Article Text
PDF
Abstract
Background Addressing increasing patient demand and improving ED patient flow is a key ambition for NHS England. Delivering general practitioner (GP) services in or alongside EDs (GP-ED) was advocated in 2017 for this reason, supported by £100 million (US$130 million) of capital funding. Current evidence shows no overall improvement in addressing demand and reducing waiting times, but considerable variation in how different service models operate, subject to local context.
Methods We conducted mixed-methods analysis using inductive and deductive approaches for qualitative (observations, interviews) and quantitative data (time series analyses of attendances, reattendances, hospital admissions, length of stay) based on previous research using a purposive sample of 13 GP-ED service models (3 inside-integrated, 4 inside-parallel service, 3 outside-onsite and 3 with no GPs) in England and Wales. We used realist methodology to understand the relationship between contexts, mechanisms and outcomes to develop programme theories about how and why different GP-ED service models work.
Results GP-ED service models are complex, with variation in scope and scale of the service, influenced by individual, departmental and external factors. Quantitative data were of variable quality: overall, no reduction in attendances and waiting times, a mixed picture for hospital admissions and length of hospital stay. Our programme theories describe how the GP-ED service models operate: inside the ED, integrated with patient flow and general ED demand, with a wider GP role than usual primary care; outside the ED, addressing primary care demand with an experienced streaming nurse facilitating the ‘right patients’ are streamed to the GP; or within the ED as a parallel service with most variability in the level of integration and GP role.
Conclusion GP-ED services are complex . Our programme theories inform recommendations on how services could be modified in particular contexts to address local demand, or whether alternative healthcare services should be considered.
- primary care
- emergency departments
- general practitioners
Data availability statement
Data are available upon reasonable request. The full NIHR peer-reviewed report is available at Davies F, Edwards M, et al. Evaluation of different models of general practitioners working in or alongside emergency departments: a mixed methods realist evaluation. NIHR Journals Health and Social Care Delivery Research (in press) https://www.journalslibrary.nihr.ac.uk/programmes/hsdr/1514504/%23/.
This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
General practitioners often work in or alongside EDs (GP-ED) with the aim of addressing demand and improving patient flow.
Available research on whether this aim is achieved is inconsistent with variation in how different service models operate, subject to local context.
Previously reported qualitative data from our work highlight the complexity in how these service models operate, both within and between GP-ED models. Our quantitative data (of variable quality) did not show a reduction in patient attendances and waiting times with variable findings for hospital admissions and length of hospital stay.
WHAT THIS STUDY ADDS
In this mixed-methods realist evaluation, we present programme theories that describe how the different GP-ED service models operate in different settings.
Inside the ED, the GP-ED service is integrated with patient flow and general ED demand and GPs take on a wider role than usual primary care. Outside the ED, the service addresses mainly primary care demand, with an experienced streaming nurse facilitating the ‘right patients’ are streamed to the GP. Other models operate within the ED as a parallel service, with most variability in the level of integration and GP role.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
These findings can be applied to local context and allow commissioners and service leads to consider how their GP-ED service could be modified to address local demand, or whether alternative healthcare services should be considered.
Introduction
Improving ED flow to reduce demand on emergency care services and improve waiting times is a key ambition for NHS England.1 However, evidence is needed to inform which healthcare service models best facilitate this.2 Establishing primary care general practitioner (GP) services in or alongside EDs was advocated by NHS England pre-pandemic as an approach to manage increasing patient demand, supported in 2017 with £100 million (US$130 million) of capital funding.3 This led to an increase in GP-ED service models in England from 81% in 2017 to 95% in 2019, despite limited evidence for effectiveness and safety outcomes.4–7
GP-ED (or primary care) service models have been described as operating: inside the ED, integrated with patient flow or in parallel to that activity; or outside the ED, on or off the same hospital site (figure 1).8 Using this taxonomy, Benger et al conducted a mixed-methods analysis of Hospital Episode Statistics for 32 type 1 English EDs with GP-ED services (6 inside-integrated; 15 inside-parallel; 11 outside-onsite) in England 2017–2019.9 They reported slight reductions in the rate of reattendance within 7 days (with negligible clinical significance) but no significant difference in attendance, waiting times, hospital admissions and mortality across all models. However, a substantial degree of heterogeneity was noted in their findings and considerable variation observed during qualitative data collection at their 10 case study sites for how the service models operate, subject to local context.9
General practitioner (or primary care) service models in or alongside EDs (published as open access in accordance with Creative Commons Attribution 4.0 Unported (CC BY 4.0) license (Cooper et al)).8
Evaluation of the effectiveness of these service models that work differently in complex adaptive socio-technical systems that vary in location, population demographic, workforce skillset and wider service provision is challenging.10 Routinely collected data may not enable detailed understanding of the complexity, and analysis and interpretation may be limited by variable quality of those data.11 Realist methods are well suited to evaluating complex interventions such as these, exploring variation and nuance in different contexts to explain what works, for whom, under what circumstances and how .12 Realist methods generate ‘programme theories’ from ‘initial rough’ and ‘refined’ theories, described by context-mechanism-outcome (CMO) configurations. These provide a means to understand the interactions that lead to outcomes of interest, and which can be a basis for making changes to improve those outcomes.12
In this paper, we integrate the findings of our mixed-methods realist evaluation the GPs in EDs study (2017–2021) including 13 GP-ED case study sites from England and Wales (3 hospitals with inside-integrated; 4 with inside-parallel; 3 with outside-onsite; and 3 control models (no GP-ED) in operation).13 (Key findings from the 17 publications are listed in online supplemental appendix 1.) We present programme theories to describe how the different GP-ED models operate and are influenced by wider system, department and individual factors. These provide more nuanced interpretation, according to each GP-ED model used, than is captured in the taxonomy of models.8 Commissioners and clinical leads can then consider further how their services are configured for their aims and local context or which factors, in particular the mechanisms of how models are operating, may be modifiable to address local demand.
Supplemental material
Methods
Study design
Realist methodology is a theory-driven approach that identifies mechanisms (M) that explain how or why contexts (C) relate to outcomes (O), describing theories as CMO configurations; definitions in table 1.12 Programme theories were developed through refining initial rough theories from a rapid realist review,6 and analysis of national patient safety incident reports,14 with previously collected and analysed qualitative and quantitative case site data.13 We followed RAMESES reporting and publication standards (online supplemental appendix 2).15
Realist methodology definitions12
Ethical approval
The fieldwork for case study site visits, local patient safety incident report analysis and staff and patient interviews were carried out after ethical approval from the Wales Research Ethics Committee on 23/07/2017 (ref 17/WA/0328).
Summary of previous work
Case site selection
Details of this work were previously reported.13 Briefly, this work began in 2017 when we recruited a purposive sample of ‘case sites’ from a national survey sent to the Clinical Directors of all Type 1 Emergency Departments (consultant-led 24 hours services with full resuscitation facilities) in England and Wales, with follow-up key informant telephone interviews.16 The included sample of 13 case sites, with characteristics listed in online supplemental appendix 3, included:
Three inside-integrated models.
Four inside-parallel models (one was reclassified following the visit).
Three outside-onsite models (outside the ED, on the same hospital site).
Three sites with no GPs.13
Qualitative data collection and analysis
As previously reported, two researchers visited all sites with a GP service (n=10; 2–4 days) and sites with no GPs for 1 day (n=3; 1 day) between January 2018 and April 2019.13 We used initial rough theories generated from the rapid realist review, for example, if GPs maintain their usual approach when working in EDs, investigation use and process times could reduce,6 to develop realist interview guides for theory testing and refining to explore how GPs maintain their usual GP approach in ED settings (online supplemental appendix 4).17 We requested local patient safety incident reports related to the GP-ED model14 and invited patients presenting with marker conditions (online supplemental appendix 5) for interviews to describe their experiences when GPs work in EDs.18 We analysed interview and observation data from multiple sources and applied knowledge from conceptual frameworks and formal theories to refine our initial theories. We then mapped CMO configurations against different GP-ED models to compare across different types of service, to describe model-related mechanisms that contribute to outcomes such as GP approach, use of investigations, process times and patient experience. We presented findings at a national event in December 2019 (n=70 attendees) for stakeholder feedback.
Quantitative data collection and analysis
Patient-level routinely collected data relating to ED attendances and subsequent hospital admissions were obtained from Hospital Episode Statistics Accident and Emergency and Admitted Patient Care datasets (via NHS Digital) for study sites located in England and from Emergency Department Data Set and Patient Episode Database for Wales (via SAIL) for study sites located in Wales as previously reported.13 The attendance-level data were summarised as time series (per site, aggregating data for each study fortnight) for the following variables:
Counts of ED attendances.
Reattendance at same ED.
ED attendance leading to a hospital admission (patient record in APC dataset).
Investigations undertaken during ED attendance.
Treatments delivered during ED attendance.
Average time (minutes) of ED attendance.
Length of stay (days) of hospital admission.
For all variables, we used standard time series analysis methods to assess the nature and extent of linear trends and seasonality in data before and after introduction of a GP-ED model at those sites.13
Current analysis: Mixed-methods synthesis and programme theory development
For this paper, we conducted a mixed-methods synthesis with two separate approaches to develop our programme theories. In approach 1, we further tested and refined our theories developed through qualitative data analysis with outcome data from our quantitative analysis (see table 2).12 For example, for GP-ED models where GPs maintain their usual approach in ED settings, did process times and investigation use reduce? In approach 2, we identified findings for key outcomes from the statistical analysis and cross-checked with the qualitative dataset (see table 3) to ensure we used both inductive and deductive approaches to maximise insights generated.19 20 Findings were used to develop programme theories describing high level individual, department level and wider system influences on the function of the different GP-ED models.12
Mixed-methods analysis approach 1—key insights from qualitative case site data analysis, explored through quantitative data analyses13
Mixed-methods analysis approach 2—findings for key outcomes from the statistical analysis, triangulated with qualitative findings13
Patient and public involvement
Patients and public members (BH, JH) were involved in the study design, as coapplicants in the funded study and are coauthors on this publication.13 They used their experience as NHS patients to contribute to this research, reflecting on interview transcripts and findings. They supported recruitment and involvement of public and patient contributors to the stakeholder events. They were involved in discussing the draft data and have a publication reflecting their experiences in this study.21
Results
Qualitative data collection at case study sites showed considerable variation between and within GP-ED models. Our published papers describe how: integration of GPs with the ED team (and resources, eg, IT systems) varied16; services varied in allocating (or streaming) patients with different presenting complaints to the GP service22; GP role and interprofessional communication varied with some services encouraging a ‘usual GP approach’ with minimal access to acute investigations or opportunity to observe patients, and others expecting GPs to use ED facilities23; some GPs took on roles supervising junior ED clinicians, including allied health professionals13; employment, contractual and governance status of GPs varied, and influenced flexibility in deploying GP and ED team members to times or places of greatest need24; there was concern that highly visible services may attract additional ‘provider-induced’ demand25; and influence patients’ expectations and experiences of using the service.26 Quantitative data (2010–2018) from the case study sites (accessed via NHS Digital and SAIL Databank) were of variable quality,13 but overall there was no GP-ED model found to be superior. GP-ED service models did not reduce attendances and waiting times (which increased across time regardless of GP-ED model (or none) used), with a mixed picture for hospital admissions and length of hospital stay.13
Mixed-methods synthesis
Tables 2 and 3 outline our mixed-methods findings across the qualitative and quantitative analyses. Due to poor quantitative data quality, links between context and mechanisms from qualitative data where were often difficult to refine with quantitative outcome data. Analysis at ‘outside-onsite’ sites, where GPs generally maintained a GP role, were limited by poor quantitative data quality. Our qualitative data offered some explanation for some patterns observed in the routine data analysis which included wider system factors such as a frailty unit potentially influencing waiting times at one case site (outside-3) and a new housing estate contributing to increasing hospital admissions at another (inside-integrated-1).
Programme theories
We focused our programme theories on highlighting how the differences in actions (largely at an individual clinician level, both within and between the GP-ED models in different settings) influenced how the three GP-ED models worked. These theories have generally been refined from the qualitative data due to the quality of routine (quantitative) data available for analysis.13 Figures 2–4 present high-level summaries of key features for each programme theory of the GP-ED models described: inside-integrated, inside-parallel, outside-onsite. In each, the patient journey is depicted with grey arrows from being selected to see a GP, receiving care (±investigation and/or treatment) and resulting in patient experience and flow (outcomes) through the department. Context is labelled in blue, mechanisms within the green arrows and outcomes in red circles.13
Programme theory to describe contexts, mechanisms and outcomes in inside-integrated general practitioner services in or alongside EDs (GP-EDs). Models where the streaming process is less influential, GPs may take on an ED clinician role and the ‘invisibility’ of the GP service limits patient expectation.Reproduced with permission from Davies et al. NIHR Journals 13
Programme theory to describe contexts, mechanisms and outcomes in outside-onsite general practitioner services in or alongside EDs (GP-EDs). Models with greater complexity ensuring the right patients saw the GP and high visibility of the service having greater impact on patient expectation.Reproduced with permission from Davies et al. NIHR Journals 13
Programme theory to describe contexts, mechanisms and outcomes in inside-parallel general practitioner services in or alongside EDs (GP-EDs). Models with the most variation in service set up and lack of clarity on GP role.Reproduced with permission from Davies et al. NIHR Journals 13
Inside-integrated sites
These services aimed to address general ED demand (rather than primary care demand) and were often funded from within the same NHS Trust as the ED. The GPs worked in the same physical space as ED staff with shared IT systems, full access to acute investigations and were valued as members of the ED team. Compared with the other models, the streaming process was less influential with effective interprofessional communication and good understanding of each other’s roles facilitating the right patient to be seen by the GP. GPs also self-selected patients that fitted their own clinical remit, which was often wider than primary care type problems. GP experience, confidence, preferences and attitudes to risk, along with ED leadership support, were described to influence whether GPs treated patients with a GP approach or adopted an emergency medicine approach. Unlike in other models, inside-integrated models allowed for GPs to take on supervision of junior ED doctors which was considered a particular asset at some sites. The ‘invisibility’ of the GP (from the ED) service meant its impact on patient expectations was also more limited (figure 2).13
Outside-onsite sites
At outside-onsite sites, quantitative data quality was poorer, with one site excluded from analysis due to poor data quality. Outside-onsite models typically aimed to deal specifically with primary care demand and had a clearly defined identity in a separate physical location, usually with incompatible IT systems to the ED. As a result, there were usually agreed criteria and pathways for accessing the service and GPs’ remit generally remained similar to that of community primary care (without access to ED investigations). Qualitative data highlighted the process of ensuring the right patient could see the GP was more complex with potential for patients to get ‘lost’ between systems or inappropriate patients being referred to the GP service. An experienced streaming nurse with clear guidance and good communication with the ED team was described to mitigate this. The high visibility and accessibility of these services were likely to have a greater impact on patients’ expectations and experiences. At these sites GPs sometimes took on a supervisory role for a wider primary care team including, for example, nurse practitioners and paramedics (figure 3).13
Inside-parallel models
Inside-parallel models showed the most variation in the way the services were set up. Some were more similar to inside-integrated models, others more distinct as inside-parallel models. The level of primary care demand (and aim of addressing this) varied at these sites and perceptions differed about how integrated the GP service and the ED were intended to be at different sites. Some services had shared IT systems, in others the IT services were incompatible. While streaming guidelines were often formalised, there was sometimes flexibility for GPs to take on a wider (emergency medicine) role, with variable access to ED level investigations, and to allow for personal characteristics of individual clinicians, including preferred ways of working, experience, confidence levels and medico-legal concerns to shape their case-mix. Techniques described by GPs as ways to mitigate risks when using the ‘GP approach’ to treat higher risk ED patients included longer consultations, different thresholds for investigation or admission, and developing supplementary emergency medicine skills. The lack of clarity around the breadth of the GP role could be a particular concern for sites with this model, although this could be overcome in settings with strong clinical ED leadership and effective interprofessional communication to facilitate understanding of each other’s roles. The visibility and accessibility of the GP service was described to influence patient expectations and satisfaction with the service (figure 4).13
Discussion
Main findings
Our study shows the complexity in these service models with variation in the scope of the GP role and scale of the service, both within and between GP-ED models, and other external influences, all of which present challenges in evaluation. Quantitative data were of variable quality (poorer quality for outside models) but overall there was no reduction in attendances and waiting times, and a mixed picture for hospital admissions and length of hospital stay. We present programme theories to describe how three previously described GP-ED service models operate: inside the ED, integrated with patient flow largely addressing general ED demand with GPs requiring a wider skillset than usual primary care and the streaming process less influential; outside the ED addressing primary care demand in a separate service with an experienced streaming nurse facilitating the ‘right patients’ are streamed to the GP; or within the ED as a parallel service with most variability in the level of integration (including IT services, investigation access) and GP role (primary care vs emergency medicine), facilitated by strong clinical ED leadership and interprofessional communication.8 13
Strengths and limitations
The analysis was strengthened by the 13 case study sites, purposively recruited for theory testing and refinement with different service models in different sized hospitals, geographically spread across England and Wales. Researchers applied a consistent realist approach, with initial rough theories developed from the literature tested through qualitative methods.17 Routinely collected quantitative data were available for analysis at most case study sites.
Limitations include all data being collected prior to the COVID-19 pandemic. Since then, other service models for accessing care, such as ‘call first’ for walk-in patients, have been introduced in EDs for patients with primary care type problems.27 There has also been increasing interest in the roles of the non-medical workforce in the urgent and emergency care system, with evaluation in progress.28 However, evidence to inform how to manage increasing patient demand following the pandemic is still relevant. We did not identify any case sites where GPs screened patients at the front door in a gatekeeper role, although there may be departments operating this service model of which we were unaware.6 The visits, at 3 days were short, limiting data collection; there was especially low recruitment for patients for interview.18 Qualitative data are subject to researcher and participant perceptions. Quantitative data availability, extraction and analysis were delayed due to pandemic constraints.
Context of current literature
Increasing demand on UK EDs is multifactorial including an ageing society with increasing multimorbidity,29 a reduction in the number of community GPs coping with an increasing workload,30 inadequate social care preventing hospital discharge for medically fit patients causing ‘exit block’31 and the impact on the general health and social care workforce following the COVID-19 pandemic and Brexit.32
Internationally, GP-ED service models have been introduced to address increasing ED attendances.6 7 Our findings highlight the challenges in evaluating the effectiveness of these service models when data quality is variable and they are part of complex adaptive socio-technical systems.10 High-level routine data analysis of Hospital Episode Statistics may not be sufficient to understand the nuanced complexity of the individual EDs when aggregated to national level. Our findings integrate a large programme of work including 17 published papers (see online supplemental appendix 1)13 and are consistent with findings by Benger et al,9 that GPs appear less influential in these overall complex processes than patient and wider system factors. The impact of GP-ED models on the community primary care workforce is unknown, as also whether other integrated healthcare and social care service models such as helping discharge medically fit patients back into the community, may have more impact on ED patient flow.31
Recommendations for policy and practice
Commissioners and service providers should recognise the complexity of system, department and individual patient and staff factors, that can influence how GP-ED service models operate.
The aim of service provision (eg, to address primary or emergency demand or staff recruitment needs), the role of the GPs within the service and the most appropriate GP-ED model should be clarified, based on local need.
The existing service model may be modified accordingly, for example, developing (or not) a separate area for primary care provision, using an experienced streaming nurse to ensure the right patients get to the GP, which role (GP or emergency clinician, including access to investigations) the GP is supported to adopt in that ED, encouraging interprofessional communications to improve understanding of the GP role and remit (and levels of GPs’ of experience to provide such ‘more primary care’ or ‘more emergency medicine’ roles), integration of IT systems, governance of employed GP staff; or whether an alternative healthcare service may be considered to meet local demand.
Quantitative data quality needs to be improved across primary and secondary care services to enable future evaluation of similar GP-ED services.
Further research
Further research is needed to determine the impact of GP-ED service models on sustainability of the community primary care workforce. With few distinct effects of GP-ED models noted for our key quantitative outcomes, cost-effectiveness analysis was not warranted. More nuanced resource use implications for the different models in relation to their costs and consequences in different contexts may be valuable to guide future service design.
Conclusion
GP-ED services are complex and influenced by individual, department and wider system influences. Our programme theories describe how the different service models operate in different settings, to inform recommendations about how services could be modified in particular contexts to address local demand, or whether alternative healthcare services should be considered.
Data availability statement
Data are available upon reasonable request. The full NIHR peer-reviewed report is available at Davies F, Edwards M, et al. Evaluation of different models of general practitioners working in or alongside emergency departments: a mixed methods realist evaluation. NIHR Journals Health and Social Care Delivery Research (in press) https://www.journalslibrary.nihr.ac.uk/programmes/hsdr/1514504/%23/.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by Wales Research Ethics Committee on 23/07/2017 (ref 17/WA/0328). Participants gave informed consent to participate in the study before taking part.
Acknowledgments
We would like to thank Nigel Pearson, Charlotte Bonner-Evans, Deborah Fitzsimmons, Shaun Harris, Saiful Islam, Rhys Pockett and Ffion Coomber for their contributions; also staff and patients at case study sites.
Footnotes
Handling editor Carl Marincowitz
X @PRIMECentre
Contributors AC, FD, PA, AC-S, MC, JD, LD, BAE, BH, JH, PH, TCH, AP, ANS, AW, HS and AE made substantial contributions to the conception and design of the work; AC, ME, FD, AW, PA and DP to the acquisition and analysis of data; and AC, ME, FD, DP, PA, AC-S, MC, JD, BAE, BH, JH, PH, TCH, AP, ANS, AW, HS and AE to the interpretation of data; AC led the initial manuscript draft which was reviewed and the final version approved by all coauthors, who agree to be accountable for this work. AC is guarantor of this work.
Funding This study was funded by the National Institute for Health Research (NIHR) HS&DR Project 15/145/04 and will be published in full in the NIHR Journals Library. Further information available at: https://www.journalslibrary.nihr.ac.uk/programmes/hsdr/1514504/%23/. The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care.
Competing interests LD declares his role on the Integrated Care Board for North East and North Cumbria. TH declares his role as NHS Clinical Lead for the implementation of the Emergency Care Data Set.
Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting or dissemination plans of this research. Refer to the Methods section for further details.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
Linked Articles
- In perspective