using informal methods (i.e. snowballing and citation tracking)103 to ensure that the scope would be adequately broad. (2) We organised a panel discussion with key content experts (details on the panel and the participants are available in Appendix 1) during which we had the opportunity to explore relevant theories and explanations. (3) We refined our findings from the literature review and the experts’ input through iterative discussions within the project team. This process revealed that the elements involved in the production of health inequalities in general practice cover a broad range of interventions, areas of interest and levels of analysis (e.g. individual or practice level), and the literature mostly focuses on outcomes (e.g. patterns of inequalities in accessing services) rather than driving mechanisms. We used our discussions within the expert panel and the research team to reflect on the potential mechanisms that may link elements of the general practice context with certain outcomes, and came up with an IPT. The programme theory reflected the breadth of the relevant elements of context, mechanisms and outcomes, and included a series of interventions that have the potential to increase or decrease inequalities. The interventions were focused on either context or mechanisms. The programme theory served as a theoretically informed evaluative framework to inform our formal literature search and evidence collection. 99 Figure 2 shows the IPT and the different areas which were later populated with evidence. The areas of interest were not organised at that stage beyond CMO groupings but, as reflected in Figure 2, they covered different domains from individual characteristics like staff beliefs, skills and knowledge to structural factors like GP distribution, funding and power hierarchies.

FIGURE 2.

Initial programme theory.

Search strategy

Our search strategy was set with the guidance of an experienced librarian (IK). Given the broad scope of our research questions, we decided to run an initial search of systematic reviews of interventions delivered in general practice and focused on CVD, cancer, diabetes and/or COPD. Driven by the IPT, we undertook electronic searches of MEDLINE, EMBASE, CINAHL, PsychINFO, the Web of Science and the Cochrane Library. The original search was undertaken on 7 April 2021, and it was updated on 23 March 2022. More details on our search strategy are available in Appendix 2. We decided that our search scope should include studies focusing on contexts outside the UK to account for contextual differences and identify the circumstances under which mechanisms can lead to specific outcomes.

All the identified titles and abstracts of reviews were screened for eligibility by AG and 20% were independently screened by JF to check for systematic errors. Wherever there was a disagreement, it was solved through discussion. The screening was conducted using Rayyan. 104 The inclusion and exclusion criteria for reviews were:

Inclusion criteria:

• reviews that used a comprehensive search strategy and appropriate quality-appraisal tool;

• reviews of interventions or care which targeted CVD, cancer, diabetes or COPD or their risk factors, namely, smoking, hypertension, diet, exercise and cholesterol;

• reviews of interventions or care delivered in general practice;

• reviews including clinical outcomes relating to the type of care (acute, chronic and preventive), function of care (diagnosis, screening and prevention, follow-up and continuity, treatment) and domain of care (effective, efficient, timely, patient-centred and safe);105

• reviews undertaken in high-income countries, as defined by the Organisation for Economic Co-operation and Development (OECD), with no language restrictions.

In a slight deviation from our published protocol,102 we decided to include review studies regardless of whether they were reporting on the differential effectiveness of care/intervention across groups or interventions aimed at disadvantaged groups. Given that at the stage of developing the programme theory we found that driving mechanisms of health inequalities are less discussed in the literature, we thought that broader inclusion criteria for the reviews could increase our chances of accessing relevant literature which could be useful for answering our research question.

Exclusion criteria:

• reviews of studies focusing on drug effectiveness or efficacy;

• reviews of interventions focusing on children, mental health or disease areas not listed above;

• reviews superseded by more recent reviews.

As a next step, two researchers (AG, RT) extracted all the primary studies included in the systematic reviews which met our inclusion criteria. One researcher (AG) screened all the titles and abstracts and a second researcher (JF) screened 5% of the articles to check for systematic errors. The inclusion and exclusion criteria for the primary studies were:

Inclusion criteria:

• interventions which reported on clinical or care-related outcomes by socio-economic group, or other PROGRESS-Plus criteria, relating to the type of care (acute, chronic and preventive), function of care (diagnosis, screening and prevention, follow-up and continuity, and treatment) and domain of care (effective, efficient, timely, patient-centred and safe);

• studies on interventions targeted at disadvantaged groups;

• general practice interventions or care targeting CVD, cancer, diabetes or COPD or their risk factors, namely, smoking, hypertension, diet, exercise and cholesterol;

• interventions focusing on clinical outcomes (e.g. mortality and myocardial infarction) or clinical measures of risk factors (e.g. change in blood pressure) or behavioural outcomes (e.g. physical activity) or care-related outcomes (e.g. screening uptake);

• studies undertaken in high-income countries, as defined by the OECD, with no language restrictions.

Exclusion criteria:

• studies focusing on drug effectiveness or efficacy;

• studies focusing on children, mental health or disease areas not listed above.

Again, shifting slightly from our published protocol102 instead of including only studies with an experimental study design, we decided to include studies of other designs (e.g. qualitative studies or surveys) in order to increase our possibilities to access data about interventions that target the SDH or IGIs.

Article selection

As per realist methodology, documents were selected according to the extent that they could contribute to the development and refinement of the programme theory – that is, they contain relevant data. 100 During screening the full text of the included primary studies, AG grouped the studies into six groups according to their focus and the country the study took place. The six groups were then classified in terms of relevance in the following order: (1) studies which focused on inequalities in the UK were deemed of the highest relevance, followed by (2) studies discussing interventions targeted at disadvantaged groups in the UK, (3) studies on interventions in the UK without an inequality focus, (4) studies on inequalities outside the UK, (5) studies on interventions targeted at disadvantaged groups outside the UK and (6) studies on interventions outside the UK without an inequality focus. We anticipated that studies focusing on settings outside the UK were likely to be less relevant but were selected for data extraction when we judged that they included information transferrable to the UK context.

Quality-assessment checklist criteria were not used as per realist methods,99 rather the rigour of the extracted data was taken into account during the coding and synthesis phase. Given that the literature did not provide rich information on inequality-producing mechanisms, we considered that even if studies were judged to be of limited rigour, they could still be used for the data synthesis if they contributed to the refinement of the programme theory. 100 To conduct the review within the study timeline, the selection of articles took place at the same time as the data extraction to speed up the process. This meant that AG uploaded the included studies by priority group in QRS NVivo (NVivo qualitative data analysis software, version 12; QSR International, Warrington, UK)106 and coded relevant useful data. When no useful data were available in the article, the article was then removed from the NVivo sources list. The data extraction stopped when sufficient iteration was achieved.

Data extraction and organisation

Key characteristics of all the included studies including the country the study took place, condition of focus, dimension of inequality (e.g. socio-economic status or other PROGRESS-Plus criteria), domain and form of intervention were extracted using an Excel spreadsheet by AG and RT. More details on these data are available in Figure 4 and Table 2.

Next, AG coded the data with feedback on the process by members of the research team (JF, GW) and the PPI group (AM, RH, SM). In the beginning of the data-coding process, a random sample of approximately 5% was additionally and independently coded by JF to check for systematic errors. Shifting slightly from the published protocol which mentioned the independent coding of a random sample of 10%, JF and AG discussed codes regularly, while another random sample of approximately 5% was independently coded by the three regular members of the PPI group to ensure the breadth and accuracy of the scope of the data extraction process.

Coding was based on the following two questions:

1. Does the text refer to any of the elements included in the IPT?

2. Does the text refer to the unequal effectiveness of care services or interventions?

Codes were chosen with an open-minded approach to reflect the themes that were emerging from the data, they were inductive (i.e. created to categorise data reported in included studies), deductive (i.e. ones that came from the IPT) or retroductive (i.e. created based on an interpretation of data to infer what the hidden causal forces might be for outcomes). They were refined regularly throughout the data analysis and were finally organised across 14 broader themes: namely, access to care, communication, community engagement, competing priorities, cultural understanding, differences between practices, interprofessional cooperation, patient education and behaviour change, patient enablement, patient perceived risk, resources distribution, the role of the GP in intervention success, time constraints, and workforce. We also created two additional categories for codes referring to theories of interventions and descriptive codes that did not refer to processes of inequalities.

Data synthesis

The data were synthesised by AG with the input of the research team (JF, GW, RH, AM) based on a realist logic and in light of the IPT. Having organised our coded data in themes, we worked on forming CMOCs (i.e. statements of causal relationships)98 within and also across themes. Given the diversity of the included articles in terms of studied condition, intervention type, outcome and PROGRESS-Plus criteria, our synthesis aimed to elicit common patterns and generalisable messages. Therefore, instead of specific characteristics of interventions, we focused on the underlying principles that informed care and interventions and the ways they can be employed to achieve equitable care in general practice. To meet this goal, CMOCs (see Figures 5–9) were refined and abstracted to a higher level so that they would enable us to understand what the broader qualities of care and interventions are that are likely to decrease or increase inequalities in general practice. This entailed using where necessary some or all of the following questions to inform the interrogation of our data:107

Relevance:

• Are the data included in an excerpt of an included document relevant to programme theory development?

Judgements about trustworthiness and rigour:

• Are the data sufficiently trustworthy to warrant making changes or additions to the programme theory?

Interpretation of meaning:

• If the answer to the previous questions is yes, can these data be interpreted as being elements of context, mechanism or outcome?

Interpretations and judgements about CMOCs:

• What is the CMOC (partial or complete) for the specific data?

• Are there more data in this or in other documents to inform CMOCs contained in this document?

• How does this CMOC relate to those already developed?

Interpretations and judgements about programme theory:

• How do CMOCs relate to programme theory?

• Are there data within and across documents that inform how specific CMOCs relate to the programme theory?

• In light of the CMOCs and the supporting data, does the programme theory need to be changed?

Summarising, the evidence synthesis was implemented by the following processes:107–109

1. juxtaposition of data sources: comparing and contrasting data available in different sources;

2. reconciling ‘contradictory’ or disconfirming data: finding explanations about why outcomes differ in apparently similar circumstances;

3. consolidation of sources of evidence: defining whether similarities between findings presented in different sources are adequate to form patterns in the development of CMOCs and programme theory.

Details of included studies

We identified 7998 review studies, of which 251 met the inclusion criteria. From the included reviews, we retrieved 6555 primary studies and proceeded with a second round of screening. In total, 325 studies met the inclusion criteria for primary studies. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) diagrams for the identification of reviews and the review of primary studies are available in Figure 3.

FIGURE 3.

PRISMA diagrams. Our PRISMA diagrams do not report reasons for exclusion of studies because often studies were excluded for multiple reasons. Our diagrams are consistent with Realist And Meta-narrative Evidence Syntheses: Evolving Standards (RAMESES) publication standards for realist syntheses. 98

The included primary studies (n = 325) covered a period of approximately 30 years, from 1989 until 2021, and most of them were conducted in the USA (n = 143) and the UK (n = 102). The studies were grouped into three categories: those focusing primarily on inequalities (n = 56), those focusing on an intervention or an aspect of care targeted at specific disadvantaged groups (n = 137) and those focusing on an intervention without an inequality focus but reporting results by at least one PROGRESS-Plus criterion (n = 132). The studies also involved a wide range of designs, with almost half of them being RCTs or other experimental design (n = 157). Most of the studied interventions concerned diabetes (n = 114), followed by studies on cancer (n = 77) and CVD (n = 49). The reported outcomes included clinical outcomes (n = 139), service uptake (n = 88), care quality (n = 73) and behavioural outcomes (n = 25). The most-studied type of intervention was screening (n = 41), followed by interventions with multiple components (e.g. tailored information material and counselling) (n = 34) and educational interventions (n = 29). In most studies, more than one PROGRESS-Plus criteria were addressed. Ethnicity was addressed most often (n = 145), together with sex or gender (n = 98) and/or socio-economic position (n = 99, measured by proxies of socio-economic status, occupational status, social capital or social class), and/or age (n = 96) and/or geography (n = 38). Other PROGRESS-Plus criteria including education, migration, disability, language and HIV stigma were addressed on their own or in combination with other criteria only in 25 studies. More detailed information about the included studies is available in Figure 4 and Table 2.

FIGURE 4.

Studied interventions.

Evidence synthesis

Our data extraction and organisation resulted in the final selection of 159 articles (details available in Appendix 3) and a series of CMOCs (n = 21) covering a broad range of topics, interventions and affected groups. In line with the IPT, our data revealed that inequalities in general practice emerge through complex processes that take place in different domains, for example, in the clinical encounter between physicians and patients but also during communication between a local practice and the population it serves, and in the process of funding allocation. What we also noticed was that although interventions seemed to focus on one aspect of care (e.g. invitations to screening programmes), the intervention effectiveness was affected by different aspects (e.g. the availability of patient contact details or the extent to which patients could engage with written material). This stressed how inequality-producing processes are not only complex but also inter-related and, at the same time, made structuring and organising our evidence a complicated task.

We realised that, on the one hand, organising our evidence vertically across levels (e.g. individual, practice, macro-level) would not allow us to effectively capture how inequality-producing mechanisms cut across levels. On the other hand, organising our findings across aspects of care (e.g. service accessibility, care quality, care outcomes) would hinder our ability to produce transferrable conclusions. To deal with this puzzle, we had to return to our conceptualisation of health inequalities and the aims of our study. As stressed earlier, we understand health inequalities as the result of power imbalance across a series of intersecting hierarchies cutting across the individual and structural levels. What we aim is to contextualise individual action and aspects of care within the power organisation to enable people working in general practice to understand the links between structural factors and their practice, and engage in meaningful action against health inequalities. Therefore, we decided that the most efficient way to organise our findings was by building on the way power is organised. To do this, we drew upon intersectional understandings of power organisation, and specifically on the Domains of Power Framework. 110

The framework, developed by the sociologist P.H. Collins, suggests that for each power hierarchy (i.e. class, gender or race), power is organised through four interrelated domains: the structural domain, which involves social structures like public policy, laws, religion, economy, and relevant institutions; the disciplinary domain, which involves all the bureaucratic procedures which aim to control and organise behaviour (e.g. protocols, routines, rules and regulations used in everyday life); the cultural domain, which involves the dominant ‘worldviews’ as well as narratives and counter-narratives regarding inequalities, their causes, and solutions, language, ideas, images and values; the interpersonal domain, which involves the experience of individuals within the power structure, including relationships and interactions of the everyday life. The framework does not promote the significance of one domain over the others but highlights the importance of their synergy in the organisation of power and the emergence of inequality in any social context. 110,111

In the following sections we present the CMOCs that emerged from the reviewed literature organised across the four power domains. We consider that this is a useful framework because, while it allows us to understand processes and their resulting outcomes in specific domains, at the same time by highlighting the synergy between the domains it captures the complexity of inequalities and the action to reduce them within general practice.

Structural domain – structures and politics

Funding of general practice

A key structural factor is the distribution of funding, including financially incentivised quality-assessment strategies, such as the Quality and Outcomes Framework (QOF) and contractual reimbursement using the Carr-Hill formula. QOF was introduced in 2004 as a system which links financial incentives with the quality of care in general practice. 112,113 It aimed to improve the overall quality of care provided and decrease variations of care across practices by resourcing and rewarding best practice. Evidence shows that in the UK inequalities in the quality of the provided care related to deprivation decreased during the first 3 years of QOF. During this period an additional budget of £2 billion was allocated to primary care, which was associated among other things with a 12% increase in the number of GPs and the recruitment of additional administrative and nursing staff. During its first 3 years, QOF seemed to reduce inequalities in the delivery of clinical care between less and more deprived patients through a standardisation of care (see Figure 5, CMOC 1). 114,115 However, later studies suggest that this impact has not been sustained across time116 and, despite the improvements in quality of care, inequalities across age, ethnicity and socio-economic deprivation persist. 29,117,118

FIGURE 5.

CMOCs in the structural domain.

Moreover, a series of studies show that a structural aspect of QOF programmes that is relevant to health inequalities concerns the prioritisation of tasks and activities in standard care (see Figure 5, CMOC 2). 29,115,119–122 There is evidence suggesting that financial incentives can effectively drive physicians to conduct certain activities. 117 For example, after the introduction of QOF, recording of smoking status increased from 55.5% to 64.3% for men (p < 0.001) and from 67.9% to 75.8% for women (p < 0.001), which linked with improvements in smoking outcomes. 117 However, within the public health domain of QOF, almost 75% of the incentivised indicators are aimed at patients who already suffer from smoking-related diseases (secondary prevention) and only 25% of the incentivised indicators are aimed at primary prevention for people who may be at risk but have not yet developed a smoking-related disease. 115 In the context of local practices, where workload is usually high and resources limited, this imbalance interferes with the prioritisation of the incentivised activities over those not incentivised. 95–98 In the case of prevention, this can maintain or even increase inequalities in preventive service utilisation. This is because poorer and socially disadvantaged people tend to use less preventive healthcare when they don’t face immediate pain or disability. 123,124 In later sections, we discuss how disciplinary aspects of the QOF also link with inequalities against socially and economically disadvantaged groups.

Moreover, the data indicate that funding allows local general practices to increase their capacity through the recruitment of additional staff (see Figure 5, CMOC 3). 125–132 One of the included studies126 shows that practices with higher numbers of full-time equivalent GPs per 1000 patients record higher numbers of patients in their hypertension registers. The authors estimated that an extra GP per thousand patients would be associated with 6% increase in detected hypertension and a 33% increase in the patients who would be able to get an appointment quickly. 126 Another study set in a single practice in an urban area of extreme socio-economic deprivation in Glasgow128 showed that funding for the appointment of an additional part-time salaried GP for five sessions per week allowed the increase of consultation time for patients with complex needs, which in turn was associated with higher patient enablement. In addition, there is evidence that patients from practices with fewer medical staff have higher haemoglobin A1c concentrations,129 while smaller practices are generally found in deprived or remote areas. 130,131 Importantly, even after the introduction of the QOF poorly performing practices are still concentrated in more deprived areas. 119,125

Service convenience and inequalities in social determinants of health

The location of services together with the availability of transport options can lead to inequalities in access to general practice through increasing or decreasing service convenience (see Figure 5, CMOC 4). 129,133–140 For example, a study on the quality of general practice in remote and very remote areas in Scotland133 for people with coronary heart disease (CHD), diabetes and stroke showed that despite the higher prevalence of CVD and diabetes, rates of statin prescribing were lower than in urban areas. A study on British Bangladeshi patients with diabetes showed that geographical barriers in the places where people live (e.g. high-rise flats with no working lifts) discourage patients from leaving their home. 138 Evidence from non-UK contexts also shows that transportation plays an important role in the effectiveness of weight-management interventions by enabling attendance. 137 This is particularly relevant for ethnic minority groups with low socio-economic status who encounter additional barriers in accessing services, such as unreliable or unaffordable transportation. 139

What is also alarming is that within disadvantaged patients, women are likely to be even more severely affected by the inconvenience of service delivery due to increased mobility or transportation barriers. The included studies do not offer evidence regarding potential gendered effects of geographical or environmental barriers and time constraints on accessing general practice. However, they highlight the fact that women and especially those belonging to deprived and/or ethnic minority communities are more likely to be financially or practically dependent for their transport on other members of their household, deal with increased time constraints due to increased caring responsibilities, and feel afraid to go out alone. 129,140

Apart from accessing general practice services, inequalities in SDH link with inequalities in the effectiveness of educational interventions aiming at behaviour change and the reduction of behavioural risk factors (e.g. smoking). 129,137,138,141–146 People living in deprived areas or ethnic minorities face practical barriers in their attempts to develop what is called ‘healthy lifestyle’, which probably affects their ability to achieve and sustain their health-related goals (see Figure 5, CMOC 5). 129,137,138,141–146 Studies in the USA146,147 and the UK111,114 highlight the impact of language barriers, insufficient or unaffordable transportation and budget or time limitations, which are only some of the SDHs that interfere with the effectiveness of educational interventions aiming to improve care outcomes among disadvantaged groups. As an answer to these constraints, interventions aiming to encourage and sustain healthy behaviours among African Americans with diabetes in the USA integrate free food prescriptions for healthy foods and free access to exercise facilities acknowledging the importance of SDH. 146,147

Cultural domain – ideas and knowledge(s)

Cultural understanding and tailoring of services

The literature highlights in different ways the importance of cultural understanding in providing effective and equitable general practice services (see Figure 6, CMOC 1). 135,138,139,142,147–153 Many of the studies conducted mostly in the USA discuss interventions focused on the cultural tailoring of provided services to African American or Latino populations. In most cases, tailoring means adjusting interventions or messages so that they align with targeted groups’ cultural norms, health beliefs and lifestyle habits and eventually become more effective. 29,115–120,142 Cultural influences are relevant to how people think about their health and the origins of their disease, what is appropriate in a clinical encounter, and the extent to which they endorse health promotion or treatment activities. 138,154,155 Further, evidence shows that culturally tailored telecommunication systems can help urban African Americans with hypertension to adhere to their medication and evidence-based guidelines for dietary behaviour and physical activity. 139 Tailored education combined with shared decision-making has been found to increase empowerment and self-care among African Americans. 146 Moreover, tailored interactive digital educational material is associated with greater satisfaction with medication information and lower levels of diabetes distress among low-income Latino and African American adults compared to non-tailored educational material. 156,157 Similarly, tailored multicomponent interventions about diabetes increase the sense of control over condition self-management among ethnic minority patients and improve their behavioural outcomes. 150

FIGURE 6.

CMOCs in the cultural domain.

However, cultural tailoring is a multi-component process. 139,158,159 It can involve a surface structure adaptation (i.e. matching intervention materials and messages to observable characteristics of a target population) and/or deep structure adaptation (i.e. developing intervention materials that are aligned with the cultural values and social, historical and psychological forces that affect how health behaviours are viewed). 139 Therefore, it implies much more than translation or name-matching and demands the developing of a deep understanding of patients’ beliefs and needs. Starting from translation, evidence highlights that translation services to be effective need to be chosen and provided carefully158 based on knowledge about the languages used within targeted communities and possible inequalities within these communities in terms of literacy in the language of translation. 159

Research in the UK focuses on what culturally sensitive care and interventions entail. A study on a lifestyle intervention on weight change in South Asian people at high risk of diabetes shows that adopting a family focus was especially important for the success of the intervention in the specific group due to the strong cultural emphasis on family life and collective eating. 158 Another study focusing on interventions to promote physical activity among South Asian women with CHD and diabetes152 highlights that instead of overemphasising barriers of cultural difference, which often promote defeatist attitudes among professionals, health-promotion strategies should focus on a positive framing of what is meaningful for this group. They should also address possible misconceptions around the benefit of exercise, respect for the women’s experience of their ‘body limits’ and an understanding of their own motivation for physical activity. 152

Similarly, a study on the beliefs and folk models of diabetes among British Bangladeshis shows that perceptions about food’s ‘nourishing power’ interfere with what Western science defines as healthy eating (e.g. white sugar or lamb are perceived as energy and health-giving nutrients), and exercise is not a relatable concept nor does it have positive connotations within these groups. 138 The authors conclude that health educational interventions need to align with people’s lay epidemiology and folk models to be effective. In the same line, gender norms are another component of cultural understanding. Often the importance of gender is discussed in relation to religious ethnic minority women, who within the context of Western societies are perceived to be (and in some cases are) sensitive about interacting with men or exposing their bodies in front of others in healthcare contexts or physical activity spaces. 152 However, our data show that what people believe about what constitutes proper male or female behaviour interferes with the effectiveness of their care. 152,153 Characteristic examples include South Asian women believing that exercise is a ‘selfish’ behaviour for a woman who should be caring for her family instead of spending time on herself,152 or Turkish men being less likely to adopt health-promoting behaviours when relevant advice is offered by women educators. 153

To increase the cultural alignment between health professionals and groups who sit outside what we perceive as ‘the norm’, interventions are often based on integrating community members in care either as link workers, peer educators, community health workers or patient navigators. 160,161 It is understood that community members can provide health practitioners with the necessary cultural understanding that the latter may lack. They can also make patient experience smoother in terms of communication and more appropriate in terms of approach. 143 It follows almost naturally that these mechanisms can enhance the trust between patients and general practices, bridge communication gaps and enhance overall care quality. 160,162 For example, it has been found that link workers traditionally working in community outreach and education improve adherence to care plans among Mexican American populations in the USA,150 as well as patients of South Asian origin in the UK. 163 In a similar manner, increasing the concordance between health workers – from physicians to receptionists – and general practice users has also been found to contribute to achieving more culturally appropriate care in various settings. 164,165

Increasing knowledge alone does not decrease health inequalities

Inequalities in care (e.g. accessing screening services) and care outcomes either clinical (e.g. type 2 diabetes) or behavioural (e.g. self-management) are associated with people’s level of knowledge around health and how this knowledge affects their behaviour. 166,167 In the context of public health, it is often assumed that inequalities in health and healthcare result from the fact that socio-economically disadvantaged groups lack health knowledge and literacy compared to more privileged groups who tend to have greater access to education and knowledge resources. 129,168,169 Therefore, educational interventions are often seen as an effective way to decrease inequalities in care outcomes, especially behavioural outcomes like self-management of chronic conditions and adoption of healthy lifestyles. 158,170,171 However, the studies included in this review draw quite a different picture. They highlight that although inequalities in health literacy and knowledge are relevant to health inequalities, educational interventions or training on their own are not enough to close the gap in health outcomes (see Figure 6, CMOC 2). 135,141,151,168 A pilot randomised trial on knowledge, awareness and self-management among South Asians with diabetes in Manchester found that interventions focusing on questionnaires and educational packages do not increase knowledge or self-management behaviours. 141 Additional evidence shows that enhanced diabetes care, including training and educational support, does not necessarily improve outcomes among South Asians in the UK, even though they172 value education and have high regard for knowledge acquired from an ‘educated person’. 135

A study on breast cancer screening among Asian women in the UK suggests that relevant educational interventions should focus on increasing awareness of the purpose and relevance of screening services because limited understanding in these domains reduces attendance in this group. 149 Also, as described earlier, the content and goals of educational interventions should be consistent with the everyday reality of the targeted populations as this is shaped by culture and structural inequalities. 138,150–152 Moreover, studies hint at the fact that the extent to which people perceive themselves as capable of achieving certain goals can also interfere with the effectiveness of educational interventions. 150,173–175 Often in literature, such perceptions are described with the psychological concept of self-efficacy. 176 However, we know that the social position that people occupy within their contexts informs the way they perceive the world, themselves and their capabilities. 90 Disadvantaged groups, beyond practical barriers, are more likely to encounter perceptional barriers in their attempts to adopt healthier behaviours and engage with their care plans. 143,146,154,177 For example, evidence shows that they often tend to have fatalistic beliefs about disease and perceive barriers in accessing services as higher than what they are. 154

Biases among healthcare staff

Evidence suggests that GPs often tend to make decisions in their everyday work routine based on heuristics or mental short-cuts (e.g. that men have a higher index of suspicion for CVD). This can lead to patients deviating from the ‘typical’ clinical picture being excluded from care because of implicit bias (see Figure 6, CMOC 3). 117,165,178–180 Meta-analyses have shown that the risk of cardiovascular mortality is two times higher among men and three times higher among women with type 2 diabetes compared to men and women without diabetes. 181–183 The gender inequality observed is attributed to less effective monitoring of risk factors among women. An observational survey among GPs in Paris178 showed that information regarding CVD risk factors was regularly less available in the files and risk assessment could be performed significantly less often for women seeing male physicians. The authors highlight that although women are generally assessed for CVD risk less often than men, gender inequalities decrease when women patients are seen by women physicians. They attribute these results partly to physicians understanding the guidelines about risk assessment differently according to their patient’s gender (i.e. men are seen as more susceptible to CVD risk), partly to women being more likely to follow guidelines more consistently.

In the UK, findings as such may be particularly relevant for gender inequalities in remote and very remote areas because practices there tend to have a significantly lower number of women GPs. 133 This has also been linked with reduced choice for patients and poor access to some forms of treatment. 133 Additionally, evidence from 14 general practices in London showed that South Asian patients had better blood pressure and total cholesterol recording than white patients, while those with no records for ethnicity had the poorest recorded outcomes. 164 The authors attributed this inequality partially to risk factor recording being considered more relevant in the case of South Asian patients because they are perceived to be at greater risk of CVD.

Disciplinary domain – rules and practices

Excluding patients from quality standards

We have already discussed how certain structural aspects of financial incentive schemes can contribute to the increase of within-practice inequalities. Here, we elaborate on how the way that such schemes are implemented in day-to-day practice is an additional driver of health inequalities in general practice. The included literature focuses on the QOF and suggests that its main caveat concerns the fact that practices are allowed to exclude from the measurement of achievement targets patients who are unsuitable for chronic disease management. 29 According to the national guideline these patients include those who refuse or fail to attend a review, those who should not receive treatment due to comorbidity, those who have been recently registered with a practice, those with drug-related limitations, patients who refuse investigation or treatment, and/or those for whom further investigation or treatment is unavailable or it is inappropriate to attend a review due to specific circumstances (e.g. terminal illness). 29 The purpose of excluding certain patients from these assessments is to avoid penalising practices for the characteristics of the population they serve.

However, the way this exclusion process is used maintains and conceals inequalities in service provision. 29,119–122,125,169,184 This is especially true for diagnostic indicators like foot and eye screening in diabetes and spirometry and record of forced expiratory volume in 1 second (FEV₁) in COPD as well as treatment indicators for CHD, stroke and diabetes which are associated negatively with levels of deprivation. 29 One of the included studies114 shows that performance as measured by the QOF is higher in practices with higher rates of patients excluded from QOF indicators. Specifically, a 1% higher rate of exception reporting was associated with a 0.35% higher rate of achievement in the second year of the framework and a 0.16% higher rate in the third year. However, practices serving the most deprived population had an exception-reporting rate that was 0.55% higher than the rate in practices serving the least deprived population in year 2, while in the third year the difference in the rate reached 0.67%. This means that approximately 630 more people are likely to be excluded in a practice of 9000 patients (which is the average list size for practices in June 202256) in the most deprived areas compared to the least deprived. Additional evidence suggests that exception-reporting rates are higher among disadvantaged groups at higher risk of diabetes complications and poorer health outcomes. 120 This evidence suggests that in more deprived areas there are either more patients who can be formally exempted from QOF indicators (e.g. due to limited attendance or increased comorbidities) or there are more patients who are treated as exceptions so that the practices can maximise their income. 120,169 In other words, it means that patients perceived as less likely to engage or respond to treatment (e.g. due to complex psychosocial problems) are less likely to be provided with ongoing support by their practice due to the effort of the practice to achieve certain performance targets (see Figure 7, CMOC 1). 29,119–122,125,169,184 It is also likely that both explanations can be true at the same time, while evidence on this remains inconclusive.

FIGURE 7.

CMOCs in the disciplinary domain (1).

Regardless of the justification for exceptions, the increased exception rates among disadvantaged groups have two serious implications. The first is that QOF through the way the exception reporting is used in day-to-day practice probably excludes disadvantaged patients from accessing necessary care including preventive and diagnostic services. 185 Importantly, this exclusion is not considered in the overall assessment of service delivery. While evidence shows that the introduction of the QOF narrowed the gap between least and most deprived populations, at least during the first three years of its use in terms of care quality,114,115 it is likely that it has sustained or even increased the gap in terms of unmet need within and between practices. The second implication is that by not accounting for the extra effort that practices in disadvantaged areas need to put to achieve quality targets and maintain income, the framework reinforces the misuse of the exception process and at the same time maintains or even widens inequalities between practices by entrenching the inverse care law. 29

Accurate patient contact details and demographic information in electronic patient records

Additional procedures that increase inequalities especially in terms of service uptake are relevant with the collection, use and maintenance of patient information. Evidence shows this is particularly true for screening services. 127,149 Maintaining accurate patient registers with up-to-date contact details enables effective contact and this links to higher rates of screening uptake and quick delivery of interventions (see Figure 7, CMOC 2). 127,149,164,186,187 However, maintaining accurate patient registries is not equally easy for all patients. Practices often have inaccurate contact details and telephone numbers of people in precarious housing, or those who live in inner city urban areas, migrants and older patients. 127,136,188–190

The optimal collection and use of patient socio-demographic data is also linked with inequalities in care effectiveness (see Figure 7, CMOC 3). 191,192 Studies on the effectiveness of CVD screening and the Health Checks programme in the UK highlight the lack of socio-demographic information especially for patients in more deprived areas. 164,193 A study across 14 general practices in Ealing published in 2010 showed that South Asian patients had better blood pressure and total cholesterol recording than white patients, while those with no records for ethnicity had the poorest data records. 164 As stressed earlier, the authors suggested that this inequality may be attributed to implicit bias that patients of ethnicity other than South Asian are not at high risk of CVD. However, other potential explanations include either that South Asian patients visit their practice more often, or there is greater ethnic concordance between South Asian patients and GPs enabling better exchange of information. This kind of evidence does not allow the effective assessment of the direction and size of inequalities but hints at the fact that in these practices there is an unspecified number of unidentified patients who may be at risk of CVD.

A prospective cohort study from 1993 to 2008 across 531 general practices in England and Wales showed that using a CVD risk algorithm which allowed adjustment for patient level of deprivation, ethnicity and clinical conditions improved the accuracy of identification of those at high risk compared to the Framingham score. 194 The authors suggested that this benefit was particularly relevant for South Asian women and for patients with type 2 diabetes and that it demonstrated the utility of linked electronic data for the development of tools that will enable physicians to make better decisions. The importance of collecting and using sociodemographic characteristics has also been highlighted in relation to the identification of lung cancer at an early stage. 195 Further, two studies on the impact of QOF on intermediate outcomes for diabetes and the quality of chronic disease management stress the importance of integrating patient data in the assessment of interventions and the need for differentiation within broader patient groups. 114,196 Both studies highlight that suboptimal use of patient data, like merging distinct ethnic groups (e.g. Indians, Pakistanis and Bangladeshis) or social classes (e.g. skilled and unskilled workers) into single undifferentiated categories (e.g. South Asians or manual classes), carries the danger of masking inequalities within disadvantaged groups.

Invitation methods to preventive services

Invitation methods to services involve additional bureaucracies through which inequalities in service uptake can increase or decrease. A study from the UK found that invitation letters with predetermined appointments are effective in increasing health checks among patients with severe mental illness and that a reminder letter has also a positive effect among non-attendants (see Figure 7, CMOC 4). 191,197,198 The authors suggest that this increase is driven by the fact that set appointments remove the difficulties entailed in making an appointment, which might be a particularly relevant factor affecting this category of patients due to increased anxiety levels and fluctuations of symptoms. In contrast, if invitations to screening programmes are sent in the form of extensive written material, they are likely to be dismissed and overlooked by disadvantaged patients and especially those with limited language or literacy skills. It is important to note that this is true even if the information is in plain everyday language and it applies whenever the communication between a practice and its patients is based on written material. 197,198

Using prompts and reminders also has a positive impact on screening uptake especially for disadvantaged groups (see Figure 7, CMOC 4). 134,154,189,190,197,199 ‘Forgetting to go to the appointment’ is a popular reason for not attending189 and reminders such as text-messages one or two days before the appointment date seem to help patients to organise their programme in a way that benefits their attendance. A study on reminders for routine cancer screening in London189 used mobile-phone contact records available in general practice registers to send text-message reminders two days before women were scheduled for routine breast cancer screening. The intervention increased overall uptake but, importantly, women living in the most deprived areas, despite having the least complete and up-to-date contact details,189 benefitted the most, with an absolute increase in attendance of 13.6% and a relative increase of 28%. However, an intervention focusing on sending text-message reminders for bowel-cancer screening with the involvement of primary care did not have an effect on gender inequalities or the socio-economic gradient in uptake. 190 Such findings highlight how intervention effects differ across disease types and patient groups.

Service hours and contact time

The studied literature shows that time is an important factor linked with inequalities in general practice; either this refers to the actual opening times of services,134,200 the length of clinical consultation, or the frequency of contact between patients and practice staff. 128,165,201,202 Evidence suggests that patients from ethnic minority backgrounds and people who cannot take time off work have increased difficulty in accessing their GP or screening services if these are available only during working hours (see Figure 7, CMOC 5). 134,200 Importantly, the challenges are the same in the case of telephone services if those are offered only during usual working hours. 200

Beyond flexible opening times, increased contact time in terms of both consultation duration and follow-ups has a serious impact on the overall communication during the clinical encounter and is associated with better patient experience and effectiveness of care (see Figure 8, CMOC 6). 128,129,142,165,201–203 Longer consultation times allow practitioners to work without feeling stressed and this improves their communication skills as well as the outcome of their communication with their patient because there is more space for information exchange. 128 Similarly, it has been shown that regular-follow up benefits information exchange about medication, diet and exercise and gives the opportunity to adjust the care plan or its goals. 129,142,203 Better communication has been found to be associated with more feelings of enablement and satisfaction for patients and confidence for practitioners, who feel that they do their work effectively. 201,202 These are beneficial qualities of care for everybody. However, they are even more so for patients who live in deprived areas and ethnic minorities. This is because these patients are more likely to deal with increased language and literacy limitations139 or have complex physical, social and psychological problems and therefore need more time to communicate their situation and understand what they need to do. 128

FIGURE 8.

CMOCs in the disciplinary domain (2).

Continuity

Continuity emerges as an additional procedural element that links with improved care effectiveness because it improves the understanding of patient needs, especially for elderly or under-served patients and those with complex physical and psychosocial problems (see Figure 8, CMOC 7). 143,204–206 A qualitative study on the lessons learned through the Health Checks programme in the UK stresses that the existence of long-lasting relationships between patients and practitioners enabled the latter to know what type of approach would work best for their population and this benefitted the delivery of the programme, especially among patients who are under-served. 206 Additional evidence128 shows that continuity of care is higher for complex consultation, hinting at the fact that complex problems can be more effectively addressed in the context of an ongoing relationship between patients and physicians either because patients may feel safer sharing information or because physicians can develop a better understanding of their patients or both.

In a qualitative study on the effectiveness of cancer detection in primary care,204 GPs were found to value interpersonal continuity as key for the effectiveness of cancer detection. According to their testimonies, trusting relationships allow GPs to identify those patients who need to be referred for further investigation from those who don’t. According to them, relationship continuity is threatened by the existence of multiple points of provision (e.g. walk-in centres) and the use of part-time or locum cover GPs. A study conducted in the USA echoes these findings and adds on the importance of continuity between primary care physicians and oncologists during cancer treatment for the better effectiveness of care and the higher satisfaction of patients in under-served communities. 207

Multidisciplinary team care

Organising care in a way that involves supportive roles is associated with the effectiveness of care and inequalities in care outcomes (see Figure 8, CMOC 8). 161,208–212 Evidence shows that programmes led by nurse practitioners improve patient management and health outcomes compared to usual care among African Americans with diabetes in rural areas. 210 Specifically, integrating in care a certified nurse practitioner who, instead of being constantly based in the practice, rotates in the rural site to provide necessary support and training works particularly well for people who face more logistic challenges in remote settings. In one of the studies, it is highlighted how nurses practically help with the use of the telehealth service for patients who face accessibility difficulties and how this links with improvements in diabetes self-management. 157 Another study from the USA also highlights the cost-effectiveness of chronic disease management by nurse practitioners and their important role in addressing health inequalities in under-served minority populations. 213

In Canada, a study showed that the presence of nurses allowed GPs to focus on their preferred domains, improved the overall communication in the practice, and increased job satisfaction among physicians, which in turn increased their desire to remain in their work. 161 Nurse practitioners were described as ‘bridges between administration staff and physicians’ because they enabled a better exchange of information. Their presence was associated with creating shared care plans, increasing access to primary care for under-served patients, and delivering new services outside the practice (e.g. home visits for frail patients). Other studies hint at the same direction by highlighting the effectiveness of interprofessional teams including nurses, health assistants and pharmacists in improving care quality for under-served people with diabetes, cancer and CVD. 207,212,214

Supporting staff to engage with prevention for traditionally marginalised groups

Availability of preventive services increases when all members of staff in a practice including healthcare assistants and nurses are supported to engage in prevention. A qualitative study on identifying best practice in England’s Health Checks programme206 found that healthcare assistants consider that training, such as information days, workshops and refresher courses, enables them to engage more effectively in the delivery of the Health Checks programme. Additionally, effective team working based on the allocation of specific tasks and responsibilities to named individuals within practices was also mentioned as a key for the successful delivery of the programme. Another study215 on the role of practice nurses in screening and preventive activities for CVD among patients with severe mental health problems showed that providing relative training to the nurses increases the level of screening and lifestyle advice given to this group of patients. It is believed that training activities for nurses and healthcare assistants decrease their misconceptions and help them understand preventive activities as a distinct part of their role (i.e. through the process of signalling). Hence, they are more likely to undertake such activities and increase the overall service availability within a practice (see Figure 8, CMOC 9). 159,170,201,206,215 Importantly, if this is targeted at disadvantaged groups it can decrease inequalities in prevention services that negatively impact traditionally marginalised people within society and healthcare. 159,215

Interpersonal domain – relationships and experiences

Relationships between practice staff and patients

Many of the included studies refer to the importance of the quality of clinical encounters and the relationship between practice staff and patients. 142,201,203,204,206 However, there are only limited studies examining specific aspects of this relationship and their impact on health inequalities. Evidence shows that when patients perceive their healthcare providers as empathetic, with good communication skills and an interest in their symptoms and their overall concerns, they feel supported to ask questions and engage with decision-making and care management. This in turn leads to more effective and long-term person-centred condition management (see Figure 9, CMOC 1). 150,170,202,216–218 We know that shared decision-making is associated with many positive outcomes and especially better control of diabetes. 146,219 This is particularly important for improving care quality among ethnic minority and socio-economically disadvantaged patients who might be less assertive, or may have limited health literacy. 138,143,206 Studies show that British Bangladeshis tend to view their GP as a busy authoritative person and tend to engage less in information exchange, while patients across different South Asian communities in Britain know less about diabetes and its management compared to white British. 138

FIGURE 9.

CMOCs in the interpersonal domain.

Further, a study on breast-cancer screening uptake among ethnic minority women in inner-city Cardiff general practices highlights the importance of trust between patients and their physicians for the decrease of inequalities in prevention (see Figure 9, CMOC 2). 136 The study shows that when women received an endorsement letter from their doctor they decided to attend the programme because of trusting their doctor and their opinion. A study in the USA points towards a similar direction, showing that physician recommendation for screening mammography can increase screening uptake among 65+ and low-income women. 220 Additionally, supportive relationships emerge as a key factor for the success of educational interventions among disadvantaged groups (see Figure 9, CMOC 3). 146,158,161,217 A study on an educational intervention for African Americans with diabetes highlights how creating a familiar feeling within the intervention groups (including group leaders) was key for the success of the intervention, which led to decreases in hemoglobin A1c and HDL cholesterol. 146 The positive impact of a supportive atmosphere is also highlighted in a study conducted in Canada on medical group visits in primary care and their impact on diabetes patients with low socio-economic status. 217

Effective team working among practice staff

In previous sections, we discussed how multidisciplinary care teams can contribute to the decrease of inequalities in care effectiveness as well as the importance of supporting all practice staff to engage with prevention activities for disadvantaged patients. Additionally, those studies stress that achieving improvements through integrating additional roles in care (including nurse practitioners) is subject to effective team working among practice staff. Effective teamwork involves effective communication regarding allocated responsibilities to individuals, trust between practitioners, and respect for the expertise and leadership skills of the non-physician members of staff. This eventually links with improved management and health outcomes, especially among disadvantaged patients (see Figure 9, CMOC 4). 161,206,208,209,212

Five qualities of equitable general practice: an action framework

The overarching question of our study was which types of interventions or aspects of care can decrease or increase inequalities in general practice for chronic diseases. In the previous section, we discussed a wide and diverse range of evidence related to inequalities across different domains. In this section, we discuss our findings with reference to the IPT, and we suggest an action framework for equitable general practice (see Figure 10). The framework identifies that equitable general practice has five key qualities (represented by the multicolour concentric rings in the figure): connected, intersectional, flexible, inclusive and community-centred. Also, it highlights areas of relevant action in the structural, cultural, disciplinary and interpersonal domains.

FIGURE 10.

Action framework for equitable general practice.

This framework is the outcome of discussions within our research team and the integration of feedback we got from our PPI colleagues and the expert panel, and through a deliberative workshop with health professionals and academic GPs as described in the methods section. It incorporates the components of the IPT that were found in the evidence and refined through CMOCs and maps them as relevant areas of action across the four different domains of power organisation. This means that the framework is not exhaustive as there might be series of other elements relevant to health inequalities in general practice that are simply not discussed in the literature. As one of the deliberative workshop participants stressed, not everything that happens is measured. Moreover, the framework provides guidance, but our analysis has expressed this at the level of abstraction that enables people involved in general practice to use it (and, where necessary, adapt it) according to the particularities of their context.

The main message we aim to convey with this framework is that action to reduce health inequalities demands shifting our focus from the GP as an individual to general practice as a heterogeneous multidimensional system. Within this system all the involved actors, from the position they occupy, need to challenge established power dynamics. It does not imply that everyone can push for change to the same extent across all the domains nor that action should always focus on all the four domains. Rather, it aims to enable people involved in general practice to become aware of their opportunities and limitations in contributing to the reduction of health inequalities, but also of their interconnections, so that they can meaningfully engage in collective coordinated action. Below, we present the action framework for equitable general practice, and we discuss what it means for general practice to be connected, intersectional, flexible, inclusive and community-centred.

Statement of principal findings

Our review revealed that research on general practice interventions that can decrease health inequalities is limited and lacks detail and in-depth focus. The studies addressing inequalities mostly document the problem and its range but they do not investigate the elements of interventions or care that link with the reported outcomes. Focusing on the transferrable qualities of interventions and building on intersectional analyses of power organisation,91,110 we suggested an action framework for equitable general practice. The suggested framework identifies five key qualities of equitable general practice as being connected, intersectional, flexible, inclusive and community-centred. These qualities should inform interventions and action across four domains. In the structural domain the areas of action include funding, premises, workforce distributions and patient life conditions. In the cultural domain, relevant areas involve patient and healthcare staff beliefs and organisational culture, biases among healthcare staff, communication and training material, translation services and inclusive language. In the disciplinary domain (organisational practices) areas of action involve working hours and contact time, financial incentives and quality assessment, patient registers and data use, and multidisciplinary care. Finally, in the interpersonal domain, areas of action cover the quality of clinical encounter, the interaction between practice administration staff and patients, as well as the relationship between practices and the broader communities they serve.

In the following sections, we provide a discussion on the interpretation of the findings in relation to existing literature, and we discuss the strengths and limitations of our study as well as the implications of our findings for research, practice and policy.

Interpretation of findings

Our review revealed that inequalities in general practice are studied more often in terms of clinical outcomes like levels of cholesterol, blood pressure and haemoglobin A1c and service uptake, including mostly the utilisation of screening and other preventive services. Fewer studies focus on inequalities in quality of care including the effectiveness of quality assessment tools like the QOF, practitioner behaviour (e.g. following guidelines) and patient satisfaction. Even fewer studies focus on behavioural outcomes involving adherence to treatment and medication, self-management, smoking, eating habits and physical activity. There were no studies focusing on inequalities in recovery from disease, life expectancy or mortality, which reveals an important gap since all the studied conditions are among the leading causes of death and drivers of inequality in early death. 222 Diabetes was the condition studied most often, followed by cancer and CVD. None of the included studies focused exclusively on patients with COPD, which is quite alarming given that poverty is a known risk factor for COPD. 223 Finally, in terms of dimensions of inequality, we noticed that disadvantaged groups in the included studies combined a multitude of categories associated with socio-economic marginalisation. These most often involved deprivation or other proxies of low socio-economic status, having an ethnic minority heritage, which was often conflated with race and/or migration, being a woman, and living in geographically remote or socio-economically deprived areas. Despite the differences that these categories imply within disadvantaged groups,72 they are rarely considered or explored in the included studies. Moreover, other dimensions associated with marginalisation and stigma like mental illness, HIV stigma, sexual orientation or homelessness fall outside the focus of the included studies.

Our findings revealed that inequalities in general practice emerge through multiple interconnected processes. To account for this complexity while identifying specific focus areas in care and interventions which link with inequalities, we organised our findings across different domains of power organisation. 110 We found that in the structural domain, which involves social structures, policy and institutions, the implementation of nationally directed financial incentives to improve quality of care increases the motivation to provide better care among practitioners. 112–115 This leads to more standardised care, which in turn decreases inequalities between practices at the initial stages of implementation,114,115 without coherent evidence regarding the sustainability of the decrease across time. 29,116–118 However, incentivising secondary over primary prevention links to the disproportionate focus on secondary prevention for those already engaged with general practice. This results in fewer primary prevention activities for disadvantaged patients. 29,115,117,119–123

Further, increasing funding for general practices especially in socio-economically disadvantaged areas may enable staff increase where appropriate and improvement in the capacity of general practices overall. This links to a series of positive outcomes, including the identification of disadvantaged patients at risk, and patient satisfaction as well as clinical outcomes. 125–132 Moreover, inequalities in SDH, including remote location of services and limited transport options, increase inconvenience and inequalities in accessing screening and other services. 129,133–140 This is especially the case for disadvantaged ethnic minority women, who also tend to be dependent on other members of their household for their mobility. 129,140 Additionally, inequalities in SDH reduce the effectiveness of educational interventions for disadvantaged groups who experience increased material barriers in participating and adhering to interventions. This eventually increases inequalities in the adoption of ‘healthy lifestyle’ and health outcomes. 129,137,138,141–147

In the cultural domain, which involves ideas, beliefs and knowledge(s), our findings showed that cultural understanding is a fundamental quality of care for the reduction of health inequalities. 135,138,139,142,147–155 Increased cultural understanding between healthcare providers and patients and tailoring of services that involves deep structural adaptation and positive framing increases the alignment of the offered services with patient worldviews and preferences. This in turn improves overall quality of care for ethnic minorities and disadvantaged patients through increased empowerment and engagement with care. 121,122,156–165 Additionally, our findings highlight that increasing knowledge among disadvantaged groups does not necessarily decrease inequalities. Designing interventions with a narrow focus on education and training often produces the opposite of the desired outcomes and increases inequalities in behaviour change and health improvement affecting disadvantaged groups. 135,141,151,168,172 This is because targeted disadvantaged groups often find such interventions not relatable and/or difficult to engage with. Finally, our findings show that when practitioners tend to take decisions based on heuristics or stereotypes, such as, for example, that men are more susceptible to CVD risk, they are likely to contribute to inequalities in effective diagnosis and clinical management due to implicit bias. 117,165,178–180 We saw that CVD risk factors were regularly less available in the files of women and risk assessment could be performed significantly less often for women seeing male physicians. 178

Our findings also highlight the importance of routines, bureaucracies and other standardised procedures within general practice in inequalities in unmet need, service uptake and care quality. These aspects of care, located in what we call the disciplinary domain, involve the use of the exception reporting in the frame of the QOF, which can motivate practitioners to exclude disadvantaged patients on the assumption that these patients are less likely to achieve targeted outcomes to maximise their local practice income. 29,119–122,125,169,184,185 Such ‘manipulations’ of exception reporting are difficult to capture; however, they are likely to increase inequalities within and between local practices in quality of care as well as unmet need. 29,187 We also found that when practices operate only during usual opening hours it can contribute to inequalities in access because they set access barriers for disadvantaged groups who are less likely to be able to take time off work. 134,200 The collection, maintenance and use of patient information is another aspect of care in this domain that is associated with inequalities. We found that accurate and up-to-date information of patient contact details increases the chances of contact with patients and this links to increased uptake of screening services among disadvantaged groups. 127,149,166,186,187 Moreover, the integration of ethnicity and socio-economic position in risk assessments improves the targeting of preventive services to disadvantaged groups through effective identification of patients at risk. 164,191–194

Increased contact time between healthcare professional and patient in terms of consultation and follow-up frequency can improve quality of care for disadvantaged patients through a series of mechanisms that involve improved communication, increased patient satisfaction and enablement. 128,165,201,202 This is especially true for those with communication and language difficulties and psychosocial problems, including neurodiverse patients. Similarly, long-lasting relationships between patients and practitioners can improve care effectiveness for disadvantaged groups because they enable a better understanding of patient needs and a safer environment for information-sharing. 143,204–207 These mechanisms lead to improved case-finding, more efficient use of services and improved quality of care and satisfaction among patients in under-served communities.

Finally, in terms of care organisation, our review revealed that when care involves multidisciplinary teams with staff from diverse backgrounds and expertise, it improves health professional retention and broadens the availability of patient support. This in turn leads to improved self-management of chronic conditions among geographically and/or socio-economically marginalised patients. 161,208–212 Related to this, supporting practice staff (including nurses and administrative staff) to engage with preventive activities targeted at disadvantaged groups can also decrease inequalities in prevention. 159,170,201,206,215 Support can take the form of training as a way to increase staff’s awareness that prevention is part of their duties and solve their misconceptions around their work and also around patients’ needs. 184–186

In the interpersonal domain, which involves individual experience and everyday interactions among people and communities, when disadvantaged groups perceive their doctor as empathetic and concerned with their overall issues, they achieve better outcomes. 150,170,202,216–218 Empathy creates space for effective communication and shared decision-making which lead to improved chronic-condition management. Similarly, trust between patient and doctor is a useful mechanism to increase uptake of preventive services among multiply disadvantaged groups. 136 For example, ethnic minority or 65+ women with low income are more likely to use screening services if they know that this is something that their doctor endorses. 220 Further, creating a familiar atmosphere in general practice and particularly in the spaces used for the delivery of interventions encourages disadvantaged groups to engage with the aims and the material of interventions and improves their health outcomes. 116,127,217 Finally, effective teamwork based on mutual trust and respect among professionals across roles and hierarchies is an additional key for the improvement of patient management. 161,208,208,209,212 This is especially true for local practices which deal with disadvantaged populations because it improves care coordination and information exchange among staff.

Our study highlights the multitude of different elements and aspects of care that link with inequalities in general practice and the interconnectedness of processes. It also makes clear that although action can be taken in specific areas and domains, to decrease inequalities we need coordinated and collective action across domains. However, coordinated and collective action within the highly diverse context of general practice is not a straightforward task. Our findings show that the context and the differences within disadvantaged groups as well as the small details of interventions can make a difference in terms of increasing or decreasing inequalities. This means that there is a need for standard high quality of care which nevertheless accounts for differences across contexts and targeted populations. Therefore, instead of suggesting specific interventions, it makes more sense to aim for common overarching principles for equitable care which will benefit everyone but even more so disadvantaged patients. 224

Equitable general practice has five key qualities. It is connected, intersectional, flexible, inclusive and community-centred. These qualities ensure the adoption of an integrative perspective with enough nuance to account for differences across people and contexts. Importantly, they are qualities of care that enable us to shift from understanding intervention effectiveness in binary terms between the individual and the population97 and rather focus more on the inherent connection between the two. Finally, they highlight that equality, inclusion and collective action are cornerstones for effective action and should be addressed at all levels and in every opportunity rather than being treated as a relevant yet distinct aspect of effective care.

Comparison with existing literature

Our findings reveal how broad the field of health inequalities is, the multitude of outcomes it involves as well as the diversity of groups affected. This once again underlines the need for a better collective understanding of the forms and the driving mechanisms of health inequalities and effective ways to tackle them. 70 In line with Attwood et al.,40 we found that although studied interventions record or even control for some of the PROGRESS-Plus criteria, only a very limited number of them report results by one or more criteria. However, our study highlights that beyond reporting results by different criteria, there is also need for research that accounts for the combination(s) of these criteria. Our study has clearly shown that disadvantage in terms of health services and outcomes emerges across the intersection of multiple aspects of social position beyond what we strictly understand as socio-economic. 72 Also, it has shown that gender is a crucial difference within disadvantaged groups, implying different forms of challenges for men and women but also different understandings of what is appropriate behaviour225 in the context of health promotion.

Our findings agree with the previous systematic review of the effectiveness of health service interventions aimed at reducing inequalities in health38 in terms of characteristics of successful interventions. We found that interdisciplinary teams improve health professional retention and increase the support for patients across a range of issues which lead to improvement in self-management and quality of care for disadvantaged patients with chronic conditions. 161,208–212 We further revealed that interdisciplinary teams need to be diverse in terms of expertise and background while working relationships between team members need to be based on mutual respect across professional roles and hierarchies. 159,170,201,206,215 Like Arblaster et al.,38 we also found that access to services, especially preventive, is linked with inequalities in service utilisation and the identification of patients at risk. Further, we revealed that access is affected through processes related to data collection and maintenance of patient records,127,149,164,186,187 invitation methods,134,154,189,190,197,199 convenience129,133–140 and working hours of services134,200 as well as with inequalities in SDH across socio-economic status and gender. 129,140 Finally, we found that community is a key principle in developing equitable general practice and thus it should cut across all the relevant interventions developed across domains from policies, dominant knowledge, bureaucracies and procedures to interpersonal interaction.

Moreover, our findings are partially in line with those of Terens et al.,41 which showed that quality-improvement strategies improve care for those older and/or less-educated and African Americans. Specifically, we found that the QOF decreased inequalities between socio-economic groups but only during the first three years of its implementation. 112–115 Further, we revealed that the QOF is likely to increase inequalities in primary prevention services due to the disproportionate number of quality-of-care financial incentives aimed at secondary prevention. 29,115,119–122 This is in line with Capewell and Graham,42 who showed that high-risk approaches to prevention increase inequality. We have shown that the implementation of the QOF and specifically the exception reporting can also increase inequalities in unmet need within and between practices. 29,119–122,125,169,184 Such findings have two serious implications. First, interventions aiming to improve quality of care or care outcomes can generate inequalities which are often undocumented. Second, IGIs25,26 can emerge due to intervention design (i.e. offering more incentives for secondary prevention) but also due to the intervention’s implementation in the every-day (i.e. manipulating the exclusion process of the QOF to maximise impact at the expense of disadvantaged patients). Finally, in contrast to previous findings which show that shared decision-making in primary and secondary care reduces socio-economic inequalities,43 our findings do not offer conclusions exclusively relevant to shared decision-making in general practice. In line with those findings, we have also found that engagement with intervention materials, patient enablement and empowerment, improved communication and shared decision-making are all mechanisms which lead to positive health outcomes for disadvantaged groups. 150,170,202,216–218

Strengths and limitations of the study

Recommendations for future research

The vast majority of studies on general practice do not disaggregate findings by disadvantaged groups, meaning that the impact of interventions on inequalities remains unknown. Research in general practice should prioritise inequalities and apply a health-inequalities lens to all research and evaluation work. Studies on intervention evaluation need to disaggregate their data by dimensions of inequality more systematically. This implies a shift from reporting average scores across intervention groups and elaborating on the differential impact that interventions may have on the basis of differences across socio-economic status, gender, age, ethnicity and also sexuality and disability. Equally important is the systematisation of this evidence and the development of platforms which will allow us to access it easily and use it effectively. In this research a large volume of time was spent reviewing the full text of included studies to identify inequalities data. Electronic databases and journals should develop easier processes to identify inequalities evidence. Machine learning techniques could be useful for the development of such platforms.

Importantly, future research should focus on effective ways to integrate and operationalise intersectionality, moving beyond the simple interpretation of intersectionality as an interaction of disadvantage. We need more studies that account for the intersection of social categories (e.g. gender and race) and explore inequalities between and also within (disadvantaged) groups. This will reveal relationships of inequalities that are systematically ignored and will offer specific guidance for action targeted at groups who experience multiple vulnerabilities. 72 Research should focus more on the role of carers (paid or unpaid) in the effective delivery of interventions aiming to tackle inequalities as well as on health inequalities affecting carers themselves. A gender focus would be particularly useful in this kind of research as women are significantly more likely to undertake caregiving roles, often unpaid, and with a significant burden on their general health. 228

In terms of using the PROGRESS-Plus criteria as a way to capture disadvantaged groups, research needs to question more the selection and operationalisation of the criteria. Selecting criteria which are relevant to inequalities in specific contexts would require some exploratory preparatory work and engaging in dialogue with community members and experts. The same applies to the operationalisation of the criteria. In the UK, the Index of Multiple Deprivation is commonly used as a proxy for socio-economic status and studies on health inequalities often report their outcomes between the most and least deprived groups. Although this approach reflects some general patterns, it is likely that it masks the accurate range of health inequalities which would be revealed if individual measures of socio-economic status were used. From another perspective, binary or categorical operationalisations that do not account for nuances within groups need also to be abandoned. This concerns gender, since studies in the field are still using the binary operationalisation which only sees males/men and females/women and excludes trans* and non-binary persons, and also ethnicity, which tends to be operationalised with a few broad categories which include multiple different ethnic subgroups (e.g. South Asian, black, white).

Importantly, the evidence base needs to be expanded with detailed information about the transferable evidence-based principles behind specific interventions. Also, it should capture upstream drivers of inequalities in SDH which interfere with access to services and intervention effectiveness. This implies that experimental designs should not be regarded as the only way to achieve robust conclusions nor should they necessarily be prioritised over other designs when it comes to funding. Mixed-methods and qualitative studies employing interviews and focus groups, and also ethnographic and observational studies, and document analyses can offer rich and in-depth information in these areas. This means that studies with non-experimental designs should also be included in systematic reviews and evidence syntheses.

Additionally, research needs to focus more on COPD as it is closely associated with levels of deprivation and two-thirds of the affected people are not diagnosed. 229 This among other factors implies studies on inequalities in the diagnosis and management of COPD, especially across gender, ethnicity and occupational status (which are strongly associated with smoking) and their intersections, as well as on ways that general practice interventions can address inequalities in housing and exposure to air pollution. 230

There is also need for more research on effective types of GP organisation from a health-inequalities perspective. There are different models of ownership, management and leadership across local general practices in the country which seem to benefit continuity of care and access to services. 231 More research is needed to expand the evidence on relevant models and also to evaluate the extent to which the achieved improvements translate into a decrease in inequalities.

Research should also focus on the cultural domain. The NHS often takes pride in being one of the most diverse institutions in the country. However, there is increasing evidence suggesting that this diversity is not reflected in governance and leadership. 232 The latest alarming findings of the British Medical Association also suggest that there is a concerning level of racism in the medical profession affecting black, Asian and mixed-race doctors and medical students which stems from fellow doctors, other NHS staff and patients. 233 An earlier report by the NHS Race and Health Observatory234 revealed overwhelming evidence for consistent ethnic inequalities across a multitude of services as well. In our study, we found that lack of cultural understanding and implicit bias can increase health inequalities in general practice. Evidence as such highlights the need for more studies on the interconnection(s) between structural racism, healthcare worker and patient experiences of discrimination, and care outcomes in general practice. There is also a need for more regular and thorough assessment of equality and inclusion strategies and their impact on inequalities in health outcomes. Further, more studies need to explore effective ways to increase cultural understanding in general practice. This kind of research should start from exploring what cultural understanding involves across different contexts, how religion and faith interfere with cultural competence, and to what extent cultural differences affecting the quality and effectiveness of care might stem from differences not in terms of ethnicity but in terms of social class and sexuality. 235,236

Further, more research is needed to evaluate existing interventions from a health-inequalities perspective. Specifically, we need up-to-date evidence regarding the long-term effects of the QOF and its impact on inequalities in chronic conditions between and within practices. Quantitative analysis across time could also reveal potential changes in rates of exception reporting in QOF, while qualitative and ethnographic research could offer evidence on the way the exception-reporting process is implemented across general practices. Case studies should also explore successful examples of engaging patients with complex psychosocial needs in care and to what extent this might be associated with changes in exception reporting and quality target achievement. Finally, more research is needed on the development of effective tools for collecting, storing and integrating patient socio-demographic information in care.

Implications for practice and policy

A major implication for practice and policy concerns exploring and developing ways in which the identified key qualities of equitable general practice can be integrated in the organisational development of future general practice. In this section we aim to integrate our framework into a series of specific suggestions for practice and policy to be considered by relevant stakeholders.

There are four main target audiences for our policy implications and practice recommendations:

1. National general practice policy-makers, such as NHS England.

2. Local and regional health systems, such as Integrated Care Systems or regional NHS England teams.

3. Workforce, training and education organisations, such as Health Education England and medical schools.

4. Individual local general practices.

While there is usually one target organisation for the policy implications or practice recommendations, there are often other organisations that are important system partners and we have sought to highlight these too.

Equality, diversity and inclusion

We tried to ensure that our study would include the voices of those who are often marginalised within the healthcare system and are affected by health inequalities in general practice. For this, we have worked consistently with a PPI group consisting of three women who have experience in patient advocacy and are themselves affected by social disadvantage that cuts across gender, ethnic minority status, age, religion, neurodiversity, socio-economic position and being a carer. Their input has been very valuable during the whole project but especially during the stages of development of the IPT, findings refinement, and the development of the framework for equitable general practice. They particularly helped us to highlight important gaps in the reviewed literature, refine the CMOCs, and link our findings to specific aspects of care in general practice. Moreover, in our deliberative workshop we had additional members of the public combining intersectional disadvantage across gender, disability, sexual orientation and socio-economic status. Their voice has helped us not only refine our findings but also acknowledge the strengths and limitations of our study. Further, in our expert panel we included individuals who work in charity organisations for homeless people and in patient advocacy.

Our research team includes people with different characteristics, and it is relatively diverse in terms of the ethnicity, language and cultural background of its members. Men and women are balanced although people identifying as non-binary or trans persons are underrepresented. The team is inclusive of sexual diversity, while a few of our members have experiences of migration and/or are members or allies of disadvantaged communities. In terms of geographical representation, although the study and meetings were mainly conducted in Cambridge, our team also includes members (SS and CB) based in Newcastle who have a long experience in research, clinical work and public health policy on health inequalities in north-east England and the north-south health divide. Their input on the interconnectedness of processes in general practice, the impact of social determinants on intervention effectiveness, and the translation of our findings into policy and practice has been particularly valuable. Also, GW works academically and clinically in London and broader south-east England. Apart from providing guidance in terms of realist methodology he significantly contributed to contextualising the findings in the setting of general practice in the UK. In terms of expertise, our team also combines people with clinical experience, public health researchers, as well as interdisciplinary social scientists working at the interface of public health, social policy, geography, sociology, epidemiology and behavioural sciences. There has been a good balance between junior and senior members of the team. AG as the most junior member has been offered opportunities to engage with general practice in south-east and north-east England and health inequalities work previously undertaken by Public Health England. She has also been supported to expand her network with international public health and health service researchers and clinicians.

Acknowledgements

This study was conducted with the support from the National Institute for Health and Care Research (NIHR) Health Services and Delivery Research Programme, award number NIHR130694. We are especially grateful to Annie Moseley, Rebecca Harmston and Sukaina Manji for their support as patient and public representatives and consistent contribution during the study. We would like to thank Fiona Head for her contribution to the study design and her comments on this report, Rachel Talbot for her assistance in literature screening and data extraction, Joao Rocha for his contribution to our research team meetings, and all the experts who participated in our expert panel and shared their input on the initial programme theory and the interpretation of the findings. Also, we would like to thank all the participants in the deliberative workshop for a fruitful discussion that helped us refine our findings.

Contributions of authors

Anna Gkiouleka (https://orcid.org/0000-0003-3543-4418) (Research Associate, Health Inequalities) carried out the review, contributed to the interpretation of the results and prepared the results for publication, led PPI activity, co-organised and co-facilitated research expert panel meetings and the deliberative workshop, and is the leading author of the current report.

Geoff Wong (https://orcid.org/0000-0002-5384-4157) (Associate Professor, Primary Care) is co-principal investigator, contributed to the study design, provided methodological guidance in realist approach, overviewed the research process, co-organised and co-facilitated research expert panel meetings and the deliberative workshop, contributed to and edited this report.

Sarah Sowden (https://orcid.org/0000-0001-9359-3463) (Clinical Lecturer and Hon Consultant, Public Health), contributed to the study design, initial programme theory, interpretation of results and editing this report.

Isla Kuhn (https://orcid.org/0000-0002-2879-4020) (Head of Medical Library Services, Librarianship), contributed to the study design, initial programme theory, interpretation of results, conducted the literature searches and edited this report.

Annie Moseley (https://orcid.org/0000-0001-9990-1834) (Patient and Public Involvement Representative), contributed to the study design, initial programme theory, interpretation of the results, deliberative workshop and editing this report.

Sukaina Manji (https://orcid.org/0000-0003-0086-811X) (Patient and Public Involvement Representative), contributed to the study design, initial programme theory, interpretation of the results and editing this report.

Rebecca R Harmston (https://orcid.org/0000-0002-4765-367X) (Patient and Public Involvement Representative), contributed to the study design, initial programme theory, interpretation of the results and editing this report.

Rikke Siersbaek (https://orcid.org/0000-0003-3223-1420) (Post-doc Researcher, Inclusion Health) contributed to the study design, programme theory, interpretation of the results and editing this report.

Clare Bambra (https://orcid.org/0000-0002-1294-6851) (Professor, Public Health) contributed to the study design and editing this report.

John A Ford (https://orcid.org/0000-0001-8033-7081) (Clinical Lecturer, Public Health) conceived the idea and drafted the initial funding application, is co-principal investigator, contributed to the study design and review process, co-organised and co-facilitated research expert panel meetings and the deliberative workshop, contributed to and edited this report; he is the guarantor.

Data-sharing statement

All requests for data should be sent to the corresponding author. Access to available anonymised data may be granted following review.

Ethics statement

No ethical approval is required for this study because it only uses secondary data.

Information governance statement

This study did not handle any personal information.

Department of Health and Social Care disclaimer

This publication presents independent research commissioned by the National Institute for Health and Care Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, MRC, NIHR Coordinating Centre, the HSDR programme or the Department of Health and Social Care.

Publication

Ford JA, Gkiouleka A, Kuhn I, Sowden S, Head F, Siersbaek R, et al. Reducing health inequalities through general practice: protocol for a realist review (EQUALISE). BMJ Open 2021;11(6):e052746.

Disclaimers

This manuscript presents independent research funded by the National Institute for Health and Care Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, the HSDR programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, the HSDR programme or the Department of Health and Social Care.

EQUALISE – all the databases

23 March 2022

Contents

Search numbers

MEDLINE

EMBASE

CINAHL

Web of Science

Scopus

ASSIA