Article Text
PDF
Abstract
Background Clinicians and managers across specialities are under pressure to review treatment and referral pathways to enable evidence-based practice, improve patient flow and provide a seamless service. This study outlines the processes and outcomes of an action research study conducted to reduce inappropriate attendances and unplanned pressures on Emergency Department (ED) staff in an English hospital during 2006–2008.
Methods Action research, comprising three action/reflection cycles conducted with participants, was used. Data were collected using retrospective patient record review (n=35 200) interviews with staff members (n=28), observation of patient pathways (n=38 patients) and measurement of team climate (n=31) with literature reviews also informing each cycle of data collection.
Results ED attendance and hospital emergency admission data were largely similar to the national picture with regards to time/day of attendance and seasonal variation. However, in the ‘adult majors’ subgroup, mean attendance on a Monday was significantly higher than the rest of the week (p<0.001) and 36% were self-referrals. Observation data revealed that patients were informally assessed by reception staff and directed to majors or minors; this practice was replaced by reinstatement of triage. Patients identified as ‘inappropriate’ were managed inconsistently, irrespective of department workload. ED attendance decreased as the project progressed and the number of attendees resulting in hospital admission rose slightly.
Conclusions Study data suggest that inappropriate attendances decreased; however, data collection exposed gaps in the existing management information systems and inconsistencies in working practices in the ED. Action research can have a practical value besides contributing to knowledge.
- (MeSH terms)
- emergency medicine
- clinical pathway
- delivery of healthcare
- emergency care systems
- emergency departments
- admission avoidance
- wounds
- research
Statistics from Altmetric.com
- (MeSH terms)
- emergency medicine
- clinical pathway
- delivery of healthcare
- emergency care systems
- emergency departments
- admission avoidance
- wounds
- research
Background
Two major policy initiatives have affected the management of emergency departments (EDs) in England in recent years. The ‘18-week journey’1 was a policy that no patient should wait longer than 18 weeks from GP referral to start of hospital treatment. Payment by results2 was a diagnosis-related group (DRG)-like system of hospital payment. These two policies placed conflicting pressures on the management of English EDs, the former providing an indirect incentive for EDs to reduce their workload, but the latter providing hospitals with an incentive to attract more patients. Furthermore, a report entitled ‘Transforming Emergency Care in England’,3 identified principles of patient care that demanded high levels of interprofessional collaboration, requiring a shift in operational and human resource management away from profession-specific sets of knowledge, skills and values towards the task of combining input from different professions3 4 so as to enhance patient care and realise, among others, the policies mentioned above.
This study evaluates a service development funded by the English Department of Health (DH) through the National Workforce Project with the aim of identifying the ways in which implementing the complex of policies mentioned above necessitated revising ED patient pathways and infrastructure at district general hospital level. The project was undertaken in the ED of a medium-sized district general hospital in a rural setting and managed by an NHS Trust, in which the staff reported tensions between professionals working in the ED. Recent studies have used action research to improve specific aspects of the patient pathway, for example continuity of care across psychiatric and emergency services,5 bed management6 and waiting times.7 In similar vein this study reports the use of action research to identify how patient flow in one ED was improved.
Although what follows describes a single ED, the problems of managing patient throughput, in the face of interoccupational tensions, so as to improve patients' access to care and to concentrate resources on the patients who most need them are important and enduring problems for most hospitals in most health systems. The action research methods used to address these problems are also of general interest for hospital management.
Aim and objectives
As often happens in action research, the original aims and objectives of this study were adjusted over its lifetime.
The original aim of the study was to identify and compare the range of training and infrastructure initiatives (including the possible opening of a Clinical Decision Unit (CDU)) available to the hospital to reduce inappropriate attendances and unplanned pressures on junior doctors and hospital staff in ED care pathways. In particular the study aimed to:
Identify cultural, training and infrastructure needs to provide a redefined health system for accessing and use of the ED.
Recommend a range of staff and types of programmes necessary to provide an appropriate skill mix in the ED.
Although these overarching aims remained constant, the nature of the ‘redefined health system’ (aim 1) required by the Trust Board changed as the project developed. Consequently, the type of data required in order to address the aims broadened.
Methods
Given this context, action research8 was the method of choice because it comprises a series of action/reflection cycles (see figure 1 below) conducted with (rather than ‘on’) study participants, with a dual agenda of seeking to influence practice during the study by feeding back to participants and of ‘systematically gathering data to share with a wider audience’.9 Action research commonly involves mixed methods, but, unlike conventional mixed methods approaches, the reflexive nature of data collection does not allow decisions about timing, weight and mixing of data10 11 to be made entirely a priori. Nevertheless, a successful action research report is characterised by a ‘shock of recognition’12 by those whose actions it describes.
Reflection (R)—action (A) spirals of data collection. CDU, Clinical Decision Unit; ED, Emergency Department.
A range of data collection methods was used (see box 1).
Research methods
Phase 1
Literature review—clinical decision units
Analysis of Emergency Department attendees
Analysis of care pathway documentation from other Trusts
Observation of patient care pathways
Interviews
Analysis of all hospital admissions
Phase 2
Literature review: telemedicine
Review of all short stay emergency admissions
Literature review: triage and ‘see and treat’
Phase 3
Measure team climate (Team Climate Inventory15)
Process evaluation (Project Board members)
A characteristic of action research is the use of a single site and multiple data sources. Methodological limitations of action research have been identified as: failure to establish validity of the findings, danger of exploiting participants, difficulties of generalising from the results and the challenge of developing theory.13 Steps necessary to avoid these limitations are identified in box 2.
Processes for establishing rigour in action research
Adhere to conventions for each method (eg, respondent validity for interviews, transparent and replicable data analysis processes).
Be alert to researcher-induced bias or (alternatively)‘going native’; seek regular critique from ‘critical friends’ to ensure any assumptions are challenged.
Agree ethical ways of working to ensure participants are not exploited.
Seek regular feedback/confirmation from participants.
Interpret local evidence in the light of policy and previous studies.
Document a decision trail to identify how participants are selected and research questions evolve.
The action-reflection spirals for the three phases are demonstrated in figure 1.
Data analysis
Existing ED attendance and admission data required significant recoding before analysis. For the quantitative data, demographic, descriptive and inferential statistics were produced as applicable, including χ2 for measures of association and independent sample t tests (t) for sample means. Approximate 95% CIs are reported where applicable. All tests for statistical significance were two-tailed with the level of significance at p≤0.05.
Observation, interview and documentary data were content-analysed using framework analysis,14 a process comprising five stages: familiarisation, defining a thematic framework, indexing, charting and mapping/interpretation. Framework analysis is particularly useful for applied or policy-related qualitative data, allowing them to be used to test prior hypotheses.14 The thematic framework was derived initially from the project brief and literature review, and refined using themes emerging from the data.
Throughout the study, findings were fed back to the participants at monthly project meetings. Mixed methods research allows data to be integrated at data collection, data interpretation or discussion stage.10 In this study data were ‘mixed’ at the interpretation stage, in order to provide interim recommendations to the project team and inform the next cycle of action and reflection. The study was conducted in three phases, each corresponding to one cycle of action and reflection.
Phase 1
A literature review was undertaken to identify factors affecting the functioning, efficiency and costs of EDs and CDUs. The search strategy used the terms: ‘CDU’, ‘A&E’, ‘Assessment Unit’, ‘Observation Ward’, ‘Admission Ward’, ‘Accident and Emergency Services’ to search Medline, CINAHL and British Nursing Index databases. The search was limited to ‘human’, English-language texts and peer-reviewed publication during 1990–2007.
A detailed case study of the ED was also undertaken including: observation of the care pathway actually followed by 38 patients, interviews with 28 members of staff and secondary analysis of ED attendance data for 1 year (n=35 276 attendees). The interviewees represented 25% of the department population, purposively selected to include all grades of medical, nursing, managerial and ancillary staff. The focus of this data collection, conducted over a period of 6 months, was to identify likely benefits and activity levels for a CDU.
Before these Phase 1 data had been analysed, the Trust Board decided that the funds earmarked for a CDU were no longer available. Consequently, other options for the management of patient pathways were explored in Phase 2.
Phase 2
Analysis of the data in Phase 1 revealed some coordination difficulties between GPs and ED clinicians, for example inappropriate referrals and unavailability of senior ED clinicians for consultation. Phase 2 data collection therefore focused on this issue, in addition to ED staff requests to identify:
What clinical interventions and investigations were undertaken for patients admitted from ED with a hospital stay of <72 h?
What potential telemedicine holds for reducing ED attendance and subsequent hospital admission?
All emergency hospital admissions staying <48 h (<72 h for those admitted on a Friday) were examined and literature related to telemedicine was reviewed.
In the analysis of Phase 1 data some potentially unsafe patient triage practices were identified. In order to ascertain how much importance should be attached to these concerns a literature review was conducted to examine ‘see and treat’ and triage models.
Phase 3
Analysis of stakeholder interviews revealed differing perspectives on the extent of teamwork so teamworking was examined using the Team Climate Inventory.15
Project monitoring data was collected by the project board during the whole 12 months of the project, including activity and occupancy levels at community hospitals and patient flows between hospital wards and community hospitals.
Ethics approval was obtained from the researchers' university (University of Plymouth); NHS research ethics bodies did not require the project to go through the full approval process.
Results
Phase 1 data
The literature review identified the ingredients of a successful CDU with regards to leadership, working relationships, policies and resource availability. The executive summary of the literature review is provided as a supplemental file on the journal website.
ED attendance (n=35 276) and hospital emergency admission data (n=17 011) were largely similar to the national picture with regards to time/day of attendance and seasonal variation. ED attendance rates were higher in summer across all groups (adults, paediatrics; majors and minors – all significant at p<0.01). Attendance rates at weekends and Mondays were significantly higher than for the remainder of the week (χ2=258.87; p≤0.001); of note, weekend attendees tended to be younger (t=−5.51; p≤0.001; 95% CI of difference −2.22 (−1.63 to 1.06). In the ‘adult majors’, subgroup mean attendance on a Monday was significantly higher than for the remainder of the week (χ2=1489.37; p≤0.001) and 36% were self-referrals.
Analysis of patient postcodes demonstrated that during the 12-month period at least one patient attended the ED from most postcode areas across Britain. Local postcodes included almost 20% from a neighbouring area with a Minor Injuries Unit.
Patients arrived at the ED from a number of sources, predominantly self-referrals (57%) and the Ambulance Service (18%). There was some evidence from the observational data that many of the self-referrals could have been dealt with by alternative resources, for example NHS Direct or in General Practice. However, field notes also revealed that patients identified as ‘inappropriate’ were managed inconsistently, with some sent away to see their GP and others treated. This was not necessarily related to ED workload. Some patients themselves justified their attendance at ED by saying that they could not obtain a GP appointment.
For all patients the average length of stay in the department was just under 2 h (1.59 min) (range 0–27 h). Children spent significantly less time in the department than adults, 1.23 h compared to 2.10 h (t=−46.62; p≤0.001; (95% CI of difference 0.46 (0.44 to 0.48)).
Content analysis of the observation and interview data identified factors that influenced the patient pathway in ED, including those occasionally causing blockages. These data were encapsulated in three themes: resource use, internal staff relationships and external interfaces. Using framework analysis, charts were developed to summarise experiences and data for each theme. Illustrative data excerpts from the chart for the ‘resource use’ theme are presented in table 1.
Chart illustrating ‘resource use’ theme in the framework analysis
The three data excerpts from administrative staff, combined with observation data, led the research team to review the evidence about triage and make recommendations about triage practices in an interim report. The literature review identified ‘see and treat’ as the optimal process to reduce waiting times while achieving high-quality patient experience. However, implementation processes must be suited to the local context and triage should be used when staffing does not allow operation of ‘see and treat’. As a result of evidence presented in the interim report, a triage system was reintroduced and additional Nurse Practitioners were employed.
Phase 2 data
A total of 177 short stay emergency admissions (<48–72 h) were reviewed. Sixty-three per cent (n=86) of adult patients were admitted via the ED and 26.2% (n=36) via the Emergency Medical Unit (EMU). Fifty-seven per cent (78/137) of patients from all admission sources were initially deemed suitable for admission to an assessment unit; a further 15 were found to be unsuitable due to comorbidities. Notes were missing for 10 patients. In 48 patient records (35%), there were discrepancies with information recorded on the patient administration system (PAS) form or in the medical notes.
Case note review also allowed compilation of case studies. The following three cases were used to provoke discussion about the appropriateness of hospital admission and to identify training needs (aim 1):
Case 1 – A 91-year-old patient was admitted from a nursing home at 0:55. The patient was transferred from ED to EMU and diagnosed with congestive cardiac failure (CCF); the patient died soon after.
Case 2 – A 77-year-old was referred by MacMillan nurses to EMU and received treatment that could have been completed in primary care.
Case 3 – An 81-year-old individual with multiple pathologies was transferred from a nursing home via ambulance to EMU and also received treatment that could have been provided in primary care.
The telemedicine literature review emphasised the importance of adequate resource and infrastructure support, the logistics and workload implications of concurrently managing a real and remote caseload and the need for optimal communication within and across teams.
Phase 3 data
Observation and interview data suggested a lack of communication between some team members:‘When the pressure's on it can get dysfunctional’. (doctor/3)
‘There are personalities that aren't team players’. (nurse/11)
‘I think what often gets in the way is politics’. (doctor/9)
There was also a culture in which individual preferences could dominate decision-making, for example in the type of investigations ordered. Team Climate Inventory data from 31 staff members across professional groups identified a number of strengths and weaknesses in the team, for example, the team shared information to a limited extent and there was a limited degree of trust within the team. Team members were clear about some of, but not all, the department objectives and were committed to providing a high standard of performance but rarely monitored colleagues' performance.
Concerns about collaborative working with external organisations were expressed in process evaluation data; when asked about barriers to the project success, members of the project board identified: ‘different agenda at different organisations and on an individual level – no agreed vision or direction’ and ‘slow engagement from the PCT [Primary Care Trust] and poor engagement from the ambulance trust’.
However, the informants described difficulties with ‘trying to raise the profile of non-elective work in a climate of payment by results’ and criticised higher level organisational barriers:From the PCT and primary care perspective there was the hope of radically reforming the whole ethos of A+E and having a primary care stream together with rapid consultant opinions, decisions and discharges. The constant meddling by Department of Health and re/disorganisation of PCTs has led to paralysis of intended action and a missed opportunity for genuine reform. (Senior Manager/A)
The informants generally agreed that communication and collaboration improved as the project progressed, although there was a sense of ‘too little too late’. However, they agreed to continue with some of the unfinished actions through a multi-agency directors' group.
Project monitoring data
Project monitoring data were presented by stakeholders at the monthly Project Board meetings with the intent of informing project development. Many of these data were not available until the end of the project, preventing the action research from making use of them. Hence, some of the data trends highlighted below were evident as the project progressed, whereas others were not. Overall, ED attendances were lower than the previous year (33 300 vs 35 200), whereas the number of attendees resulting in hospital admission rose slightly (9900 vs 9200).
During the year there was an increase in breaches of the 4 hour ED waiting target (2.67% of all attendees) with a sharp rise from September 2007. Interviews with stakeholders suggested that the focus on the ‘18-week journey’ led to more patients waiting for admission in ED. There were no serious incidents during the project; however, near misses and untoward incidents rose during the project, with a sharp rise in the final 2 months. This coincided with the request to report all patients who were managed on a trolley in the corridor (rather than in a cubicle) as a clinical incident.
Discussion
Examination of the current patient pathway through the ED at the beginning of the project suggested key decision points and issues that affect the efficiency of the care pathway. They include: inappropriate referrals by primary care professionals, and an ill-informed public who attend the ED out of lack of understanding or frustration at the lack of alternative resources, especially out of hours. Communication networks with primary care warrant attention to reduce inappropriate admissions and improve discharge planning. The informants in this study also claimed that lack of engagement with the Ambulance Service and PCTs were barriers to the success of the project. Patient pathways are central to current debates regarding optimising the ED patient experience; the Map of Medicine is one solution proposed to facilitate working across boundaries, providing an early filtering system between primary and secondary care.16 Within the ED, the handover studies currently under way at the Australian Commission for Safety and Quality in Health Care are expected to generate findings relevant to this problem.17 However, the identification of inadequacies with the initial pathway into majors or minors also allowed this situation to be rectified during the lifetime of the project, with employment of additional Nurse Practitioners and changes in practice in the reception area. Relationship issues with the rest of the hospital may be resolved by more internal rotation of staff to improve insight, understanding and empathy.
Despite the lack of engagement with PCTs and Ambulance Trust during the project, the rising percentage of ED attendees that resulted in hospital admission by the end of the project (project monitoring data) might indicate more appropriate use of the ED, although the number of admitted patients who could have been managed outside of hospital, for example via chronic disease case management,16 was not estimated. However, this also emphasises the importance of ED attendance data for hospital bed and resource management. Although considerable time had been dedicated to the collection of these data at the study site, significant recoding was necessary to summarise and make sense of the data. In the authors' opinion, it is remarkable that such data were not routinely easily available to ED managers. The accuracy of national performance monitoring data for ED has been questioned previously16 18; review of data from a single site for this study highlights considerable problems with data management that may also be present in other hospitals.
The range of data analysed during the study enabled recommendations to be made about the type of service provision (including short stay decision unit) that would provide a sensible ‘fit’ for the patient profile. These data also highlight a number of aspects to be explored in other EDs (see box 3).
Aspects to explore in other Emergency Departments
Pathways for patients admitted for short stay.
Completeness of patient records.
Analysis of attendees by postcode.
Modelling of patient pathways to identify ‘blocks’.
Usefulness and accessibility of patient pathway data.
Multiprofessional teamworking within the ED and hospital.
Regular meetings between PCT/Ambulance Service/ED leads. to review patient pathways and impact of policy changes.
Conclusions
In order to provide a redefined system for accessing and using the ED, the present data suggest that it is essential to have useable management data and agreed demarcation between roles of primary healthcare and the ED. Some of the frustrations reported and observed in this study suggest that the accretion of multiple policy initiatives can be counterproductive when these policies either conflict, or set up conflicting imperatives on ED staff. In particular, the pressures on the ED resulting from a non-ED policy initiative (the 18-week journey) were in direct conflict with the goal of improving patient flow through the ED. Nevertheless, this study provides another piece of evidence suggesting the value of clear guidance about the appropriateness of ED referrals and caseload, and the value of routinely available management information about caseloads and their sources.
The action research data enabled recommendations to be made regarding changes in practice and staffing necessary to improve emergency care pathways. Although these recommendations were relevant to the local situation, findings from action research studies can also provide indicators for other settings provided sufficient attention is paid to establishing rigour. The process of action research provided a framework in which staff from different occupations could collaborate on (so to speak) neutral ground at problem-solving. The attempts to collect data to identify problems and then monitor interventions that the staff as action researchers had devised also exposed gaps in the existing management information systems, and inconsistencies in working practices in the ED. In these respects, action research can have a practical value besides contributing to knowledge.
Footnotes
Funding Department of Health, New Ways of Working programme.
Competing interests None.
Ethics approval This study was conducted with the approval of the University of Plymouth.
Provenance and peer review Not commissioned; externally peer reviewed.
Linked Articles
- Primary survey