Case Studies Chapter 9 Care Coordination and Continuity in Health Care Setting and the Community

Design

Our methodological strategy reflected the nested character of the structures through which care is co-ordinated. We started from an analysis of patients' experiences of continuities of care, and then traced the organisational-level co-ordination mechanisms that produced those experiences. From the provider level we worked 'outwards' again to the care network mechanisms for co-ordinating care across different providers. Using Øvretveit, Hansson and Brommels' ideas³³ but reversing the direction of analysis, we worked back from 'consequences for clients' to 'consequences for personnel' to 'coordination structures'; an 'inside-out analytic strategy'²³⁸ moving from a particular to an ever-broader view of the local health economy at each study site.

This strategy required three main pieces of knowledge:

1. Patients' experiences of care co-ordination, from their own accounts, their health-care records and administrative data. (Assessment of clinical outcomes was beyond the scope of this study.)

2. What care co-ordination mechanisms were available to that care group in that setting, and what barriers to co-ordination existed, knowledge obtainable from the practitioners and managers responsible for the care group.

3. What effects each of the observed co-ordination mechanisms (or barriers) has on the continuities of care.

This study empirically examines points 1 and 2 in NHS and NHS-like settings. Chapter 2 gave an overview of the extensive though still incomplete evidence base regarding point 3.

To discover what care co-ordination mechanisms exist in a given setting, how each mechanism produces continuities or discontinuities of care and how patients experience them requires multiple qualitative methods; our study design follows Alexander and Bae's recommendation⁷⁹ for researching the similar context of the US patient-centred medical home. Our multiple-methods design combined five main methods:

1. Assembly of an analytic framework by non-systematic review of existing research. This contributed directly to answering RQ 4 and indirectly to answering the other RQs.

2. A framework analysis of patients' experiences of the continuities of care in the same sites as the organisational case studies (see below), using patient interview data and, as validation, quasi-quantified (patient records) data, and comparing the findings with the co-ordination mechanisms described in the organisational case studies. This contributed to answering RQs 1 and 4.

3. A systematic comparison of organisational case studies, applying the above analytic framework to a maximum-variety sample of care co-ordination mechanisms (organisational structures, care network structures, models of care, managerial practices). This contributed to answering RQs 1, 2, 3 and 5.

4. A cross-country comparison of care co-ordination mechanisms found in our NHS study sites with Swedish polyclinics whose care co-ordination structures are absent from the NHS. This contributed to answering RQs 1, 2 and 4.

5. Synthesising the above empirical findings so as to answer all four empirical RQs and correct our initial hypothesis about the relative merits of integrated organisations and care networks as means of care co-ordination (RQ 4).

Stage 1 above corresponds to the theory-building and modelling stages for evaluating complex interventions, our empirical work to the exploratory phase.^{239 , 240} Table 2 shows the methods, data and analyses for each RQ.

TABLE 2

Research questions, methods, data and analyses

We also compared our own findings with relevant empirical findings emerging from research studies in the Health Reform Evaluation Programme, Health Services and Delivery Research (HSDR), NICE and Department of Health Health Policy Programmes.

The underlying methodology was realist; not realist evaluation, as our RQs did not require us to test the programme theories underlying current NHS policies for care integration,²⁴¹ but realist in the sense of attempting to expose, characterise and compare alternative organisational and network mechanisms for care co-ordination.²⁴² It was also falsificationist,^{243 , 244} in that we first formulated a set of theories and hypotheses about care co-ordination and then compared these with our new evidence, revising and refining those explanations as that evidence dictated. In that way we also confronted the issue of reflexivity; that is, of being aware of the preconceptions (explanatory assumptions) that we brought to the data, and looking for errors in those.

We involved patient representatives, recruited and supported by the Patient and Public Involvement Team from the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research and Care (CLAHRC) for the South West Peninsula (PenCLAHRC). We worked with them before the research started on both a group and an individual basis, looking at the form and content of the patient information and interview schedules, and we will involve them again in the post-project dissemination activities. We will also involve study site managers in disseminating our findings, focusing at the outset on the NHS research and development offices for our case study sites and our original key informants.

Initial explanatory and analytic framework

We assembled our initial analytic framework (see Chapter 2) from a review of existing research, whose findings we categorised, linked and thematised 'by a process of gradual abstraction'.²⁴⁵ The review was non-systematic because it addressed three tasks for which systematic reviews are not designed; that is, to:

1. assemble theories about what organisational arrangements produce care co-ordination

2. identify care co-ordination mechanisms that appeared relevant to our focal care group but are not used in the NHS

3. operationalise our RQs, and hence focus the empirical research.

The review was a realist synthesis in that it attempted to synthesise what is so far known about the mechanisms for achieving care co-ordination, their contexts and outcomes. However, it was not free standing. Its findings were intended to inform, and to be tested through, the other research methods noted above. So, after Chapter 2 they are not presented separately from those produced by the other methods we used. With that qualification, our methods conformed to the Realist And MEta-narrative Evidence Syntheses: Evolving Standards (RAMESES) recommendations.⁷⁷

We initially scoped the literature by searching for the most recent systematic reviews in the relevant fields (e.g. about care plans, polyclinics, etc.). Where recent (≤ 10 years old) reviews existed, we focused our search for primary studies on the period following the reviews. Otherwise, we placed no time limits on our search. Except for very specific keywords (e.g. 'care plan' or 'primary care team'), hand-searching journal contents lists and snowballing back from one study to another often revealed more studies than did automated searches (e.g. of PubMed or Google Scholar) of this conceptually and terminologically ill-defined literature. We hand-searched journals dedicated to primary care co-ordination (e.g. International Journal of Integrated Care), those which focus heavily on clinical practice (but not necessarily on care co-ordination) in primary care (e.g. British Medical Journal, British Journal of General Practice), and those which focus on health policy and its implementation in fields relevant to care co-ordination (e.g. Social Policy and Administration). We also hand-searched policy and think-tank websites (e.g. Department of Health, The King's Fund) for peer-reviewed studies.

Given our 'inside-out' analytic strategy, we started assembling the theoretical framework in Chapter 2 by conceptualising (as others have²⁴⁶) a patient's experience of complex care as a 'journey' along a care pathway (a de facto care pathway, even when it is not formally documented): that is, as a process of production. This starting point gave us three ways of using existing research findings to formulate and define the hypotheses listed at the end of Chapter 2.

1. For some topics (e.g. conditions for effective interdisciplinary teamwork), sufficient studies – mostly qualitative – existed for some overall empirical patterns to be recognisable.

2. Beyond the health sector, certain more general organisational theories (reviewed in an earlier NIHR study²⁴⁷) report which characteristics of a production process constrain, and which are constrained by, the organisational structures used to manage it. Among these characteristics are how uniform and standardised the process is; how far it can be mechanised; whether it is continuous or episodic; what skill mix it demands; and whether its component subprocesses occur sequentially or in parallel. Different combinations of these characteristics imply that a specific kind of organisational structure will 'fit' best for managing each kind of productive process.^{163 , 248} Research on interorganisational networks reports what patterns of interorganisational linkage appear conducive to effective and efficient interorganisational collaboration in health care.^{32 , 188 , 249} Some of the studies cited in Chapter 2 appeared to describe instances of these wider organisational or network patterns. One could then infer corresponding hypotheses about what effects these specific characteristics of primary care processes, organisational structures and interorganisational networks would have on primary care co-ordination and continuities.

3. Where there were few empirical studies and none of the wider theories seemed applicable, hypotheses had to be formed by analytic (qualitative) generalisation,²⁵⁰ the weakest of these three methods.

Given our RQs, we above all formulated hypotheses about what effects their context (integrated organisation vs. network) was likely to have on the operation of mechanisms intended to produce care co-ordination and continuity. By the time the data analysis and synthesis were completed, we had revised our initial framework in several ways, reported in Chapter 11, Revisions to the initial analytic framework (an 'open-ended' analysis²⁵¹).

Case studies of organisational and care network co-ordination mechanisms

Applying the above analytic framework, we systematically compared case studies of a maximum-variety sample of care co-ordination mechanisms (organisational structures, care network structures, models of care, managerial practices). This contributed to answering RQs 1, 2, 3 and 5.

Sampling

Although we intended to trace back from patients' experiences of the continuities of care to the co-ordination structures that produced them, our aim was to contrast different types of co-ordination structure. We therefore selected a theoretically based, purposive maximum-variety qualitative sample of study sites, chosen as follows to instantiate the main types of co-ordination structures identified in our RQs:

1. Hierarchy, public ownership. Ideally, this would be a single, integrated organisation providing primary medical care, CHS, mental health services and social care. However, no such organisation existed in the NHS at the time of our fieldwork.

   1. The nearest NHS equivalents were trusts or foundation trusts that combined CHS provision with hospital services and/or mental health services and/or social care.

   2. In Sweden, however, such organisations did exist: the vårdcentral described in Chapter 10.

2. Two types of care network ('federated') structure pursuing the objectives originally proposed for English 'polyclinics' by combining:

   1. General practices and CHS.

   2. CHS and social care providers (sometimes with GP participation).

3. Hierarchical, privately owned commercial or joint-venture primary care providers. We selected a corporate private practice whose owning company also provided other primary and CHS in the same locality.

   1. A hybrid partnership, hierarchical and care network structure: concretely, the standard tripartite NHS model, as a reference point.

   2. A privately owned medical practice networked with public CHS. In the event, the study site mentioned at (3) also satisfied this criterion, since that general practice was networked with NHS-owned CHS and other providers. For further comparison we also examined (in less detail) a Swedish equivalent (see Chapter 10).

When our fieldwork began, no routinely published national data reported general practice ownership or the care co-ordination structures in each PCT (as local health economies then were). We therefore identified possible sites using the professional press (especially Pulse, GP Daily News, Medeconomics and the Health Services Journal), NHS Networks websites and mailing lists, Department of Health and think-tank websites, and national media (especially the BBC and The Guardian websites), supplemented with the researchers' own knowledge and networks. In consequence, we refined the sampling strategy stated in the study protocol to distinguish two models of federated (networked) primary care co-ordination: networks of professional partnership general practices (e.g. CCGs, general practice 'federations')⁴³ and networks of hierarchically organised public providers (confusingly, also called 'federations' in some places). During this period of greatly increased workloads for general practices, commissioners (and, therefore, NHS research governance administration) were also being restructured and restaffed. In all we approached 10 sites, and at the expense of much time and effort gained access to five, which together represented the main co-ordination structures that we needed to study.

These methods yielded the following sample of study sites in England. Three sites were whole counties, one was part of a county, and the fifth was part of a larger conurbation. We have pseudonymised the sites. Fuller details are in Appendix 3, but, briefly:

1. Eastmere instantiated the standard tripartite NHS co-ordination structure of separate primary medical care (professional partnership general practices), NHS trusts (CHS; acute hospital; mental health) and social services. The CCG divided its territory into 10 GP localities. The county council divided the same territory into six localities. Towards the end of the study period the county council was awarded integration pioneer status for its plans to join up health and social care services, and one study general practice was given a RCGP quality award. This study site was, therefore, a well-developed example of the standard model.

2. Whinshire was selected for its large, elaborate managed network for co-ordinating care across provider organisations. CHS and mental health services were integrated into a single NHS trust which, until 2012, also provided social services on behalf of the county council (under a §75 agreement). Thereafter, the site reverted to colocated, separately managed services, which had separate data holding but retained a shared referral model. It hosted a personal health budget pilot scheme. Its care network originated from a collaboration between the local practice-based commissioning group and the PCT, and developed into a community interest company. By the time of our fieldwork this network involved nearly all of the general practices in the county, all of the other health-care providers, and social services.

3. Sedgeton was selected for its virtual 'polyclinic' housing networked colocated services – that is, organisationally separate community health, acute hospital outpatient and mental health services, all in one building, with mental health inpatient facilities next door. Links were developing with social care. The general practices referring to all of these services were part of a 'federation' including these and other providers. Our case study practices were two of those that referred most patients to the colocated services.

4. Tarrow was a vertically and horizontally integrated health and social care provider (hierarchical governance) with an acute hospital, CHS and adult social care (under a §75 agreement), all managed by one chief executive. Mental health services were provided by a separate trust. When the §75 agreement ended, adult social care services reverted to separate county council management. Two virtual ward schemes were being piloted and, later, integrated health and social care personal budgets.

5. Fivecastles contained a corporate general practice belonging to a company that provided other primary and CHS in the county (hierarchical governance). The county also had a combined hospital and CHS trust. The general practice, integrated community and acute trust and mental health trust were, thus, three separate hierarchies, externally networked to each other.

Every site had a NHS treatment centre, a NHS ambulance trust and commercial and third-sector service providers, vertically integrated mental health services and, of course, other professional partnership general practices.

In each site integrated organisations and care network structures coexisted, and the organisations were components of care networks. Following the care process logic of our analytic framework (see Chapter 2), Table 3 summarises which structures coexisted. In each row, the bodies listed in the 'Care network' column co-ordinated the providers listed on either side of them. Where a combined organisation provided normally separate services during part or all of the study period, Table 3 brackets them together.

TABLE 3

English study sites: organisations and care networks. Services bracketed together were provided by a single organisation

In each site we assembled a sample of key informants based on our patient interviews and discussions with a practice manager or GP, who in turn suggested further informants relevant to the study. This snowballing method yielded the sample of 80 informants summarised in Table 4.

Data collection

Data were collected from informant interviews (using the schedule given in Appendix 4) and managerial documents. Unless the interviewee requested otherwise, all interviews were digitally recorded and professionally transcribed, and interviewees were offered the chance to correct their transcript. Our findings chapters (see Chapters 5–9) identify these informants by the relevant case study prefix (E–W), followed by the letter P (for 'practitioner') and a numerical identifier. Our collection of managerial documents was guided by our interviewees, as they were best placed to say which documents were locally seminal. We also collected professional press reportage.

Initial analysis

We collated data about each organisation's and each network's care co-ordination mechanisms against the relevant headings in the analytic framework (Table 5), adjusting these categories when necessary.

TABLE 5

Data collation framework: organisational case studies

Collating data against a prior framework revealed where any mechanisms described in the analytic framework were absent. It also provided a structured means of data triangulation so that we could supplement our data with ad-hoc e-mail and telephone enquiries when the data had gaps or ambiguities, or if we had reason to believe that circumstances had changed at a particular site. From any remaining data we inducted patterns, supplementing the analytic framework as necessary. Following the care process logic outlined above, we differentiated the care co-ordination structures and mechanisms found in the following:

1. first-responder organisations at each site

2. care networks and their managing bodies

3. main second-responder provider organisations, insofar as they contributed to the care of our focal care group.

Because of how the NHS is structured, this approach approximately corresponded to and analytically separated three contrasting governance structures for care co-ordination: professional partnerships (most general practices), networks (referral hubs, virtual wards) and hierarchies (NHS trusts, social services).

Patients' experiences of continuities and discontinuities of care

Sampling

As we wanted to trace the connections (and disconnections) between patients' experiences of the continuities of care and local organisational and network structures, we sampled patients in the same study sites as the organisational case studies. We also needed to sample patients who had recently experienced complex care and hence who satisfied the following inclusion criteria:

1. being 65 years of age or older

2. having complex health-care needs, defined as having at least two of the following conditions: arthritis; heart failure; COPD; stroke; mental health problem (depression, dementia, psychosis, schizophrenia or bipolar disorder); and sensory deprivation (blind or deaf)

3. receiving care for at least 1 year before the study from at least two provider organisations

4. living in own home or with family.

Exclusion criteria were:

1. having a moderate or severe learning disability

2. receiving terminal care

3. being in the criminal justice system

4. being unable to give informed consent

5. being too frail to give an interview

6. living in residential care.

Our nine chosen GP practices each drew a sample of patients conforming to these criteria and sent out invitations, study information and reminders on our behalf. Patients who were interested in participating were then able to contact the research team using the reply slip and prepaid envelope supplied or, on occasions, by e-mail, telephone or their GP surgery. Because we intended to characterise patients' experiences of the continuities and discontinues of care qualitatively, we interviewed small numbers of patients in depth until we approached data saturation (n = 66; Table 6). Interviews typically lasted about 45 minutes and, given the patients' age and interest in the topic, often became somewhat discursive.

TABLE 6

English study sites: patient interviewees

The wide inclusion criteria meant that patient characteristics varied across the study sites (see Chapter 5). The decision to take the patient as the starting point for the research meant, however, that they had not necessarily used the care networks that practitioners and managers subsequently described. For example, although the Sedgeton patients had all used services at the 'polyclinic', none of the Tarrow patients had used the virtual wards, as many lived beyond the relevant catchment area. Similarly, whereas the Eastmere interviewees were the most limited group in terms of activities of daily living and received the most social care support, Fivecastles was, in contrast, an affluent area with patients who were:

[. . .] very good at going and helping themselves [. . .] well supported by helpful, intelligent and well-resourced families.

GP F03

None of the Fivecastles patients had current social services or care agency support. Several had used private health care or local private hospitals. Patients in Eastmere, Tarrow and Fivecastles lived in 'predominantly' or 'significantly' rural areas, but the Whinshire patients lived largely, and Sedgeton patients lived entirely, in urban areas.

Data collection

Patient interviews were conducted using a semistructured interview schedule (see Appendix 5). With patients' consent, their interviews were digitally recorded and transcribed professionally. When patients wished, their carers (for one, his 93-year-old sister; for two others, their spouses; and for another, his son) were present and contributed. We extracted data from the interview transcripts into a standard pro forma (Table 7), a deductive coding frame.

TABLE 7

Data extraction template: patient interviews

Free-text answers were thematically coded by the researcher who had conducted the field interviews [initially using NVivo (QSR International, Warrington, UK)] and then passed for validation to the principal investigator. One case study site was initially also coded by a third member of the research team to test for consistency of approach. Patient-level data were then collated into a site-wide form of Table 7, with additional categories suggested by the inductive analysis, so that patterns across patient experiences, and differences, could be seen. We also made an emergent (inductive) coding of the responses that the initial deductive coding framework could not accommodate, which included many of the more unstructured, 'free' responses. These additional codes (e.g. patient willingness to co-ordinate own care) and the coding of any ambiguous responses were agreed between the principal investigator and the researcher(s) most acquainted with the study site. We thus conducted both a framework and an inductive ('ground-up') analysis of the interview data.

Information on health care and referral pathways across the past year was perhaps the most difficult for patients to recall:

Interviewer:

Who called the ambulance for you?

Patient:

I don't know. When I woke up the ambulance was here.

Interviewer:

Did they take you to [hospital name]?

Patient:

[Hospital name].

Interviewer:

So how long were you in [hospital name] for?

Patient:

I can't say – I can't remember.

E13

As validation and to obtain further data we therefore undertook a content analysis of the medical record for each patient for the preceding year, extracting data about three basic activities from the patient's general practice record into a standard pro forma (Table 8):

TABLE 8

Data extraction from general practice medical records

1. encounters between general practice and patient (e.g. consultations, treatment and review in the surgery, home visits, telephone calls)

2. actions in which the practice engaged with external providers on the patient's behalf (e.g. referrals, requests for advice, case meetings), usually without the patient being present

3. patient encounters with other providers (transfers), typically documented as reports, reviews, treatment, information, advice and discharge notices.

In extracting data we focused on pro tem transfers of responsibility for care between the patient, clinicians and organisations. We did not extract data about requests for repeat prescriptions, did-not-attend notices or practice requests to make appointments for, for instance, chronic condition reviews; we extracted data only about the resulting encounter. These data describe the activity captured by general practice medical records (and hence information readily available to GPs) but will tend to under-report community and social care, and will omit undocumented activities. In one practice, for example, CHS staff had read/write access to the GPs' records, and so CHS activity was more fully recorded there. Practices had different ways of recording information depending, for instance, on which electronic system they used, practice protocols and individual idiosyncrasies. (One GP might, for example, write very full notes including a description of a patient's holiday; another might write sparse notes relying heavily on abbreviations.) As with practitioner interviews, extracts from patient interviews are indicated by a case study prefix (E–W) and numerical identifier. Medical records are indicated in the same way, but with the additional letters MR ('medical record').

Initial qualitative analysis

Our analytic framework (see Chapter 2) listed observable features that would be evidence of continuity and co-ordination of care: the presence of a care co-ordinator and of a care plan; an ongoing relationship between care co-ordinator and patient; reviews; access to services; and information transfer. Combining the interview and medical records data for each patient, we checked for the qualitative presence or absence of each of these features in the patient's experiences. Applying the same headings and collating within each study site, we wrote for each study site a case study of care co-ordination there, as collectively reported by its patients. We noted any differences between patients' accounts, any aspects of care co-ordination not anticipated in our original framework, and the broad (qualitative) patterns of care co-ordination reported in the patients' medical records. By comparing these case studies, we could then qualitatively compare study sites.

Initial statistical analysis

Quantifiable data from patient interviews and the matching medical records were tabulated and coded onto a spreadsheet, including coding for additional (e.g. organisational) variables constructed that might explain any variations between study sites or sampled practices. Where data sets were large enough, we ran simple tests of correlation between activities and sites [using IBM SPSS Statistics version 21 (Armonk, NY, USA)], declaring p-value ≤ 0.05 as the significance level. [The findings chapters (see Chapters 5–9) report which data sets were large enough to be so analysed. In those chapters we use the terms 'significant difference' or 'no significant difference' where statistical tests were made. Without that qualification, terms such as 'difference' refer to qualitative differences.] Simple measures of central tendency and cross-tabulations ('crosstabs') were used. Depending on data characteristics, the tests applied were Pearson's chi-squared test, Fisher's exact test and analysis of variance. This analysis contributed to answering RQs 1 and 4.

Cross-country comparison

Design

Given our RQ, the ideal comparator against which to compare care co-ordination by care networks would have been a single, integrated organisation providing primary medical care, CHS, mental health services and social care. The NHS had no such providers at the time of our fieldwork, but they did exist in Sweden. A comparison of care co-ordination mechanisms available in these two settings would contribute to answering our RQs 1, 2 and 4. For comparability, we used the same analytic methods as for the English case studies and reused secondary data from surveys of Swedish polyclinic directors about their management of four chronic diseases (coronary heart failure, depression, diabetes and asthma). We did not analyse patient records but did interview a small number of patients (not reported here).

Sampling

Using the researchers' own knowledge and networks to identify study sides, we selected the following PHCCs (polyclinics) whose organisational structures had no NHS equivalent:

1. three publicly owned, hierarchically structured polyclinics (vårdcentral) in Stockholm (Lisebergs, Kista, Djursholms Iäkarmottagning)

2. TioHundra AB, a vertically and horizontally integrated, publicly owned provider of hospital, primary care and social care (Norrtälje)

3. a not-for-profit trust providing primary medical care and other primary care services (Stockholms Sjukhem)

4. a corporation (private for-profit hierarchy) providing primary medical services (Solna Capio).

All were financed by Stockholms läns landsting (SLL: Stockholm county council). SLL directly line managed the three PHCCs under heading (1) above and commissioned the others through contracts. In size and function the PHCCs were similar to the larger NHS general practices.

Data collection

Data were collected (2011–13) at county and municipal (i.e. commissioner) and at provider level by mixed methods from the following sources:

1. Interviews and discussions with key informants. These included three representatives of SLL (polyclinic line managers and commissioners), the polyclinic head and two nurses at each provider, and at Norrtälje a half-day meeting with the polyclinic head for mental health services and three clinicians to discuss how care was co-ordinated across services. We interviewed six patients at the directly managed polyclinics, three at Norrtälje, four at Capio and two at Stockholms Sjukhem.

2. Grey material including official regulations and guidance.

3. Ad-hoc enquiries from individual experts.

4. Published research found by hand-searching journals.

Interviews were recorded and transcribed. Material from the Swedish study sites was obtained partly in Swedish and partly in English. The former material was translated or summarised for us by native Swedish speakers.

The surveys (2005 and 2011) replicated a US survey to discover the use of recommended care co-ordination practices and of IT for co-ordination in primary care.^{252 – 254} The response rate was 37%, with a bias towards the larger centres. Whether or not there was a higher proportion of public than private PHCCs replying is unknown, as there are no reliable national data about PHCC ownership.

Initial analysis

To enable comparison we made the same initial analysis of the Swedish as of the English organisational case studies. Chapter 10 presents these findings.

Data synthesis

In all, we assembled the data summarised in Table 9.

We synthesised the initial analyses in four stages. The first three were systematic, in the sense of using the same analytic framework to structure and compare different sets of data. We continued to adjust the analytic framework in the light of unforeseen findings.

Nested framework analyses

Given our 'inside-out' study design, the framework analyses were nested in the following order:

1. We started from the findings (see Chapter 5) about patients' experiences of the observable features of continuity and co-ordination of care (see Chapter 2, Processes and structures for care co-ordination).

2. Next, we compared the pattern of patient experience at each site with the initial analysis of care co-ordination mechanisms in the first-responder organisations at that site. Chapter 6 presents these findings. Where they matched, the organisational case study provided a prima facie explanation of patients' experiences. For example, patients often reported difficulties accessing 'their' preferred GP, whereas the general practices reported overloaded appointment systems and high GP workloads. Where patients reported discontinuities of care for which our case studies contained no apparent organisational explanation, we sought explanations in non-organisational causes²⁵⁵ such as patients' own resources and decisions. However, our health worker informants more often reported care co-ordination mechanisms, or the absence of such, not mentioned in the patient experience data (e.g. case reviews which the patient was unaware of). We consider in Chapter 11 what that signifies.

3. We repeated step 2 for the case studies of the 'second-responder' provider organisations at each site. Again, we interpreted organisational case study findings that matched patients' experiences as being prima facie explanations of those experiences. Chapter 7 presents the findings.

4. Next, we read off from our informants' accounts how care networks (and other factors) had influenced the providers' organisation and management. At this stage of the analysis the most informative data were usually our informants' narratives (a local NHS history) of how care networks had developed in each study site, the rationale for those developments and the practical constraints on them. Chapter 8 presents these findings.

5. Finally, we traced the ways in which the governance of local health economies had shaped the working of the care networks. Here, too, informants' narratives of recent NHS history in the study site were often the most informative data. These findings are in Chapter 9.

At each stage draft analyses were circulated among the researchers acquainted with a particular study site, to check for accuracy, completeness and face validity. We systematically compared findings across study sites to reveal similarities and differences.

Systematic comparison of organisational case studies

We systematically compared the initial analyses of organisational and care network structures in order to find common patterns in care co-ordination mechanisms across sites, and differences between sites that might reflect differences in organisational and network structures, including any differences between care networks and organisationally integrated providers. We therefore compared:

1. sites where acute and community services were integrated into one organisation (Fivecastles and Tarrow) with sites where they were separate organisations (Eastmere, Sedgeton and Whinshire)

2. sites where community and mental health services were integrated into one organisation (Whinshire) with sites where they were separate organisations (Fivecastles, Eastmere, Sedgeton and Tarrow)

3. sites where community services and social care had been integrated into one organisation (Tarrow) with sites where they were separate organisations (Fivecastles, Eastmere, Sedgeton and Whinshire).

In this way we addressed RQs 1 and 3. By comparing these findings with our original analytic framework and hypotheses, we also began to address RQ 4. We furthermore systematically compared the corporate general practice (Fivecastles) against the professional partnership general practices in the other four sites, thereby addressing RQ 2.

Cross-country comparison

We extended the systematic comparison of English organisational case studies by adding in our initial analysis of the Swedish polyclinics. The methods and process of analysis were essentially as described above.

Inducting patterns of managerial discretion

Given our RQs, we filtered our organisational data for any that suggested ways in which managerial decisions had produced or frustrated continuities of care. For this purpose, 'bad' managerial decisions (reducing continuities of care) are as informative as good. We collated these findings across sites, thereby answering RQs 3 and 5.

Ethics and research governance

Ethical approval was obtained for the study from the NHS Research Ethics Committee system (reference 10/H0206/71), subject to informant anonymity and obtaining research governance approval from each research site, which we did for all the NHS sites. University of Plymouth ethical approval was obtained for non-NHS sites. All English sites, organisations, informants and patients are, therefore, given pseudonyms in this report. As noted above, obtaining permission for study site access was laborious and took many months. The growing importance of evidence-based policy-making contrasted with these obstacles to accessing the necessary evidence. No such difficulty arose in Sweden.

burgeswastand98.blogspot.com

Source: https://www.ncbi.nlm.nih.gov/books/NBK311231/

Share this post