This chapter describes the research design, data collection and analysis used in this study.
Research design
The study used a mixed-methods sequential design consisting of three integrated work packages (WPs) (). The first WP (WP1) comprised a literature review and four citizens’ panels with service users and health-care professionals. It aimed to describe and explain how patients and the public, providers (professionals and managers), and shapers (commissioners and policy-makers) define and make sense of the urgent care landscape (objective 1). The second WP (WP2) used serial qualitative interviews to examine the role of sense-making in patient help-seeking strategies accessing and navigating available urgent (and emergency) care services (objectives 1 and 2) and to identify and describe the ‘work’ for people of navigating and using urgent care (objective 3). In WP3 we integrated our analyses of these data to construct a conceptual model of urgent care help-seeking behaviour that explains urgent care utilisation and identifies modifiable factors that affect urgent care patient decision-making (objective 4).
Diagram of the research objectives and study design.
Methods were integrated in two ways. The first was developmentally, such that the findings from one WP informed the design and analysis of subsequent components. The literature review complemented and shaped the focus of the study, as well informing the conduct of the subsequent data collection. The meanings and definitions of urgent care identified in the review were explored in the empirical data collection with service users (interviews in WP2) and with health-care professionals (the citizens’ panels in WP1). The literature review also informed the conceptual model (WP3). Second, the results were integrated by exploring convergence and contradiction in the findings derived from different methods, using this process of ‘crystallisation’ to provide a more comprehensive account than that offered by a single method.25
Work package 1: literature review and citizens’ panels
Literature review
A structured review of the published literature from 1990 onwards was undertaken with the primary aim of generating meanings and definitions of urgent care from the multiple perspectives of policy-makers, service providers and patients/the public. This was informed by Weick’s sense-making perspective21,26 alerting us to the possibility of contested definitions and meanings of urgent care, which in turn can have implications for people’s attitudes and behaviours associated with the use of urgent care services.
The review included policy documents (and related grey literature) as well as empirical research literature published since 1990 relating to urgent and emergency care. Although not an exhaustive systematic review, it sought to examine the evidence in relation to the following questions:
How do service users (patients and the public), providers (professionals and managers) and shapers (commissioners and policy-makers) define and understand urgent care differently?
Does the way in which patients and professionals perceive urgency influence the way in which patients seek help for urgent care problems?
Conducting the review
A review team was established to develop and manage the review (JT, GM and JP). We employed the expertise of an information specialist (Karen Welch) to undertake literature searching, and, with input from the team, Karen Welch developed the search strategy [using both free-text and database Medical Subject Headings (MeSH) terms where appropriate]. Joanne Turnbull, Gemma McKenna and Jane Prichard identified, screened and critically appraised the literature identified. Critical appraisal was guided by the Critical Appraisal Skills Programme (CASP) tools appropriate to different types of research designs.27
Identification and review of key policy documents and grey literature
Our review focused on policy documents and other grey literature since 1990 because many of the developments in urgent care services occurred from the 1990s onwards (see ) and we focused on the literature that could shed light on policy for urgent or unscheduled care in the UK that also included relevant references to emergency care policy. The search of the SIGLE (System for Information of Grey Literature in Europe) database focused on formal, government-level policy and local-/practice-level policies and guidelines. Specific governmental and health websites were also searched. Key sources are shown in . Reports from individual urgent care centres (UCCs) and wider policy documents dealing with health-care delivery were included when relevant (e.g. NHS Plan in England, White Papers). In total, 60 documents and information sources were reviewed (last searched November 2017). The 44 sustainability and transformation partnerships proposals were not examined in detail but were briefly examined to confirm that they did not contain important information about urgent care that was not represented in the other documents obtained.
Key sources of policy documents
Primary documentary research methods were used28 to identify and compare policy and service provider meanings of urgent and unscheduled care. Members of the research team read the documents to identify definitions, and a spreadsheet containing research questions about the policy context and definitions of urgent care was used to extract, examine, summarise and synthesise key information and concepts ().
Questions underpinning the review
Identification and review of the research literature
Structured searches of the research literature were conducted in two stages. First, a detailed search on MEDLINE used a wide range of search terms and a combination of free-text and MeSH terms as well as appropriate subheadings (see Appendix 1). This initial search retrieved a large number of results and was refined to reduce the number of results. One early modification was to include the term ‘ambulatory care’ only if it was linked to urgent or unscheduled care as the term retrieved a large number of results related to hospital outpatient care rather than urgent care. The strategy combined terms relating to urgent care and non-urgent use of emergency care services (‘Set 1’) with terms that focused on patient experiences, for example patient help-seeking and decision-making (‘Set 2’), to produce a final search.
Targeted searches were then undertaken of MEDLINE In Process & Other Non-Indexed Citations, EMBASE, Web of Science, Cumulative Index to Nursing and Allied Health Literature (CINAHL), ProQuest Sociological Search and PsycINFO around urgent and emergency care use and help-seeking (see Appendix 1). Searching was undertaken in November 2015, and updated in September 2017 to identify additional relevant literature. Studies were included if they met the criteria outlined in . International evidence was included to consider alternative models of urgent care in comparable health-care systems.
Inclusion and exclusion criteria for the literature review
The final search retrieved 7634 results after deduplication. Twenty-six additional papers were identified through other sources (e.g. reference lists, general searching) (). In total, 144 papers were included in the final review. Key information about the content of the papers was summarised in tables as part of the critical appraisal process against the questions in the CASP checklists, augmented with additional questions from . This aided the synthesis and identification of main themes.
Flow diagram of search process using PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2009.
Citizens’ panels
A number of approaches to seeking public views is identified in the literature, including citizens’ juries and citizens’ panels. We drew on a modified citizens’ panel approach. One of the defining features of a citizens’ panel is to:
bring together a small group of people . . . and present them with a policy question. The panel listen to expert witnesses, examine the evidence, deliberate on the issues and arrive at a policy decision or set of recommendations.
p. 78829
The aim of this type of panel is to ensure deliberative and inclusive involvement directed at executing high-quality citizen contributions that can inform the policy-making process.30 Citizens’ panels permit participants to ‘engage with evidence, deliberate and deliver recommendations on a range of complex and demanding topics’ (p. 6).31 They provide an opportunity for citizens to challenge managerial and professional viewpoints and offer a chance for alternative perspectives to be explored.32 We used citizens’ panels to examine urgent and emergency care policy and to deliberate on the provision of urgent and emergency care, and we asked the participants to help develop agreed definitions of urgent care.
Participants and recruitment
Guided by discussions within the research team and with our advisory board, we agreed that our main criteria for the selection of participants should maximise variation in the people and professions involved, drawn from the urgent and emergency care network of actors. Four separate panels were convened to debate and offer direction about how to define and conceptualise urgent health care. Our ‘citizens’ were drawn from (1) the Polish community, (2) a wider general population, (3) health professionals and (4) members of Clinical Commissioning Groups (CCGs). We purposively sampled to represent a range of perspectives and to include public, provider and policy-maker perspectives.
In our original research proposal, three panels (general public, health-care professionals and NHS commissioners) were envisaged. However, following a review of the literature, and discussion within the advisory group and research team, another panel was included comprising Polish participants. This offered an opportunity to include a group who could be characterised as more ‘marginalised’,29 and whom few previous studies had consulted on the topic of urgent care. It was also a chance to explore the views of recent migrants who have the unique experiences of two very different health-care systems. Our public panels (general population and the Polish community) reflected the populations also chosen for the more detailed qualitative interviews in WP2 (see Work package 2: the qualitative interviews).
Participants in the public panels were recruited via local community groups and networks (e.g. through community centres, local groups and organisations) and reflected the diversity of study setting (). They were recruited to act as citizens rather than as expert patients or patient representatives, and so they had varying experiences of health-care need and service use. Health-care professional and provider participants were recruited primarily from a single local NHS trust, along with some who were recruited from health-care education programmes at the university (see ). The participants in the panel that was designed to elicit the views of those engaged in local policy-making and shaping were approached via CCG contacts and were representatives of local commissioning bodies (see Report Supplementary Material 1 and 2).
Characteristics of citizens’ panel members
In total, 41 participants took part in the panels. The shapers and commissioners’ panel had three participants because of some late decisions not to attend, but the three other panels had between 12 and 14 participants. Public panel members ranged in age from 18 years to ≥ 75 years. The panel format was face-to-face deliberation over 6 hours in a single day, except for the shapers and commissioners’ panel, which took 4 hours, reflecting the smaller number of members (see ). Public panel participants were offered £120 to take part [calculated in line with National Institute for Health Research (NIHR) INVOLVE rates]33 as recompense for the significant time commitment required to participate.
Conducting the citizens’ panels
Each panel took place over one day and was designed to explore a set of questions informed by the literature review, which included:
How would you like to see urgent care described and defined?
Are there circumstances in which urgent care services are particularly appropriate (or inappropriate)?
What benefits and risks do you think that urgent care services have for (1) patients and (2) health-care providers?
What principles would you wish to see underpinning developments in the provision of urgent care services?
What do you think are the key differences between urgent and emergency care?
What range of services do you think come under the heading of ‘urgent care’?
Do you think urgent care services – and perceptions of urgent care – have changed over time?
As a group how would you describe and define ‘urgent care’?
The research team prepared a set of resources (drawing on web links, videos and other visual resources) and prepared activities designed to facilitate discussion (). Prior to the panels taking place, four individuals (drawn from the public panels) acted as a panel advisory group. These individuals were asked for their feedback about some aspects of the panel (e.g. the use of some resources in some of the activities, and the format of some activities, such as whether or not to use small discussion groups). We sought their views (through a series of telephone and e-mail discussions) on the best ways to support and facilitate citizens’ discussions during the panel to ensure that panel members would be comfortable with particular activities and tasks.
Urgent care citizens’ panels: process model.
All participants were sent introductory material including policy statements and findings generated from the literature review. Activities on the day included:
brainstorming words associated with urgent care
discussion of hypothetical scenarios or case studies derived from the literature review
discussions about perceptions of different services prompted by pictures of urgent care and emergency services
discussion and debate centred on the diagram of urgent care presented in the Keogh review.
1
Two members of the research team attended each panel to facilitate group discussion. One of the team (GM) led the facilitation and one of the team (JT) supported the day’s events. Members of the research team adopted a neutral role, facilitating participation to ensure that the discussion stayed on topic, and to derive recommendations and reach a consensus. Each panel commenced with an introduction to the study, and then introducing panel members to each other. Some activities took place in small groups, facilitated by Joanne Turnbull and Gemma McKenna. The panel were encouraged to question the researchers about the evidence. Activities also included a large group discussion and agreement on definitions of and terms for urgent care by the end of the day, facilitated by Gemma McKenna and Joanne Turnbull. Data generated from the panel were recorded in contemporaneous notes taken by team members and by audio-recordings, as well as some written data (flip chart summaries, diagrams, and sticky notes used in the discussions). Notes and transcripts of discussions were anonymised.
Work package 2: the qualitative interviews
Qualitative interviews were used to examine, in depth, the role of sense-making in help-seeking strategies and how the respondents accessed and navigated services, and to identify and describe the ‘work’ entailed in navigating and using urgent care (objectives 1–3). To obtain a rich description, qualitative semistructured interviews were undertaken with three carefully selected groups of service users who reflected a diverse range of experiences of urgent care need (see Selection and recruitment of interview participants). A second interview was conducted with a sample of participants to explore the items raised in more detail and to overcome some of the weaknesses associated with ‘one-shot’ interview studies. This use of serial qualitative interviews proved to be effective in building rapport and relationships between interviewee and interviewer, and to generate the kinds of private accounts that may not have been revealed in a single interview.34 This design added a prospective dimension to the study, offering the respondents and researcher time and space to reflect on and revisit topics from the initial interview, and capture changes between the two time points. The first interview probed how interviewees distinguished between routine, urgent and emergency care needs, and understandings of service availability, and examined attitudes and beliefs about urgent care services. The second interview (conducted between 6 and 12 months after the first) examined, in more detail, interviewees’ experiences of using urgent care services in the intervening months (if at all) and explored the ‘work’ entailed when navigating and accessing care.
Selection and recruitment of interview participants
Participants were sampled from the large geographical area served by a single NHS 111 provider (South Central) which covers four counties (Oxfordshire, Berkshire, Hampshire and Buckinghamshire) that are diverse in their geographic and demographic characteristics. By selecting an area covered by a single NHS 111 provider, we were attempting to recognise geographical boundaries that also ‘made sense’ within the structure of the NHS service provision. Although this setting is not the most socioeconomically deprived compared with other parts of the UK, it includes pockets of deprivation, and some lower layer super output areas are in the most deprived quintile nationally (e.g. parts of Portsmouth, Southampton, Reading and Milton Keynes). It also contains areas that are in the most affluent categories (e.g. Wokingham, New Forest and Aylesbury), as well as major cities (e.g. Portsmouth, Southampton and Oxford), and a mix of urban and accessible, and more remote rural areas.
We purposively sampled from three populations that represented particular facets of urgent care need and a range of participants in terms of socioeconomic and demographic characteristics. Two groups were chosen to reflect populations with known high use of emergency care (people aged ≥ 75 years and those aged 18–26 years), and a third group, people from the East and Central European community, was chosen to represent a population known to be increasing locally as a result of recent migration, and who may, therefore, be vulnerable because they are less familiar with NHS services.35
Older people (aged ≥ 75 years)
This group represents a key demographic change experienced in the UK whereby the ageing population has led to a significant increase in the population aged > 80 years.36,37 The research literature suggested that this group had higher rates of attendance at EDs and made greater use of urgent care38 than other age groups. However, this is a group for whom we lack evidence about help-seeking and decision-making around health service use.39
Younger people (aged 18–26 years)
This group was selected because research evidence suggests that younger adults have the highest rates of ED attendance.37 Adolescents (aged 15–19 years) are also more likely to attend UCCs than general practice40 and younger adults (aged 20–29 years) tend to access UCCs because they offer convenience and ease of access.41
East European communities
People from the Accession 8 (A8) countries (Poland, Slovakia, Czechia, Slovenia, Hungary, Latvia, Lithuania and Estonia) were granted rights of free movement across European Union Member States in 2004, and there have been relatively high rates of A8 migration, particularly people from Poland, into the major towns and cities in our chosen setting. Some of the rural areas in our setting have also experienced their first international migration of people from these countries.42 This is a new and possibly growing population, but we have little research evidence about their health needs, or their knowledge of, and use of, urgent and emergency care.43,44 There is some evidence that some ethnic and migrant groups are less likely to use urgent care and more likely to use emergency care,45 although this is contested.46 We refer to this group as East European people, recognising that not all the participants in this group were from the A8 countries (and that the designation of Eastern Europe is contested).
As with all qualitative research, the goal of sampling is not to enable statistical representativeness but to provide a detailed and nuanced understanding. From past experience we are aware that purposive samples allow us to access a range of experiences and to capture rich data about beliefs, attitudes and experiences, and reported behaviours. To achieve an adequate final number of completed serial qualitative interviews, we aimed to conduct 105 first interviews (± 10%), recruiting approximately 35 people from each of the three population groups, and then conduct a second interview with 50% of these (± 10%). In total, we conducted 93 first interviews with 100 people (some in pairs) and 41 second interviews ().
Number of interview participants by population group
We adopted three recruitment strategies to ensure maximum variation: (1) recruitment from nine NHS urgent and emergency care services, (2) recruitment from the general population via community networks and local advertising and (3) snowball sampling via participant networks following interviews (). In our original proposal, we had expected most of our participants to be recruited via NHS services. Although we recruited nine NHS organisations to act as participant identification centres (one NHS 111 service, five EDs and three UCCs/WICs), it proved very difficult to identify and recruit participants from these sources, and only 13 participants were found via NHS sites. Poor recruitment via NHS sites may have been due to challenges in identifying individuals that matched our three population groups (there were particular sensitivities about identifying and approaching East European people). In addition, potential participants were sometimes reluctant to engage in conversations about research when attending, or calling, for an urgent care problem. We were largely reliant on staff at sites to approach people about the research on our behalf, which took place in often busy and pressured environments.
Guided by discussions with our advisory group, we widened our strategy and recruited another 87 participants using a combination of community-based advertising and local media advertising to meet sample targets. To encourage greater uptake of interviews, we successfully applied to the Health Research Authority (HRA) for amended ethics approval to offer £15 in gift vouchers (per interview) to participants as an incentive to take part. (To ensure equity, we contacted those who had already taken part and also offered them this reward.)
Conduct of the interviews
The topic guide for the first interviews was informed by the literature review and the citizens’ panels, and the topic guide for the second follow-up interview drew particularly on notions of patient ‘work’ and the analyses of interview 1 (see Report Supplementary Material 3). Service users were encouraged during interview 2 to explore the networks and resources that support them in their help-seeking or illness tasks. This was aided by using a simple diagram of concentric circles on which participants captured and mapped their social networks (family, friends, groups, professionals, services and third-sector organisations) in order of importance. The location of the interview was determined by interviewee preference, and most were conducted in people’s homes, with a minority taking place at participants’ places of work or study. All interviews were digitally recorded (after consent was obtained) and transcribed, and they typically lasted between 35 minutes and 1.5 hours. The majority of interviews were conducted by the same researcher (GM) supplemented by one other (JT). Participant information sheets and consent forms can be found in Report Supplementary Material 4 and 5.
Work package 3: integrating the analysis to build a conceptual model
Data analysis began alongside data collection, initially focused on the data generated during the citizens’ panels. We undertook a thematic analysis of these data, broadly following the stages described by Braun and Clarke,47 familiarising ourselves with the data, generating initial codes and categories and then identifying themes. To facilitate analysis and discussion among the team, grids and matrices were used to chart and compare the data, and this involved the wider research team, comprising fieldworkers, researchers and clinicians.
Qualitative interview data were analysed using a data clinic approach to share and interpret data collectively, building emergent themes and developing narrative and interpretive summaries. Core team members (GM, CP, JP, AR and JT) initially read and open coded a sample of transcripts independently, and these early codes were discussed and refined to form the basis for a coding scheme that was refined and developed and applied to all transcripts. We drew on a framework analysis approach,48 looking across cases and exploring similarities and differences (paying attention to contradictory cases). These analyses were informed by conceptual ideas drawn from previous research and theorising, including work on sense-making and patient work. As the analysis developed, the themes were refined using matrix techniques to facilitate further comparisons and interpretations. Emerging themes and interpretations were shared with patient and public involvement (PPI) and advisory group members to check the credibility of these and further refine our analyses.
To build a conceptual model to explain urgent care utilisation, we drew together the findings from the literature review, citizens’ panels (WP1) and qualitative interviews (WP2) by examining codes that:
described and explained different conceptualisations/definitions of urgent care
identified, characterised and explained sense-making strategies that influenced help-seeking choices and behaviours
identified and characterised the ‘work’ involved in understanding, navigating and choosing to utilise urgent care.
We created lists, and taxonomies, of influences that shaped choices and reported behaviours. We recognised that interaction with urgent care services was not simply produced by individual help-seeking behaviour but was also a collective phenomenon (such that narratives and processes are shaped by the views and behaviours of multiple participants), and we used this understanding to underpin our analysis. We then explored the identified factors, exploring how they were formed related to each other and to the contexts in which they operate. We continued to use comparative analysis to identify factors that appeared common across different data sources and different care contexts. As the analysis progressed we began to use mind maps, decision trees and logic models to map our interpretations. Our data clinics allowed us to revisit the data from WPs 1 and 2 to test emerging hypotheses concerned with how sense-making and help-seeking related to each other and to identify factors that might be modifiable, which provided the basic material for a framework from which a conceptual model of relational choices about engagement with urgent care was developed.
Ethics approval
The citizens’ panels comprised members of the public (recruited from community groups and local public networks) and health-care staff. This component of the study did not require NHS ethics approval, and was approved by the University of Southampton (Ethics and Research Governance Online, number 20217). NHS/HRA ethics review was required for the qualitative interviews (WP 2) as some participants were recruited via NHS organisations (REC reference number 16/EM/0329).
