General practitioners working in or alongside the emergency department: the GPED mixed-methods study

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 15/145/06. The contractual start date was in June 2017. The final report began editorial review in November 2020 and was accepted for publication in June 2021. 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’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Disclaimer

This report contains transcripts of interviews conducted in the course of the research and contains language that may offend some readers.

Copyright statement

Copyright © 2022 Benger et al. This work was produced by Benger 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 adaption 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.

2022 Benger et al.

Overview

In this chapter, we set out the background, context and aims of the general practitioners working in or alongside the emergency department (GPED) study. We show how GPED has developed in response to increasing emergency department (ED) demand and how it relates to previous initiatives, drawing on a range of published literature. A taxonomy of GPED models, developed with the other research team undertaking related work and supported by the same funding call, is described, and this has also been published. 1 By considering the key determinants of GPED, we identify a lack of clarity and consistency regarding the expectations of GPED and the effects it will have on patients, staff, the ED and the wider health-care system. The central roles of streaming and governance are briefly considered before we set out the aims and objectives of the research, the expertise of the study team and the structure of this report.

Background and policy context

Health-care systems across the globe are facing unprecedented challenges. This is particularly apparent in urgent and emergency care, which is experiencing steadily increasing demand. 2 In 2019, attendances at EDs in England reached record levels; 2018–19 saw an increase of 4.4% compared with 2017–18, and an increase of 21% since 2009–10. 3

Although attendance numbers decreased dramatically in March 2020 because of the COVID-19 pandemic (decreasing by 29.4% in March 2020 compared with March 2019), the proportion of patients requiring admission reached its highest ever level. 4 COVID-19 required the NHS to change its approach to managing demand, for example by reducing capacity in EDs to accommodate social distancing on the assumption that all patients might have COVID-19. 5 Despite an uncertain future, the Royal College of Emergency Medicine (RCEM) and other authors have stated that EDs need to be transformed to avoid returning to previous experiences of crowding and to provide the best care for patients. 6,7

High levels of attendance result in high occupancy, which is often referred to as ‘crowding’. Crowding occurs when there are insufficient resources to adequately meet patient demand. 8 ED crowding can undermine patient safety, clinical outcomes and quality of care because of increased waiting times9–11 and delayed service delivery,12 with associated increased mortality13 and reduced patient and clinician satisfaction. 14

Waiting times to be seen and treated in England also reached record levels during 2019. Only 79.8% of patients were seen and discharged, admitted or transferred within 4 hours of arrival in December 2019, which was the lowest recorded performance since the ‘4-hour target’ was introduced in 2004. 15

Alongside an acknowledgement that overall ED attendances are increasing, one hypothesis is that more patients are attending the ED either ‘inappropriately’ (i.e. with medical conditions that are not sufficiently urgent to require ED care) or for routine care that could be better delivered elsewhere. It has been estimated that between 15% and 40% of patients attending the ED could be treated in general practices,16–18 particularly children17,19 and those aged 16–44 years. 20,21 However, the lack of a standardised definition of ‘inappropriate’ can be problematic. Using criteria developed by a panel of experts from the RCEM, Morris et al. 22 found that the proportion of avoidable attendances was 19.4%, lower than previously reported. 16

Explanations for avoidable ED attendances include the convenience of access, a sense of anxiety on the part of a patient or carer about the severity of the patient’s condition, a lack of self-care skills,23 the availability of prompt and effective care and access to hospital-based investigations within the ED, a lack of awareness of alternatives to the ED,18 patient dissatisfaction with their own primary care provider24 and a belief that the ED is the correct place to source health care. 25 In a qualitative metasynthesis, Vogel et al. 26 identified four reasons that patients visit the ED over primary care:

1. perceived urgency of the medical condition

2. barriers to accessing primary care

3. advantages of the ED

4. fulfilment of medical needs and quality of care in the ED.

Similarly, a team from the University of Sheffield27,28 identified six mechanisms that explain why patients make clinically unnecessary use of emergency and urgent care:

1. need for risk minimisation

2. need for speed

3. need for low treatment-seeking burden

4. compliance

5. consumer satisfaction

6. frustration in accessing general practitioner (GP) appointments.

Some authors believe that this is exacerbated by reduced access to primary care. 2 However, others report that poor access to GP services does not appear to influence patient attendances to the ED29 and this issue remains unresolved.

Many initiatives have been introduced in the UK to address this challenge, including telephone advice and guidance (NHS Direct/NHS 111) and the provision of alternative facilities, such as walk-in centres and urgent treatment centres, at which patients can access primary care for non-urgent conditions. 2 In addition, over the past decade, EDs across the UK and Europe have implemented a range of new models of care co-locating general practice services in or alongside the ED. By 2015, some form of general practice co-location existed in 43% of UK EDs. 18

Rationales for introducing GPs in or alongside the ED, in addition to addressing the rising demand in attendance, include bringing vital primary care skills and expertise into the ED, reducing admission and investigation rates, improving patient care and reducing costs. 30 However, there is a lack of clarity regarding the mechanism(s) through which these benefits might be achieved, and the hypotheses that underpin the deployment of GPs in the ED are often vague and unclear. 31

In 2014, four Medical Royal Colleges produced 13 recommendations to improve urgent and emergency care. 32 The first of these was that ‘Every emergency department should have a co-located primary care out-of-hours facility’ (copyright © The Royal College of Emergency Medicine, Royal College of Physicians, Royal College of Paediatrics and Child Health and Royal College of Surgeons). 32 This recommendation was echoed in 2015, when a review of urgent and emergency care in the NHS in England proposed that selected patients should be directed or ‘streamed’ to an alternative health-care provider who could better meet their needs to reduce ED attendances. 33 However, the RCEM reported in 2015 that ‘less than half of EDs in the UK have fully implemented co-located primary care out-of-hours facilities and a third have no co-located primary care facility at all’ (copyright © The Royal College of Emergency Medicine). 34

In 2017, these recommendations were translated into policy. In March 2017, the Next Steps on the NHS Five Year Forward View was published and set out a ‘comprehensive plan to reduce the growth in minor cases that present to A&E [accident and emergency] departments’ (contains public sector information licensed under the Open Government Licence v3.0.; URL: www.nationalarchives.gov.uk/doc/open-government-licence/version/3/). 35 An accompanying statement said that ‘[e]xperience has shown that onsite GP triage in A&E departments can have a significant and positive impact on A&E waiting times’. 36

To support this intervention, the UK government announced a capital fund of £100M in the Spring Budget 201737 to which hospitals in England could apply to support or introduce GPED. 36,38 This announcement occurred at the same time as our team was awarded funding to complete the research that is reported here, and this initiative has influenced our study and our findings significantly.

Although the UK government and NHS leaders have proposed the adoption of front-door streaming and co-located care, data to support the benefits of GPED are limited.

Some evaluations of early adopters in the UK and Europe suggested that situating GPs in the ED has the potential to be a promising innovation39 that may reduce resource use40,41 and increase patient satisfaction. 42 Carson et al. 30 reported that the proportion of cases seen by GPs varied and that clinical and operational governance was often disjointed. In a survey of patients, Bickerton et al. 43 found that, although co-located care offered patients a greater range of health-care provision, it also increased the risk of duplication of services and repeat attendance. Similarly, evaluations of nurse-led walk-in centres co-located with the ED found ‘no evidence of any effect on attendance rates, process, costs or outcome of care’. 44 More recently, in a relatively small study, Uthman et al. 45 found that GPs who saw patients in the ED used fewer resources without increasing reattendance and referred more patients to follow-up services. In addition, service users appreciated simplified health-care provision from a single point of access, although this required clear communication between health service staff and patients. 46

Critiques of GPED’s ability to address high levels of ED occupancy include the arguments that eligible patients are often quick and easy to manage, do not breach the ‘4-hour target’ and do not contribute to crowding. 30,47 A recent realist review concluded that, despite GPs in ED being associated with a reduction in process time for non-urgent patients, this does not necessarily increase capacity to care for the sickest patients. 48 The main cause of ED crowding was perceived be because of congestion in the flow of sicker patients into the hospital and a lack of beds, rather than absolute attendance numbers. 49 In addition, GPED may encourage patients to present to the ED with a primary care problem, thereby increasing attendance numbers. 47,50

Several recent reviews have examined GPED in more detail. In a rapid review, Fisher et al. 51 concluded that there was insufficient evidence to support policy or local system design. In an update of Fisher et al. ’s51 rapid review, Turner et al. 52 reported that ‘the evidence base to support development of this model of care was weak and based on poor quality studies’ (contains public sector information licensed under the Open Government Licence v3.0). These findings were supported by a narrative review by Ramlakhan et al. ,50 which recommended a robust evaluation to inform future policy.

A 2012 Cochrane review53 concluded that there was insufficient evidence regarding GPED to make recommendations for policy or practice. However, the review was based on three non-randomised studies that did not assess patient safety or outcomes and were judged to be of low quality. An updated review54 in 2018 still found insufficient evidence from which to draw conclusions for practice or policy.

Based on current evidence, the most recent guidance55,56 from the National Institute for Health and Care Excellence (NICE) concluded that there was insufficient evidence to reach a recommendation on co-located GP units. NICE found inconclusive evidence that such units reduce the demand on the ED or improve clinical or economic outcomes. 55,56

Providing a reliable solution that applies across the complex range of ED services is challenging and requires adaptation to the specific context. Nevertheless, co-location of GPs in or alongside the ED (i.e. the GPED intervention) remains a preferred option. 57

To date, the consequences for the NHS workforce, both GPs and hospital staff, have not been well considered. It is not clear what the impact is for GPs, who are already under considerable pressure, and GPED may not be the best use of their time and skills. It is also unclear how GPED might have an impact on existing primary care services that are themselves experiencing a steadily increasing demand, with particular challenges in recruitment and retention, and how the additional costs of employing GPs, which may exceed the employment costs of other types of staff, will be met. 58–61

Some of the apparent impact of GPED on EDs may simply be relabelling of the same work, with no real benefits for patients or the NHS. Furthermore, any improvements reported may be attributable to employing additional senior medical staff rather than the fact that they are specialist primary care practitioners. 62 Co-located GP services may stimulate an increase in demand at hospital sites, transferring the problem of overcrowding from EDs to co-located GP urgent care centres (UCCs).

In response to these challenges, and the lack of reliable evidence, this study is one of two commissioned by the National Institute for Health and Care Research (NIHR) [the other being Health and Social Care Delivery Research (HSDR) project 15/145/04; URL: https://fundingawards.nihr.ac.uk/award/15/145/04 (accessed October 2021)] to evaluate the clinical effectiveness, safety, patient experience and system implications of the different models of GPED.

Taxonomy of GPED models

Departments that were early adopters of GPED were pioneers in the design and execution of co-located GP services. These EDs provided a blueprint for those that followed, with some used as exemplars. Foremost among these was Luton and Dunstable Hospital (L&D) (for more information on the L&D model, which was included as one of our case study sites, see Chapter 3).

There are several models currently in use without any evidence regarding which is the most effective and efficient model of care. 30 An evaluation of 13 EDs in the north of England found that sites varied considerably with regard to their models of care. 59

In 2010, three types of service were identified in a published taxonomy: off-site GP services, a service co-located with the ED and GPs fully integrated with the ED team. 30 It should be noted that the primary purpose or intention of these models may differ; for example, off-site and co-located services may be implemented to move a cohort of patients with problems suitable for general practice out of the ED altogether (potentially reducing the recorded number of ED patients) and integrated GPs may be intended to improve the care and experience of ED patients.

In 2017, in a study of 13 EDs in the north-west of England, Ablard et al. 59 described three distinct models: primary care services embedded within the ED, co-located UCCs and GP out of hours (OOH). In a concept paper published jointly by our study team and the other NIHR-funded team described above,1 a four-model taxonomy was described (Figure 1).

FIGURE 1.

Taxonomy of GPED. (a) Inside – integrated; (b) inside – parallel; (c) outside – on site; and (d) outside – off site. Reproduced with permission from Cooper et al. 1 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Cooper et al. 1 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

The four models are based on whether access to the GP service is inside the ED (patients access a primary care service within the ED) or outside the ED (patients access a primary care service that is distinctly separate from the ED). GP services located on a hospital site with no links to an ED are not included in this taxonomy:

1. Inside – integrated: the primary care service is fully integrated with the emergency medicine service.

2. Inside – parallel: there is a separate primary care service within the ED for patients with primary care problems.

3. Outside – on site: the primary care service is elsewhere on the hospital site.

4. Outside – off site: the primary care service is off site (may include telephone advice via NHS 111, or pharmacies, dentists, opticians, UCCs or registered in-hours or OOH primary care services).

Although this describes the structural location of the GPED service, there is still considerable variation in the way that GPED models operate across sites.

Factors influencing the implementation of GPED

This study was designed to examine the impact of GPED and the relative clinical effectiveness of the most common models of GPED currently in use. It is therefore important to recognise that these models can differ quite substantially. This might reflect the rationale for introducing GPED; local demands from both service users and community health-care provision; and practical issues, such as geographical space, financial constraints, and governance and managerial responsibilities.

Study aim

We aimed to explore the effects of GPED on patient care, the primary care and acute hospital team, and the wider urgent care system, and to determine the effects of different service models of GPED.

Study objectives

Report structure and overview of study plan

This was a mixed-methods study, comprising three work packages (WPs).

Work package A mapped, described and classified current models of GPED in all EDs in England. This classification was based on an updated taxonomy (see Figure 1) created in collaboration with the other NIHR-funded study to ensure consistency of terminology and, therefore, the comparability of findings. Our research team interviewed key national and local system leaders, staff and patients to identify the domains of influence and hypotheses underpinning GPED, and the potential mechanisms for benefit and disbenefit. We examined these further through WPs B and C.

Work package B used a retrospective analysis of routinely available HES data. 66 This was complemented by a detailed mixed-methods analysis in 10 case study sites. Our primary outcome measure was originally planned to be the number of ED attendances; however, in practice, this proved impossible and the changes that we made to the protocol as a result of this and other challenges arising during the study are considered further in the Chapter 2. Additional outcome measures included waiting times, admission rates, zero-day admission rates, reattendances, patient satisfaction and mortality.

In WP C, we reported a detailed mixed-methods analysis in 10 case study sites that were about to implement (six sites) or had already implemented (four sites) a new GPED model of care. In each of the 10 case study sites, we used survey and interview techniques to collect and synthesise quantitative and qualitative data to ascertain the views and experiences of GPED from staff working across the case study sites and from patients and carers attending the service.

The findings of all study WPs are reported in Chapter 3 and discussed in Chapter 4. The conclusions and implications for future research and policy are mapped out in Chapter 5.

Summary

In this chapter, we have set out the background of and context for GPED initiatives in England. We have described an underpinning taxonomy of GPED models of care and noted uncertainty regarding the anticipated impact of GPED at all levels of the system. ED streaming and models of GPED governance are identified as key determinants of success. We have stated our study aims and objectives and described the structure of the report that will follow.

Overview

In this chapter, we set out our research methods and show how these addressed each of the nine study objectives. We describe the protocol changes that were required as the research progressed, and the reasons for these modifications. Of particular note is the fact that we found GPED to be more widely adopted than we had anticipated, highly variable within and between sites, and extremely sensitive to local context. For each of the three WPs, we describe our approach to data sampling/recruitment, collection and analysis. GPED has effects across multiple levels of the health-care system and, therefore, we have adopted an approach that recognises macro levels, meso levels and micro levels of influence. The chapter concludes with a description of our approach to mixed-methods data synthesis, the ways in which patient and public involvement (PPI) have contributed to and enhanced this research and our approach to knowledge mobilisation.

Study design

As described in Chapter 1, this study adopted a mixed-methods approach with three WPs. The links between the nine study objectives and the three WPs are shown in Table 1. Each WP is described in more detail below and is illustrated in Figure 2.

FIGURE 2.

General practitioners working in or alongside the ED: efficient models of care – flow diagram. (a) WP A: mapping different models of GPED in England; (b) WP B: retrospective quantitative analysis of national data – costs and consequences; and (c) WP C: prospective and retrospective mixed-methods evaluation in 10 case study sites. a, Includes 4-hour performance, unplanned ED reattendance within 7 days, patients leaving the ED without being seen, emergency hospital admission, zero-day admission, length of stay, re-admission to hospital within 28 days and in-hospital mortality. CQC, Care Quality Commission.

Ethics and research governance permissions

This research study was approved by:

• East Midlands – Leicester South Research Ethics Committee (reference: 17/EM/0312)

• University of Newcastle Ethics Committee (reference: 14348/2016)

• Health Research Authority (HRA) (Integrated Research Application System: 230848 and 218038).

Changes to the study protocol

The study protocol has been published previously. 31 However, some important changes were required as the study progressed in response to ongoing research activity and the emerging findings.

Work package A: mapping different models of GPED and interviews with key informants to understand the hypotheses that underpin GPED and the experience of implementing these models of care

Work package B: quantitative analysis of national data to measure the clinical effectiveness of GPED using retrospective analysis of Hospital Episode Statistics

Statistical analysis

Each outcome was analysed separately and two alternative methodological approaches were considered: (1) a pooled analysis of all ED attendances to hospitals reporting GPED start and end times, in which differences in GPED opening hours across hospitals were used to separate the effect of GPED from the general effects of time of day; and (2) a regression discontinuity design (RDD) in which outcomes for patients attending the same ED shortly before or after the GPED service starts are compared to identify hospital-specific effects of GPED. Figure 3 provides a stylised representation of these analytical approaches.

FIGURE 3.

Stylised representation of alternative methodological approaches to quantify the impact of GPED using variation in GPED service opening hours.

All analyses use a critical value of α = 0.05 to determine statistical significance. No adjustment was made for multiple testing.

Pooled analysis

Our main approach was to pool data on all patients treated in English NHS hospitals in a single regression model and to use differences in GPED service availability across hospitals and times of day to identify the impact of GPED on outcomes. Nearly all hospitals for which we had data from WP A operated a GPED service for a specific, continuous subset of hours within each day, but the specific hours differ across providers. For example, a service might begin at 08.00 and end at 23.00 in one hospital, but run from 12.00 to 18.00 in another hospital. We quantified the effect of GPED by comparing outcomes for patients arriving at a given hour of the day in hospitals that (1) operate a GPED service at that time (the ‘treatment group’) or (2) do not operate the GPED service at that time (the ‘control group’).

We estimated separate ordinary least squares (OLS) (outcomes 1–7, i.e. linear probability models) and Poisson (outcome 8) panel data regression models for each outcome. Our main variable of interest was a binary indicator for whether or not the hospital operated a GPED model at the hour of the day when the patient attended the ED (yes/no). We also analysed the impact of different GPED models (inside –integrated, inside – parallel, and outside – on site; these models are referred to as ‘integrated’, ‘parallel’ and ‘on site’, respectively, in the rest of the report) by including a set of interaction terms between the GPED model and the binary indicator of service availability. All models controlled for hospital and hour-by-day of week fixed effects, month of year and, in the case of the OLS models, additional patient characteristics, such as their age (in 5-year bands), sex and socioeconomic deprivation profile, which may have acted as confounders. The hospital fixed effects captured time-invariant differences in performance across EDs that reflect hospital-specific factors, such as management quality, building infrastructure and the quality and availability of substitute health-care services within the local health economy. The hour-by-day fixed effects captured difference in service availability and patient acuity over the course of the day that follow a common pattern across all hospitals in England. For ease of interpretation, all regression coefficients are presented as average marginal effects on the original scale of the outcome variable. The estimates’ quantities refer to either percentage point changes in the likelihood of an event (for OLS models) or changes in the counts of events (for the Poisson model).

The scope of this pooled analysis was limited to hours of the day when there was variation in GPED service availability across hospitals. Patients attending EDs during hours of the day when all/none of the hospitals operated a GPED service contributed to the identification of hospital fixed effects and the influence of case-mix characteristics, but did not contribute to the statistical identification of the effect of GPED services on outcomes.

Regression discontinuity design

The pooled analysis estimated the average effect of GPED, calculated over all hospitals and hours of the day in our sample. To explore potential heterogeneity in the clinical effectiveness of GPED services across hospitals, we used a RDD that permits estimation of hospital-specific effects. In each hospital, there are some patients who attend the ED while the GPED service is in operation, whereas other patients arrive before or after those core hours and, therefore, are not treated by a GP. Patients attending at markedly different times of the day are likely to be systematically different in terms of their observed and unobserved characteristics, which precludes a direct comparison of outcomes for patients attending when the GPED service is operational with outcomes for patients attending at other times. However, we assumed that the exact arrival time of patients at the ED is exogenously determined (e.g. by the time of onset of their medical problem) and is not affected by the availability of GPED services. This implies that patients arriving shortly before or shortly after the time GPED starts are, in effect, randomly allocated to the control and treatment groups within the same hospital. The same principle holds for patients arriving shortly before (treatment group) and after (control group) the end of GPED each day. Consequently, we expect patients in both groups to have, on average, similar observed and unobserved characteristics so that any differences in outcomes between groups are unlikely to be biased by confounding. Put another way, the GPED start/end times create discontinuities in treatment allocation that the RDD exploits to obtain hospital-specific estimates of the effect of GPED on the relevant set of outcomes.

Under random allocation, the difference in outcomes between patients in the treatment and control groups can, in principle, be established using appropriate statistical tests (e.g. two-sample t-tests for continuous variables). However, residual imbalances in observed patient characteristics across treatment groups may remain, in which case inference can be further improved through regression adjustment. We therefore estimated the effect of GPED on outcomes 1–6 using logit regression models that include an indicator for treatment status and adjust for a set of observed patient characteristics. The volume of activity was not analysed with the RDD approach, as ED demand is known to fluctuate rapidly over the course of the day, so comparisons of volume across adjacent time intervals are not valid. As same-day discharge (outcome 7) is defined as discharge from an inpatient setting on the same calendar day as admission, time of ED attendance is strongly related to the probability of same-day discharge. Therefore, this outcome was also not analysed, as results would provide a partial picture only, strongly driven by the distribution of times when GPED starts and ends. We estimated separate models for each hospital and for each outcome around the start and end of GPED under the assumption that the effect of GPED may be partly driven by the hour of the day it was in operation, that is the effect at the start and end of the service may not be symmetrical. This resulted in up to 80 hospital- and time-specific estimates per outcome. We then summarised these estimates, with associated measures of uncertainty, using forest plots and random-effects meta-analysis, with subgroups defined by the type of GPED model in place (integrated, parallel and on site).

Some patients who arrive outside GPED opening hours may still be treated by a GP, for example because they are required to wait in the ED before being seen and a GP begins their shift in the meantime. These patients would be misclassified as belonging to the control group rather than the treatment group. We therefore excluded patients arriving during the hour before the start of GPED. Similar adjustments were applied around the GPED end time. These adjustments (described as 1-hour ‘doughnut holes’) may weaken the argument of random allocation and we tested for differences in observed characteristics of treated and control patients with a series of logit models, in which treatment status is used as the outcome variable and patient characteristics are used as covariates. Assuming that patients in both groups are similar in terms of observable variables, we would expect the overall explanatory power of these models, represented by the R-squared statistic, to be low. A cut-off value of 0.05 was used to exclude hospitals that showed poor balance, and, therefore, evidence of non-random treatment allocation, from the analysis. Based on this rule, three hospitals were excluded, leaving a sample of 37 hospitals.

Work package C: detailed mixed-methods case studies of different GPED models, consisting of non-participant observation of clinical care; semistructured interviews with staff, patients and carers; and workforce surveys with emergency department staff

Mixed-methods data synthesis

In addition to the individual analyses that were conducted for each of the GPED project’s three main WPs, we conducted a higher-level synthesis to integrate the study findings. This mixed-methods approach was pivotal to answering our research questions and for understanding and describing the complexity of GPED.

Integrating our findings was an iterative process that was undertaken throughout the GPED study, requiring regular discussion and sharing of findings between members of the research team. For example, the qualitative coding framework, pen portraits and WP A database were shared with the WP B team and were used to inform their knowledge of GPED models at sites that were included in their analyses. In some cases, this led to sites not being included in the WP B analysis because of uncertainty surrounding GPED operating hours.

In addition to this, we held three formal ‘mixed-methods’ roundtables, which were conducted during the final stages of the WP B and WP C data analysis. These meetings were held virtually as restrictions were in place because of the COVID-19 pandemic. The purpose of the meetings was to agree on how findings from each of the WPs could be integrated to demonstrate the impact of GPED from a mixed-methods perspective. Discussions were therefore centred around exploring complementary and divergent themes across each of our data sources. For example, how could our qualitative data be used to provide explanation and context to our quantitative findings? The inter-related nature of our findings meant that instead of presenting our results using separate chapters for each WP, we have chosen to report our study findings by theme in Chapter 3.

We then used NPT79 to reframe the overall data synthesis to understand the adoption of GPED into routine practice. NPT conceives making changes in established routines as a complex and dynamic enterprise, and proposes a model that explains the way in which new practices are adopted and absorbed by individuals into existing behavioural conventions and routines. The theory views implementation as a turbulent and unpredictable process that NPT, through its four core constructs (i.e. coherence, cognitive participation, collective action and reflexive monitoring), seeks to understand. NPT provides a theoretical framework that supports the integration of our qualitative and quantitative data and highlights the extent to which GPED has become a part of routine practice. Therefore, it is used to summarise and interpret our findings at the start of the Chapter 4.

Patient and public involvement

Patients and members of the public were involved throughout the development and delivery of the GPED research project. This began with an initial group of public contributors that was formed to provide input on the development of the grant proposal. This group formed the nucleus of the public involvement group that has worked with us since that time.

The group consisted of 10 public contributors with a wide variety of experiences of using ED services. Some had interacted with ED services as carers for children, partners or older parents, and others had used ED services as patients in their own right. Our public contributors came from a diverse range of social backgrounds and were of white British ethnicity.

Throughout the study, the group was kept informed of progress and was involved in regular workshops and meetings. Two members of the group were also full members of the SSC.

Early public involvement in the study focused on the preparation of data collection instruments and the ethics committee applications.

Knowledge mobilisation

Summary

This chapter has described our research methodology, showing how the three WPs combined to address the nine objectives of the project and explaining the protocol changes that were required as the research progressed. We have set out our approach to data analysis and synthesis, described the central role that PPI played throughout and described our approach to knowledge mobilisation.

Overview

In this chapter, we describe our research findings. We begin by reporting elements of data collection from all three WPs, individually, and then prepare for our mixed-methods analysis by considering the policy context and describing eight ‘domains of influence’ identified during WP A and WP C. The main section of the results then considers each of these domains in turn, combining quantitative and qualitative data from all three WPs to understand the potential effects of GPED and the factors that influence these.

Data collection from work package A

From our WP A database, we attempted to determine the GPED model(s) provided by all 177 type 1 EDs (consultant-led 24-hour services with full resuscitation facilities) in England at two time points: September 2017 and December 2019. Models were classified according to our iteratively developed taxonomy: inside – integrated; inside – parallel; outside – on site; and outside – off site (see Figure 1).

Data were obtained from 163 out of 177 (92%) of all the type 1 EDs in England:

• 149 out of 177 (84%) at September 2017

• 160 out of 177 (90%) at December 2019

• 139 out of 177 (79%) at both time points.

The GPED models in place in September 2017 and December 2019 are shown in Table 4. Capital funding was awarded to 87% (142/163) of the participating EDs.

Between September 2017 and December 2019, 23 sites commenced and four sites ceased GPED provision. Three of the sites that ceased provision chose to discontinue an inside – integrated model. The most common service change (20 sites) was from an inside – integrated model to an inside – parallel model. In addition, 11 sites moved from an outside – onsite to an inside – parallel model.

We found no significant difference between group means by the type of GPED model adopted and the observable characteristics of included EDs (e.g. annual number of new attendances, deprivation, rurality, receipt of capital funding).

Data collection from work package B

The hours of operation for sites included in the WP B analysis are shown in Table 5.

Appendix 2, Table 24, summarises the main findings of our pooled analysis. The table reports results from all sites for which data were available, and separates these by GPED service type (integrated, parallel and on site). This pooled analysis uses the fact that GPED services operate at different times of the day, so the overall effect is identified by comparing the outcomes of patients arriving at a given hour of the day to hospitals operating or not operating a GPED service at that time. These results are considered further in Mixed-methods synthesis.

Data collection from work package C

Researchers collected data at three time points (0, 6 and 12 months) from six sites, and at two time points (0 and 12 months) from the other four sites. This reflected our original plan to divide these sites into ‘prospective’ and ‘established’ categories; however, as described in Chapter 2, this distinction was not identified in practice (Table 6).

A total of 10 interviews were conducted at the macro level and 57 interviews (representing 64 EDs) were conducted at the meso level. Ten EDs were recruited as case study sites at the micro level. See Table 3 for a summary of our data collection and participant characteristics. To protect the anonymity and confidentiality of sites and participants, we have summarised our data collection in a combined characteristics table. The qualitative data reported here were collected during WP A and WP C (see Figure 2).

A total of 472 WFS responses were received across the 10 GPED case study sites. We endeavoured to determine the response rate to our WFS in each site; however, this was complicated by the fact that the survey was disseminated through both paper and electronic media on several occasions. In three out of our 10 case study sites where we were able to reliably determine a denominator, the response rates from staff were 9.4%, 25.5% and 36.4%, giving an average of 24%. A small number of surveys (n = 23) could be matched to individuals using a unique identifying code. However, given the limited number of these longitudinal data, and the decision to abandon any distinction between prospective and established sites, the analyses focused on cross-sectional data derived from surveys obtained from (1) established sites and (2) the baseline time point at prospective sites. These data were viewed as the most comparable and readily combined, producing a total analytic sample of 383 surveys.

Characteristics of the cross-sectional analytic sample are displayed in Appendix 3, Table 26. The majority of survey respondents were female (64.8%, n = 248), with the largest proportion of respondents being aged 18–34 years (41.8%, n = 160) and reporting having worked for their current NHS employer for 1–5 years (36.6%, n = 140). The largest proportions of respondents reported being nursing staff (41.3%, n = 158) or ED specialists (26.4%, n = 101), and indicated that they delivered care in the ED (67.1%, n = 257). Eleven per cent of the respondents (n = 42) indicated that they delivered care in the GP service related to the ED and 4.7% (n = 18) reported being GPs. The small number of GPs in the survey sample is arguably proportional to the number of GPs employed in EDs under the GPED model.

A summary of the WFS results and NoMAD findings can be found in Appendix 3. For the purposes of this primary report, a limited range of survey results is reported and selected according to mixed-methods considerations. That is, survey results were prioritised for reporting if they either aligned with or provided a potential contrast with findings from the qualitative data (thus allowing a form of triangulation between the data sources) and are incorporated into the mixed-methods synthesis below.

Mixed-methods synthesis

GPED domains of influence

We present our main results according to eight domains of influence, or areas that participants felt would be affected by the introduction of GPED. From the outset of the study, we were aware of a lack of consensus as to the desired or anticipated effects of GPED and, therefore, we sought to identify a series of hypotheses that could be tested in our research. However, as the work progressed it became apparent that in many areas even the direction of the effect was contested; thus, rather than establishing unidirectional hypotheses, we have chosen to identify more general domains of influence. These domains were identified through analysis of 228 semistructured interviews with national policy-makers, local service leaders, health professionals, patients and carers (WP A and WP C), and were developed through a series of roundtable discussions undertaken by the research team. The domains were refined and revisited during researcher meetings on an ongoing basis throughout the data collection and analysis, and are used as a framework for structuring our discussion of the impact of GPED.

In our WFS, most participants understood the potential value of GPED in principle: 78% of staff (n = 299) who responded agreed or strongly agreed with the statement ‘I can see the potential value of GPED for my work’, and 65.5% (n = 251) agreed or strongly agreed that participation in GPED was a legitimate part of their role. The overwhelming majority agreed or strongly agreed that they were open to working with colleagues in new ways to use GPED (91.6%, n = 351). However, reflecting the varying local needs and contexts and the different ways in which GPED had been introduced, there was disagreement at the individual, stakeholder and organisational levels about the purpose and anticipated impact of GPED. For example, only half (54%) of staff agreed or strongly agreed with the statement ‘the staff have a shared understanding of the purpose of GPED’.

In the subsequent sections, the impact of GPED on each domain will be explored using quantitative and qualitative data, as available. However, evidence for some domains will be either solely qualitative or quantitative because of the data that are available and the nature of the domain. This is summarised in Table 7.

Table 8 summarises the arguments that stakeholders posed regarding the potential impact of GPED, with most domains including arguments for positive, negative or no effects. Further details of how the domains of influence were identified and descriptions of the anticipated effect of GPED on each domain have been published separately. 94

In the following sections, we consider each of the eight domains of influence in turn.

Domain of influence 1: performance against the 4-hour target

There was considerable uncertainty regarding the expected impact of GPED on the ‘4-hour target’, that is that 95% of ED patients should be admitted, discharged or transferred within 4 hours of arrival. Although this was quoted as one of the potential benefits of GPED, not all stakeholders were convinced of the likely impact, given the site-specific patient mix, the selected GPED model and significant uncertainty as to whether or not GPED-streamed patients should be included in ED performance measures such as the 4-hour standard.

If GPED patients are counted in the ED group, then the premise on which GPED is based predicts that GPs in the ED will process these patients more efficiently and quickly (e.g. less testing, fewer unnecessary admissions) than ED staff. However, it was noted that annual growth in ED workload may mask the visibility of any improvements. If GPED patients were not included in the ED count, then this would reduce the number of ED patients, with the potential to reduce the number of patients exceeding the 4-hour standard. However, by diverting people with minor conditions, the acuity of those patients remaining in the ED is likely to increase. These more complex patients may require longer management, which could negatively affect performance against the target.

In our pooled analysis of HES data66 (WP B), we found no impact of GPED on the overall 4-hour performance (Table 9).

Figure 4 shows a forest plot derived from our RDD analysis of 4-hour performance in 35 EDs, reflecting ED throughput and some aspects of patient experience. It illustrates 4-hour performance in these 35 EDs around the times that GPED services start and finish, excluding a 1-hour period before and after the start and finish time (‘doughnut hole’), as described in Regression discontinuity design. Effects to the right of the line indicate better performance (shorter waits), whereas those to the left of the line indicate worse performance. In all forms of GPED, the start of the service each day is not associated with any significant change in the percentage of patients who stay in the ED for > 4 hours. Following the end of the service each day, there appears to be a slight reduction in the proportion of patients remaining in the ED for > 4 hours, but considerable heterogeneity remains. It is also important to note that, although the results of the RDD analysis (see Figure 4) are helpful in demonstrating the degree of heterogeneity that exists, the RDD summary measures are considered less reliable than the pooled analysis (Table 10) for the reasons described in Chapter 2.

FIGURE 4.

Forest plot illustrating the effect of GPED on 4-hour performance. Percentage discharged within 4 hours of arrival, with 1-hour doughnut hole at (a) GPED start; and (b) GPED end. REML, restricted maximum likelihood.

It is perhaps unsurprising that no clinically significant change was observed in performance against the 4-hour target in our pooled analysis. Differing approaches to counting patients across sites and the factors influencing time spent in the ED are likely to be working against each other, drawing any potential differences towards the null. In addition, the flow of patients through the ED is heavily influenced by factors other than GPED, such as the availability of inpatient beds, and diagnostics and timely reviews by specialist inpatient teams:

The complicated factors because we refer a lot of patients to specialty, some of the breaches are actually not due to us, but due to waiting long times to see specialties.

GP (Teak) 040, time 3

In the Linden case study site, a reduction in the 4-hour wait had been noted; however, this was due to additional steps being taken to ensure that complex patients who were not eligible for GPED were taken out of the ED system within the 4-hour target period:

That’s one of the reasons why we’re successful, and we’re talking about KPIs [key performance indicators], we are successful at the 4-hour target. Because we look at it from a quality perspective rather than a process perspective, rather than just the target. We’ve got to get them out, we’ve got to get them out. It’s about, it’s about the entire process. It’s about streaming, it’s about seeing patients within the first hour. It’s about diagnostics in the second hour, it’s about decisions in the third hour, by the fourth, between the third and the fourth hour the bed must be allocated. There must be a direction for where they’re going.

ED system lead (Linden) 031, time 1

The 4-hour target was also considered a blunt tool for evaluating impact; clinical indicators such as time to CT or time to receiving antibiotics were suggested as more relevant measures of good clinical practice.

Domain of influence 2: use of investigations

There was a lack of consensus as to whether or not GPED models should give GPs access to investigations and diagnostic tests, such as radiographs and blood tests. These differences in opinion were a source of tension and disharmony between staff working in EDs and GPED, and had implications for what people considered to be the purpose and impact of GPED. To this end, decisions regarding GPs’ access to investigations and diagnostic tests were often made during the design of GPED models at a given site, and were related to the service configuration and system of governance in place.

A number of site staff perceived giving GPs access to investigations as crucial to the model’s clinical effectiveness and worried that the potential scope of GPED would be limited if GPs were not able to undertake investigations and refer to specialties. Participants also felt that not giving GPs access to investigations reduced the clinical effectiveness of GPED by limiting the range of presentations that GPs could treat:

So, some people with abdomen pain and things, we sometimes do a little bit of the investigations as well . . . Try to take the pressure off the guys in majors really . . . So we don't do all the investigations, we don't do all the bloods, we don't wait for the results.

Locum GP (Birch) 016, time 1

In contrast, others believed that GPs should work in the same way as they believed that GPs to work in a community setting. These beliefs were based on perceptions that giving GPs access to investigations and diagnostic tests requires GPs to work outside their clinical competency and encourages a ‘hospitalisation of GPs’, which causes them to become middle-grade ED or ambulatory care doctors and lose their unique GP skills. However, there was a sense that, even if investigations were available, these were not used by GPs to the same extent as, perhaps, they would be in the ED and that the risk of acculturation was minimal:

But my thought would be while ever I’m organising a blood test and keeping a patient here that doesn’t need to be here because they’ve not got an urgent or emergency care need, they have got something that could go back to their own GP, I’m causing delays for other patients. I’m also reinforcing the behaviour of ‘if you go there they will do it today, you’ll not have to wait’. I’m also creating an unseen risk because, OK, I might do somebody’s blood test for their full blood count today and it might be just below the normal range and I’d say that’s probably alright. But without the context of having been to your own GP and got a trend of results that actually this has fallen over the last four blood tests and I’m worried about other stuff now. You’ve took that . . . that patient has got no awareness at risk and actually the clinician might be blind to that risk as well.

GP lead (Redwood) 013, time 3

Similarly, some ED staff felt that GPs did not possess the skills required to interpret investigations, particularly radiographs and scans. These concerns were heightened by fears that GPs may not be supported to use diagnostics and so would feel pressured to undertake ED work and/or be unable to recognise higher-acuity patients, with a consequent risk to patient safety and outcomes:

Actually looking at X-rays and ECGs [electrocardiograms] is, it becomes a bit of a, a dying art in general practice, if you’re not looking at those sorts of things on a daily basis, and what we provide again is allowing GP’s the ability to keep those sort of clinical skills up and running, when I think that, and I think that’s the attractiveness about doing this.

UCC clinical lead (Teak) 017, time 1

It was acknowledged that giving GPs access to investigations and diagnostic tests may provide a more convenient and efficient process for patients. However, patients may choose to inappropriately attend GPED as opposed to traditional general practice to gain access to further investigations and tests, thereby increasing the department’s workload and ED service pressures:

I suppose a bit more than a, than a clinic, because we are able to do these things [investigations]. That’s probably another reason why people do come here as opposed to, um, go elsewhere.

Band 5 ED nurse (Teak) 013, time 1

The potential for patient confusion when accessing a GP in the ED was noted, relating to the extent of the service that patients expect and the crossover of the general practice and ED. GPs were often faced with patients who did not understand why they could not access investigations that were available on site. It was considered that there needed to be clarity and consistency about the services on offer, which should then be communicated with patients. However, sometimes GPs themselves were unclear about which investigations were available to them and whether or not it was appropriate to use these investigations:

People turn up at a hospital and if you say ‘right, I’m a GP, but I can’t get an MRI [magnetic resonance imaging] scan for you’ [they say] ‘but you’re in a hospital, why can’t you?’, you know . . . in their heads, they can’t understand that. So people don’t turn up in the GP surgery requesting an MRI scan. Why? Because they know there are no MRI scans there, there’s no X-rays there, you know, so you can say to them ‘look, we don’t do that here’ and they can really accept it, and you come here, there’s a different thing. You have to work out what it is you want to do with your service, and then figure a way to keep it right; if you don’t want people coming in unnecessarily, then you, it mustn’t pay, it mustn’t benefit them to do that . . . I work here as a GP pretty much the way I work in surgery.

GP (Chestnut) 003, time 1

Domain of influence 3: hospital admissions

Staff at Linden highlighted that diverting people with minor conditions, which are, theoretically, quick to resolve, to GPs had screened minor issues out of the ED and resulted in ED admission rates that were now above the national average:

Obviously, with us streaming a proportion of our minor patients away it does make our admission percentage look a lot higher than the national average, um, because, obviously, we haven’t got those patients in our denominator, um, so that has been, that has been a, a query that we’ve had to answer on a, a number of occasions, though I think everyone’s got the message now.

ED system lead (Linden) 032, time 1

This respondent identified that the apparent change in admission rate is due only to reclassification and a consequent change in the denominator used in reporting statistics, rather than being a genuine effect.

When looking at our case sites more broadly, the impact of GPED on admissions was considered variable and, to some extent, associated with the different approaches to risk demonstrated by GPs and ED staff, and also individual views on whether or not GPs should have access to investigations and diagnostic tests when working in the GPED model. GPs were perceived to frame health and illness in a different way from ED staff, with the ‘wait and see’ culture of primary care leading many to view GPs as more ‘risk tolerant’ and more appropriately qualified to care for lower-acuity patients than their ‘risk-averse’ ED colleagues. This, in turn, was thought to be beneficial for GPED because GPs were viewed as less likely to admit or refer lower-acuity patients, order investigations and admit a patient unnecessarily. GPs were, therefore, perceived to be better placed to deal with lower-acuity patients or those with long-term conditions who need skilled holistic management rather than specialist investigations. Although some ED staff exhibited caution about what presentations to refer to GPED, ED staff were criticised for overinvestigating through fear of missing something important. GPs’ risk behaviour was generally viewed as beneficial; however, several GPs and ED staff expressed frustration that they were unable to order investigations and treat a wider range of patient presentations:

But what we are good at doing as GPs is identifying risk thresholds of, ‘this is something I can sort out and I’m happy to take that risk’. This is something why it’s beyond the risk that I’m willing to accept as something that I’m going to manage. What the A&E consultants struggle to grasp or chose not to, I’m not sure which it is, is that if I’m out there in a GP surgery and I see a patient is very poorly whether it’s because I don’t know what to do next or because I’ve decided you’re really unwell and need to go to A&E is never questioned. I can pick up a phone and I can dial 999 and somebody comes along in an ambulance and deposits them in the A&E department, at which point somebody sorts them out. When that threshold is met by a GP or an advanced clinical practitioner working here in the department, some ED consultants will not accept that transfer without a barrier to that by wanting to know all the ins and outs of what’s wrong with the patient. Trying to suggest helpful things that you might do to manage this patient as opposed to just accepting that.

GP lead (Redwood) 013, time 3

Our pooled analysis showed no effect of any GPED model on patient admission or non-same-day discharge attendance from the ED (see Table 10):

Figure 5 shows a forest plot derived from our RDD analysis. It shows the proportion of patients admitted from 32 EDs around the times that GPED services start and finish, excluding a 1-hour period before and after the start and finish time (a ‘doughnut hole’), as described in Regression discontinuity design. It confirms substantial heterogeneity but no overall or consistent effect of GPED on hospital admissions from the ED.

FIGURE 5.

Forest plot illustrating the percentage of ED attendances admitted as inpatients. Percentage of ED attendances admitted, with a 1-hour doughnut hole at (a) GPED start and (b) GPED end. REML, restricted maximum likelihood.

Domain of influence 4: patient outcome and experience

Overall, our patient interviews found that a positive patient experience of GPED was more likely if a patient was seen in a timely manner and if they felt that their complaint was taken seriously and was addressed in a reassuring and professional manner. Some staff expressed concern that patients would prefer to be seen by an ED doctor and reported feeling that they were in ‘the firing line’ when patients objected to seeing a GP. However, there was variation in patients’ opinion and knowledge of the expertise of GPs: some were happy to see a GP as they regarded them as having more experience than ED doctors, whereas others felt that ED doctors were the more senior and experienced staff. Some patients were aware of the role of other health-care practitioners [e.g. advanced nurse practitioners (ANPs)] and were happy to be seen by them, if appropriate. Indeed, many patients did not appear to know the type of clinician they had seen:

No. All I get told is it’s a doctor. That’s all you get told.

Patient (Rowan) 16, time 3

Patient expectations were potentially harder to meet when patients were aware of GPED within the department. Staff experienced both sides of the coin: patients unhappy about being sent to a GP because they felt that their condition was serious enough for ED and/or they wanted investigations, and patients who did not want to be sent to the ED because they believed that the waiting times for the GP would be shorter and it was less chaotic than the ED.

In general, however, patients were positive about GPED, with the majority of patients we interviewed being indifferent to the type of health professional they saw as long as they received appropriate care. Equally, most patients were happy to consult with a GP and, potentially, wait longer to be seen as they recognised that this was because their symptoms were not considered to be too serious. Patients therefore valued GPED and believed that it was beneficial to have GPs in the ED to enable less unwell patients to be seen and ‘relieve the pressure on the emergency doctors to actually be dealing with the real emergency cases’ [Patient (Poplar) 015, time 1]. It made sense to patients to have GPs in EDs as they were doctors, just with a different specialty.

Patients were also of the opinion that GPs in EDs could offer better care and a more thorough service than a patient’s own primary care GP. Patients felt that it was an easy option in terms of accessing care and guaranteed that they would be seen in a more timely manner than traditional general practice. The 4-hour target was not a high priority to patients; although they appreciated being seen more quickly, when possible, they were also happy to wait longer knowing that they would be seen at that visit. This arbitrary indicator was of far greater relevance to hospital managers than to patients:

Yeah, I think that’s really important, I think given the way the hospital performs with the government’s 4-hour target, I think it’s a source of pride for the hospital, for the Chief Exec[utive].

ED consultant (Linden) 029, time 1

Although patients were largely receptive to GPED, there were concerns from staff that patient flow could be negatively affected and a backlog created by requiring patients to disclose clinical information on multiple occasions before seeing a GP. Some staff reported that patients presented inconsistent medical information during their ED/GPED journey. Patients who attended for a second opinion after seeing their primary care GP or those who were referred to the ED by their primary care GP were less satisfied with GPED:

Because I was sent, I went to the GP, then it’s referred to the hospital, so in the circumstances today, I think I would prefer to see a specialist then, rather than a GP.

Patient (Poplar) 23

Some patient safety issues were also identified. The NHS requires urgent (‘2-week wait’) referrals for potential cancers to be provided by the patient’s own GP. 95 Therefore, patients may not be appropriately referred to secondary services or delays may occur if the GPED service advises them to go back to their own GP. Across sites, GPED had no review or follow-up system, so there was no way of checking whether or not patients had attended their own GP. Referring back to the patient’s own GP for onward referral was also considered highly inefficient by staff, given the unnecessary doubling of health-care attendances. Although it went against policy, some GPED staff did provide urgent referrals to prevent patients from slipping through gaps in the system. Likewise, GPED GPs gave ‘safety-net’ advice to patients, explaining what they should do if their condition deteriorated but, again, systems were not in place to follow up patients. This was mitigated somewhat by the production of electronic discharge letters to inform the patient’s own GP of GPED attendances and any follow-up/safety net advice given.

Although staff were concerned that knowledge of GPED was leading to an increase in patients attending the ED to see a GP specifically, most patients stated that they would still use their primary care GP as their first port of call and use the ED in an emergency only. However, they did voice concerns that other patients might use the ED more often to see a GP if they knew that the service was available.

Our pooled analysis identified no effect of GPED on the proportion of patients who leave the ED without being seen or who die within 30 days of attendance (Table 11). There was, however, a very small but statistically significant reduction in the rate of unplanned reattendances at the same ED within 7 days (–0.3%; p = 0.015) for patients attending during GPED service hours, but the clinical significance of this was judged to be negligible. This finding appears to be driven by integrated GPED services (–0.6%; p = 0.028), with estimated nil effects for the parallel and on-site GPED services (see Table 11). Note that these results may still reflect chance findings owing to the large number of tests conducted. Using a more conservative critical value of α = 0.0016 (based on the Bonferroni adjustment procedure and allowing for 32 tests), the results of the outcome of unplanned reattendance would not be considered statistically significant.

This is further illustrated by the forest plots arising from our RDD analysis, which are shown in Figures 6–8.

FIGURE 6.

Forest plot illustrating the percentage of patients attending the ED who leave without being seen, with a 1-hour doughnut hole at (a) GPED start and (b) GPED end. REML, restricted maximum likelihood.

FIGURE 7.

Forest plot illustrating the percentage of patients attending ED who have an unplanned reattendance to the same ED within 7 days, with a 1-hour doughnut hole at (a) GPED start and (b) GPED end. REML, restricted maximum likelihood.

FIGURE 8.

Forest plot illustrating the percentage of patients who die within 30 days of an ED attendance, with a 1-hour doughnut hole at (a) GPED start and (b) GPED end. REML, restricted maximum likelihood.

Figure 6 shows considerable heterogeneity at GPED start, with no consistent overall effect. At GPED end, there is a more consistent trend towards an improvement (i.e. decrease) in the proportion of patients who leave the ED without being seen; however, the effect is very small, and not confirmed in the more reliable pooled analysis (see Table 11).

Figure 7 confirms a statistically significant reduction in the rate of unplanned ED reattendance to the same ED within 7 days, which is associated with GPED. However, the clinical significance of this difference (< 1% change in 7-day unplanned reattendance rate) is negligible. This is considered further in Chapter 4.

In keeping with Table 11, Figure 8 shows no statistically significant effect of GPED on patient mortality at 30 days. The pooled estimates for both GPED start and finish are extremely close to zero, with very narrow confidence intervals. This is reassuring from a patient safety viewpoint, but is also expected, as there is no reason to anticipate that GPED would affect patient mortality.

Domain of influence 5: service access

Increasing use of emergency departments

There was variability with regard to patients’ knowledge of the health-care and ED systems. Reflecting the different stages of GPED implementation, patients from only a small number of case study sites reported having previously used the ED and expected to be streamed to GPED. Indeed, even those with prior knowledge of the system expressed uncertainty as to how the service worked or who it was aimed at. Some patients knew that they could see a GP in the ED, but the majority of those interviewed did not, which is perhaps surprising given that most sites had been implementing a form of GPED model for some time:

I think, from what I’ve found working in A&E for the last 4 years, there are still quite a lot of patients in the area that don’t know the service exists. When I’m assessing patients in triage and saying, ‘I’m going to send you to the out of hours’ they say ‘I didn’t realise you had that service.’ I explain ‘if you call 111, they can, if needs be, send you to us or if they feel it’s appropriate they can book a GP out-of-hours appointment’. A lot of patients don’t realise the service . . . They just come straight here. They don’t realise there’s an out-of-hours service, which is quite surprising, really, because it’s been here quite a few years.

ED nurse (Hawthorn) 010, time 3

Staff were worried that easy access to a GP through the ED would encourage further attendance and that the greater the supply, the more the demand would increase (i.e. supply-induced demand). There was also a perception that, if GPs refer to the ED patients who are subsequently seen in GPED, patients will then start to attend GPED directly, rather than their own primary care GP. Staff were anxious that GPED would become a ‘victim of its own success’, with patients attending frequently; hence, they did not want patients to be aware of the GPED service:

Do you know [what] I mean? But ED’s frightened to send anything away, so everything comes in. So I don’t blame the public for attending if they can see a GP within 3 hours, rather than having to wait 6–7 days or 2 weeks for an appointment. But I just wonder if it’s made a demand for it, because you get people coming back to see the GP again in ED.

ED nurse (Rowan) 037, time 2

At most sites the service was not advertised to avoid encouraging attendance for GP-type presentations:

We’re not shouting it from the rooftops because the minute you start shouting from the rooftops that you’re going to have a GP service within the ED then people are gonna be already camping outside waiting with their, you know, verrucas and things.

ED senior nurse (Juniper) 003, time 1

Staff perceived there to be an increase in the number of patient attendances at the ED both generally and between our research team’s site visits. Interviews with patients and carers suggested a number of reasons as to why individuals present at the ED, such as not being able to get a GP appointment or not being registered with a GP, wanting further investigations they knew would not be available at a primary care general practice or simply finding the ED more convenient than a primary care GP appointment. These reasons are summarised in Table 12.

TABLE 12 - Reasons for patients’ attendance at the ED

C, carer; P, patient.

Reason for ED attendance	Summary	Quotation
Difficulty accessing GPs in the community	Most patients stated that they attended the ED because they could not get a timely GP appointment with their primary care GP. There was variability as to whether or not patients had tried, on this occasion, to get a primary care GP appointment. Some patients were not registered with a primary care GP because they were a temporary resident in the area or did not need medical care previously	Just because I’ve been to my GP before and they’re not very helpful and it’s a long waiting list and it’s now, I’m not sleeping at night. I can’t function at work properly and the past couple of days it has been unbearable and so I needed something like a quick relief rather than waiting for another appointment at the GPP (Nutmeg) 013, time 3Um, my daughter phoned the doctors ’cos she was feeling unwell. It’s been going on for a little while. But the doctor said ‘there are no appointments available’. So he suggested we either ring 111 or come up to you here at the [general practice] clinic and be seen here, so that’s what we decided to doC (Linden) 009, time 1
Advised or sent to the ED by other urgent care service	Patients were advised to attend the ED by NHS 111, GPs and other health-care specialists. GPs advised ED attendance if they felt that the problem was outside their remit or if they had no available appointments. NHS 111 referrals were a cause of frustration for staff, who often felt that patients were referred unnecessarily	I have got a really bad toothache with an abscess on my tooth, which is coming up next week. The pain got a lot worse so much it spread up to the right-hand side. And I tried to persuade the dentist to drain the abscess to, sort of, relieve the pain. And the dentist told me that they don’t do that surgery. And they referred me, they said I should go down to A&EP (Juniper) 056, time 3I went to the GP about 10 past 10 because I’d fell over in the garden about 4 weeks ago. She had a look and she thought, well, it’s best if I come to the Birch for an X-rayP (Birch) 002, time 1
Attending the ED as it is convenient	GPED was easier than booking a primary care GP appointment and was seen as more beneficial than walk-in centres that were nurse led Although most patients did not know it was possible to see a GP in the ED, those that attended the ED frequently often knew this was possible; it was these patients who were more likely to attend the ED as they knew they would be seen and, for some patients, because they knew it was a convenient way to be seen by a GP specifically	Because my GP is overcrowded. Never get appointments and I know if I come here, if I need anything, I’ll be able to go to hospital straightaway instead of going there and back all the time. I can just get seen hereP (Chestnut) 014, time 3OK, just because of history and the familiarity of it, I know that I can get my issue resolved here. They’ve got my records so it’s a lot easier for them just to see what’s going on and be able to direct me from thereP (Teak) 007, time 1
Patient expectations	Some patients bypassed their GP or ringing NHS 111 because they expected a certain level of care/investigations that they believed that only the ED could provide. This was often because they believed that further tests or radiographs would be required	I mean, I don’t think, I hardly come in to A&E and the only reason I’ve come here is because I know that if I went to the GP they would have told me to come here, so it’s saved me the trip, so I just came straight here . . . To be honest, with me, I look tired but kind of made that deduction myself that I should actually come here rather than go to the GP and save myself timeP (Nutmeg) 011, time 3Yes. I think I need an X-ray, is why I came. Just to rule out not having fractured or broken my wristP (Hawthorn) 03, time 1
Poor relationship with GP and obtaining a second opinion	A number of patients attended the ED because they did not feel valued by their primary care GP (e.g. they felt rushed and not listened to). Others did not agree with what the primary care GP had said or their symptoms had worsened since their primary care consultation	So, when I saw my GP yesterday I was, I came away feeling reasonably reassured, we had quite a long conversation, but [sighs] come lunchtime afternoon the pain got worse and worse and worse and by the time I got home it was, you know, it was making me weep, and then I suddenly thought actually he hasn’t, at no point has he actually examined me, you know, and because we’re going away at the end of the next week, I thought I can’t wait to go through the whole process of going back to see him again and, you know, asking for a more thorough consultation or even for a referral or something and that’s why I, and I was just so distressed last night I thought the only way I can do this is, to get something immediate is to come up hereP (Juniper) 054, time 3

Our quantitative analysis did not identify an association between the absolute and relative volume of attendances and GPED (Table 13).

TABLE 13 - The effect of GPED on the absolute and relative volume of ED attendances

Volume of attendances	All EDs	GPED model			Number of patients
		Integrated	Parallel	On site
Absolute volume
Coefficient	0.001	–0.015	0.034	–0.0002	336,882
p-value	0.970	0.673	0.414	0.996
Relative volume
Coefficient	0.001	–0.018	0.039	–0.001	336,882
p-value	0.931	0.635	0.380	0.972

Health service literacy

Staff believed that patients lacked health service literacy and required more education about the purpose of ED services. However, they acknowledged that the constantly changing organisation of the systems and services at which patients could seek care would be confusing for patients and have a negative impact on their working knowledge of the optimal use of available services. This was witnessed at some sites, with patients thinking that they were attending a walk-in centre rather than an ED. Frequently, these patients had been misinformed or directed to the ED by NHS 111, or the service provided had changed since their last attendance. As we have shown, this confusion could be compounded by sites consciously not making it clear what types of services were available, with the intention being to reduce the perceived risk of generating additional demand. Indeed, Table 13 indicates that concerns relating to an increase in demand caused by the introduction of GPED services had not come to fruition, possibly because of a combination of these factors. There was very little evidence from the qualitative data to suggest that patients were presenting to the ED with the specific intention of seeing a GP.

Opinion was divided, particularly among GPs, on whether or not staff should ‘educate’ patients about appropriate attendance. Some felt that this was not their role and were uncomfortable with this suggestion, whereas others happily told patients that they should not have gone to A&E:

. . . or their perceived importance of their condition, that they need to be seen urgently. When, in fact, it’s a routine thing, and they can wait 3 days to see their own GP or even 2 weeks to see their own GP. We are seeing cohorts of patients that shouldn’t be going to A&E in the first place. How to square all of that and educate the patients, well, we’ve tried to do that for 30 years, haven’t we? [Laughter] It’s a bit of a tricky one.

GP (Rowan) 040, time 2

Although staff did recognise that it was often difficult for patients to know the severity or acuity of their complaint and, therefore, which service to access, they remained prepared to criticise the decisions that patients made regarding ED attendance. Local demographics also influenced what was perceived as an ‘inappropriate’ attendance. There was a view that this was more of a problem when the population surrounding the hospital had a lower socioeconomic status, if it was a tourist area or if it had a high population of immigrants or students. However, it is interesting to note that even highly experienced nurses did not always find streaming patients between GPED and the ED straightforward, yet expected patients to be able to make these decisions:

I think it’s down to, obviously, your training, but also how risk averse you are, and some people are very risk averse and will just have a much lower threshold for streaming people into ED and then also the urgent care centre, rather than directing appropriately, you know, taking that risk.

Paramedic (Chestnut) 022, time 1

The decision to use the label ‘urgent care’ was thought to legitimise a patient’s attendance and create further confusion about the difference between ED ‘minors’ and urgent care. A couple of local service leaders remarked that they are not ‘a fan of just putting a GP in an A&E department’. Although GPED was considered a good idea on paper, a single access system for unscheduled care with OOH provision in the ED was considered a more effective approach that would respond to consumer culture by ensuring that all services were accessible at one site. It was felt that a clearer distinction between unscheduled health-care demand and emergency care was needed, along with services responding to patients’ behaviours and expectations rather than expecting patients to fit into the needs of service providers.

Equally, the failure of previous national initiatives to encourage patients to attend elsewhere, coupled with the fact that, historically, patients have always attended the ED for primary care, meant that some local service leaders saw the solution as placing GPs where patients attend and investing in EDs rather than in primary care. Patients commented that GPED provision gives out mixed messages: on the one hand, patients are encouraged to use their own GP, but, on the other hand, GPED makes access easier and may promote ‘inappropriate’ attendance at the ED.

For some patients, GPED was thought to alleviate some of the anxieties around navigating services as they did not feel that they were taking up valuable emergency resources and were happy to consult with a GP. Despite staff acknowledging that navigating existing services was complex, staff accused patients of ‘playing the system’ to access care more quickly by bypassing their own GP. Staff felt that patients believed that by going to the ED they would be seen more quickly and could have an instant referral rather than having to wait for their own GP to make the referral from primary care. This perception was fuelled by those patients who arrived at the ED expecting or requesting further tests or investigations:

. . . they [patients] don’t tell you the same story. So they, and patients are savvy as well, so they’ll tell you what they think they want you to hear in order to get them into the service they want to be seen by.

Band 6 ED nurse (Linden) 006, time 3

Many patients, before attending the ED, had sought advice from other health services, such as NHS 111, a pharmacy or their own primary care GP. Often these patients were still viewed as ‘inappropriate attenders’ and streamed to see a GP. ‘Inappropriate’ referral by NHS 111 to the ED was a significant cause of frustration for staff. Furthermore, NHS 111 was believed to give patients the impression that the ED was expecting them, that they would be seen within a certain time frame and that the hospital would have their medical records. This led to expectations on the patient’s part and heightened tensions when these expectations were not met. This highlights the point that NHS services themselves can have different understandings of the roles of other services.

Despite these perceptions, our pooled analysis of patients who could potentially be seen in other health-care settings (i.e. ‘unnecessary attendances’) did not show any association with GPED (Table 14). This was confirmed in a RDD analysis forest plot of the same parameter, which showed no consistent association with GPED model, or GPED overall (Figure 9). In common with the other RDD analyses, however, considerable heterogeneity was observed between sites.

TABLE 14 - Pooled analysis of ‘unnecessary attendances’ at the ED

Analysis	All EDs	GPED model			Number of patients
		Integrated	Parallel	On site
‘Unnecessary attendance’
Coefficient	0.003	0.0003	0.009	–0.004	4,302,347
p-value	0.497	0.964	0.291	0.525

FIGURE 9.

Forest plot illustrating the percentage of ED attendances identified as ‘unnecessary’, with a 1-hour doughnut hole at (a) GPED start and (b) GPED end. REML, restricted maximum likelihood.

Domain of influence 6: workforce recruitment and retention

Staffing issues dominated discussions about the impact of GPED and posed a major threat to its success. Local services leaders and site staff expressed concern that GPED could draw GPs away from primary care and cause competition for GP staff. Consequently, GPED was perceived to have the potential to worsen primary care staffing issues:

Primary care is in crisis, there are not enough GPs and asking every ED in the country to hire GPs will make the existing situation worse. Somewhat strange policy for these reasons. The danger is that we are over resourcing this for the return that we get back from it and that we ship out GPs from primary care and put them into hospital, feeling that in this area not enough demand for this type of service.

Local service leader 01

It is important to note that this is occurring in the context of staff shortages more generally in both EDs and general practices; for many participants, this was identified as a key issue that had an impact on patient flow and waiting times. Several sites highlighted difficulties in the recruitment and training of doctors, and populating the middle-grade positions and the consultant rota alongside the financial restrictions on this:

I’m four doctors down due to people ringing in sick and they are in key pathway and middle-grade level, has an impact, short of doctors and also short of nurses despite repetitive recruitment drives, still have same number of nurse vacancies as when I started.

Service leader 31

GPED was considered an attractive prospect for GPs who were seeking portfolio careers and wished to expand their practice, knowledge and skills. Traditional primary care practices were seen as more stressful and less attractive workplaces than newer service models. This was because of several pressures, including the increasing volume and complexity of the workload and depleted community and social care provision. These findings were consistent with the results from GPED GPs who responded to the WFS, and the findings can be compared with those from all GPs that were reported in the Ninth National GP Worklife Survey. 75 These findings should be interpreted cautiously, and in the context of the relatively small numbers of GPED GPs who completed the WFS (n = 18). However, relevant comparisons indicated that GPED GPs were substantially less likely to report high or considerable pressure as a result of factors including ‘having insufficient time to do justice to the job’ (GPED GPs, 27.8%; all GPs, 85.3%), ‘long working hours’ (GPED GPs, 27.8%; all GPs, 82.6%) and ‘increasing workloads’ (GPED GPs, 33.3%; all GPs, 92.3%).

There was some debate as to how the flexible hours associated with GPED would have an impact on job satisfaction. For example, some participants found that this flexibility made it easier to fill rotas, whereas others felt that shift working goes against one of the main reasons that people choose to become a GP:

What appeals to me is that I can do a bit of acute general medicine, trauma, etc., and I’m trained in that, but, equally, I can also lapse into what was my comfort zone. I mean, some days if I’m feeling a bit tired then, actually, it’s much less stressful for me to go into my normal comfort zone and just go through stuff that I’m really competent with and, you know, I can sit down and I can bang through lots of info and that works really well, whereas when I’m feeling a bit more, sort of, ‘right, come on, you know, I can get into resus and I can learn a new thing’ and I really enjoy that.

GP (Juniper) 024, time 1

Some GPs used GPED as a form of semiretirement, enjoying the continued clinical role without the complexity of partnership arrangements and ongoing practice responsibilities. In this way, GPED enabled experienced GPs who would otherwise have left the workforce altogether to continue to contribute to medicine. Many actively chose locum work over salaried GPED positions, which were seen as less flexible, and some felt that moving to salaried positions would result in less experienced GPs being utilised, which might prove detrimental to the service. However, some GPs preferred salaried working and recognised the need for a consistency of provision within GPED:

What attracts many of us is that it is flexible. We really like that. Actually, not that many of us would particularly want to do it and be salaried, but that may make headaches for people in that, probably, the hospital would like people to [have a salaried post]. At the moment, many of us have done 25+ years in medicine. Yes, most of us were saying that we wouldn’t actually want to do this. If they made this salaried, we don’t know if we would do it. It’s that whole flexibility. Some of them are doing it in addition or they’re taking semiretirement and doing it and, actually, it’s having that flexibility. I don’t know, but if you went to a salaried model, you might end up with some of the people who were just coming out of medicine and into general practice, rather than actually, it might be then very different, because a lot of us have got a lot of experience and so are quite confident.

GP (Rowan) 003, time 1

There were several reasons why GPs thought that GPED was unattractive to GPs. As well as the pace of work, risks associated with the ED, and potentially sicker and more acutely unwell patients, practical issues such as travel, parking at the hospital and the pay scale were identified:

I’ve got to, I’d have to ask myself ‘why do I commute an hour when I could commute 20 minutes and do that job?’ . . . and I’m financially, you know, financially, I’m being paid on a sort of middle-grade salary, sort of fairly low, low pay rate . . . I don’t think that they would attract anyone to the role unless they did up the funding for it.

GP (Juniper) 024, time 1

These findings were potentially consistent with results from the small number of GPED GPs who completed the WFS. For example, only 38.9% (n = 7) of these GPs reported being ‘satisfied’ with the opportunity to use their abilities (with item scores ranging from 5 to 7, which was above the neutral mid-point on each scale), relative to 66.6% of all GPs who reported satisfaction with the same question in the Ninth National GP Worklife Survey. 75 Furthermore, only 55.6% (n = 10) of GPED GPs reported being satisfied with the amount of variety in their job, relative to 71.7% of all GPs in the Ninth National GP Worklife Survey. 75

Additional barriers to effective employment and GP working included IT and enforced contractual breaks arising from funding issues, which led to negative working practices:

So one of the problems, I had to take 2 weeks’ break in contract because they don’t want to employ me permanently in case the funding runs out, I guess. And it’s a 6-month contract that’s renewed every 6 months, which is a real pain because then when I come back my IT stuff doesn’t work, and my cards don’t work, my passwords don’t work, it’s just a pain in the arse. That’s what I’ve spent most of this morning doing until you rang. So I’ve seen one patient in the past 2 hours, it’s really annoying.

GP (Birch) 018, time 1

Of additional concern was the perceived lack of suitably qualified GPs with the necessary skills and experience. Site staff placed importance on making GPED an attractive place to work and ensuring that GPs felt valued, supported and appropriately remunerated for effective implementation. However, conflict was noted around the recruitment and retention bonuses that were sometimes required to encourage GPs to fill positions compared with the salaries that could be afforded for ED staff, which were described as ‘a slap in the face’.

Emphasis was also placed on ensuring that GPs felt protected and supported to work within their scope of practice. As a result, some participants felt that GPs needed to be upskilled or would require extra training. To compensate for this, some participants emphasised the importance of recruiting experienced GPs who had previously worked in the ED or employing GPs who were trained at their site as juniors. GPs felt supported when they had a sense of belonging and contributing to the wider team. Practical issues such as trusts contributing to GP indemnity were seen as facilitative, as previous arrangements in which GPs paid their own indemnity meant that costs were prohibitively high for OOH work:

Indemnity has changed now. A lot of out-of-hours requests by us is high risk, obviously, then it would be too expensive to do out-of-hours. As from April this year, it’s changed so NHS is taking over the indemnity side of things. Yeah, you still pay, like, you know, a fair amount, but like the big thing has been taken over by the NHS.

Locum GP (Birch) 016, time 1

Many staff perceived GPED to provide training and educational benefits for junior doctors, whom it was felt would, in some models, become more confident about discharging patients and build up their primary care knowledge. Equally, diverting patients with minor conditions to GPED was seen to have benefits for ED juniors and trainees by exposing them to more acutely ill patients.

Domain of influence 7: workforce behaviour and experience

Most staff could see how GPED differed from their usual ways of working (68.9%, n = 264) and felt that they were able to easily integrate GPED into their existing work (68.4%, n = 262). The clinical effectiveness of GPED services largely depended on effective communication, trust and confidence between streamers and GPs. Streamers were confident streaming to GPs whom they knew and trusted. Similarly, it was important for GPs to trust streamers and have confidence in their abilities to direct patients appropriately and safely. However, there was significant variation between both individual GPs and streamers in terms of what presentations streamers were prepared to refer to GPs and what presentations GPs were prepared to accept. Practices such as a dependency on locums and part-time GPs and the rotation of the streaming nurses were seen to reduce consistency and limit trust. This was reflected in the NoMAD survey, in the responses to which only 60% of staff strongly agreed or agreed that they had confidence in other people’s ability to conduct their roles in GPED (59.5%, n = 228) or that work was being assigned to those with skills appropriate to the role (58.0%, n = 222). Proactive behaviour on the part of both GPs and streamers was viewed as beneficial in enhancing working relationships and increasing the clinical effectiveness of the service:

We are familiar with probably two or three GPs who we would consider very well trained and able to deliver what it says on the tin. And that’s great when they’re on. The others are an unknown quantity and, to be honest, we have such a huge volume of locums through, they’re often an unknown quantity.

ED consultant (Chestnut) 024, time 1

However, tensions between GPs and streamers were identified at all sites. GPs were seen to ‘pick and choose’ their patients – ‘they’ll only see what they choose to see, you know, they won’t see this or they won’t see that’ [Streaming nurse (Birch) 001, time 1] – whereas nurses were criticised for streaming patients inappropriately. This increased when the ED was busy. Similarly, GPs were seen to pass patients back to the ED to reduce GPED workload. There was clear dissonance between the types of patient that ED staff and GPs perceived as being ‘GPED appropriate’. In this way, the relevant workforces acted to protect their own working environment:

There were some obvious frustrations around communication, and rejection of patients perceived as suitable for streaming to the UGPC [urgent general practice centre]. This included, for example, phone calls back and forth between the ED front desk and the primary care centre explaining/questioning the rationale for streaming certain patients.

Band 7 ED nurse (Linden) 001, time 1

As a consequence, and later in the study, one case site had developed plans to respond to excessively busy workloads from a broader perspective:

. . . it’s out-of-hours escalation plan that if we have an increase in workload that we might have to look at the activity that comes through ED and consider ‘have we got the capacity to take the streaming patients?’ and it’s part of the escalation plan. So we would review the activity as a bigger picture in terms of what’s the activity going through ED, what’s the activity coming through 111 and look at the kind of pathways, managing the levels of demand in line with acuity and everything else.

Band 5 nurse (Hawthorn) 015, time 3

However, individual streamers’ and GPs’ tolerance of risk and clinical ambiguity also had an impact on the clinical effectiveness of GPED and affected streaming decisions. This was linked to concerns regarding patient safety and was associated with not only the characteristics and behaviour of staff at an individual level, but also broader organisational and professional support:

Here in Hawthorn, we’re very risk averse, so not as many go to out-of-hours GP as they could. If I’m honest, it’s because the wider NHS complaints and concerns system will nearly always appear to take the patients’ position.

Band 5 nurse (Hawthorn) 018, time 1

Protocols had been developed in case study sites, with the aim of standardising both streaming decisions and GPED acceptance criteria. However, these protocols were disseminated and utilised with varying clinical effectiveness, and streamers and GPs adhered to these to differing degrees:

It should be fixed, but, as I said, depending on who you speak to, it does waver slightly on what practitioners and GPs are willing to see. So, it’s a bit of a grey area really. It depends who you’re working with really. I don’t . . . yes. So, it’s not fixed. It should be really.

ANP (Redwood) 002, time 1

Effective working relationships were also hindered by divergence in ways of working between primary and secondary care. Culturally, GPs and ED clinicians worked in different ways and this had an impact on the working relationships and integration between the clinical silos. Although there was physical integration, ED staff often felt that GPs lacked collegiality:

So, the GP will, sort of, arrive, go straight into their room and then stay in the room unless you call them out for huddle or something like that, whereas A&E nurses and all of our doctors are all quite social. We’re a team; we’re really visible to each other.

Band 6 nurse (Nutmeg) 015, time 1

However, findings from the WFS indicated similar rates of ‘satisfaction’ with colleagues and fellow workers among GPs (83.3%, n = 15), compared with ED doctors (84.2%, n = 85) and nurses (79.7%, n = 126). Likewise, very few staff indicated in the NoMAD survey that GPED disrupted their working relationships (13.3%, n = 51). Initially these data appear at odds with the qualitative findings; however, this is likely to reflect strong intra-ED staff and intra-GP staff collegiality, which was apparent from the interviews and observations.

Table 16 presents item-level descriptive analyses of the 10-item measure of job satisfaction for the sample as a whole and in comparison with relevant findings from the Ninth National GP Worklife Survey;75 Table 17 presents a further breakdown by job category. These tables include mean ratings for each job satisfaction item, as well as the number and proportion of respondents who provided a response above the neutral mid-point on the scale (item scores ranging from 5 to 7), which were, therefore, classified as ‘satisfied’.

As shown in Table 16, when compared with the national sample of GPs, hospital staff involved with GPED generally reported lower levels of job satisfaction in relation to physical working conditions, freedom to choose their own method of working and remuneration. By contrast, our sample of hospital staff reported higher levels of job satisfaction associated with work hours and overall ratings of job satisfaction.

Comparisons conducted across employee groups (see Table 17) indicated that job satisfaction was generally highest among ED specialists; the small number of GPs in this sample reported levels that were typically either similar to or lower than those of ED specialists. With some exceptions, the lowest levels of job satisfaction were typically reported by nursing staff.

Although the majority of staff understood how GPED had affected the nature of their own work (78.1%, n = 299), a far smaller number strongly agreed or agreed that they valued the effect that GPED had had on their work (56.7%, n = 217). Despite this ambiguity, most staff (across all staff groups) stated that they would continue to support GPED (86.4%, n = 331).

Domain of influence 8: resource use

Respondents could generally see the potential for overall cost savings associated with GPED, in principle, if all of the assumptions were met [i.e. having a positive impact on hospital admissions, reduction in patient investigations and more effective use of ED staff resources (seeing more unwell patients)]. However, much scepticism remained around whether or not these goals would be achieved by GPED alone, and it was often felt that broader investment in the number of acute beds and enhanced social care would also be necessary:

So the problem is not latching GPs onto here, it’s further back, it’s much, much closer to the root of the problem that they need to start looking, really, but they know that, but they’re not going to do it because it costs them too much money.

Band 7 ENP (Linden) 002, time 1

Staff and patients predicted that GPED would incur higher costs because of the cost of GP employment, and placed importance on ensuring that staffing and resources are carefully matched. As ED staff and local service leaders considered GPs a costly resource, they felt that GPs needed to demonstrate their clinical effectiveness and they would require that GPs saw adequate volumes of patients in a timely manner and were also on shift when demand was at its peak:

Eventually, I suppose, it would be to look at, are there days and times of the days where it’s not cost-effective? I think that has to be taken on board for the future.

GPED system lead (Rowan) 031, time 1

Staffing issues, such as struggles to recruit to GPED and the need for senior nurses to stream, also dominated discussions about the potential impact of GPED on resources. The employment of locums and agency staff to fill these positions led to higher costs. Some argued that the money could be better spent improving primary care provision, resulting in a reduction in the number of primary care patients attending ED in the first place.

It was also noted that in situations in which a GPED-streamed patient needed to be referred back to their own GP, an additional cost to the broader NHS would be incurred, as the patient was being seen more than once for the same illness episode. This was viewed as both being costly and generating additional unnecessary workload. For GPs, there was also a tension between service policies and associated funding and effective and safe patient care. As described in Domain of influence 4: patient outcome and experience, an example of this was ‘2-week wait’ for possible cancer referrals. For GPs, patient safety was paramount and this over-rode funding decisions. Consequently, GPED GPs would refer patients directly to the 2-week wait service, rather than back to their own GP, regardless of the cost implications:

I think it’s really important that we, again, it’s all about efficiency, that, actually, if we’ve picked up that the patient’s having palpitations, we should just send them to the cardiologist from here. It’s ridiculous to be sending them back to the GP, who then sends them, and there’s all sorts of excuses that everyone uses in here, it’s all to do with the funding and no one will pay for it . . . So, I think that those are the sorts of things that just really wind me up about the health service that actually, just because of the funding it means that actually there’s a really inefficient pathway in there and we need to try and change that. So, I think GPs have a role of being here and standing up for general practice and saying ‘hang on a minute, that’s not efficient, let’s refer him then go back to the GP for ongoing care’, why can’t we do that, it’s crazy, you know? It’s just making work that doesn’t need to be made.

GP (Juniper) 024, time 1

General practitioners felt that initiatives to broaden access to health care were, in reality, often confusing for patients and actually created a duplication of workload, resulting in increased costs. Although patients were often aware of traditional routes to care, such as primary care and the ED, they were less sure about other ways to access care. This led to patients using tried-and-tested access routes, only for them to then be signposted to different services. GPs saw this as a waste of resources:

I think there are a lot of resources out there that patients don’t use and don’t know about; so, for example, the minor eye condition service where patients with minor eye conditions can actually be treated by their optician and get payment rather than coming to their GP for their dry eyes or conjunctivitis or painful eye, but patients have no idea about that so they present to the GP anyway. That uses up an appointment, just being told ‘well you can go to [service name]’. So, you know, NHS Choices [now nhs.uk; www.nhs.uk (accessed 28 November 2021)], brilliant, it’s a fantastic website, but no one knows about it. 111 people are starting to know about, it’s taken a long time, I think we need to have some sort of national programme that really educates people about what services are available in their community and where to look for it.

GP (Juniper) 024, time 1

Incidental costs, such as paying for training and the set-up and management of new IT systems, constituted added expenses and time burdens that staff felt had not always been taken into consideration. Staff reported receiving little training when joining GPED and there were continued frustrations with IT and the lack of integration between GPED and ED systems. Findings from the NoMAD survey indicated that only one-third of staff felt that sufficient training was provided to enable staff to implement GPED (33.4%, n = 128) and that sufficient resources were available to support GPED (32.6%, n = 125).

Staff at case study sites described several different approaches to how capital funding had been used. These included structural changes, whereby new structures were created or existing spaces were adapted; governance and contractual changes that often included working in new partnerships across commissioners and providers of health and care; the integration or introduction of multiple IT systems; and operational changes, such as revised patient pathways or the reconfiguration of existing services. Changes made with this funding were, in some cases, minimal, but could also include physical changes, such as building a purpose-built UCC or dedicated waiting area, or alterations to management and governance. This was exemplified at one site, where staff were unclear how the new GPED service would differ from their established GP OOH service, apart from providing additional funding or new triage rooms. Table 18 summarises the key issues raised in relation to GPED resource use.

Patient and public interpretation

During the workshops held towards the end of the study, public contributors noted that many factors influenced the performance of GPED, making it difficult to attribute cause and effect. They further noted that any general pronouncement on the clinical effectiveness of GPED might not adequately take account of the potential benefits of GPED in specific circumstances. However, they felt that the quantitative data were ‘strongly suggestive of no effect’, which, in turn, led the public contributors to question whether or not GPED represented a good use of public money and GP time.

Summary

This chapter has used a mixed-methods synthesis to present our research results within the overarching policy context and eight identified ‘domains of influence’. We have found considerable heterogeneity and complexity, combined with a high degree of local variability, which makes it impossible to identify any consistent or sizeable effect or any evidence of general cost-effectiveness. We will consider the implications of this further in Chapter 4.

Overview

In this chapter, we summarise and interpret our research findings using NPT79 and discuss each of our nine objectives in context. We then consider the ‘success factors’ that influence the implementation of GPED, and the implications for clinical practice and local service delivery. A description of the strengths and limitations of our study is followed by our recommendations for future research.

Summary of research findings using normalisation process theory

Discussion of research objectives

Success factors for introducing GPED

This section identifies a series of ‘success factors’ that may inform how services choose to implement future GPED models. Examples of these success factors include streaming, staff training and IT access.

Implications for clinical practice and service delivery

Following from the previous sections, we offer a summary of the key implications of our work for clinical practice and GPED service delivery:

• GPED services are highly context specific and could be planned and implemented in a way that is sensitive to local circumstances, service provision, workforce and demand. One size does not fit all.

• It is essential to be clear as to the intended benefits and outcomes that will follow GPED implementation; these benefits should be actively monitored and measured to understand the extent to which they have been achieved over time.

• Clear governance and oversight arrangements could be established from the outset, in particular the expected scope of practice of those clinicians providing the GPED service. Access to diagnostic investigations, specialist reviews and hospital admissions could be agreed and consistent for all GPs.

• GPED services are most likely to be successful in circumstances where significant numbers of eligible patients attend the ED, where a stable GPED workforce can be recruited and retained, and where the physical environment allows the planned service to be delivered successfully.

• Effective streaming is central to an effective GPED service. Streaming staff are usually experienced nurses, and could be trained and supported to ensure that they are able to deliver a clear, safe, consistent and rewarding streaming process.

• Staff engagement is required from the outset, at all levels and across all professions. Strong and visible management and clinical leadership are required over an extended period of time, and particular efforts could be made to foster effective communication and develop positive interprofessional relationships and a spirit of team working.

• Care could be taken to ensure that supporting IT and administrative systems are put in place.

• The GPED service should be subject to regular multidisciplinary audits and reviews, including a review of service performance and clinical incidents.

• Enduring service change takes time, and results will not be realised immediately; commitment and perseverance are required from all parties.

Strengths and limitations

To the best of our knowledge, this is the most complete and detailed study of GPED provision in England to date. Our analysis used robust modelling approaches and individual patient data from type 1 EDs linked to in-patient and mortality data to assess the differences in hospital performance and patient outcomes between GPED services and normal services. The HES data set66 allowed us to adjust for a range of potential confounders, including age, sex, ethnicity, level of deprivation and arrival method. Crucially, given the strong associations between time of attendance and patient volume and acuity, we were also able to take account of the time of attendance in relation to the different timings of GPED services.

We used two different statistical approaches to quantify the impact of GPED on outcomes, which rely on different sources of variation in the availability of GPED services across hospitals or time. Both methods yielded similar findings, which greatly increases our confidence in our estimates. Our telephone interviews were conducted prospectively and through these we were able to collect the views of national policy-makers and local system leaders, based on their recent experiences. The ‘domains of influence’ that we have identified in this report were generated from a large evaluation that used ‘big qualitative data’ (n = 228 interviews) and the views of multiple stakeholders. This provided a rich and nuanced understanding of the complexity surrounding a current national policy: GPED. Our 10 case studies included participants from a wide range of EDs in terms of size, location and existing use of GPED, and we collected a very large number of interview, observation and survey data to provide a comprehensive understanding of GPED provision and address our study aims.

However, our study has a number of limitations. These are:

• The reliability of the data sources used to populate our database (WP A) varied and required some interpretation by the research team. In addition, data collection relied on self-reporting and respondents’ ability to accurately describe their model of service provision. As a result, we included a smaller sample of EDs in our quantitative analysis than we had planned.

• We were unable to complete our quantitative analysis as intended, and were required to adjust our analysis plan and exploration of the costs and consequences of GPED accordingly. We were also limited in our ability to collect data from the primary and urgent care systems surrounding our WP C case study sites, which significantly limited our ability to quantitatively evaluate the effect of GPED on the wider health-care system as planned. (This is considered further at the beginning of Chapter 2.)

• There were several limitations to the quantitative analysis of the data –

  • We relied on data that were routinely collected by ED staff and were, therefore, dependent on the complete and accurate recording of patient episodes. Different key variables had different levels of missingness within the data set and between hospitals, and there is the potential for missingness not at random if, for example, GPs record information in different ways from ED staff or if ED staff change their recording behaviour in response to GPED.

  • There is the potential that patient episodes may be misclassified as GPED/not GPED, as streaming activity and GP availability may not have corresponded exactly to official GPED start and finish times, particularly for patients arriving close to these times or undergoing extended waits. Related to this, our analysis assumes that GPED was always operational during designated hours and, therefore, does not account for physician absence. To reduce the risk of misclassification, we excluded patients arriving in the hour before GPED start/finish times in our RDD analysis. Including these patients in a sensitivity analysis produced similar results, tending slightly more towards zero impact.

  • We lacked detailed cost data for GPED services, so we used crude estimates based on average GP salaries and estimates of hours of operation. Therefore, associated costs are missing from our analysis, as are any offsetting costs relating to compensatory reductions in other forms of staffing. Related to this, we were not able to assess ‘role substitution’ (i.e. whether GPs were directly added to normal services or they replaced another member of staff). In addition, we did not assess the full opportunity costs of GPED in terms of alternative uses of GP salaries by EDs (e.g. employing additional consultants and/or nurses) and reductions in GP availability in the community, which is already characterised by shortages.

  • Because of the national roll-out of GPED services and the lack of information on implementation dates, we were unable to conduct our intended analyses that would have provided a cross-sectional comparison across hospitals with and without GPED services (i.e. ‘intervention’ vs. ‘control’ hospitals) or a within-hospital comparison over time, using an interrupted time series (before and after) approach. Instead, for hospitals where a GPED service was already in place, we compared patients arriving at times when GPED was operational with patients arriving when GPED was not operational. Therefore, we could not account for any broader effects of GPED on hospital procedures and patient outcomes that occur outside the operational hours of the service.

• In our interviews with system leaders who were implementing GPED initiatives with capital funding support, we were able to interview representatives from only 36% of all eligible EDs and the respondents may not have been representative of all EDs. This weakness was compounded by the fact that follow-up interviews were completed in only 47% of the EDs in which staff we were able to interview initially.

• Our case study sites were purposively selected to be as representative as possible; however, the participation of all 10 sites, and of the staff and patients within those sites, was voluntary. We could interview only those who agreed to take part and, although our data span a very wide range of individuals and views, they are unlikely to be exhaustive.

• In particular, staff choosing to complete the WFS may not have been representative of all of the staff involved in GPED at that site. The WFS return rate was 24%. This is at the lower end of the range reported from a meta-analysis of 59 methodological studies106 designed to estimate the magnitude of non-response bias in statistics of interest. This lower level of return reflects the challenges of engaging a wide range of staff and the relative length of the survey. In some sites, local concerns regarding information governance limited our ability to disseminate the survey effectively. Although the low response rate is a potential study weakness, it should be noted that response rate is not necessarily predictive of non-response bias validity and, therefore, low response rates should not be cited routinely as a reason to disregard survey results. 107 Given the close alignment between the results of the survey and our other approaches to data collection, we believe that the survey provides a valuable and complimentary contribution to the other study findings.

• The analyses of data from the WFS involved comparisons that should be interpreted cautiously, for the following reasons –

  • Comparisons with data from the Ninth National GP Worklife Survey75 should be considered in relation to the very small number of GPED GPs who were represented in the survey sample.

  • As far as we are aware, comparable items from the Ninth National GP Worklife Survey75 (e.g. regarding job satisfaction) have not been widely administered to other staff groups (e.g. nurses).

  • This report presents a selection of results from the WFS that were chosen on the basis of mixed-methods considerations (i.e. they reported survey findings that either aligned or contrasted with qualitative data to provide a form of methodological triangulation). However, this approach means that some results from the WFS are not included in the main report, and these can be found in Appendix 3.

• Our data apply to England only, so they may not be generalisable to other countries and health-care settings.

• It could be argued that the data we present here represent the inherent uncertainty and resistance to change that most health-care policies encounter prior to or during early implementation, so the data are representative of typical ‘teething problems’. However, although there is evidence that it takes 2–6 years to achieve lasting service change and it is, therefore, assumed that such issues will improve over time,108 recent research suggests that issues that are identified early in the implementation process often persist long after establishment. 109

Recommendations for future research

This is a comprehensive mixed-methods evaluation of GPED in England and, when combined with the study that was commissioned under the same funding call (HSDR project 15/145/04; URL: https://fundingawards.nihr.ac.uk/award/15/145/04; accessed October 2021), which we anticipate will examine cost-effectiveness in more detail, it seems unlikely that further similar research will yield substantial additional knowledge. We therefore offer the following recommendations for future research, listed in priority order:

1. As identified in Strengths and limitations, although some forms of GPED have existed for decades, a co-ordinated national initiative is relatively new and expected to develop over time. If these service models continue, further research after a period of 3–5 years could be considered to understand the longer-term effects and implications.

2. We identified particular ambiguity and uncertainty in relation to streaming in the ED; further research to better understand streaming and clarify the optimal approach to streaming in terms of patient outcome, safety and experience, and associated staff experience is recommended.

3. Part of the underlying premise of GPED rests on assumptions regarding differing attitudes to risk tolerance and risk management between GPs and emergency physicians; however, empirical evidence to support this hypothesis is limited. Additional research to explore attitudes to risk and how these vary between and within medical specialties, and the impact that this has on practical clinical decision-making, could help to ensure that staff are trained, deployed and supported in ways that maximise clinical effectiveness, efficiency and patient safety.

4. Research addressing the ways in which national policy initiatives should be developed, piloted and implemented in the most clinically effective and cost-effective manner is surprisingly limited, with a lack of consensus across the health and care system. We have considered this issue in our report, and believe that further research is required to explore and understand the contrasting views of stakeholders, and provide empirical evidence as to how national health-care initiatives should be implemented (rather than whether or not a particular initiative is effective).

5. Following on from the above, comprehensive evaluations of new models of service delivery often lag behind practical implementation. This undermines the purpose of evaluation as by the time commissioned research has been completed the change has been enacted. This is a particular problem if the evaluation does not support the new model of service delivery. We, therefore, recommend that a set of standards is developed for health policy implementation, including an outline evaluation protocol to ensure that the necessary data are available and the necessary design is feasible. This should be supported by methodologies that allow rapid and ‘real-time’ evaluation of new models of workforce and service delivery in health and social care.

6. The relationship and interface between primary and secondary care within the wider context of the health and care system remains central to the future delivery of urgent and emergency care. Further research is required to understand this relationship and guide future models of service development and delivery in the context of rising health-care demand.

Summary

In this chapter, we have summarised our findings using the framework of NPT and discussed each of our nine objectives. We have considered the practical implications of our findings, described the strengths and weaknesses of our study and listed our recommendations for future research. A short conclusion chapter now follows (see Chapter 5).

Contributions of authors

Jonathan Benger (https://orcid.org/0000-0001-6131-0916) (Chief Investigator) took responsibility for the integrity of the data and the accuracy of analysis, was involved in the study conception and design, interpreted the data, and drafted and critically revised this manuscript. He is the guarantor.

Heather Brant (https://orcid.org/0000-0001-9608-7451) (Project Manager) collected and interpreted the data, provided project co-ordination and support, and drafted and critically revised this manuscript.

Arabella Scantlebury (https://orcid.org/0000-0003-3518-2740) (Qualitative Researcher) was involved in the data collection, analysis and interpretation, and critically revised this manuscript.

Helen Anderson (https://orcid.org/0000-0002-6945-0590) (Qualitative Researcher) was involved in the data analysis and interpretation, and critically revised this manuscript.

Helen Baxter (https://orcid.org/0000-0002-3320-2915) (Knowledge Mobilisation Expert) was involved in the study conception and design, supported knowledge mobilisation and dissemination, and critically revised this manuscript.

Karen Bloor (https://orcid.org/0000-0003-4852-9854) (Statistician and Expert in the Analysis of Routine Health Data) was involved in the study conception and design, supervised data collection and interpretation, and critically revised this manuscript.

Janet Brandling (https://orcid.org/0000-0003-3352-7979) (Qualitative Researcher) was involved in the data collection, analysis and interpretation, and critically revised this manuscript.

Sean Cowlishaw (https://orcid.org/0000-0002-8523-3713) (Statistician with Expertise in Cross-Sectional and Longitudinal Observational Studies) was involved in the study conception and design, supervised data collection and interpretation, and critically revised this manuscript.

Tim Doran (https://orcid.org/0000-0001-7857-3704) (Statistician and Expert in the Analysis of Routine Health Data) was involved in the study conception and design, supervised data collection and interpretation, and critically revised this manuscript.

James Gaughan (https://orcid.org/0000-0002-8409-140X) (Data Analyst) undertook statistical data analysis and interpretation, and critically revised this manuscript.

Andrew Gibson (https://orcid.org/0000-0002-4641-2583) (PPI Expert) was involved in the study conception and design, supported PPI and critically revised this manuscript.

Nils Gutacker (https://orcid.org/0000-0002-2833-0621) (Statistician and Expert in the Analysis of Routine Heath Data) was involved in the study conception and design, supervised data collection and interpretation, and critically revised this manuscript.

Heather Leggett (https://orcid.org/0000-0001-8708-9842) (Qualitative Researcher) was involved in the data collection, analysis and interpretation, and critically revised this manuscript.

Dan Liu (https://orcid.org/0000-0002-1891-9352) (Data Analyst) undertook the statistical data analysis and interpretation, and critically revised this manuscript.

Katherine Morton (https://orcid.org/0000-0003-2413-8867) (Project Manager) collected and interpreted the data, provided project co-ordination and support, and critically revised this manuscript.

Sarah Purdy (https://orcid.org/0000-0002-3445-986X) (Expert in Primary and Urgent Care Research) was involved in the study conception and design, involved in the data analysis and interpretation, and critically revised this manuscript.

Chris Salisbury (https://orcid.org/0000-0002-4378-3960) (Expert in Primary and Urgent Care Research) was involved in the study conception and design, involved in the data analysis and interpretation, and critically revised this manuscript.

Anu Vaittinen (https://orcid.org/0000-0001-8471-9204) (Qualitative Researcher) was involved in the data collection, analysis and interpretation, and critically revised this manuscript.

Sarah Voss (https://orcid.org/0000-0001-5044-5145) (Researcher in Emergency Care) was involved in the study conception and design, supported data collection, interpreted the data and critically revised this manuscript.

Rose Watson (https://orcid.org/0000-0003-4549-3462) (Qualitative Researcher) was involved in the data collection, analysis and interpretation, and critically revised this manuscript.

Joy Adamson (https://orcid.org/0000-0002-9860-0850) (Expert in Mixed Methodology and Ethnographic Approaches) was involved in the study conception and design, supervised the data collection and interpretation, and critically revised this manuscript.

Publications

Morton K, Voss S, Adamson J, Baxter H, Bloor K, Brandling J, et al. General practitioners and emergency departments (GPED) – efficient models of care: a mixed-methods study protocol. BMJ Open 2018;8:e024012.

Cooper A, Edwards M, Brandling J, Carson-Stevens A, Cooke M, Davies F, et al. Taxonomy of the form and function of primary care services in or alongside emergency departments: concepts paper. Emerg Med J 2019;36:625–30.

Brant H, Voss S, Morton K, Cooper A, Edwards M, Price D, et al. Current provision of general practitioner services in or alongside emergency departments in England. Emerg Med J 2021;38:780–3.

Scantlebury A, Brant H, Anderson H, Leggett H, Salisbury C, Cowlishaw S, et al. Potential impacts of general practitioners working in or alongside emergency departments in England: initial qualitative findings from a national mixed-methods evaluation. BMJ Open 2021;11:e045453.

Scantlebury A, Adamson J, Salisbury C, Brant H, Anderson, H, Baxter H, et al. Do general practitioners working in or alongside the emergency department improve clinical outcomes or experience? A mixed-methods study. BMJ Open 2022;12:e063495-e063495.

Data-sharing statement

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

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report 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.

Topic guide for national-level system leaders interview

What is your current role and what has your role been regarding the introduction of GPs into EDs?

Tell us the background to the concept of introducing GPs into EDs as you see it:

• Who have been key stakeholders in the idea?

• What do they hope to achieve?

• Where did it originally come from?

• How does it fit with other services, e.g. walk-in centres, 111, out-of-hours GP?

• Have lessons learnt from the experience of introducing other services been incorporated?

What is your sense of the ‘buy-in’ from GPs?

• Sustainability (lack of GPs).

• Desirable role for GPs.

• What are the challenges/benefits for GPs in this role?

• Terms and conditions (e.g. employer, indemnity).

Why do you think the government has decided to invest in GPED?

Describe the different models of GPED care that you are aware of having been/going to be implemented:

• Have you got a sense of which might work better (according to what outcomes)?

• What do you think patients think about the idea in general?

What do you think the likely impact will be (do you have evidence for this)?

• On ED care delivery from perspective of ED department.

• On primary care delivery across the community.

• On patient care.

• Have you considered unintended consequences (e.g. it will increase demand)?

Topic guide for site-level system leaders initial interview

ED context:

• How many new adult patients does your ED see each year?

• How many new paediatric patients does your ED see each year?

What model of working with GPs/primary care operates in your ED currently (if any)?

Tell us about the GPED model you are planning to implement.

Can you tell us the background to that decision?

• What are hoping to achieve?

• What discussions took place?

• What options were considered?

• What major factors impacted on decision-making?

• (If do not mention might want to prompt on waiting time, cost, numbers.)

How is it different from the model you have in place now (is it clearly distinct)?

• Structural requirements for proposed model.

• Organisational requirements for proposed model.

• How will changes (if any) be achieved?

• Timetable for change (date).

Do you think this model makes sense/is the right thing for your department?

Do you think staff value the proposed model of service provision?

What do you think the impact will be to your department on?

• Performance (4 hours, hospital admission rate).

• Staff (which staff in particular, in what ways).

• Division of labour.

• Interaction between different professional groups.

• Resources.

• Consultations with patients.

Will staff require additional training before implementation?

• Which staff and what training is planned/available?

• How will you judge the success/impact of the new mode of service delivery?

• What data might be available for research purposes?

• Mechanism for staff feedback about the intervention.

• Can the intervention be adapted on the basis of experience?

What are your thoughts on the decision to fund these models of service delivery?

• Does the idea of GPs in ED make sense in general?

Any other comments to add about GPED?

Topic guide for site-level system leaders follow-up interview

What model of working with GPs/primary care is now operating in your ED (if any)?

Tell us what has happened over the last 12 months in relation to GP working in or alongside your ED:

• What structural and organisational changes have occurred?

• Is it running as initially planned or were changes made – if so, why?

• What was the final timetable for implementation?

• Did the changes require any training, if so, what form, was it sufficient?

• What changes have you noticed in your day-to-day practice?

• What were your expectations regarding the change, have they been met?

Do you think the current model is better or worse than where you were 12 months ago? Why do you think this?

Is the model that you now have in place the right thing for your department?

Do you think staff value the new model of service provision?

Do you think patients value the new model of service provision?

What has the impact of this change been on?

• Performance (4 hours, hospital admission rate).

• Staff (which staff in particular, in what ways).

• Division of labour.

• Interaction between different professional groups.

• Resources.

• Consultations with patients.

Do you have any evidence of these impacts (figures or case studies)?

What are the main enablers and barriers that you have encountered to GPs working in or alongside your ED?

What are your thoughts on the decision to fund these models of service delivery?

• Does the idea of GPs in ED make sense in general?

Any other comments to add about GPED?

Pooled model

Table 24 reports results from all sites for which data were available, and separates these by GPED-service type (integrated, parallel and on site). This pooled analysis uses the fact that GPED services operate at different times of the day so that the overall effect is identified by comparing the outcomes of patients arriving within a given hour of the day at hospitals operating or not operating a GPED service at this time.

We found no statistically significant effect of GPED on the volume of attendance, nor was a statistically significant effect apparent for most performance indicators, including 4-hour waits, patients leaving without being seen, unnecessary attendances, hospital admissions or mortality at 30 days. Our analysis did, however, identify a very small, but statistically significant, reduction in the rate of reattendance within 7 days, and this is discussed further in Chapter 4, Discussion of research objectives, Objective 3.

Topic guides

Workforce survey

Pen portraits

Birch pen portrait

FIGURE 10.

Birch pen portrait. ENT, ear, nose and throat.

GPED service times unclear

Currently, patients register at the ED reception and, if ambulatory, they are sent to the ambulatory assessment unit for triage to a number of clinical streams, including GPED. Patients are triaged, rather than streamed, by a band 5/6 nurse and vital signs are recorded. There is no specific protocol for identifying which patients are suitable for GPED; instead, the process is based on an established triage assessment. Patients with ‘GP-type’ presentations or subacute problems are directed to GPED. A card is put in the GPED tray (or another appropriate clinician/majors, tray), which is then picked up and the patients are seen. However, the indication is that this process does not always take place and the times/staffing of GPED is unclear. GPs can redirect patients whom they assess to be inappropriate for GPED back to the ED and they can refer patients on to a number of specialities. GPs are expected to see an average of two patients per hour and between 22–25 patients per shift. The total number of patients attending the ED is approximately 300–380. If GPED is quiet, GPs may identify patients that they think are suitable for GPED on the tracking screen and see these patients (after discussion with an ED consultant). GPs have no access to patients’ primary care records, but they are able to order investigations such as radiographs and blood tests; this is seen as a hybrid GP/ED role. There is no designated GPED area or room.

There is one salaried GP and three or four GPs undertake regular locum shifts to cover the service. However, it is unclear what the established GPED service hours are and the service is patchy and sporadic. There appears to be a 07.30–18.00 shift (although some say it is 08.00–16.00/17.00), but sometimes the service runs from 10.00–18.00 and sometimes there is no service at all. There may be one or two GPs covering a shift. It is unclear whether or not the service runs at weekends:

Erm, [hesitantly] it’s, at the moment we are supposed to have our target or the common target, we are supposed to have a GP service to allow us a day for 7 days a week, so – no, at the moment we have one GP who basically is employed by us . . . so he mainly worked 08.00–16.00, 08.00–17.00, I think it’s 5 days, 9 days out of 10 days, so he takes every tenth day as an off day. And, in addition to him, we have some GPs who come and do basically GP locum work for us. So, at the moment, the person who’s working with us is a fully qualified GP who’s been a GP for years and years . . . So there might be a day we might have fully diverse coverage, or 12 hours coverage, but usually only Wednesday, Thursday. Some weekends we do not have any GPs, so . . .

Staff (Birch) 14, time 1

The current model has been in place for around 1–2 years. Previously, the service was run by a private company/CCG GP collective and its target was to see a preagreed number of patients per day. This target was not met and the service ceased. It was felt that the service remit and acceptance criteria were too narrow. There was then a gap during which the trust had no contracted service, but had an ad hoc arrangement with locum GPs instead. However, the trust saw value in a GP service and decided to take it in house, employing GPs directly (rather than employing a company to provide GPs as before). The new GPED model involved various stakeholders, including ED clinicians and nursing staff, as well as input from the wider trust. GPs are audited by the trust. The next stage in the process is to embed the service and increase coverage.

National context

Staff contextualised the GPED service against the backdrop of a highly challenged primary care service. However, they also felt that improvements could be made to general practice more broadly and that the nature and organisation of primary care was challenging, with the traditional ways of running practices not matching modern demands. In addition, patient behaviour and expectations were seen as a challenge, with patients’ expectations of the service increasing and demands for free prescriptions influencing patient behaviour. The ageing population was also seen to contribute to ED and secondary care problems, with poor community service provision failing to keep elderly patients out of hospital and in their own homes.

Local context

Staff felt that the ED is under pressure and struggling to meet targets against the backdrop of the high volume of patients. Variability among general practices meant that although some practices could accommodate patients, others struggle to offer appropriate appointments. Staff compare their service to other local hospitals that are seen to be better resourced, with GP availability 7 days per week and nurse practitioners available 10.00–22.00. Because there is a local childrens’ hospital, no patients aged < 16 years are seen at this hospital. Local commissioning groups have acted to address patients attending for free prescriptions by limiting the prescriptions that can be bought over the counter, which is something that staff seem to approve of:

But luckily some of the, the, err, Commissioning Groups have said ‘we won’t pay for over-the-counter medications, so you can’t prescribe it’. I said that to patients ‘you need to get some paracetamol, ibuprofen’, 25p in any supermarkets, round about, and they go ‘can you not prescribe it?’, ‘I’m not allowed, I’m afraid, I’m not allowed’.

Staff (Birch) 1, time 1

Trust emergency department and urgent care culture

Staff can be actively redirecting patients away from the ED, to resist the flow of arriving patients, or providing a service in response to, and recognition of, the fact that patients have attended with health concerns that need to be addressed. Staff may feel that they must see patients and that the responsibility rests with them to provide health care.

No patient interviews were carried out at this site as there were no GPs available at time 1 data collection.

Staff saw the cultures of the ED and GP as being quite separate, with the ED seen as working from the starting point that the patient is sick until proven otherwise, whereas for GPs it was seen as the opposite way around. This affected risk management. Staff recognised and valued the different roles and skills of ED and GPED clinicians. GPs were seen as better placed to deal with longer-term, lower-acuity presentations that required clinical reasoning rather than specialist investigations:

A&E likes what A&E likes. If it’s acute and it’s new, and it’s something which, you know, which can be dealt with, it can be really sick or it can be really well, but we like it to be new. And we like it to be something which requires immediate help. For long-term chronic conditions, actually, and conditions where decision-making can be done without high-tech investigations and where you can use your experience to say ‘it’s this, go home’, GPs are invaluable.

ED Staff (Birch) 8, time 1

The system worked particularly well with one GP, who was seen as proactive in seeking out patients and who staff felt was embedded in the ED team. However, GPED as a service was seen as being very selective about which patients/presentations would be accepted and there was an ‘us and them’ feeling, despite the departments recently becoming more integrated, with some staff unsure as to what GPED does and the rationale for the service. Some staff were concerned that patients were becoming increasingly aware of GPED, which could encourage attendance. When this happened, staff felt that they could not turn patients away or tell them that they had attended inappropriately. There was a feeling that, at a societal level, patients’ expectations had changed and the wider service needed to accommodate this.

Patients’ reasons for attending the emergency department

Patients attended for a number of reasons. Some stated that they could not get an appointment with their own GP or that the wait for an appointment was too long. Some patients had been advised by someone else to attend the ED (e.g. a carer who called an ambulance, a GP who thought that the patient needed a radiograph). One person attended for a second opinion after being seen in their own primary care practice. Although most patients had attended the ED previously, two had not.

Presentations included sore leg, burn, foot swelling, blocked ear and worsening headache, and a fall 4 weeks ago (for which the GP advised the patient to attend the ED for a radiograph).

Staff thought that patients may attend because they either could not get a GP appointment, or had failed to get an appointment previously and so had stopped trying to book GP appointments and instead came directly to the ED. They had the expectation that the ED would not turn them away and that their problem would be addressed that day. Staff felt that primary care should be the first point of contact, but decreased funding and increased patient volume led to GPs behaving in ways that helped to manage their workload, but had a negative impact on patients and the workload of the ED. For example, primary care GPs can refer patients directly for radiographs or to specialties, but are thought to send patients to the ED instead as this is a quicker process for GPs.

Staff felt that there were variations in what patients perceived as an emergency, but acknowledged that it is sometimes difficult for patients to know the level of severity/acuity and which services to access:

It’s easier for us ’cos we’re health-care professionals and, you know, like, ‘well, actually, I feel a bit ill, but it’s not that serious because I’m A, B, C, D and E’, whereas most people haven’t got that, have they?

Staff (Birch) 4, time 1

GPED patients were considered to be between acute ED patients, who need to attend the ED, and more chronic primary care patients, who probably attend their own general practice and have a good relationship with their GP. GPED caters for those who do not fall into either group. Some staff felt that patients from abroad used services differently and would attend ED if they could not get a GP appointment. Staff also implied that service efficiency encouraged patients to attend, sometime repeatedly, which was something they actively worked to prevent:

So it’s being recorded anyway but I don’t see anyone in less than an hour unless they’re ill because if you see them too quickly . . . we had one lady, she came back three times in a day, err, and that’s when we thought we really need to, err, we need to stop this somehow. We had a few keep coming back because it’s quicker to come here than it was to see their GP.

Staff (Birch) 18, time 1

Service literacy

The majority of patients were unaware of the GPED service, with only one having prior knowledge, although they lacked certainty about how the service worked and who it was aimed at. Because of this, there was no expectation that they would see a GP in the department.

Staff felt that patients had different expectations of primary and secondary care, so that although they may have been happy to wait for investigations in primary care, the expectation was that secondary care would provide this immediately. Consequently, patients were also perceived to attend the ED as a way of circumventing the system and being seen more quickly. Although staff had some sympathy with the difficulties patients faced in getting primary care GP appointments, they also felt that patients did not try to book a GP appointment and this was made worse by the availability of GPED:

I mean, we all know how difficult it is to get in and see your GP or to get and go and queue up at the walk-in centre and things like that, but I do believe that sometimes now it’s got to the point where people don’t even try because they know they can come and see a GP up here. Yeah ’cos I’m sure that that’s why people were coming what we used to call ‘inappropriately’ . . .

Staff (Birch) 4, time 1

There were undercurrents of frustration at this behaviour and although some staff challenged patients’ behaviour and tried to catch patients out by contacting the patients GP to check, others did not see this as their role:

And a lot of people say they can’t get hold of their GPs. I’m not very argumentative, but I’m not stupid either, if they say something I will double check it. So I normally ring their surgery and find out whether they did really have an appointment or not, usually they haven’t and they haven’t even bothered trying, they just turn up here because it’s easier.

Staff (Birch) 18, time 1

It was also considered that patients would attend the ED for a second opinion after seeing their GP. GPs themselves were sometimes reported to be the cause of a patient's ED attendance:

I’ve had a guy this morning that’s had a 2-year history of a cyst on his head and he’s come today to get it drained because his GP said ‘you’ll get it done quicker if you come to A&E’. So, people will come to A&E with the expectation that we can fix any chronic problem, and that’s what happened there.

Patients were seen to require education around appropriate ED attendance, and it was felt that it was the GPED GP’s role to do this. Staff also felt that patients were attracted to the Birch ED as it was well known and well regarded in the local area and, as a consequence, sometimes patients attended from outside the ‘catchment’ area.

It was highlighted that patients often used internet resources to research their problem prior to attending the ED. This was something one clinician found beneficial as they felt that it meant they could spend less time explaining differential diagnoses to patients.

Implementation

No patient views were available. Members of staff at all levels were involved in devising the GPED service, ranging from ED clinicians (including nurses) to trust- and executive-level leaders. Implementation was limited by staffing difficulties related to the lack of suitably qualified GPs and other clinicians, as well as a perception that GPs do not want to regularly staff GPED, but would rather work on a locum or ad hoc basis. This inconstancy in workforce supply reduced the ability of GPED to integrate with the wider ED system. The aim was for GPED to cover 12 hours per day for 7 days per week, but 12-hour cover was usually achievable only on Wednesdays and Thursdays, with patchy cover at other times. Some weekends had no GP cover. This inconsistency in service provision meant that the ED did not always send patients to GPED as they were unsure when it was staffed. There was a lack of integration between the ED and GPED, which staff felt was problematic. ED staff appreciated how beneficial GPED was, but only really noticed when it was understaffed:

When it’s here it works well, but when it’s not we miss it. It’s sporadic at the moment, and it’s too ad hoc. There needs to be a bit more of a structure and there needs to be a bit more integration with working. We’ve got one GP that we work with who’s amazing. He will pull, he will filter, he will be like ‘just come to me, just come to me.’ But, then others tend to be quite sedentary, sort of, working alone and we don’t even know if there’s a GP on or not.

Staff (Birch) 3, time 1

There were also practical implications for GPED staff working as locums. The model is that GP locums are employed on 6-month contracts, which are renewed after a short break between contracts (i.e. 2 weeks) to prevent invoking a permanent contract as funding is for a finite period only. This break in contract had unintended consequences, such as IT access and passwords no longer working.

It was suggested that, for effective implementation, and to increase the number of GPs, GPED needed to work reciprocally, that is GPs needed to feel valued, supported and appropriately remunerated. The lack of a designated space for the GPED service was seen as contributing to the lack of value placed on GPED. Although the relationship between GPED and the ED was generally considered to be good, some staff who were interviewed were unclear of the remit of GPED, the IT system that GPED runs from and whether or not GPED has access to primary care records. However, some staff did feel that the ED could learn from GPED GPs. They considered GPs to frame health and illness in a different way, which resulted in a difference in managing and tolerating risk. This was seen as beneficial in reducing inappropriate admission rates:

As I said, I think the GPs in A&E are a very good service. They could probably teach our doctors a lot regarding risk management and regarding long-term condition management and changing the boundaries. As a trainee ACP [advanced care practitioner], I am still at the point in my naive career where I see every patient as a ‘why can’t I send you home?’, whereas every GP sees a patient as ‘do I really have to send you to hospital?’ and I think getting the mindset of ‘you’re well until proven sick’, as opposed to ‘you’re sick until proven well’, and primary care is all well until proven sick and A&E is all sick until proven well. I think that would be very helpful, I think you could probably actually cut a lot of admissions by having a higher level of balls, which GPs have to have in their community role. Does that make sense?

Staff (Birch) 8, time 1

The lack of consistency in the streaming process was challenging. As well as variation in the presentations that individual GPs were comfortable with, there was no streaming pathway or protocol. This was seen as more problematic when the triage nurse was more junior. The streamers’ lack of general practice knowledge also made streaming more difficult in these circumstances. Although staff could feed back any concerns to the matron, the effectiveness of this was limited as the matron was not responsible for clinical staffing. There was also inconsistency in the clinical effectiveness of GPs. GPED sees 20–25 patients per day, but there is no specific/designated consultation time and there is an implication that GPs could see a greater number of patients. Some staff felt that there is not enough GP-appropriate work at Birch because there is a separate children’s hospital.

Perceived impact

For patients, being seen quickly and by an appropriate practitioner was key and they were not concerned whether that practitioner was an ED doctor or a GP.

Staff considered that patients were happy to be seen at GPED and their priority was to be seen quickly. It was thought that patients perceived GPED as different from their primary care GP service and thought that ED doctors were consultants and, therefore, senior to GPs (although it was also acknowledged that some patients were unaware of the difference between ED doctors and GPs). Staff also thought that patients saw GPED as a way of speeding up referrals to specialties.

Some staff considered GPED a positive addition to the ED in that it filtered patients who did not need to be seen in ED majors, improving waiting times for both patients with GP-type presentations and patients with major illness and injury. However, staff were sceptical that GPED would reduce patient volume and felt that it did not reduce the ED workload, but, instead, encouraged more people to attend with conditions that were seen as inappropriate for the ED. It was felt that GPED would change the way patients seek care and GPED was seen as encouraging patients to bypass their own GP and attend the ED instead. It was predicted that this may extend to patients presenting with chronic and acute problems. GPED was also thought to cause confusion for patients by providing a mixed message that encourages non-emergency attendance. The associated likely increase in footfall in the ED was seen as potentially reducing the pressure on primary care GPs, but conversely increasing ED attendance numbers. Patient confusion about the service was also demonstrated in reports of patients attempting to register with GPED as the patient’s own GP.

The filtering/triaging of patients depended on the experience of the nurse carrying out this role. The Birch GPED model had no formal streaming process or established protocols, but more junior staff could seek support from senior members of the team. Staff had seen the benefit of GPED models elsewhere and were aware of other models where streaming staff could redirect patients away from the hospital; however, this was not the policy at Birch.

GPED was seen as providing a more appropriate service to those patients suitable for GP management and, consequently, was thought to free up ED clinician’s time to see more seriously/acutely ill patients. GPs were considered to order fewer investigations for lower-risk patients and appropriately refer patients to the ED if they required more extensive work-up. It was expected that the planned co-location of GPED in its own purpose-built area would be beneficial to the ED as it could then concentrate on accommodating majors. GPs felt that GPED was already more merged with the ED than other departments in which they had worked. One GP enjoyed working with the ED team as they felt that it expanded their scope of practice, knowledge and skills. Similarly, ED staff considered GPs a valuable teaching and learning resource. However, the level at which GPs integrated with the ED was variable and this had an impact on working relationships. Although GPs could not quantify the impact of GPED, they felt that they were contributing to meeting the 4-hour KPI. They were also aware that GPED activity was audited and GPED patient outcomes were tracked.

Although there were some positives and GP input was valued, staff generally felt that GPED did not work as effectively as it should. For example, the service would benefit from greater consistency between GPs and the capacity to cover all required shifts, which was not possible all the time. When on duty, GPs were considered to see a reasonable number of patients and their clinical effectiveness was demonstrated by the service being missed when no GPs were available. The lack of consistency and the variance in GPs’ willingness and proactivity with regards to seeking out GP-type work were seen as major factors in reducing the impact of GPED. Some staff had worked at other trusts with a more comprehensive GP service and associated this consistency of service with a reduction in the amount of GP-type patients seen in the ED.

Expectations of GPED (time 1)

No patient stakeholder feedback was gained.

Staff felt that greater commitment to a fully staffed, consistent service was necessary to increase service impact. It was felt that GPED would increase the numbers attending the ED and reduce the pressure on primary care GPs. Staff were unsure whether or not moving GP resources from primary care to the ED would make any difference as it would not actually increase the resources, but simply redistribute ED and GP provision.

Chestnut pen portrait

FIGURE 11.

Chestnut pen portrait. HCA, health-care assistant.

Current model

• All patients are streamed at the front door. There is a quick assessment by a streaming nurse before they sign into the department; the streaming nurse also takes some very basic observations, that is oxygen saturation, blood pressure and heart rate (but no blood samples are taken).

• UCC deals with minor illnesses and minor injuries:

  • It has access to radiographs.

  • It is staffed by GPs, health-care assistants (HCAs), nurse practitioners, usually one or two GPs at a time and ENPs. Issues identified around this include GPs generally not being trained enough to deal with injuries – nurse practitioners mainly deal with these injuries.

• Majors in the ED deals with major illnesses and major injuries.

• There is a separate paediatric ED; the UCC staff see suitable paediatric patients in the paediatric ED.

• The OOH general practice is co-located in the UCC, but this is a separate system based on appointments. OOH GPs also do home visits and telephone consultations; these are not done as part of the UCC. They are funded separately, but both are funded by the same CCG. If the UCC gets really busy some patients can be transferred to OOH GPs if there is capacity, but the UCC and OOH general practice run on separate computer systems so this can be challenging.

• Some staff move between departments; in particular, nurses and HCAs might move between the ED and UCC.

• Some difficulties in getting enough GPs to staff the UCC were reported; the UCC is currently staffed from a local GP collaborative, with locums but not many permanent ED GP staff.

• There were some issues with ED staffing as well: it was often short staffed and this was further exacerbated by needing to send staff from the ED to support the UCC when they were short staffed or too busy.

• After streaming, patients sign in with reception in either the ED or UCC:

  • If it is the ED, they are then triaged by a nurse before seeing a doctor.

  • If it is the UCC, they are then assessed, usually by a HCA, before seeing a doctor.

Streaming/patient flow

• All patients are streamed via the streaming nurse opposite main reception.

• Patients can be streamed to:

  • On site – majors, UCC.

  • Redirection on site – maternity triage, eye clinic (on site), specialty or ambulatory care.

  • Redirection off site – pharmacy, eye clinic, own GP, genitourinary medicine clinic (off site or on site), dentist, OOH dentist, GP hub or another hospital with an ear, nose and throat (ENT) specialty.

  • GP hub – open 08.00–20.00, appointments available on the hour from 08.00–19.00 (note that patients can also be sent here from UCC after streaming to UCC).

• Patients get frustrated when they have to go to another location on site or off site, and when they have to repeat their symptoms to different staff members.

• There are strict streaming criteria; however, experienced and higher bands of nurses were reported as being better at streaming.

• Streaming nurses occasionally call on ED staff to provide a second opinion on where to appropriately stream patients.

• Patients can be referred to tertiary departments from the ED or UCC:

  • Issues were reported with patients from the UCC being referred to, for example, fracture clinics, but with no fractures.

• NHS 111 referrals are made to the OOH GP service, but sometimes they get referrals from out of area. When this happens, the UCC duty manager has to call the patient and tell them not to come to the ED and make an appointment at the appropriate OOH service for them, as well as calling the NHS 111 service to let them know.

Future plans for GPED

• A new purpose-built centre for the ED and UCC is being built.

• Construction was not completed by the predicted end date and staff seem unsure about when it will actually happen.

• The new building will see the ED and UCC located together without so much distance between them.

• The new building will also have more space for the ED and more beds.

History of GPED

• The UCC was previously operated by a private provider (2012–17) that did not renew their contract. It then came back under trust management. This history has caused some tension between the ED and UCC staff historically.

Physical environment

• The ED and UCC have been in ‘temporary accommodation’ for around 2 years.

• The reception and ED are in Portakabins^® (Portakabin Ltd, Huntington, UK), with limited space. The UCC is also in a temporary building down the corridor, with limited space.

• The streaming booth is opposite the main reception desk and it has a shoulder-height wall/small divider. It is a very small space, with room for a small desk with a computer, a chair for the streaming nurse, a chair for the patient and a small sink. There is limited privacy as the waiting area seats are directly behind the low wall.

• Staff point out places where the floor is coming away and has been taped up or where temporary desk legs are loose. One nurse explains that they feel sorry for the cleaners as it does not matter how clean they make it, the building still looks shabby.

• There is a shortage of space to see patients in the ED; sometimes there are discussions with patients in corridors and, when it is busy, there are patients in beds in the corridors.

• The UCC is a couple of minutes’ walk from the ED, and there are green footprint stickers on the floor to direct people from reception.

  • This ‘geographical dislocation’ (C.S.022.int) was highlighted as a potential patient safety issue as if a patient needs to be transferred to the ED, a bed must be used to transport them from one location to another.

• The paediatric ED is along the corridor and there are blue footprint stickers on the floor to direct people from reception.

• Patients can get frustrated at being sent around the site and off site.

Time 3 summary

Pen portrait

The urgent treatment clinic (UTC) was run by a private provider until April 2019; it is now dual run by the GP collaborative and the hospital. Since the change in provider, two GPs now work in the UTC. In general, there is a GP present 7 days per week. The two GPs split the shifts, working either 11.00–19.00 or 4.00–13.00. In addition to the GPs, there are four permanent nurses present in the UTC from 20.00/20.30–22.00/22.30. Overnight, the UTC is staffed by junior doctors rather than GPs and ANPs, partly because of difficulties in recruiting GPs for OOH shifts. Staff reported this change to the rota as a significant change at this site since time 1 (or, possibly, time 2). GPs in the UTC can order tests at this site, if required, but they are considered part of standard clinical practice for GPs working at this site.

Since December 2018, the ED/UTC has been co-located in a new building. For > 12 months prior to this, the UTC and ED were housed in a temporary set-up in which they were physically separated (this was when time 1 interviews took place). The new building was designed/built when the ED/UTC was run by a private provider. It is noted by staff that there is less space (i.e. treatment rooms) in the new-build department. Its design may have been different (i.e. to facilitate cohesion and partnered working between the EDD/UTC) had it been known at the time that the trust would take over from the previous provider.

Patients arrive via either an ambulance or the walk-in entrance and are streamed by the nurse (band 6 or 7) at the main reception; brief clinical details are taken to determine where the patient will receive the most appropriate care and the receptionist checks the patient in. The streaming system is based on the Manchester triage. 110 A low score would indicate that a patient could be suitable to be seen by the UTC. Previously, streaming was conducted by a non-clinician and, because of this, the system was incredibly rigid, so that any hint of acuity would be seen by the ED.

Patients are seen in order of their arrival. The UTC/ED at this site has permanent HCAs who complete vital signs checks with patients within 15 minutes of rotations being booked. These checks help to quickly identify if a patient is seriously ill. A nurse then triages all patient (e.g. observations, urine, blood samples) to assess the urgency of the illness or injury. Triage is conducted on the basis of perceived urgency. Patients are then signposted to the most appropriate care:

• resuscitation for people with life-threatening illness or injury

• majors for people with urgent illness or injury

• the UTC for people with an illness or injury that can be treated in one step

• seen by an ACP and then directed off site to either the GP hub or their own GP.

From there, patients may be required to stay in hospital or may be discharged from the ED. Patients are treated here if they are aged > 17 years; younger patients are directed to the paediatric ED, whether or not they need to see the UTC/primary care staff. Both the ED and UTC share the same computer system; however, the system is designed for the ED, so it is not a good fit for the UTC.

Local context

There is an assumption among some patients that the health services in this locality are poor because they are based in the south of England. There is a strong acknowledgement from patients (and staff) that the GP service and wait-time is no better, and, perhaps, worse, than attending the ED/UTC. There is also recognition that the health service needs more financial resources for patients to receive medical care quicker.

National context

. . . it is interesting to go around the country and see the different units and see the way they cut the cake if you like in terms of who sees what and how we decide who sees what. The problem I generally have with that is the complexity of the rota and how these units work in the way that, you know, sometimes there’s a GP available to see these things and sometimes there is not and then they run two entirely different service models depending on who’s around that day or now, oh because Dr Smith has gone home, oh well in that case we’ll do it this way. I just do not like that whole set-up.

S 009, time 3

Trust culture

Team working was apparent at this site. It appeared to be common for GPs to be approached by other staff members (e.g. F1 doctors, ANPs) and asked for their opinion on the most appropriate course of care for patients who presented as complex cases. Some patients appeared to be resistant to ANPs’ diagnosis and requested to see a GP for a second opinion when the ANP did not give the patient the antibiotics they were expecting (i.e. because there was no infection present). In such cases, the GP felt that it was important not to undermine the diagnostic decisions of their colleagues. The new build, which had physically merged the ED and UTC, was perceived by most staff to have facilitated the interaction and integration of staff.

Staff appeared frustrated by patients’ need to be seen by a medical practitioner for non-urgent care whenever patients felt that this was required.

Interviews with both staff and patients identified a number of scenarios in which patients were using the ED/UTC services inappropriately. A prominent reason was patients being unable to get a timely appointment at their own GP, which meant that people often used the ED/UTC as a first port of call, rather than attempting to get an appointment with their GP. Staff noted that they would ask patients whether or not they tried to get an appointment at their GP before attending the ED/UTC if the staff did not feel that the patient’s issue was urgent. Although staff were aware of the protocols of redirecting patients away from the service when required, they did not necessarily agree with them. This was, in part, because staff felt that the service was there to be used, and it was considered more efficient to treat patients there and then, particularly when their issues could be resolved within minutes, rather than redirecting them away from the ED/UTC to another service in the community. It was also recognised that most patients were right to attend the ED/UTC when they did, as they required a responsive primary care service, a service that many GPs can no longer provide because of the lack of timely appointments.

The integration of the UTC within the ED and the implementation of the 4-hour wait target appealed to patients and, in turn, increased the number of attendances at the ED/UTC.

Service literacy

Most patients expected to have to wait a few hours to be seen at the ED/UTC; however, they anticipated this and it did not appear to be an issue. This was primarily because they knew that they would be seen on the day, whereas they would probably not be able to get an appointment at their GP surgery for days or weeks. In most cases, patients were seen by a clinician quicker than they anticipated.

Some patients attended the UTC knowing that they would be seen by a GP; others were not aware that GPs were now located in the ED. Nevertheless, most patients were satisfied with being seen by any medical practitioner at the UTC, whether that be a doctor or an ANP, as long as it meant they would receive the medical attention/treatment they felt that they needed.

However, some patients were dissatisfied with the practitioner they saw if it did not result in the outcome they were expecting, such as receiving antibiotics. In such cases, if the initial examination was completed by an ANP, patients were more likely to request to see a GP for a second opinion. Although staff facilitated this, GPs noted that it was vital that they were not seen to be undermining the decisions of their colleagues.

Patients attended the ED/UTC for reasons such as it was closer to their home than their own GP surgery, they were attending for very minor issues that did not require urgent/immediate attention (‘I see people come in with a cut finger . . . another had a bad splinter. Splinter! It’s not like its life-threatening’ Chestnut, Patient 012), being unaware of the standard pathways of referrals, and incorrectly using the ED/UTC in an attempt to be referred quicker. Several patients also attended the ED/UTC in an attempt to gain information about test results that had been requested by either their GP or private health services. One patient explained that they visited the ED/UTC for a practitioner to interpret test results, as they felt that ED/UTC staff were more ‘qualified’ than their own GP. It was noted by staff that some patients made an assumption that primary and secondary care is a fully integrated health service; however, this is not the case. For example, the ED/UTC do not have access to GP-based medical records.

It was perceived by some patients to be less strain on the NHS to visit the ED/UTC rather than their own GP surgery.

Patients’ reasons for attending the ED/UTC were as follows:

• finger injury, mallet injury, male patient

• injured wrist, female patient

• lethargy, described as a complex case, male patient

• suspected bronchiolitis, male patient

• finger injury due to attack, female patient

• injured wrist, male patient

• painful back, requiring report of back scan to be read

• broken finger

• back pain

• hip pain.

A recurrent reason for patients attending the ED/UTC was that they could not get an appointment with their own GP. Some patients did not attempt to get an appointment with their own GP before attending the ED/UTC, as it would be days/weeks before one would be available. For many patients, this was not their first time attending the ED/UTC. Some patients noted they would rather attend the ED/UTC than their own GP surgery for convenience, because it was closer to home or because attending the ED/UTC meant that they could have tests completed immediately that their own GP would have had to refer them for (e.g. radiograph), essentially ‘cutting out the middle man’.

Staff commented that some patients had replaced their own GP with the ED/UTC and that patients knowing that GPs were present in the ED/UTC encouraged its use.

One patient perceived that, regardless of whether patients were seen at the GP surgery or in the ED/UTC, the need for any medical care put a strain on the NHS.

Implementation

Facilitators

The UTC, including the nurses, GPs and ED doctors, worked closely and the service felt altogether more integrated in terms of daily working. The GPs’ varied skillset made them well placed to work within the ED/UTC to deal with patients who present with primary care issues. Their skill set meant that they could balance risk, they tended to order fewer tests and discharged people home quicker than ED doctors.

Barriers

The hospital struggles to recruit locum GPs owing to changes to rotas that now see the shifts run through the day. As a result, the GPs who would usually work OOH after spending the day in their own practices are unable to pick up the shifts. Having to use temporary staff and being unaware of their skill set proved problematic when trying to rota a multiskilled team. As a result, the UTC is staffed with junior ED doctors overnight.

Although it was felt that a GP’s skillset was ideal for treating minor illness, not all GPs had the correct knowledge to treat minor injuries. This was noted to be because GPs see very few minor injuries on a daily basis in their own surgeries, as patients with injuries are more inclined to seek medical care from the ED. Staff noted that in other UTC departments around the country GPs are responsible for treating patients with minor illness and ANPs are responsible for treating those who present with minor injuries. However, there was a reluctance at this site to rota staff in such a way because of the planning and logistics involved. It was the department’s preference to employ multiskilled staff, such as GPs, who can deal with both minor illness and injuries, although this was not always possible when having to cover shifts with locum staff whose skill sets were unknown.

Staff did not respond well to the transition from the ED/UTC being ran by a private provider to the trust, and some are still adjusting. When ran by the private provider, the ED and UTC were more autonomous and less integrated; this ethos is noted to be particularly prevalent among staff who have been at the site for many years and is seen to be a barrier to the full integration of the UTC/ED as a department. By encouraging staff to be more integrated in their thinking, the UTC can more effectively alleviate pressure from the ED. For instance, staff discussed how tests were not part of UTC care when it was run by the private provider. Now, tests (e.g. blood tests, ECGs, radiographs) can be requested if practitioners in the UTC believe that it will give them greater certainty regarding the correct treatment and care within the UTC and a timely discharge for the patient. However, some staff have not adjusted to these changes.

The trust would like to increase the number of patients seen by the UTC to further reduce the demand on the ED. To do this, an alternative streaming system would need to be put in place (currently, band 6 or 7 nurses are responsible for clinical streaming). GPs are cautious that this would increase the number of patients incorrectly streamed to them when the patients actually need emergency care.

Perceived impact

The new building has facilitated better integration and multidisciplinary working between the ED and UTC in comparison to previous set-ups where the ED/UTC were located much further apart. It has also helped to refine the streaming process and transfer patients from the UTC to the ED if needed.

The UTC was seen by both staff and patients to alleviate pressure on ED.

Changes to the rota in April saw the UTC being covered by ED doctors rather than GPs overnight, which reduced the department's compliance with the national 4-hour standard until staff adjusted to the set-up.

Patient safety was considered to have improved since the non-clinical streaming had been taken over by band 6 and 7 nurses.

Expectations

That’s the question. Because the question is, what are the GPs here to do? And the answer is no one is actually sure, as far as I can see, because I have sat in, you know, NHS England meetings where the argument is both. It’s only people who turn up with primary care issues that it would be far more sensible to put GPs in there to deal with those primary care issues because they’re the most skilled at doing that. They’re the most experienced. They know the pathways. They’re the ones who are able to deal with comorbidities and complex medical conditions, etc., the old party line and that’s fine. But then, on the other hand, you’ve got the same or similar level of seniority people saying that you need to be redirecting these people away, they shouldn’t be using secondary care, it’s an expensive way of doing it. So they’re turning up with a primary care problem, you should be redirecting them more appropriately to which to me, is two completely opposing things. As I say, it depends who you ask which answer you get and the answer is, at the moment, for example, we’re still commissioned to redirect an awful lot of stuff, you know, a lot of people would suggest it’s probably a lot more efficient to just see them at the point of care. Try and educate them for next time, although good luck with that! But, really, just see them and deal with the problem they’ve got, but we’re commissioned to send them away. So someone comes in asking for a repeat prescription, as long as it’s not an emergency, you know, if they’ve run out of insulin, they’re type 1 diabetic, then fine. But if they want their routine blood pressure medication then we just don’t give it to them. We’re commissioned to send them away, but it would take me 2 seconds to do that, but I’m not allowed to and I’m not supposed to. I mean, you can start talking about the fact that a lot of prescriptions are embargoed because the patient hasn’t come for their blood pressure check and if we start handing them out then they’re bypassing the system. But there are other drugs they come for where they just, you know, people just vote with their feet. In [place] there’s about a 3-week wait for a GP appointment. They come here. They know it’s going to take usually a 2.5-hour wait, on average, so they take their choice.

Chestnut, Staff 009, time 3

Hawthorn pen portrait

Current model

• GPs are in operation Monday–Friday, 18.00–08.00 the following day. There are two GPs 18.00–23.00, one GP between 19.00 and 23.00 and one GP 23.00–08.00. There is coverage by GPs 24 hours per day at weekends. There is currently a HCA who undertakes patient observations in the OOH service at weekends.

• GPs are part of an OOH service and so patients can be triaged from the ED, or they can book telephone or face-to-face appointments via NHS 111. GPs also perform home visits. The majority of patients who GPs see are booked via NHS 111. Patients who have a booked appointment through NHS 111 receive priority over ED-streamed patients or patients who ‘drop in’. Patients who arrive at the OOH service without an appointment are told that they are not supposed to do this and are seen, but have to wait for a time to become available. The analogy given by a GP was that ‘it is an OOH [service] with streaming tagged on.’

• In the ED, there are two rooms dedicated to triaging patients. One has been purpose built as a result of the capital funding and was in place at our 6-month visit; the initial plan was to have two nurses triaging at all times, but, despite employing two nurses for this purpose, staffing issues have not allowed this to happen. However, there is usually one nurse triaging patients, with the other room used for a minor injuries nurse or ENP. During very busy periods or when the demand for triage is too high for one nurse, a second nurse may be called from the ‘shop floor’ and will start triaging from the second triage room. Sometimes an ED doctor may also sit in with the nurse who is triaging to try to discharge patients at this point when possible.

• After booking in at reception, a card with the patient’s chief complaint is placed in a box by reception; the triage nurse uses this card to call patients for triage. Patients are seen in chronological order. The minor injuries nurse will either see patients after triage or select patients from the box who have presented with a minor injury, and so some patients may not be triaged.

• There is also now an investigation and treatment room, which is staffed by a HCA and nurse. Patients who need further observations, for instance an ECG, will be sent here before being seen by a doctor, who will decide that the patient is either fit to wait in the waiting room or needs to be admitted to a cubicle. Patients can be sent to the investigation and treatment room for assessment/monitoring from either an ambulance or triage. For instance, patients who present with chest pain at triage may be sent for an ECG at investigation and treatment.

Streaming patient flow

• The triage nurse was able to perform ECGs, but this has recently been removed from the triage room.

• Observations at triage are limited to blood pressure, temperature and blood sugar.

• There are no strict streaming criteria.

• When a GP is not in operation, patients can be streamed to ED minors or majors. When a GP is in operation, patients can be streamed to minors, majors or the OOH service.

• The triage nurse is not able to see how busy the OOH GPs are or the number of GPs who are on duty. This is a particular problem as the majority of GP patients are booked through NHS 111. If the GP is unable to see any ED patients because they are on a home visit or too busy, the only way to inform the ED is to physically go around to the ED.

Future plans for GPED

• There are no further plans to expand the hours for GP service beyond or outside the GP OOH that is currently in place.

• There are no immediate plans for any structural changes to the ED and the GP OOH set-up in terms of consultation rooms/layout of the service.

History of GPED

• Prior to the GP OOH service being located around the corner from the main ED, the GP OOH service was located further down the corridor.

• In terms of hours of service over the past 12 months, the service has not changed from what was described above, with the exception of the new paediatric/child waiting area and additional streaming rooms.

• At one point, the service had a GP in the ED, but the staff felt that this did not work, which is part of the problem now because they want to keep the services separate and feel that it did not work last time.

Layout

• The space used for GPs is not purpose built. GPs are based in the fracture clinic, with the same reception, rooms, etc. GPED is in operation when the fracture clinic is closed.

• Since the GPED capital bid funding, a new paediatric ED has been built.

• There are two rooms for triage in the ED. One has been purpose built since our first visit. This was to have two nurses streaming patients to the GP. However, currently, there is one nurse triaging and one room is used for an ENP/minors nurse.

• The OOH service is managed by the trust, but there is a different manager for the ED and OOH service.

• There is nothing in either the GP OOH service or the ED that states the number of GPs who are on duty or which GPs are on duty. Staff commented that this is problematic as some GPs will not see certain types of patients. The GPs also do not usually come to the ED and introduce themselves to the streaming nurse at the start of their shift.

FIGURE 12.

Hawthorne flow diagram. Black text indicates the baseline visit, dark-blue text indicates the 6-month follow-up and light-blue text indicates the 12-month follow-up.

Nutmeg pen portrait

The GPED/UCC service was very recently (in December 2018) relocated on site to an area integrated with the ED, which is where the fracture clinic was previously located (i.e. connected to the ED), and this is where the service was operating during the time 1 data collection. The walk-in entrance opens to the large waiting area, with a corridor leading to the majors department to the left of the entrance.

FIGURE 13.

Nutmeg layout.

Service times

Although there is an aspiration to operate an ‘all-hours’ service, this is not being achieved currently owing to a shortage of GPs. A GP normally arrives at the urgent treatment unit at 08.00 or 09.00 and sees selected patients until the end of their shift, which is of variable duration.

Model and patient flow through GPED

This is a joint-led service for ambulatory patients. From 8.00 until 12.00 the GP service provides staff to look after, or to see and treat, appropriate primary care presentations. Outside of these times, the ED looks after this group of patients in conjunction with the GP service. The ED sees the patients that are unsuitable for the GP service, for instance those that are injured.

Interviewees described the service as changing from co-located to an integrated/embedded model. The primary care service in the UCC is provided by a private provider [the same provider as was involved in the previous co-located service (see Linden Pen Portrait)]. The site has a protocol which the band 6 and 7 nurses use to stream people from the front door if they meet the protocol to wait to see the GP. This means that some patients are not seen by an ED nurse through first contact; they are instead taken straight to chairs to wait to be seen. Subsequently, one of the GPs (or HCAs) will take the patient in to complete observations and a short assessment to check that they are suitable to wait in the queue. Then the GP will see and assess them. The ED staff have little contact with these patients, except at night when the GPs and other primary care staff are not present. If a patient brought in by ambulance is not considered suitable for the ‘majors’ area of the ED by the ambulance crew, they can be reassessed and directed to the GP service.

Rowan pen portrait

FIGURE 14.

Rowan flow diagram.

Baseline

Layout

The GPED service is co-located within the ED and operates from the same area as the ED minor injuries unit.

Service times

The GPED service is available 7 days per week from 08.00 to 23.00 and there is one GP available at any one time. The OOH GP service operates from the same area outside the GPED times (this service is accessible by appointment only).

Model and patient flow through GPED

Patients walk in through the main ED reception (they may also have been directed to the ED by NHS 111). Generally, patients report to main reception, where a triage nurse queries their reason for attending, asks additional questions and, subsequently, directs the patients to the main ED waiting area or to the GP or minors service. If they are directed to the GP service, the patient is given a GPED sticker. For main ED patients, the clerk registers the patient on the system at the main ED reception, whereas for GP or minors the patient is registered at the GPED/minors reception. Frequently during data collection patients presented at the minor injuries reception with GPED-appropriate concerns and the clerk had to direct them back to the main ED reception for triage. If they were triaged for the GP then they would return to be registered at the GPED reception, where the clerk would query whether or not they would allow the GP in the ED to access their GP notes. The patients were then asked to wait to be seen by the GP. If the GP deems it necessary, they can redirect patients back to the main ED queue. The GPs do not have access to any routine specialist referrals; patients who require this need to see their own GP to get an appropriate referral. Paediatric patients who register at the main ED reception after being triaged are asked to wait in the specialist paediatric ED waiting area whether they are waiting to see the paediatric ED or the GP.

Staff/workforce

The triage is undertaken by the band 6/7 ED nurse located at the main ED walk-in reception; currently there is only one primary care specialist nurse in post to work within the GPED service. There is one GP on shift at any one time and, during baseline data collection, there were some days when the patient care specialist nurse and GP were unable to get to the ED either entirely or for part of their shift because of the weather. There is currently a large pool of GPs (> 50) from the GP federation, which the trust works with in delivering the GPED service, who are undertaking shifts for GPED. The GP shifts are released on a certain day each month/at a certain time and GPs access this system and choose the shifts they would like to work. In addition to this, the GPs who saw the most patients each month were allowed to pick their shifts before other GPs.

Future plans

No major structural changes are planned, but there are plans for the development of the service and more coverage for the primary care specialist nurse.

Six months

Layout

There were no changes to the layout of the GPED service, but screens had been assembled at the main ED reception and the minors’ reception.

Service times

Service times were the same as at baseline.

Model and patient flow

The model and patient flow had not changed from baseline and the process had remained the same.

Staff/recruitment

There were no changes at this time; however, the site was in the process of recruiting more band 7 nurses to work as primary care specialist nurses within the GP service.

Future plans

There were no major future plans indicated at this stage, apart from the plans outlined above to recruit more primary care specialist nurses to share the work of the primary care nurse specialist.

Twelve months

Layout

There were no major changes to the existing layout structurally; however, in terms of the space from which the GPED service operates, there had been some minor changes. The service is still located in the same place as it was previously, but, at baseline and at the 6-month check-in, the primary care specialist nurse worked from a consulting room opposite the main ED waiting area, whereas at the 12-month check-in the nurse was more permanently located in one of the consulting rooms in the minor injuries unit area, which is also where the GPED service and GP consultation room were situated. This meant that the minor injuries nurses had one fewer consultation room to use when the GPED service was operating.

Service times

There were no changes in service times from baseline and the 6-month check-in, and the service is available 7 days per week, 08.00–23.00. There is still only one GP available at any one time and the daily service is generally divided into three GP shifts, albeit GPs sometimes work double shifts.

On a limited number of occasions during the winter, there had been two GPs on shift simultaneously, but this was not something the site is looking to adopt long term as it is not considered cost-effective and it is difficult to consistently predict the times when there will be a high volume of patients. The project manager did note that they were considering potentially starting the service a little later to provide a period of overlap between the two GPs, but this had not taken place yet.

Model

There were no major changes from baseline or the 6-month check-in, and the site continues to utilise the Manchester triage. There had been a few minor changes to the triage criteria (e.g. the conditions for which the GP does not see patients) at the 12-month check-in.

Patient journey and flow

The patient journey and flow through the service were the same as at previous data points. In observations, it was noted that patients still often presented directly to the minor injuries unit reception, from where they were directed to the main ED reception to be triaged first. After this, they were asked to return to the minor injuries unit/GPED reception if they were appropriate for either of the services. GPED stickers continued to be given to the patient, which they were to hand to the reception clerk.

In terms of patient journey, the only difference was that patients who required observations or urine tests prior to seeing the GP (or being directed back to the main ED, if required) now see the primary care specialist nurse, who completed these observations and then liaised with the GP when required. After this has taken place, the results were given to the GP and the patient went back to the waiting area and were seen by the GP in the ED in due course. If there were any issues regarding observations or urine tests that indicate that the patient needs further tests or their condition is more acute, then the patient care specialist nurse could move them into the main ED queue and direct them to the main ED waiting area to wait to be seen.

Staff/recruitment

Since baseline and the 6-month check-in, there were now four primary care specialist nurses, instead of one single nurse, working the primary care specialist shifts alongside the GP. There were no more nurses on shift at any one time, but the shifts were shared across the four nurses, who also work in the minor injuries unit as part of their role.

Future plans

No significant further changes to the system were planned.

Teak pen portrait

The Teak ED has been operating their model for > 12 years. The department was refurbished and reopened for use in March 2018.

Patients walk in and present to the front desk. Here, they are streamed by a nurse (band 6 or above). The streaming nurse is able to stream to a number of different places within and outside the department. The department is split into these areas, but all are in close proximity to each other. Inside the department, patients are able to go to the UCC, paediatrics, triage, majors or resuscitation. External directions for patients include the patient’s own GP, the GP hub or the pharmacy. The OOH GP service, which is on site but external to the department, is offered presently. This is via an external provider and has been in place for the last 2 weeks (i.e. early November 2018), operating between 18.00 and 22.00. This provides 20 appointments, but the maximum number filled so far has been 13.

Once a patient has been streamed into the department, they are given a number. When their number is called, they go to a registration desk to book in and, afterwards, they are directed to the appropriate seating area.

The UCC operates 24 hours per day, 7 days per week. Most staff rotate through the department, except for the GPs and ENPs. There are a number of GPs who are regular providers and are well known to the team. Other shifts are filled by locum doctors. The GPs are employed by the trust on a 0-hour contract basis.

Once a patient goes to the UCC, they are seen by a nurse, who conducts an initial assessment. They then wait to see either a GP or an ENP. ENP nurses see patients with injuries, whereas GPs see patients with minor illnesses and work within a protocol. This means GPs do not provide extended investigations.

FIGURE 15.

Teak flow diagram.

Force-field analysis

A force-field analysis80 is conducted to identify the forces that drive change and those that resist change to understand the current state of equilibrium (Figure 16). For change to occur, the driving forces must be strengthened or the resisting forces weakened. It was intended that the findings of the study, if shared effectively, could become an additional driving force for change.

FIGURE 16.

Knowledge mobilisation force-field analysis.

Theoretical basis for the knowledge mobilisation approach

Two theories of knowledge mobilisation were drawn on to inform the approach and strategy for the study. Both theories explain how knowledge is shared through everyday interaction and highlight where opportunities to introduce new knowledge may lie. By adopting the knowledge mobilisation approaches that are likely to be the most effective for the key stakeholder groups identified (i.e. clinicians and policy-makers), the impact of the study findings on practice could be maximised, as well as the opportunities for practice to influence the potential study outputs. Previous research looking at how policy-makers and clinicians share and mobilise knowledge have highlighted the importance of personal relationships, face-to-face interactions and trustworthy sources of information. 87,92,93,111