What evidence is there on the effectiveness of different models of delivering urgent care? A rapid review

Article history

The research reported here is the product of an HS&DR Evidence Synthesis Centre, contracted to provide rapid evidence syntheses on issues of relevance to the health service, and to inform future HS&DR calls for new research around identified gaps in evidence. Other reviews by the Evidence Synthesis Centres are also available in the HS&DR journal.

The research reported in this issue of the journal was funded by the HS&DR programme as project number 13/05/12. The contractual start date was in October 2014. The final report began editorial review in April 2015 and was accepted for publication in June 2015. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HS&DR 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.

Declared competing interests of authors

none

Copyright statement

© Queen’s Printer and Controller of HMSO 2015. This work was produced by Turner et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Hypotheses tested in the review (research questions)

The NHS England review6 has set the agenda for urgent care, with recommendations on how the urgent care system and the services within it need to change. We have used the key themes identified in this review as the framework for this rapid evidence review to provide both focus and context for evidence appraisal and the identification of evidence gaps, which will be of direct relevance to future developments. The five key themes identified in the NHS England review are:

1. providing better support for people to self-care

2. helping people with urgent care needs to get the right advice in the right place, at the first contact

3. providing highly responsive urgent care services outside hospital so that people no longer choose to queue in accident and emergency (A&E) departments

4. ensuring that those people with more serious or life-threatening emergency care needs receive treatment in centres with the right facilities and expertise to maximise chances of survival and a good recovery

5. connecting all urgent and emergency care services, so the overall system becomes more than the sum of its parts.

The first theme, focused on providing better support for people to self-care, encompasses the much broader areas of health care related to reducing the need for urgent care. This theme warrants a separate review, as it involves complex issues such as management of long-term conditions, health promotion and injury prevention. As it targets an alternative health-care vision outside of urgent care; the potential scope was considered too broad and diffuse to be included within the constraints of this review. We have therefore excluded this theme and concentrated on the other four themes directly related to delivery of urgent care.

Within each of these four key themes the NHS England review sets out more specific proposals for service change and delivery, and these will form the focus of the primary scope for individual elements of this review. We have also added an additional underpinning theme, which was not identified as a separate issue by the NHS England review. In order to develop services that are responsive to the needs of the population using them, it is essential to understand the characteristics and drivers that underpin demand for services and the choices people make about how they use those services. Without this it is difficult to ensure alignment between service development and patient need. We have therefore included within our review a brief overview of a fifth theme focused on patterns and characteristics of the demand for urgent care (including change over time), and the factors that influence decisions about when and how to access urgent care.

Although these key themes provide focus, each one still potentially includes a range of issues. To keep the review process manageable within the time and resources available we have therefore restricted the research questions for some themes to a particular service area highlighted as of particular importance in the NHS England review.

Research questions

The research questions examine the evidence relating to the following:

1. To what extent does evidence on existing and proposed approaches to the delivery of urgent care support the development of four key themes in the NHS England review of urgent care?

   1. Helping people to get the right advice in the right place, first time. This theme could potentially cover a range of services in terms of what care is eventually accessed. However, the process of providing advice and directing people to the right service when they first try to access care is firmly grounded in the NHS England review as the NHS 111 telephone service. This service is seen as the gateway to directing requests for emergency care to the right service. We have therefore focused on telephone-based access services in this review.

   2. Providing highly responsive urgent care services outside hospital. This theme also potentially includes a range of community-based services; however, it was beyond the scope of this review to search and synthesise all of the potential literature about community-based urgent care. The 2013 NHS England review and related action plan make a clear statement that the ambulance service is considered a key provider in achieving this objective. We have therefore focused on the evidence for developing the ambulance service to manage more people in the community setting in this review.

   3. Ensuring that people with serious or life-threatening emergency care needs receive treatment in appropriately staffed and resourced facilities. This theme is concerned with the provision of ED care, including both major regional facilities and local EDs. There is already a substantial evidence base about the impact of providing regionalised services (e.g. for stroke, heart attack), so there is no value in repeating this here. Furthermore, service pressure is greatest in general EDs (and major regional facilities also function as ‘local’ EDs). We have therefore focused this review on the evidence about different models and processes for delivering ED care to keep the review relevant to current NHS challenges.

   4. Connecting urgent and emergency care services. The NHS England review sets out a clear view that the way to achieve this objective is through the development of urgent care networks to develop and manage local urgent care systems. We have focused this element of the review specifically on evidence about models of urgent care networks.

2. What evidence is there on characteristics of demand for urgent care, and why and how people access urgent care, that may help future service planning?

We have conducted and reported a rapid review for each of these five themes. For each review we have considered two additional questions:

1. What is the quality of that evidence?

2. What are the main/significant evidence gaps?

Overview of rapid review methods

This was a rapid framework-based evidence synthesis which needed to be completed within the relatively short time frame of 6 months to produce a review that met the HSDR programme’s needs. We have used rapid review methods to ensure the efficient identification and synthesis of the most relevant evidence. The multiple dimensions covered by the review questions posed a considerable challenge to the rapid review process. This challenge was further complicated by the fact that emergency and urgent care does not involve discrete populations or conditions, but encompasses whole populations and a heterogeneous mix of conditions and acuity, and care is delivered by a range of services. As a consequence there was a potentially huge pool of related literature.

Given the large scope and time and resource constraints we have not taken a standard approach to this review. Our aim was to provide a broad overview of the existing evidence base for each theme and any associated limitations. We have therefore applied the following criteria to structure the review process:

• We have concentrated on identifying and synthesising the key evidence using a focused, policy-relevant framework to keep the task relevant and manageable. Framework-based synthesis has been identified as an efficient method for synthesising evidence to inform policy within relatively tight time constraints. 7

• The review did not attempt to identify all relevant evidence or to search exhaustively for all evidence that meets the inclusion criteria; instead we have used a structured searching approach to identify the key evidence.

• The data extraction and quality assessment have focused on the most critical information for evidence synthesis rather than aiming to exhaustively extract and critique all the available information in individual papers.

• We have not appraised the evidence in terms of how future services should be provided, nor made recommendations about service configuration.

Framework

As the focus of this review is on models of care, that is service and system delivery, we have not searched for, or considered, evidence related to specific clinical interventions for specific conditions. We have also only included primarily evaluative research of actual interventions (although the definition of intervention can be broad and encompass changes to organisation, changing professional roles, new services, etc.) in order to provide an overview of what may or may not work in practice. For this reason we have purposely excluded the more theoretical literature, for example relating to organisational behaviour, professional development and clinical competence, work psychology, patient decision-making and behaviour. Where additional review in these related areas is of value, these have been highlighted in the individual review chapters as specific areas for further in-depth review and analysis.

For each of the four themes related to the NHS England review we have considered three main areas:

1. evidence on efficiency and clinical effectiveness and cost-effectiveness of service delivery for any identified operating models, including individual service and whole-system perspectives

2. evidence on associated workforce issues where this is primary research evaluating the effectiveness of changing or developing new professional roles in the delivery of urgent care and workforce planning

3. evidence on any related patient experience outcomes.

Urgent care provision in England is a rapidly changing environment. The NHS England review has prompted a range of work programmes8 and professional bodies, for example the Royal College of Emergency Medicine,9 to regularly publish recommendations about delivery of services. Where relevant, we have used key policy documents published before October 2014 specifically related to the implementation of the NHS England reform of urgent care to develop the review framework.

The additional fifth theme on understanding demand and use of services has focused on primary research that:

• reports analysis of not only level of demand but also the characteristics of that demand (e.g. age profiles, condition profiles, whole-system demand for different types of service)

• reports patient-derived explanatory research concerned with decisions to access urgent care.

This framework has provided a clear structure with which to guide the review while retaining the flexibility that has allowed the development of each individual theme in terms of defining the scope of the search strategies, defining inclusion and exclusion criteria to specify what types of studies will be included in each theme and evidence synthesis.

Search methods

A variety of search methods were undertaken in order to identify relevant evidence for each of the review questions and themes in a timely fashion. We have used a number of different search strategies for this review while using a general structure of combining relevant terms, such as:

• Population

  • Users of the range of services within the emergency and urgent care system (ambulance services, ED, other urgent care facilities, telephone access services, primary care-based urgent care services).

• Outcomes

  • Processes – ED attendances, emergency admissions, ambulance calls, dispatches or transports, demand, appropriateness of level of care, adverse events.

  • Patient outcomes – patient experience and satisfaction, decision-making, cost consequences and cost-effectiveness.

Searches were conducted in two stages:

1. Stage 1 – general search on MEDLINE.

2. Stage 2 – targeted database searches around telephone triage, ambulance, demand, organisation of EDs and networks. To increase efficiency, where appropriate, we have utilised existing search strategies from related research we have conducted within the School of Health and Related Research (ScHARR) or from existing related systematic reviews.

Database search strategies

Review process

Introduction

The main focus of this rapid review is assessment of the evidence relevant to the NHS England review of urgent and emergency care. However, to provide context we have presented a short overview of the current state of knowledge of the characteristics and drivers that underpin demand for services. This may be of use in terms of future planning, the ability to develop services that are responsive to the needs of the population using them and ensuring alignment between service development and patient need, and so is of relevance to the later review about urgent care networks.

Increases in demand for ED care are well documented. In England, demand for ED care doubled from an estimated 6.8 million first attendances at type 1 (24-hour, consultant-led service) EDs to 13.6 million over the 40 years from 1966/7 to 2006/7 – equivalent to an increase from 138 to 267 first attendances per 1000 people per year. Since 2006/07 attendances at type 1 EDs have further increased to 14.3 million in 2012/13 and at the same time there has been a rapid increase in the use of minor urgent care services [type 3 – not 24 hours, may be run by nurses or general practitioners (GPs), limited facilities such as radiography], with attendances increasing by 46% from 4.7 million in 2006/7 to 6.9 million in 2012/13. 11 Similarly, demand for 999 ambulance services has also steadily increased from around 4 million calls per year in 1994/5 to 9 million in 2012/13 (an increase of 160%), with utilisation rising from 78 to 171 calls per 1000 people per year over the same time period. 1 People with health problems also access urgent care via NHS 111 and primary care, but NHS 111 is a relatively new service and there is a lack of national data on urgent care contacts with primary care, so it is difficult to assess whole-system demand for emergency and urgent care in England.

More detailed analysis of UK trends in demand is available in reports from the NHS Confederation1 and Nuffield Trust. 12 In this report we have examined the empirical evidence that may help explain why demand is changing.

Methods

The main inclusion and exclusion criteria, search strategies and review process have been described in Chapter 2. We have conducted previous reviews in this area and are aware of the relative scarcity of related evidence. In addition, this topic area is not concerned with interventions or service delivery and hence the effects on processes or patient outcomes. We have therefore included literature reviews that were not systematic reviews but which have described a structured search strategy. Search dates were from 1995 to 2014. For this review, specific additional inclusion and exclusion criteria were applied to studies investigating the following:

Review process

Studies were identified from updated and expanded database searching using a search strategy from one of our own previous reviews in this area1 and a review of the evidence on callers to the 999 service with primary care problems from an NIHR doctoral research fellowship currently awaiting publication in the NIHR Journals Library (Dr M Booker, University of Bristol, 2015, personal communication). As the aim of this part of the review was to describe an overview of current evidence to provide context for the more detailed rapid reviews on service delivery, we limited the studies included in three ways:

1. We previously conducted a scoping review of potential reasons for increases in ambulance demand and, as this is already in the public domain and available for reference, we have not considered papers included in this review. 13

2. We did not conduct a double 10% random sift of the results of the database searches. These were sifted by one reviewer (JT) and supplemented by potential inclusions identified in the 20% random sample from the general search, also sifted by the same reviewer.

3. Data extraction of individual papers meeting the inclusion criteria was only conducted for papers not included in relevant review papers identified in the searches.

The results of the review sifting process are given in Figure 2.

FIGURE 2.

A Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram for emergency and urgent care demand searches.

Results

We identified four relevant systematic or rapid reviews, and an additional 39 primary studies not included in the systematic reviews. Of these 39, eight related to demand and 31 were patient-based studies exploring reasons and choice in terms of accessing urgent care. The characteristics and findings of the included reviews and primary studies are summarised in Tables 1–3.

Summary of findings

This review has been conducted to provide a brief overview and context for the subsequent more detailed reviews. We have therefore only identified key themes that have emerged from the available evidence.

Conclusions

Despite serious concerns about rising demand for emergency and urgent care and the impact this has on health services, there is remarkably little empirical evidence that can fully explain why this trend in behaviour has occurred. The evidence included here has highlighted a range of factors that may influence demand, but much of the research has focused on either individual services or populations, such as the elderly. Most of the evidence presented here has come from Australia and there were no UK-based studies. However, there is scope to replicate some of the Australian studies using UK data.

The four review articles helpfully brought together a broad literature of discrete studies on individual factors that may influence demand, and there is a substantial patient-focused literature that has examined the reasons why people choose to use emergency and urgent care and how they decide which service to access. The evidence gaps and challenges identified from the existing evidence concern the need to:

• Examine demand from a whole-system rather than single-service perspective and gain better understanding of the relative proportions of demand for different parts of the system and the characteristics of patients within each sector. This more detailed understanding would allow more accurate assumptions to be made about changing care delivery, for example what is a realistic proportion of patients who could be managed out of hospital.

• Build on the existing knowledge about the range of factors that may be influencing demand and develop more sophisticated research studies that can comprehensively map these issues into a coherent model. The conceptual model describe in the review by He et al. 15 is a useful starting point.

• Explore the relationships and associations between individual influences and demand so that the relative contribution each one makes can be assigned to a map or model. This would then support the development of service design and planning to meet the needs of local populations19 and the development of predictive models that can be used to estimate likely changes in population needs for emergency and urgent health care. 22

• All of these will require the development of robust, system-wide information systems to support these complex functions.

Introduction

A key principle of the NHS England urgent and emergency care review is provision of a telephone-based service that can act as a gatekeeper and direct people with urgent care needs so that they can get the right advice, in the right place, at the first point of contact. 6 NHS 111 is the cornerstone to this function and already provides a free, national, 24-hour, telephone-based service to triage, assess and signpost requests for urgent health care. However, the introduction and implementation of NHS 111 has not been without problems, and the NHS England review recognises the need to substantially enhance this service in order to deliver the intended benefits within the wider urgent and emergency care system. We have reviewed the broader evidence on use of telephone-based triage and advice services for access to and management of urgent care in order to summarise the current state of knowledge and identify evidence gaps that may need to be addressed to enable further development of this key service.

Methods

The main inclusion and exclusion criteria, search strategies and review process have been described in Chapter 2. Search dates were from 1 January 1995 to 11 November 2014. For this review specific additional inclusion and exclusion criteria were as follows.

Review process

Studies were identified from database searching and through expert knowledge of the field of interest. In addition to the studies identified through database searching, the study team were aware of 10 further studies, which were assessed for inclusion in this review. The results of the review sifting process are given in Figure 3. A randomly generated 10% sample of studies from the database searches (n = 124) was double sifted by another member of the study team. The reviewers initially agreed on inclusion or exclusion for 88% of studies and agreed to exclude the remaining 10 studies after discussion. A total of 87 papers were included in the review, representing 83 individual studies.

FIGURE 3.

A PRISMA flow diagram for telephone triage and advice services searches.

Systematic reviews included

Eighty-seven papers representing 83 studies were included in this review and of these eight57–65 were systematic reviews. The results from the eight reviews (10 papers57–65) are shown in Table 4. Many of the references included in the reviews overlapped. This overlap is shown in Table 5. Formal data extraction was not undertaken for the 33 papers66–98 meeting the inclusion criteria and included in systematic reviews, as this would have been conducted during the review process. Table 6 lists these 33 papers (32 studies66–98) and the systematic reviews in which they were included. A total of 43 studies (44 papers99–142) did not feature in any of the included systematic reviews and underwent full data extraction (Table 7).

Summary of findings

A summary of the main characteristics of all included studies, excluding systematic reviews, is given in Table 8.

Staff type comparisons

Seven studies (nine papers) reported on telephone triage performed by different staff types. 72,92–94,99,117–119,125 There was one randomised controlled trial (RCT),93 one controlled trial,92,99 three prospective observational studies117–119,125 and one retrospective study. 72

Two studies (three papers) compared the appropriateness of nurse telephone referrals with other referral sources. 117–119 Both studies found that GP or physician ED referrals were more appropriate than either nurse telephone-referred or self-referred patients when measuring appropriateness using clinical criteria.

Two studies directly compared nurse and doctor telephone triage. 72,93,94 One study reported that nurses were more likely to recommend higher-intensity advice than doctors, but that when the appropriateness of this advice was reviewed by experts the nurse advice was more likely to be appropriate. 72 The other study found that callers were significantly more likely to be satisfied and have higher compliance rates if their call was dealt with by a doctor rather than a nurse, whereas repeat calls for advice were significantly higher in the nurse group at both the 4- and 72-hour time intervals. 93,94 A different study looked at introducing second-level physician triage in a nurse telephone system. 125 Adopting this approach halved the number of OOH referrals.

One study looked at using telephone advice given by nurses and paramedics for low-urgency ambulance services. 99 They found that nurses were more likely to triage calls as not requiring an ambulance. The 330 cases triaged by a nurse or paramedic as not requiring an emergency ambulance were less likely to attend ED, but 9% of these patients were admitted to a hospital bed. The only included study of a service using non-clinical call handlers supported by clinicians was the evaluation of NHS 111 pilot sites,105 but no comparison was made of decision-making by non-clinical call handlers and clinicians.

Safety

Eighteen studies reported on safety of telephone triage. The related systematic review by Huibers et al. 62 included 10 of the papers identified in this review66,73,74,77,78,82,89–92 and reported that a high proportion (97%) of all OOH telephone triage contacts were safe but that the proportion of safe contacts decreased when looking at high-urgency patients. From 10 studies using high-risk simulated patients, the authors found that just 46% were safe, bringing into question whether telephone triage and advice is as safe for higher-urgency patients.

Of eight additional studies68,92,103,104,133,140–142 not included in the Huibers et al. review,62 one study also found that more safety issues were associated with higher-urgency calls140 and two studies, one set in primary care68 and one set in emergency medical service (EMS) care,103 both reported no adverse outcomes or differences in suboptimal outcomes for high-urgency calls.

One study looked at characteristics of calls coded as low-urgency that resulted in death. 104 The authors found that these calls related to older people who did not make the call themselves. The calls were short, did not always conform to the system protocol and non-life threatening symptoms were reported. This concurs with the findings from a primary care-based study,137 which found that telephone consultations were shorter and did not include sufficient information to exclude serious illness. However, a study by Snooks et al. 142 reported that the overall proportions of patients who receive insufficient advice were low (1.4–3.3%). Based on the information from a study that analysed the incident reports from errors, 25% of errors related to incorrect assessment. 141

Compliance

Twenty-one studies reported on compliance with telephone triage advice recommendations. There were also two systematic reviews (i.e. the review of Blank et al. 57 and Purc-Stephenson and Thrasher58) that reported on compliance, which contained 13 of the studies identified in this review. 67,76,77,79,81,83,85,87,88,91,93,94,96 Blank et al. 57 reported that compliance ranged from 56% to 98%, and Purc-Stephenson and Thrasher58 reported an overall compliance rate of 62% but that compliance rates differed by the type of advice received, with 63% complying with advice to attend A&E, 44% complying with a primary care referral and 78% complying with self-care advice. Of the studies not included in the systematic reviews, three studies reported compliance that was within the range identified by Blank et al. 66,111,132 One study143 identified a 20% refusal of telephone triage recommendations, whereas another123 reported that callers may interpret advice differently from how it is intended, resulting in unintentional non-compliance. Compliance with advice provided by callers to NHS 111 was 86%. 106 A qualitative study reported that compliance may be improved when callers feel involved in the decision-making. 129

Satisfaction

Twenty-two studies (23 papers) reported satisfaction data. 66,67,69–71,79,86,87,93–95,97,101–103,106,109,110,120,121,127,129,137 Satisfaction with telephone triage and advice services ranged from 55% to 97%. Dissatisfaction ranged from 2.3% to 18.3%. Dissatisfaction was higher for EMS-based studies where patients may have expected to receive an ambulance and instead received telephone advice. Qualitative research suggests that patients value feeling involved in the decision-making, that they appreciate the manner of the call handler being calm, friendly and respectful121,129 and that they feel it is a useful backup service where they can avoid being labelled as ‘time wasters’. 121

Where telephone advice provided by different staff types was compared, patients reported less satisfaction with nurse-led services than with doctor-led services. 93,94

Costs

Nine studies reported costs or cost savings. 75,79,86,101,108–112,126 The most common method of calculating and reporting costs was to report costs saved from callers’ change in subsequent health-care-seeking behaviour as a result of the call. 75,79,108,110 These were not consistently reported and were reported as average per patient savings, cost savings per call recommendation type, annual net savings and overall reduction, making comparison difficult. Two studies reported costs related to introducing a nurse telephone line for low-urgency ambulance patients. 101,112 One of these studies112 reported a cost reduction of £4,520,000 from reduced ambulance dispatch and one101 reported reductions in costs owing to reduced ambulance cycle times and fewer admissions to hospital. Only one study reported a cost increase and this related to the cost of the general practice telephone bill after the introduction of a telephone service. 86 A cost–consequence analysis conducted as part of the evaluation of NHS 111 pilot sites estimated that the service might potentially save the NHS £2.4M a month, although this was based on limited cost data and considerable assumptions. 144

Service impacts

Seventeen studies reported impacts on their own service or other services, either from increased or reduced service use and or increased or reduced staff workload68–70,86,95,98,99,101–103,105,109,112,114,122,126,134 One study found that NHS Direct referred a higher proportion of patients to other services when compared with general practice. 134

A RCT by Lattimer et al. 68 reported a 38% reduction in primary care appointments and a 23% reduction in home visits, whereas another study reported no change in primary care appointments but home visits reduced by 18%. 70

A number of studies reported that doctor workload fell after the introduction of a telephone service. One study reported a reduction in workload of 54% compared with the previous 3 months69 and that nurses were able to handle 26% of requests for GP appointments by telephone, whereas another study reported a 39% reduction in demand. 86 However, other studies have found that, while a telephone advice service may reduce demand for face-to-face appointments in the first instance, patients consult their GP more frequently in the subsequent weeks. 95 In particular, older patients were more likely to have follow-up contacts after receiving telephone advice. 114 The evaluation of NHS 111 pilot sites found no impact on attendances at EDs and contacts with urgent primary care services but a statistically significant increase in ambulance incidents. 105

Five studies found that the introduction of a telephone advice service for low-urgency ambulance service patients resulted in fewer ambulance responses being required. 99,101–103,112 However, some studies reported high call-return rates, where calls are passed back for an ambulance response, or that not as many calls as expected could be resolved by telephone. 101,102

Accuracy and appropriateness

Twenty-six studies reported data on the appropriateness or accuracy of telephone triage and advice services. 72,73,75–77,80–82,84,88–90,100,101,115–119,124,131,135,138–140 The systematic review by Blank et al. 57 also reported on appropriateness and 11 of the studies identified here were included in the review by Blank et al. 73,75–77,80–82,84,88,90 This review identified that triage decisions rated as appropriate varied between 44% and 98%. However, many different definitions of appropriateness were used, which leads to difficulty when trying to compare results.

Of the 15 other studies,72,89,100,101,115–119,124,131,135,138–140 four studies looked at the appropriateness of ED referrals. In terms of referral sources, no difference was found between telephone and self-referred patients,118 but doctor-referred patients were the most appropriate referrals. 117 One study identified 33% of ED referrals as non-essential139 and another identified that a high proportion of ED referrals should have been handled in primary care. 116

Most studies reported that a high proportion of referrals, often in excess of 90%, were appropriate. 131,140 Triage advice and referrals were more likely to be unnecessary than insufficient. 142 However, one study identified that acute and non-urgent scenarios were more likely to be classified correctly than urgent. 115

Where low-urgency ambulance calls were passed for telephone advice there was a high return rate for an ambulance response. 101

Quality

There is a large evidence base on telephone triage and advice, but studies are dominated by retrospective studies and observational designs, with only a small number of randomised trials. 8 This reflects the difficulties inherent in conducting randomised studies in this environment, particularly if the service being investigated is in a single site. Individual studies and systematic reviews assessed a range of different outcomes and only one systematic review conducted a meta-analysis for a single outcome, compliance. 58 Broad, relevant quality issues identified by the authors of systematic reviews and individual studies include:

• different definitions of appropriateness and methods for measuring this outcome

• different methods for measuring patient satisfaction

• different clinical assessment systems used to triage calls

• differences in operational design – 24-hour services versus those only available out of hours

• differences in the wider health-care settings services have operated in with consequent variation in the availability of alternatives to ED care, such as primary care services.

The consequence is a wide variation in the range of reported outcome results, for example the systematic review by Blank et al. 57 found assessment of appropriateness ranged from 44% to 98%, which makes it difficult to make comparisons between studies and draw firm conclusions about the strength of evidence for some specific outcomes.

Conclusions

There is an existing, substantial evidence base about the operational and clinical effectiveness of telephone-based triage and advice services for management of requests for urgent health care. We have identified 87 relevant papers from 83 studies, including eight systematic reviews, reporting a range of outcomes. The most commonly reported outcomes were concerned with accuracy and appropriateness of decision-making, patient compliance with advice, safety of decision-making and patient satisfaction. Despite the limitations outlined above, some general conclusions can be made about these aspects of telephone triage and advice services:

• Accuracy of decision-making is generally high with respect to minimising risk. Similarly, despite wide variation in reported assessments of appropriateness overall, the proportion assessed as appropriate was high in the majority of studies. Where calls were judged to have been inaccurately or inappropriately triaged, the tendency was to assign calls to a higher level of care than needed rather than a lower level. What is less clear is whether this is a function of the triage assessment systems, the individuals using the system or a combination of the two. With the exception of one early NHS Direct study, we have found no research that has compared the performance of different telephone triage clinical assessment systems.

• A consequence of this apparent inherent risk-averseness is that, overall, telephone triage systems appear to be safe, although they may not be efficient. In the small number of studies where adverse patient events or outcomes have been measured, these were very low.

• Reported compliance is also variable but overall tends towards a high level, although this varies depending on advice given, with self-care or ED visits appearing to have higher levels of compliance than advice to contact primary care. This may reflect patient perceptions and preferences and would align with the evidence described in Chapter 3 on patient perception of perceived difficulty in accessing primary care.

• Patient satisfaction with telephone triage is generally high and the service well liked by users. Although only a small number of studies compared nurse-led services with doctor-led services, satisfaction appears to be higher in doctor-led services.

Other outcomes or aspects of service delivery have been less well investigated and warrant further study:

• Only 17 of the included papers assessed the impact of telephone triage and advice services on other services associated with emergency or urgent care. The majority of these studies only assessed impact on a single other service (e.g. ambulance responses, ED attendances, primary care attendances). With the exception of one US study109 that assessed impact on both ED attendances and primary care activity, only the UK-based studies of NHS Direct66 and NHS 111105 have attempted to assess the impact of telephone services on multiple services within the emergency and urgent care system. Whole-system impact is a key outcome if one objective of telephone-based services is to contribute to improving system efficiency.

• The included studies were predominantly concerned with nurse-led telephone services. Only nine papers from six studies explored the effects of different types of staff delivering these services. Most of these compared nurse- or doctor-led services, one compared nurses and paramedics and two papers from one study assessed lay operators in a GP OOH setting. The findings from these studies were mixed, with some reporting no differences in appropriateness of decisions between nurses and doctors while others found nurses more likely to refer to higher-level care than doctors. Acute problems and non-urgent problems appear to be more accurately identified than urgent problems, suggesting that the latter are more difficult to assess, and there is a suggestion that higher-level clinician assessment may improve accuracy of urgent problem identification. Only one study investigated two-level triage (nurse assessment followed by doctor assessment) for a subset of calls. The broad area of optimum skill levels and mix of clinicians and non-clinicians has not been addressed.

• Few studies have examined the costs associated with telephone triage and advice services, and these have taken different approaches to estimating and reporting costs, which makes comparison difficult. The financial benefits of these services both at an individual-service and whole-system level remain unclear.

• Given that a primary objective of telephone triage and advice services has been to improve and simplify access to urgent care, there is remarkably little evidence about the impact of these services on access. Only five included studies examined impact on access.

These four evidence gaps are of particular relevance to the continued development of the NHS 111 service within the NHS. Assuming that NHS 111 will continue to be a principal entry point for access to urgent care and that the existing clinical assessment system (NHS Pathways) will remain unchanged, the key challenges are based around improving the existing service so that it can fulfil the intended objectives. It is worth remembering that the scale and scope of the NHS 111 service is much greater than that of the services included in the existing evidence, including its predecessor NHS Direct. It is also the case that there have been significant technological advances over the 20+ years that the evidence spans, therefore evidence from early studies may lack relevance now. Previous services have mostly been small in scale and intended to provide advice and some direction about what service people should access.

The vision for NHS 111 is that it provides broader functions, including health information, signposting to other services to which patients can refer themselves, direct referral to other services on behalf of the patient (including ambulances, where needed, and appointments made in real time) and advice. Good decision-making at the time of the call is therefore key to an efficient and effective service. The most pressing research issue is therefore concerning staffing the NHS 111 service. NHS 111 is unique in that it is a national telephone urgent care-access service where non-clinical call handlers make the first assessment. The only evidence on impact of using non-clinical call handlers is the existing NHS 111 evaluation studies of pilot services. Qualitative research examining the role of non-clinical call handlers found the work had more in common with clinical work than other call centre work and that the requisite skills need to be carefully incorporated in to training. 145 These call handlers are supported by nurse or paramedic clinicians, and the NHS England review has stated that the addition of senior (doctor) clinicians to NHS 111 will be considered as part of the ongoing development of the service. Changes to the staffing profile of NHS 111 call handling services will provide a unique opportunity to address the questions around impact of different types of clinicians on accurate and appropriate call assessment and the proportions needed to provide an efficient service. Additional specialist clinical skills, such as mental health specialists and pharmacists, could also potentially be added to the mix. Adequate follow-up of callers would be needed in order to assess whether or not care pathways are shortened and care really is provided in the right place, at the right time, from the first call, as well as whether or not access improves. The early research on NHS 111 found it did not substantially change the way the population accesses urgent care; however, this may have changed since it became a national service. 107

Alongside this, the impact of different staffing models on whole-system activity can be measured. NHS 111 has been criticised for referring too many patients inappropriately to higher-level ambulance and ED care. The early evaluation of pilot sites did show an increase in ambulance incidents after the introduction of NHS 111 (the reasons for which remain unexplained), but not ED attendances. 105 More detailed evaluation of the accuracy and appropriateness of call assessment decisions would answer some of these questions and identify where there is scope to improve it, either through changing the staffing profile or modifying the clinical assessment system.

Changes to the staffing profile, particularly the addition of doctors and/or increasing the proportion of clinical advisors, and increased IT infrastructure to support real-time referrals will incur costs which will need to be offset by improving whole-system efficiency. Robust economic evaluation alongside other changes would provide evidence on the associated costs of providing this complex service, whether or not it provides value for money and the necessary cost shifting within the emergency and urgent care system needed to plan and support the service in the future.

Finally, evaluation of improvements in the NHS 111 service itself, together with measurement of whole-system changes in volumes of activity in different associated services and some specific population-based research, would also begin to address issues about access and whether or not the service produces real changes and improvements in the way people access urgent care.

Introduction

Another key principle of the NHS England review is to provide more urgent care services away from hospitals and so encourage people not to choose A&E services. Enhancing urgent care outside hospital involves a range of services, including GPs, primary care and community services, community pharmacists and ambulance paramedics. There was not scope to consider all of these services within the time frame for this review. Particular emphasis has been placed in the NHS England review on development of the ambulance service by extending paramedic training and skills so they can manage more people on-scene and avoid unnecessary journeys to hospital. We have therefore focused our rapid review for this principle on identifying and summarising the existing evidence around initiatives implemented by the ambulance service to support this objective.

Methods

The main inclusion and exclusion criteria, search strategies and review process have been described in Chapter 2. Searches were conducted for the years 1 January 1995–26 November 2014. For this review, specific additional inclusion and exclusion criteria were used to identify studies investigating the areas detailed below.

Review process

Studies were identified from database searching and through expert knowledge of the field of interest. In addition to the studies identified through database searching, the study team identified 10 additional studies and these were assessed for inclusion in this review. A randomly generated 10% sample of studies from the database searches (n = 462) was double sifted by another member of the study team. The reviewers initially agreed on inclusion or exclusion for 97% of studies. The remaining 12 studies were discussed with a third reviewer; of these, six were included and six excluded. The results of the review sifting process are given in Figure 4.

FIGURE 4.

A PRISMA flow diagram for ambulance management in the community searches.

Results

We identified seven relevant systematic reviews and, within these, 64 papers from 57 studies were included, although only 21 of these papers met the inclusion criteria for this review. We also identified an additional 12 primary studies not included in the systematic reviews. Formal data extraction was not undertaken for the 21 papers meeting the inclusion criteria and included in systematic reviews, as this would have been conducted during the review process. Table 9 lists these 21 studies and the systematic reviews in which they were included.

Studies fall in to two broad groups: those concerned with a focus on investigating expanded professional roles for pre-hospital care providers (predominantly ambulance paramedics) and those focused on investigating alternatives to transport to ED, which encompasses broader professional groups rather than just those with enhanced skills.

We have summarised the characteristics and findings of the seven systematic reviews146–151,173 (Tables 10 and 11). Data on the 12 papers174–185 not included in the systematic reviews were extracted in to summary tables (Tables 12–14).

Summary of findings

The main characteristics of the 33 primary papers (21 papers included in systematic reviews and 12 additional papers) are summarised in Table 15.

The majority of related research has been conducted in the UK and eight of the included papers were outputs from two large studies by Mason et al. (i.e. Mason et al. 2007165 and Mason et al. 2012168). There were only two randomised studies165,181 and one relevant economic evaluation associated with a larger trial of a community falls service. 186 The main outcomes studied were rates of transportation to EDs, accuracy and safety of decision-making and patient satisfaction. Over 50% of studies included the broad emergency ambulance population, with specific populations dominated by older people who have fallen. The small number of studies that have considered costs were also associated with the three randomised trials highlighted above.

Effectiveness

Six of the systematic reviews146–148,150,151,187 found evidence that prehospital professionals can successfully reduce the number of patients transported to EDs and provide autonomous care and appropriate referral in the community setting. The evidence is predominated by schemes to provide ambulance paramedics with extended and enhanced clinical skills. Two studies160,165 showed that hospital admissions were also reduced in patients managed at home by paramedic practitioners and that patients managed at home had no more subsequent health contacts. 167

Four of the UK studies154,157,161,168 involved emergency care practitioners (ECPs), a role open to other health-care professionals such as nurses, although in reality dominated by ambulance paramedics, and practised in settings other than ambulance services. The review by Evans et al. ,148 which was concerned only with paramedics with extended skills, excluded studies of ECPs. The title ‘ECP’ has created problems as the generic ‘practitioner’ part of the title proved confusing. It conflicts with the regulators’ ability to ensure patients know who is treating them and was opposed by paramedic professional bodies. 188 The title ECP has been replaced (for paramedics) with titles listed in the paramedic career framework, such as specialist and advanced paramedic. 189

A smaller number of studies have investigated whether or not ambulance clinicians other than those with advanced skills can manage patients out of hospital and reduce transports to EDs. Snooks et al. 190 evaluated ‘treat and refer’ protocols for paramedics and found these did not increase the number of patients left at home. More recently, a trial of computer decision support software to aid paramedic on scene decision-making about referral to a falls service has shown some promise in terms of increasing the number of patients referred and left at home, although the numbers included in the study were small. 181

Where measured, the evidence on patient satisfaction and acceptability of initiatives to manage patients out of hospital and closer to home is positive. The UK studies on paramedic practitioners and ECPs165,168 found that patient satisfaction was higher in the groups receiving care from the advanced practitioners than those receiving usual care, and other studies of ECPs also found high levels of patient satisfaction191 and that patients found advice and explanations were clearer. 161,191 A positive effect for carers has also been reported. 192 Another study found that patients left at home following an episode of diabetic hypoglycaemia and referred to a specialist nurse reported high patient satisfaction and increased confidence in managing future episodes. 185

Only a small number of studies have examined costs but, where an economic evaluation has been done, management of patients by ambulance clinicians out of hospital has been reported as having lower costs and as being likely to be cost-effective when compared with conventional hospital care. 167,181,186,193

Safety and decision-making

Reducing transports to hospital is linked to safe and appropriate decision-making. Studies of advanced practitioners have shown that decision-making is generally safe and appropriate. 161,166,169 A number of studies have also examined the accuracy of ambulance clinician decision-making around whether or not to transport to hospital for the wider workforce and not just paramedics with advanced skills. The systematic review by Brown et al. 173 assessed the evidence on paramedic decision-making about need for transport and concluded that there was insufficient evidence to establish whether or not paramedics can accurately determine who needs transport. However, only five papers were included in this review.

More recently, one UK study has assessed the ability of ambulance staff to predict which patients need hospital admission and found that they showed reasonable accuracy in determining who needed admission and could correctly identify patients who did not. 177 A US study assessed emergency medical technicians’ ability to identify intoxicated people who did not need transporting to hospital and found there was potential for emergency medical technicians to accurately triage patients to alternatives to ED. 178 Another recent UK study evaluated the utility and safety of a triage support tools (Paramedic Pathfinder) for deciding which medical and trauma patients needed transporting or could be left at the scene. 179 Sensitivity was high, with reasonable specificity; however, the tools were used in shadow form (all patients were still transported) and there is currently no published evaluation of the tool in real practice.

A small number of more recent qualitative studies have provided some valuable insight into the difficulties associated with the decision-making processes faced by paramedics and advanced paramedic practitioners when deciding whether or not to transport patients. A detailed study on safety and decision-making by ambulance staff highlighted that, as the need to reduce transports to hospital increases, the requirement for appropriate training and education to equip ambulance clinicians for this role also increases. 176 Two studies on decision-making by ambulance staff for patients with epilepsy174 and elderly fallers175 both identified a need for additional training and support to aid safe decision-making, as did one Canadian study that focused on out-of-hospital management of patients with long-term conditions. 182

Implementation of initiatives

Some studies have highlighted the challenges, facilitators and barriers to implementation of initiatives to enable ambulance clinicians to manage more patients out of hospital, although these aspects of service provision are less well reported. The systematic review by Bigham et al. 146 highlighted issues associated with identifying patients suitable for advanced paramedic practitioner intervention and appropriate dispatching at the time of the emergency call as a potential barrier to using these practitioners effectively. As discussed above, training to support decision-making has been identified as a facilitator, but also, where it is felt to be insufficient, as a barrier. Issues about effects of integration with the wider urgent and emergency care system on successful implementation and, in particular, the need to have clear information about patients and about access to alternative care pathways for patients have also been raised. 150,174,176 Other relevant issues identified which impede patient management out of hospital include the conflicting demands of maintaining operational performance and meeting response-time targets (which reduces opportunities for training), barriers between professional groups, a continuing perception that the ambulance service is still primarily a transport service and fear of litigation which is associated with complex decision-making. 174,176

Quality

Although there is a substantial evidence base associated with the development of the ambulance service as a provider of out-of-hospital care, as evidenced by the seven systematic reviews included in this rapid review, there is a lack of robust, high-quality RCTs. Much of the relevant evidence included in the systematic reviews is derived from the two UK studies comprising a cluster randomised trial of paramedic practitioners165 and a community intervention trial of ECPs. 168 There is also a single randomised trial of use of computerised clinical decision support (CCDS) software for management of fallers. 181 The rest of the evidence is based in a range of study designs that introduce a number of limitations identified by study authors, including:

• use of small convenience samples in some studies, which introduce potential bias from self-selection of participants and may not be representative of the intended clinician group

• use of service evaluation approaches with no control group and, therefore, reported results do not account for any potential bias or confounding effects

• studies of decision-making where clinicians have recorded intended rather than actual decisions (i.e. patients have still been transported to hospital), which limits applicability to actual clinical practice. Compliance rate in completing questionnaires is often low

• important outcomes such as adverse events and subsequent health-care utilisation are, with a small number of exceptions, not recorded or reported.

More generally, most studies have been small scale, in single sites and usually involved a subpopulation of patients and clinicians, thus limiting generalisability to wider clinical practice.

Conclusions

Extended paramedic roles have been implemented in various health systems and settings and, based on a small number of high-quality studies, appear to be successful at reducing transports to hospital, making safe decisions about the need for transport and delivering acceptable care out of hospital, as well as being potentially cost-effective. The evidence on safety of decision-making about need for transport to hospital in the broader paramedic workforce without advanced skills is less clear and questions remain about safety. Studies concerned only with decisions about whether or not to transport to hospital also tend to ignore the important issue: that a decision not to transport requires good clinical care to be provided at the scene as an alternative. Good-quality care out of hospital is about more than just the decision to transport.

The authors of the seven systematic reviews and 12 individual papers included in this rapid review have made a range of recommendations for further research, including:

• further work on the safety of decision-making and development of better guidelines, decision support systems (including CCDS) and training to support decision-making

• further assessment of the risk of under-triage associated with decision support tools

• barriers to implementation

• cost-effectiveness of investment in training and potential cost savings across different settings and patient groups

• further development and exploration of out of hospital care pathways for a range of conditions to support decision-making as effectiveness of initiatives to reduce transport to ED will depend on availability of alternative pathways and services

• assessment of the impact of prehospital interventions on ED crowding.

These areas warranting further research have direct application to the NHS England review and the review’s expectations of an enhanced role for the ambulance service in managing more patients with urgent health problems in the community setting. It could be argued that at least some of these areas, for example resolving issues around safety, cost-effectiveness and wider system impact, are precursors to any widespread adoption of increased out-of-hospital management of patients by ambulance clinicians. However, pragmatically, it is the case that ambulance services in England have been progressively reducing the proportion of patients it transports to EDs over a number of years, both through managing more patients at the time of their call (‘hear and treat’, discussed more fully in Chapter 4, Costs and Service impacts) and outside hospital (‘see and treat’). Currently, the mean non-conveyance rate to ED is 37%, although this varies from 22% to 55% between individual ambulance trusts (www.england.nhs.uk/statistics/statistical-work-areas/ambulance-quality-indicators/ambulance-quality-indicators-data-2014-15/).

The reasons for this variation are unclear and the HSDR programme has already funded research to explore this. 194 However, this does illustrate that the ambulance service already manages a substantial number of its calls outside hospital. Given this and the clear important strategic role of the ambulance service in the NHS England review implementation plans,8 the future questions relevant to the continued development of the ambulance service as an out-of-hospital care provider are not about if it should be providing this service but how it can continue to expand and improve this aspect of service delivery. There are four key evidence gaps that could be addressed to support this process:

1. Clearly a key component to achievement of the NHS England plans is development of the paramedic workforce. The available evidence has shown the potential of advanced paramedic practitioners, but these have been relatively small scale and are likely to have involved highly motivated individuals willing to advance their practice. There is no evidence about the implications of developing the paramedic workforce at scale to achieve national objectives. One of the issues highlighted in the existing evidence is the possible ‘gap’ in decision-making skills between paramedics and paramedics with advanced skills. The paramedic workforce is changing, with a much clearer professional framework,189 and is rapidly moving towards a graduate-based profession;195 however, at present it is unknown to what extent this change may reduce the gap. There is a pressing need to begin to investigate the necessary skill mix of the paramedic workforce and the proportions of paramedics and paramedics with advanced and specialist skills needed to provide a safe and high-quality, community-based service for patients. The implications for the educational and training needs of the paramedic workforce that will be needed to fulfil this enhanced role, at scale, will need to be considered at the same time.

2. More broadly, and to support paramedic workforce planning, some clarity is needed about volume of activity and, specifically, the proportion of patients who could be managed outside hospital. Although the NHS England review has a clear view that ambulance services can manage more patients out of hospital, it stops short of stating how many patients this might encompass. As stated earlier, we already know that there is substantial variation in the non-conveyance rate to hospital across ambulance trusts in England, and it is reasonable to assume that there is scope for services with low non-conveyance rates to move towards performance of services comparable with higher rates of non-conveyance. What is uncertain is whether or not there is a threshold for non-conveyance beyond which the risks of leaving people at home begin to outweigh the benefits.

3. The qualitative research studies that have explored in more depth the barriers and facilitators to developing management of patients in the community have highlighted the need to develop robust clinical care and referral pathways for a range of different conditions as alternatives to transport to hospital. Without these pathways and confidence that onward referrals will be actioned,176 efforts to manage more people at home will be thwarted. Development of pathways is linked to development of robust guidelines and tools to support clinical decision-making at-scene. This will require wider system collaboration rather than small-scale, individual, local initiatives. It may be the case that not all patients who can be left at home need an advanced practitioner if they can be referred to an appropriate alternative provider. The existing evidence suggests that there is potential for CCDS software and triage tools (such as Paramedic Pathfinder) to improve and support decision-making; further research in this area is needed, but this will need to be aligned to workforce development, as graduate paramedics with higher-level skills may circumvent the need for this type of triage tool. However, it is likely that, as the range of alternatives increases, so the need for decision support tools increases.

4. The existing research evidence is based in small studies which, although showing local effects on reducing transports to ED, have not been at sufficient scale to assess system impacts across a range of services, both in terms of activity and associated costs. Expanded care by ambulance services in community settings may reduce ED attendances and hospital admissions but increase contacts with primary care, community and social services. Ambulance services themselves may need additional resources (staff, vehicles and equipment) and have increased training costs to support this expansion of their service. Future evaluation of initiatives to develop paramedic practice and increase the number of patients managed in community settings needs to assess the impact of these initiatives at scale and at a whole-system rather than individual-service level.

Introduction

The NHS England review6 has a focus on managing more urgent conditions outside acute hospitals, including, for example, urgent care centres. It also recognises the need to provide higher-level care for some health-care problems and sets out a strategy for emergency care that can deliver treatment in the right type of facilities and maximise patient outcomes. The model envisaged includes two types of facility: (1) major emergency centres, with highly specialised services; and (2) emergency centres, which can treat most emergency conditions and, where appropriate, refer on to a major emergency centre. A core principle for both centres is the presence of senior clinicians 7 days a week. This ‘model’ has been interpreted to some extent as creating a two-tier service but, in reality, reflects current provision, and NHS England states it expects the number of centres to remain broadly the same as the current number of EDs.

Reviewing the evidence on delivery of ED services poses significant challenges. Unlike the two preceding reviews, which had a focused, well-defined method (telephone triage) or aspect (paramedic management of patients out of hospital) of service delivery, provision of ED care encompasses the much broader set of issues associated with whole-service delivery. These include organisation and operations, service dynamics (flow), clinical roles of different professional groups, demand, clinical complexity of a heterogeneous population of users and relationships with other parts of the health service. Consequently, there is also a vast evidence base of related research.

The commissioning brief for these reviews has a focus on models of delivery. We have not reviewed the evidence on highly specialised and regionalised care (acute cardiac care, stroke, major trauma) as these have been extensively reviewed elsewhere. 196 Instead, we have focused this review on the evidence on the organisation and operations associated with delivery of general ED services (which are also provided in major centres). The Royal College of Emergency Medicine produced a report in 2014, which identified 13 recommendations for changes needed to build a sustainable and resilient emergency and urgent care system. 9 We used some of these recommendations, specifically regarding delivery of ED care (rather than whole-system issues) to further refine the focus. We have therefore conducted two linked but separate rapid reviews:

1. A key recommendation is that every ED should have a co-located primary care OOH facility. We have updated an existing related rapid evidence review published in 2010 to assess the current evidence for this model of care.

2. We have also reviewed the broader evidence on ED organisation and operation and, on advice from a senior clinician, have kept the emphasis on ED flow (including front-door assessment and back-door discharge), the related area of clinician role and the key growing area of managing the complex elderly patients in EDs.

Co-location of primary care and emergency departments

Results

Nine papers were included in the review:10,143,197–203 two systematic reviews,10,197 which included 28 individual papers, and seven additional papers. 143,198–203

The results of the review sifting process are given in Figure 5.

FIGURE 5.

A PRISMA flow diagram for management of primary care in ED searches.

A summary of the characteristics and findings of the included papers are given in Tables 16 and 17.

TABLE 16 - Summary of systematic reviews on management of primary care conditions in ED

EP, emergency physician; NP, nurse practitioner.

Author, year	Study design	Population/setting	Purpose	Main findings
Fisher, 201010	Rapid review of the literature and narrative analysis	Patients of all ages seeking primary care attending a service: integrated within an ED co-located with an ED located within the grounds of the hospital located within close proximity to a hospital	Assessment of evidence on interventions for managing patients with primary care-type conditions, either within or allied to the ED Outcomes included: attendance at primary care, attendance and/or reattendance at ED, adverse events, patient satisfaction, investigations and referrals requested and prescriptions issued	Twenty-eight studies included. A GP working in the ED may result in fewer referrals for admission and fewer tests being undertaken. Cost benefits may exist but the evidence is weak Redirect away from ED has variable results in terms of predicting future attendances. Assessments of safety of this intervention have variable results. Primary care appointments made for patients redirected from the ED may not be kept Educational interventions do not show a change in attendance patterns There is a paucity of evidence available to support the current system
Khangura, 2012197	Systematic literature review	Patients who present to hospital EDs with illness or injury suitable for primary care	Effect of primary care professionals working in hospital EDs (including GPs, NP, optometrists and dentists) on admissions, length of stay, resource use, costs, follow-up care and health-care utilisation and adverse events	Three studies included. Two studies reported that GPs used significantly fewer health-care resources than EPs, with fewer blood tests, radiography, admissions to hospital and referrals to specialists. One study reported no difference in number of prescriptions made by GPs compared with EPs, and one that GPs prescribed significantly more than EPs. These two studies showed marginal cost savings from introducing GPs in hospital EDs The third study failed to identify a significant difference in the number of investigations or admissions to hospital, but identified a significantly greater number of referrals to specialists and prescriptions made by GPs compared with EPs. There is insufficient evidence to draw conclusions about the effectiveness and safety of GPs vs. EPs for non-urgent patients in the ED

TABLE 17 - Summary of individual studies on management of primary care conditions in ED

CI, confidence interval; ECAP, emergency care access point; HGP, hospital-integrated general practice; IQR, interquartile range; OR, odds ratio; PCR, primary care related; PCR-ED, primary care-related emergency department.

Author, year, country	Study design	Population/setting	Purpose	Main findings
Boeke, 2010,198 the Netherlands	Before-and-after comparative study of a new care method in the ED of an Amsterdam university medical centre	Adult patients who self-presented to ED on weekdays during the day (10.00–17.00 hours)	Evaluation of impact of introducing GP services into ED to manage primary care conditions on patient satisfaction, time from arrival to departure, time from start of consultation to departure and referrals for aftercare	Patient satisfaction with the treatment increased significantly The mean process time decreased from 93 to 69 minutes (p < 0.001). The mean treatment time decreased from 60 to 35 minutes (p < 0.001)
Bosmans, 2012,199 the Netherlands	Before-and-after design	Self-referrers who attended the ED on weekdays during the day (10.00–17.00 hours)	Evaluation of impact of adding a GP to the ED to manage primary care problems on patient satisfaction, process time and costs	Significantly reduced process time, significantly higher patient satisfaction and no statistically significant difference in the number of correct diagnoses after intervention introduced. Total costs per patient were €217 in the new service compared with €288 before (mean difference –€71, 95% CI –€121 to –€23)
Enard and Ganelin, 2013,200 USA	A non-equivalent comparison group, quasi-experimental study design	Patients older than 18 months and younger than 65 years who frequently use the ED for primary care and are publicly insured	Impact of a primary care patient-navigation referral programme on PCR-ED attendance and cost per person of a PCR-ED visit	Patient navigation intervention was associated with decreased odds of returning to the ED in less-frequent PCR-ED users. For patients who returned to the ED for PCR reasons, the pre-/post-mean visits declined significantly over 12 months but not 24 months. Savings associated with reduced PCR-ED visits were greater than the cost of implementing the navigation programme
Salisbury, 2007,201 UK	A controlled before-and-after study	Patients consulted, admitted or discharged monthly from 6 months before the opening of walk-in centres to 6 months after	Evaluation of the impact of opening walk-in centres located alongside EDs on attendance rates, duration of visits, costs and outcome of care	At most sites, the walk-in centres did not have a distinct identity and there were few differences in the way services were provided compared with control sites. Overall, there was no evidence of an increase in attendance at sites with walk-in centres, but considerable variability across sites. Proportion of patients managed within the 4-hour target improved at intervention and control sites. No evidence of any difference in reconsultation rates, costs of care or patient outcomes at sites with or without walk-in centres
Sharma and Inder, 2011,143 Australia	Observational study of ED waiting times over 1 year in the state of Victoria	Emergency (category 2) and non-urgent (category 5) patients in 38 hospital EDs	Analysis of differences in patient waiting times (difference between arrival time and time first seen by treating doctor) at EDs with and without co-located GP services	The degree of choice for non-urgent patients has a non-linear effect: more choice for non-urgent patients is associated with longer waits for emergency patients at lower values and shorter waits at higher values of degree of choice. Increasing choice of ED for non-urgent patients related to longer wait for emergency patients in EDs. Waiting time for emergency patients in hospitals with co-located GP clinics was 19% (mean 1.5 minutes) less than in hospitals without co-located GP clinics
Thijssen, 2013,202 the Netherlands	Longitudinal observational study 3 years before and 3 years after the start of an ECAP	All attendances at ED and GP co-operatives over the 6-year period of 2006 to 2012	Evaluation of the impact of introducing an integrated after-hours care model on ED and GP co-operative self-referrals, non-self-referrals and follow-up contact	ED attendances decreased by 13% after the start of the ECAP (59,182 before vs. 51,513 after). Self-referred ED patients decreased by 99.5% (OR 0.003, 95% CI 0.002 to 0.004) Referred patients increased by 213.4% and ED hospital admissions increased by 20.2%. Planned outpatient follow-up increased by 5.8% (OR 1.968, 95% CI 1.870 to 2.071). Regional general practitioner co-operative consultations increased by 26.0% (183,782 vs. 232,246)
Wang, 2014,203 USA	Before-and-after study	Consecutive patients attending ED with ambulatory care conditions from 1 October 2007 to 12 November 2007 before intervention (n = 451) and routed to HGP from 6 April 2010 to 14 May 2010 post intervention (n = 342)	Evaluation of the effects of implementing of a new HGP service on time from admission to discharge and time intervals between different stages of care	Median time from admission to discharge was 120 minutes in the ED (IQR 80–165 minutes) vs. 60 minutes in the HGP (IQR 40–90 minutes) (p < 0.001) The adjusted OR of receiving any additional diagnostics was 1.86 (95% CI 1.06 to 3.27; p = 0.032) for ED doctors vs. GPs controlled for age, sex and injury-related medical problems

The characteristics of primary studies included in the two systematic reviews and the seven additional studies are described in Table 18.

TABLE 18 - Characteristics of studies on primary care in the ED

Characteristics	Number of studies, n (%)
Country
USA	14 (40)
UK	6 (17)
Other Europe	14 (40)
Other	1 (3)
Study design
Before-and-after study	21 (60)
Randomised control trial	5 (14)
Prospective	3 (9)
Observational study	4 (11)
Systematic review	1 (3)
Retrospective audit	1 (3)
Patients
Adults and children	11 (31)
Children	7 (20)
All	12 (34)
Adults	5 (14)
Setting
Co-located	10 (29)
Not co-located	25 (71)

Summary of findings

Two systematic reviews10,197 examining the evidence on the clinical effectiveness and cost-effectiveness of primary care services associated with EDs included findings from 28 studies. Both concluded that, although there was some evidence that locating primary care services within or close to EDs could reduce ED attendances may save resources by using fewer diagnostics and referrals for admission and potentially could reduce costs, the evidence base to support development of this model of care was weak and based on poor-quality studies. Our search identified a further seven more recent studies which, overall, presented more positive findings on the potential benefits of integrating GPs within EDs. Two studies in the Netherlands, where GPs were located within the ED,198,199 found process and treatment times were reduced and patient satisfaction increased after the introduction of this service, and that there were cost savings. However, these were both uncontrolled before-and-after studies, so the influence of other potential confounding factors is unknown. A study of a primary care patient navigation service significantly reduced the likelihood of ED readmission for less-frequent primary care-related ED users200 and produced cost savings, while a hospital-integrated general practice service was shown to have potential to significantly reduce the time from admission to discharge and resource use as primary care doctors used fewer diagnostics than ED physicians. 203 A broader study examined longitudinal trends in ED attendances over 3 years after introducing integrated OOH GP co-operatives. 202 The findings were that, overall, ED attendances, and in particular self-referrals, decreased but referrals from GPs to EDs increased, as did ED admissions. Unsurprisingly, GP co-operative consultations increased and what is unclear is the net effect of patient movement between these services. Again, all of these studies were uncontrolled. The only comparative trial was conducted in the UK and investigated the impact of introducing walk-in centres in the vicinity of EDs. 201 This study found that, while ED attendances decreased in some centres, in others there was no change. Overall, there was no evidence of impact on reconsultations, patient outcomes or costs. The lack of success may have been owing in part to the walk-in centres not having any clear identity and hence not being visible to the public as an alternative to EDs. Of the available evidence, the most convincing seems to be from models where GPs or primary care professionals are located within the ED itself rather than alongside it or as stand-alone units. Intuitively, this makes sense, as it creates a single ‘front door’ which may reduce confusion for the public and allow more efficient streaming between ED and primary care, with a single entry and assessment point. It is also possible that this intervention may be most suitable for simple single episodes and that complex conditions and frequent users need more specialist and sophisticated care. 200 Boerke et al. 198 make an important point about the influence of general societal and cultural factors on health-care-seeking behaviour, and suggest that interventions to reduce the amount of ‘inappropriate’ use are likely to fail and that strategy might be better focused on enhancing adequacy of care than on trying to match health-care-seeking behaviour to the services.

Quality

The quality of the available evidence is generally poor. Of the 35 included studies, there were only five RCTs and a high preponderance of uncontrolled before-and-after studies, which cannot control for other system effects. For example, the study by Thijssen et al. 202 acknowledges that closure of hospitals during the study period may have had some influence on the findings. The authors of the two systematic reviews identified a range of limitations to the studies included in their reviews, including small sample sizes in the form of low response rates, lack of data on patient waiting times, length of hospital stay, adverse effects or mortality and a lack of consideration of patient safety issues. 197 The review by Fisher et al. 10 also identified the poor research design and reporting of studies, including variability in study design and length, sample size, period of follow-up and definitions of primary care/non-urgent type condition. Individual authors of included studies have acknowledged the difficulties of conducting randomised trials in this difficult environment. 199 The majority of studies were set in single centres, which limits generalisability. This is exacerbated if the focus is a particular group, such as the uninsured, the elderly or children, and may also be limited by the health-care system in which an intervention operates. 200 The trial by Salisbury et al. 201 identified some additional limitations and important considerations, particularly the low quality of available routine data, the need to allow new services time to mature (and hence to design studies of sufficient length to detect real effects) and the need to conduct studies at scale and assess true impact across a whole system rather than single sites.

Emergency department organisation and operation

Results

Twenty-two systematic reviews were included. 204–225 The results of the review sifting process are given in Figure 6.

FIGURE 6.

A PRISMA flow diagram for delivery of ED services search.

A summary of the characteristics and findings of the included systematic reviews are given in Tables 19–21.

TABLE 19 - Summary of systematic reviews on managing ED flow

CI, confidence interval; CPOE, computerised physician order entry; LOS, length of stay; LWBS, left without being seen; NP, nurse practitioner; RR, relative risk; SOU, short-stay observation unit; TLP, triage liaison physicians; TNO, triage nurse ordering; WMD, weighted mean difference.

Author, year	Study design	Population/setting	Purpose	Main findings
Abdulwahid, 2015204	Systematic literature search with assessment of relevance and bias using the Effective Public Health Practice Project bias tool	ED, senior doctor triage	To review and synthesise studies with the primary aim of investigating the effect of senior doctor triage on ED quality indicators: waiting time, LOS, LWBS, left without treatment complete	A total of 25 studies with a control group were included. Most studies showed that using senior doctor triage improved ED performance measure results. Pooled results from two out of four RCTs showed a significant reduction in LOS for medium acuity patients (WMD –26.26 minutes, 95% CI –38.50 to –14.01 minutes). Two other RCTs showed waiting times significantly reduced when using senior doctor triage (WMD –26 minutes, 95% CI –31.68 to –20.65 minutes). LWBS was reduced in two RCTs (RR 0.79, 95% CI 0.66 to 0.94). This was similar to most pre–post study designs. Senior doctor triage had no effect on adverse events or clear benefits for patient satisfaction or cost-effectiveness. Nine studies were of moderate quality, four were of strong quality and 12 were of weak quality
Bullard, 2012205	Narrative synthesis, grouped by outcome	ED, rapid assessment zones/pods	To synthesise evidence on the use of rapid assessment zones/pods to reduce overcrowding in adult EDs	There were four included studies (one RCT, one controlled trial and two before-and-after studies), with sample sizes between 200 and 12,305, representing 23,189 participants. Study quality was assessed as weak for three of the studies The review reported that rapid assessment zones/pods have the following benefits: Two studies showed a reduction in ED length of stay. This ranged from –20 minutes (95% CI –47.2 to 7.2 minutes) to –192 minutes (95% CI –211.6 to –172.4 minutes) Three studies showed reductions in physician initial assessment ranging from 8 minutes (95% CI –13.8 to –2.2 minutes) to 33 minutes (95% CI –42.3 to –23.6 minutes) Two studies showed a reduction in LWBS, ranging from a RR of 0.93 (95% CI 0.77 to 1.12) to 0.68 (95% CI 0.63 to 0.73). Low-acuity patients were found to benefit most from rapid assessment zones, but there is limited evidence to support implementation
Cooke, 2003206	Qualitative systematic review	ED, assessment/admission wards	To review the advantages and disadvantages of the use of assessment/admission wards	A total of 17 studies included. Most suggest assessment or admission wards are beneficial; they improve patient satisfaction, are safe, reduce length of stay, provide earlier senior involvement and reduce unnecessary/inappropriate admissions. May be particularly useful for certain diagnostic groups, but have little effect if care is already good. Observation wards may reduce ED workload and improve patient flow. Reduced pressure on hospital staff allows more time for important decision-making such as diagnosis, ensuring appropriate admissions that may reduce costs. Some studies advise that patients should be transferred out within 24 hours to prevent inappropriate use
Daly, 2003207	Systematic review (comparative studies)	ED, SOUs	To conduct a systematic review of how SOUs affect the efficiency of health-care delivery and the quality of services provided	Reported potential benefits of SOUs are reduced length of stay, improved ED efficiency and reduced cost, but there was variation between studies regarding reported benefits Length of stay: one study reported reduced LOS for patients treated with a ‘rapid’ chest pain protocol in a chest pain unit to exclude the diagnosis of myocardial ischaemia. Average LOS for ‘rapid’ protocol patients was 15.4 hours (± 12.2 hours) compared with 54.6 hours (± 12.6 hours) for routine care Recontacts: two papers examined the impact of SOUs on ED recontacts for patients with asthma. In one study, repeat ED visits within 72 hours increased from 3% to 5% after the introduction of an observation unit for asthma patients. The other reported no difference Cost: a cost comparison showed observation unit management for asthma patients over 7 days was significantly lower (mean US$1203 ± US$1344) than those admitted to inpatient care (mean, US$2247 ± US$1110). Costs were higher than standard care for patients who were subsequently hospitalised
Flores-Mateo, 2012208	Systematic review (n = 48)	ED	To systematically review the effectiveness of organisational interventions designed to reduce ED utilisation	Most studies were of poor methodological quality. Results were reported in the following categories: Interventions addressing the supply and accessibility of services (n = 25): 10 studies demonstrated that lower ED visits are associated with increased numbers of primary care centres or doctors. Of three studies concerned with hospital admission, two reported a decrease and one reported no change. Two out of nine studies reported less ED use after increasing OOH provision, but results for other studies reported less consistent results and some increase in use. Analysis of all studies showed that ED use is only reduced if primary care is provided by the same team Service demand interventions: Educational interventions (n = 6). Two RCTs found no differences in post-intervention ED use. Others reported specific interventions associated with lower ED use were monthly educational group meetings, health education, patient advice, counselling, use of care facilitators. Two studies found a significant reduction in the number of hospital admissions for the intervention groups Barrier interventions (gatekeeping/cost) (n = 17). Seven out of eight studies reported that ED cost sharing reduces ED utilisation, and the other reported a dose–response relationship between co-payment and ED visit reduction. No increases in hospitalisations or mortality rates were reported Overall, evidence suggests that ED cost-sharing and interventions to increase access to primary care are successful in terms of reducing ED visits. Other interventions had variable results
Georgiou, 2013209	Systematic review of quantitative studies	ED, CPOE systems. ED settings or ED outcomes	A systematic review of the quantitative literature related to the effect of computerised provider order entry systems in the ED	Twenty-two studies met the inclusion criteria. CPOE was associated with an increase in time spent on computers (up to 16.2% for nurses and 11.3% for physicians), but with no significant change in time spent on patient care. CPOE with decision support systems showed significant decreases in prescribing errors (ranging from 17 to 201 errors per 100 orders), potential adverse drug events (0.9 per 100 orders) and prescribing of excessive dosages (31% decrease for a targeted set of renal disease medications). There was a high possibility of publication bias towards studies with positive results reported
Harding, 2011210	Systematic review, no study design restrictions. Data were analysed using descriptive synthesis	ED, triage systems	What effect do different triage systems have on patient flow?	The authors identified that triage systems may improve quality of patient care across multiple settings through reduced waiting time, shorter waiting lists and reduced LOS. Some evidence that services providing triage and initial treatment have a positive effect on patient flow, but conflicting evidence on patient flow for systems providing triage without initial treatment Eight studies conducted in hospital EDs with a doctor at the triage desk in conjunction with standard nurse triage found some evidence of benefit for reducing LOS, number of patients who left without being seen and ED waiting time Three studies evaluated use of a multidisciplinary triage clinic and found over 20% of patients were discharged at triage and waiting times were reduced Half of the included studies were considered poor quality
Holden, 2011211	Critical review	EDs introducing Lean methods of working	To critically review 18 articles describing the implementation of Lean in 15 EDs within the USA, Australia and Canada	Lean is a process improvement approach designed to eliminate waste. Within ED this relates to patient flow and over-triage/unnecessary investigations. Patient care usually improved after implementation of Lean, with many EDs reporting decreases in LOS, waiting times, and proportion of patients leaving the ED without being seen. Few null or negative patient care effects were reported, and studies typically did not report patient quality or safety outcomes beyond patient satisfaction. The effects of Lean on employees were rarely discussed or measured systematically, but there were some indications of positive effects on employees and organisational culture. Success factors included employee involvement, management support and preparedness for change
Hoot and Aronsky, 2003212	PubMed search and synthesis	General ED	To identify and report on articles that study causes, effects or solutions of ED overcrowding	Causes of crowding were: non-urgent visits, frequent attenders, influenza season, inadequate staffing, inpatient boarding and hospital bed shortages Effects of crowding included: increased patient mortality, transport delays, treatment delays, ambulance diversion, patient elopement and financial effect Solutions of crowding included: additional personnel, observation units, hospital bed access, non-urgent referrals, ambulance diversion, destination control, crowding measures and queuing theory
Katz, 2012213	Systematic review	ED and post-ED treatment plans; urban, rural, multicentre, single centre	A systematic review on the clinical effectiveness of ED-based care co-ordination interventions	Of 23 included articles, 14 were RCTs and nine were quasi-experimental studies. The many different types and settings of EDs made it difficult to compare results. Most care co-ordination interventions have been implemented in single-centre EDs, with good results. Two-thirds of studies described interventions that were clinically effective in improving their primary outcome. A total of 19 studies developed post-ED plans and 12 were clinically effective in improving follow-up rates or reducing repeated ED visits. However, four studies found higher ED visit rates Of the four studies that used educational services for continuing care, two were effective. Of the two studies evaluating information transfer, one was effective
Lee, 2008214	Systematic review; acute ED presentations	ED	This review systematically evaluated the literature on ED consultations with other specialities	Twelve studies included, with little new research published since the mid-1990s. Eleven out of 12 studies were from the USA or Canada, so may not reflect EDs in other countries. Overall, four studies examined ED consultation proportions, six identified the rate of consultation for special populations of ED presentations and two examined interventions to improve consultations. Only two published studies on interventions to reduce or control ED consultations. Consultation rates varied from 20% to 40% for all patients, usually reported for patient subgroups and varied by setting. Limited research on how to improve the ED consultation process and the relationship with overcrowding
Paul, 2010215	Systematic review	ED, overcrowding	The aim of this systematic review was to highlight the contributions of simulation studies to our understanding of ED overcrowding and to explore how simulation can be better used as a tool to address crowding	Most studies aimed to identify potential improvements in costs and competition, efficiency, re-engineering and quality of service Studies examined resource-, process- and environment-related scenario Resource-related scenarios found: increased ED bed capacity did not reduce waiting times or LOS, just shifted delays holding areas for patients waiting for beds found positive results in terms of minutes saved per patient staff: nurse scheduling does not significantly impact on waiting times. Few studies looked at configuration of resources and equipment and most found there was no optimal solution. One study found that installing laboratory and radiographic facilities in the ED was equivalent to an additional nurse and doctor Process-related scenarios to reduce ED crowding showed introducing a fast-track path for low-urgency patients was effective at reducing waiting times without compromising care quality. Changes to triage protocols was also associated with time saved Environment-related scenarios focused on external causes of overcrowding, such as unavailable inpatient beds. Studies found that increasing the number of hospital beds to which ED patients can be admitted can reduce ED stay and costs Although simulation is useful for identifying areas for improvement, the practical use is limited
Porter, 2014216	Systematic review. Evaluation of innovation without comparator, or comparison of innovation with usual care	ED, adult patients; clinician-led rapid assessment teams to streamline patient handover and flow in the ED	The systematic review was undertaken to identify rapid assessment models used to expedite patient care and treatment in EDs and evidence about their effects	Studies reported that rapid assessment models reduced waiting time, length of stay and numbers of patients who left without being seen. Interventions tended to benefit all patients in the ED, not just target conditions, as reducing time in ED for low-acuity patients freed resources for patients with more urgent needs. Staff and patients reported positive reactions. Methodological and reporting quality of papers was mostly moderate or weak, so evidence about the effect of these interventions should be treated with caution
Roberts and Mays, 1998217	Systematic review	ED, patients with minor illness or injury	To assess the extent to which primary–secondary substitution is possible in emergency care	A total of 34 studies of a range of interventions were included. Variable methodological quality. Demand for ED may be reduced by increasing access to primary care or introducing barriers to ED care (e.g. charges), but cost-effectiveness of these approaches is unknown. An alternative, more cost-effective approach is to use primary care professionals within ED to treat patients with non-urgent needs. At the time of the review there was little evidence on impact and cost-effectiveness of telephone triage, minor injuries units and GP OOH co-operatives The results showed no evidence of any significant differences in patient satisfaction or health outcomes between GP and hospital doctor management of patients, and suggests that restricted hospital access carries low levels of risk and was acceptable to patients
Egleston, 1998218	Systematic review	ED, patients with minor illness or injury	A systematic review of the literature on the cost-effectiveness of alternative models of A&E care for the primary care attender	A total of 17 papers were included, 12 of which were from the USA. There was variation in study inclusion criteria, sample size, study length and the definition of a primary care patient. The studies considered different interventions so no conclusions could be drawn on the most cost-effective model of service provision. Summary of the evidence: interventions that divert patients from ED to primary care may be effective but not always reduce the number of patients with primary care complaints attending ED primary care provision within ED may increase patient satisfaction, particularly if waiting times are reduced the introduction of a co-payment scheme reduced primary care ED attendances by 14.6% patient education by itself does not produce a measurable reduction in primary care ED attendances the unit cost of treating a primary care patient in ED is similar to treating that patient in primary care, and providing GPs within EDs could potentially reduce costs
Rowe, 2011219	Systematic review	ED, adult only (> 17 years) or mixed (i.e. child and adult) populations	A systematic review to examine the effectiveness of TLPs on mitigating the effects of ED overcrowding	Twenty-eight studies were included: 13 journal publications, 12 abstracts and three web-based articles. Most studies employed before-and-after designs; 23 of the 28 studies were considered to be of weak quality. Pooling of data from two RCTs found TLPs resulted in shorter ED LOS compared with nurse-led triage (WMD –36.85 minutes, 95% CI –51.11 to –22.58 minutes). One RCT showed a significant reduction in the time to physician assessment associated to TLP presence (WMD –30.00 minutes, 95% CI –56.91 to –3.09 minutes); the other RCT showed no change in LWBS rates (RR 0.82, 95% CI 0.67 to 1.00)
Rowe, 2011220	Systematic review	ED, adult only (> 17 years) or mixed (i.e. child and adult) populations	A systematic review to examine the effectiveness of TNO on ED overcrowding, ED LOS and time to physician initial assessment	Fourteen, mostly single-centre ED, studies included. Most studies had methodological issues. The evidence showed TNO appeared effective at reducing ED LOS, particularly for cases of injury or possible fractures. For injured patients with suspected fracture, three RCTs indicated TNO interventions produced a significant reduction in ED LOS (pooled mean difference –19.7 minutes, 95% CI –37.5 to –1.9 minutes), but result has high heterogeneity (I2 = 92%) TNO interventions comparing emergency physician and emergency NP ordering radiography found a significant reduction in ED LOS (mean difference –37.2 minutes, 95% CI –44.1 to –30.3 minutes)

TABLE 20 - Summary of systematic reviews on ED workforce

NP, nurse practitioner; PA, physician assistant.

Author, year	Study design	Population/setting	Purpose	Main findings
Carter and Chochinov, 2007221	Systematic review (MEDLINE and CINAHL)	ED; NP	To assess the use of NPs in the ED setting in relation to waiting times; patient satisfaction; quality of care and cost-effectiveness	Quality of care: doctors and NPs were equally competent and accuracy was associated with experience regardless of profession Patient satisfaction: consistently high for both NPs and physicians, but often higher for NPs. NPs explained procedures better than physicians (14.3% for NPs and 6.1% for residents) Waiting times: introducing a NP in a minor injury unit, ED or free-standing unit associated with reduced waiting times Cost-effectiveness: costs of NPs are higher when compared with resident physician costs, but there are no data comparing the cost of NPs with hiring additional medical professionals
Wilson, 2009222	Systematic review	ED; patients aged > 16 years with minor injuries	A systematic review to determine the clinical effectiveness of ED NPs in assessment, treatment and management of minor injuries	Nine studies included in the review: five experimental studies and four descriptive surveys No significant differences in terms of significant errors in care or number of patients followed up for emergency NPs compared with junior doctors or standard care (two RCTs, n = 1652). There was no evidence of statistical heterogeneity for these analyses. The review findings support the use of NPs for management and treatment of minor injuries in adults, but more high-quality studies are needed
Doan, 2011223	Systematic review	EDs employing physician assistants	To describe the role and impact of PAs in the ED	A total of 66 studies, mainly of limited methodological quality, were included. More academic medical centres report using PAs in the ED (65–68% vs. 13–18%). Most of the evidence shows that PAs are able to assess some medical complaints and perform procedures, and are accepted by both ED staff and patients. Evidence is limited as to whether PAs improve ED flow or are cost-effective

TABLE 21 - Summary of systematic reviews on management of frail elderly in EDs

CI, confidence interval; CGA, comprehensive geriatric assessment; RR, relative risk.

Author, year	Study design	Population/setting	Purpose	Main findings
Conroy, 2011224	Systematic review of RCTs	Acute hospital setting, discharge of frail older (> 65 years of age) patients, rapid discharge (< 72 hours)	To review the evidence on the impact of CGA on outcomes for frail older patients attending ED	There were only five trials of sufficient quality to include. There was no clear evidence of benefit for CGA interventions for mortality (RR 0.92, 95% CI 0.55 to 1.52), readmissions (RR 0.95, 95% CI 0.83 to 1.08) or for subsequent institutionalisation, functional ability, quality of life or cognition. Study quality was poor and more research is recommended
Graf, 2011225	Systematic review (MEDLINE)	ED, geriatric	To review the literature on the use and value of CGA in ED for evaluation of older patients and its influence on adverse outcomes	Eight controlled or matched studies using CGA efficiency and 14 on screening tools. Use of CGA in ED decreases functional decline, ED readmission and potentially nursing home admission in high-risk patients. Validated screening tools can be applied to detect high-risk patients who will benefit most from CGA There is no evidence that undertaking CGA in ED had any effect on mortality. It is more efficient than age-based screening and, if it is followed by appropriate interventions, will lead to improvements in patient outcomes

Summary of findings

Quality

Quality issues were generally the same as those discussed in the section on quality in the earlier section in this chapter on GP services in ED. Overall, the quality of the available evidence is generally poor, with a paucity of RCTs. Only one of the 22 systematic reviews we have included utilised solely RCTs in the review;224 another included 14 RCTs in a review of post discharge interventions, but there was a broad range of interventions and settings so general conclusions were difficult to form. 212 Variable metrics used to assess performance precluded meta-analysis in most of these reviews. As previously, there was a high reliance on before-and-after studies and the authors of the reviews included here identified a similar range of limitations as previously discussed, such as poor study design and small sample sizes. Very few studies considered the cost consequences and cost-effectiveness of interventions, and inclusions were dominated by single-centre studies with consequent implications for generalisability; none considered whole-system effects and the impact they may have on the success or failure of particular interventions. Four of the systematic reviews we have included are now quite old (two published in 1998217,218 and two in 2003206,207). Delivery of emergency and urgent care continues to develop and operates in a rapidly changing environment so, although not without some value, at least some of the studies included in these reviews and the inferences drawn from the results may have limited relevance to current service provision.

We have also conducted a simple quality assessment of the 22 systematic reviews included in this section,204–225 using four basic methodological criteria: adequate search, assessment of risk of bias in included studies, use of an appropriate method to synthesise the studies (e.g. meta-analysis or narrative synthesis) and whether or not the authors’ conclusions reflected the evidence presented. Critical appraisals from the Database of Abstracts of Reviews of Effects (DARE) were used, if available. For other reviews, quality was assessed by one researcher, using the full text of the review. The results are given in Table 22.

Overall, the reviews performed reasonably well against the four criteria. All except two of the reviews212,225 had an adequate search scope (covering at least two databases and/or other sources). These two reviews only searched MEDLINE or PubMed, therefore they may not have located all relevant studies.

Risk of bias of included studies was assessed using recognised checklists or scales in 13204,205,208,210,213,214,216,217,219,220,222,224 out of 22 reviews. 204–225 In the remainder, risk of bias was either not assessed or no results were reported. In the absence of a risk of bias assessment, it is more difficult to assess the strength of the evidence underlying the review’s conclusions.

Seventeen reviews204,205,207–213,215–220,222,224 were judged to have used an appropriate method to synthesise the studies. In most cases a narrative approach was used because the included studies were too heterogeneous for meta-analysis. Reviews that provided a narrative discussion of the included studies without any description of methods of synthesis have been classified as ‘unclear’.

In most reviews (14204,205,208,209,211,214,216–221,223,224 out of 22204–225), the evidence presented was judged to support the authors’ conclusions. Reviews where the DARE appraisal considered that the authors’ conclusions were potentially unreliable or should be treated with caution have been scored as ‘unclear’.

Conclusions

This rapid review attempting to assess the evidence on different models of delivering ED services has been particularly challenging. The broad focus on a whole service means there are a whole range of potential areas to assess that are of pressing concern, including operational processes at different stages as patients progress through an ED and the issues associated with that (front-door assessment, flow through the department, exit block); workforce issues (who does what, when, how many); and the associations with the wider emergency care or hospital system that impact on performance. The more focused review on co-location of primary care within the ED did identify a relevant and manageable set of existing systematic reviews and primary research studies to allow some synthesis of evidence. Practical constraints meant that for the second part we have only been able to conduct a ‘review of reviews’ but, given the extensive evidence available, we have probably not done this topic area justice and there will be many relevant aspects that we have not considered at all. In addition, given the complexity, it has been difficult to identify clear and unambiguous evidence gaps. However, the review has highlighted some areas that are worth further consideration and can guide future research in this area:

• In the process of conducting these rapid reviews we have conducted extensive searches and have created a substantial library of relevant research articles. Given the complexity of the topic area, more focused reviews could be carried out utilising the work already done here with, where necessary, targeted focused additional searches. Because of the complexity and because of the rapidly changing environment and pressure on services to implement change, there are already policy and guidance documents from a range of sources, such as the Royal College of Emergency Medicine, setting out interventions that should be implemented,226 many of which may not have a solid evidence base. Before further, more focused reviews are carried out, it may be prudent to consult with the Royal College of Emergency Medicine and identify which specific topic areas would benefit from a more detailed analysis.

• Although management of the frail elderly is seen as a key area for development9 there is remarkably little research on interventions to improve care for this population. There is scope to identify more recent primary studies from our existing search library to supplement the two existing systematic reviews, as both were published in 2011, and to further review this important area.

• One clear evidence gap identified from the research we have appraised here is lack of studies that take a broader system approach to identifying, implementing and evaluating interventions to try to improve emergency and urgent care. This has important implications for future service design and planning. Future ED-focused studies could be broadened out to encompass these wider system issues. If this is ignored there is always a risk that some specific interventions may appear to be inefficient or not deliver intended benefits at an individual service level and be abandoned when in fact the benefits may be being achieved elsewhere or some external system issue may be inhibiting an intervention from working optimally.

Introduction

The final element of the NHS England review6 is focused on the development of broad emergency and urgent care networks with the aim of connecting services together as a system. The vision for these networks is that they will reduce boundaries between hospital- and community-based services and support information flow and expertise across sectors so that patient care can be delivered in the most appropriate and convenient setting. The review initially envisaged that major emergency centres would take the lead in this development, but NHS England has subsequently outlined a wider model based on a combination of strategic and operational functions with no specification on lead (www.england.nhs.uk/2014/05/30/jonathan-benger-2/).

The emphasis on providing emergency and urgent care has seen a shift towards a consensus that a whole-system approach is needed, and emergency and urgent care networks are seen as being central to the achievement of system-wide objectives. In this review, we have attempted to identify whether or not there is any empirical evidence on the organisation, function and effectiveness of emergency and urgent care networks in improving the delivery of emergency and urgent care.

Methods

The methods and search strategy have been described in detail in Chapter 2. Searches were conducted from 1 January 1995 to 21 January 2015. For this review the specific inclusion and exclusion criteria were as follows.

Exclusion criteria

• Descriptive studies with no assessment of effect.

The results of the searches are given in Figure 7.

FIGURE 7.

A PRISMA flow diagram for emergency and urgent care networks searches.

Results

We found no relevant papers that fulfilled the inclusion criteria; that is, that provided a description of an operating emergency and urgent care network model with some empirical evidence of how it performed and produced an impact on health-care delivery. There were 21 papers that were related to networks and the delivery of health care, but these were predominantly related to the following study types:

1. Descriptive studies of clinical networks for specific conditions or patient groups, but with no assessment of effect. For example, a clinical network to provide care pathways for aboriginal and Torres Straights populations with acute coronary syndromes,227 a model of a trauma network in the USA228 and, perhaps more relevant, a network for integrating ambulance services and acute hospitals in the Netherlands. 229 The description for this network included a statement that government policy had been put in place to monitor development, but no outputs have been identified.

2. Studies reporting data on activity within clinical networks for specific conditions such as stroke,230,231 trauma,232,233 primary or community care234,235 and ST elevation myocardial infarction. 236 Most of these studies claimed that clinical networks had improved care by, for example, increasing numbers of patients admitted to specialist facilities,230,232 receiving thrombolysis for stroke231 and improving mortality from trauma,233 but all of them reported current activity and had not made any assessment of differences in care compared with before the network became operational or, therefore, whether the network had improved care.

Perhaps more useful in terms of providing insights into factors that aid or inhibit network function, and so provide lessons for emergency and urgent care network development, are the small number of qualitative studies that have examined this area in more detail. One study of health-care professionals’ perceptions about the functioning of a gerontological network highlighted three key characteristics: (1) the central role of GPs in co-ordinating care, (2) geriatrician’s intervention in assessment of older patients’ needs and (3) the interface between hospitals, GPs and homecare. Lack of collaboration hindered network function. 237 A UK study of the network governance lessons from a programme aimed at improving care for older people and reducing emergency bed-days identified some helpful lessons. 238 The study found features associated with effective network governance included the selection of a small number of evidence-based services, co-ordination by a network-based strategic group with clear responsibility for implementation at operational level and a ‘joined at the top’ implementation group.

There were a small number of papers that did have an emergency and urgent care network focus but were of limited value in terms of providing evidence on successful and effective network models. One UK paper describes attempts to develop an early emergency care system in Surrey,239 although this is now old, the NHS environment has changed substantially since then and there appear to be no subsequent outputs demonstrating success or further development. Two papers emerged from a US consensus conference on integrated networks of emergency care. One sets out a conceptual framework for how categorise, designate and regionalise services within a network and identifies a range of research opportunities needed to support this process, including contributing to design, processes, utility and impact on patient outcomes. 240 The other sets out a research strategy for developing methods for defining and measuring success in emergency care networks and the infrastructure needed to support it, including identifying key metrics, scaling, network data systems and data linkage. 241 These two papers set out an ‘aspirational’ view of future emergency care networks and there may be some value in considering the conceptual approaches they have described when thinking about network development; however, they do not provide any information on how these aspirations might be operationalised into a functioning network.

Conclusions

We have found no evidence base on how to best organise and operate an emergency and urgent care network with associated empirical evidence to demonstrate the effectiveness of a network model in managing and improving delivery of care. There is some related evidence from other types of clinical networks but these are very focused in their remit. While they may provide some useful insight for network development, strategies that work in improving care for specific conditions or groups of patients with clear and well-developed care pathways may not be transferable to the complex, broad population-based, whole-system-based focus of emergency and urgent care networks. Literature on networks within the specific topic area of emergency and urgent care is scant and confined to descriptive articles or ‘think pieces’ about the conceptual and aspirational frameworks for network-based models rather than examples of functioning network models. We conducted some descriptive research on this topic for the Department of Health in 2007, which described a small number of embryonic emergency and urgent care networks in existence at the time,242 but which subsequently disbanded or changed with multiple reorganisations. This report did suggest some principles for network development based on the experiences of the early study sites, which have been translated into the NHS England suggested model, but we did not measure network effectiveness.

Given that emergency and urgent care networks are a key component of the delivery of the NHS England review, there is remarkably little evidence and information available to inform this development. Anecdotally, health economies are building networks, but there is no information available at a national level about the structure and operating models of existing networks, nor indeed about how many there are. It is likely that networks are developing in isolation, which limits opportunities for sharing experiences about network model design, development and implementation.

There are some key and pressing research activities that could support emergency and urgent care-network development:

• A more detailed and targeted rapid review could be undertaken to further explore the more theoretical literature about network model development and implementation to supplement the small descriptive evidence we have identified. More detailed synthesis could provide evidence on central components and strategies for successful network development that could be used to underpin network development now taking place.

• Of course, it may be the case that this additional evidence review would come too late and that nationally network development is already under way, although there may still be some value in modifying existing models. There is a paucity of information about activity in an area viewed as being of such strategic importance in terms of delivering a whole-system approach to emergency and urgent care. Some rapid scoping research to identify and map current emergency and urgent network development nationally could provide useful information about numbers, size, models, state of maturity, etc. This could be supplemented by more qualitative work to identify the more detailed issues associated with developing and implementing a network model, including challenges, barriers and facilitators, which can then be shared.

• A programme of research to evaluate emergency and urgent care progress using case study sites identified from the work suggested above. This would need to be a long-term strategy involving comparing different models of networks to assess processes, operation and management and, ultimately, to measure effectiveness and the extent to which different models change and improve the delivery of urgent care and achieve the objectives set out in the NHS England review plan.

• Information will be a central component of network function. Chapter 3 set out the importance of whole-system information systems that will allow better understanding of demand and inform planning of services needed to meet that demand. Chapters 4, 5 and 6 have highlighted the need to evaluate service developments in terms of whole-system as well as individual service impact. Development of robust and linked patient-based data systems will be needed to support these activities. Research to better understand the information needs of emergency and urgent care systems and networks and map this onto existing information-system development by other parts of the NHS, for example the Academic Health Science Networks, the Health & Social Care Information Service, NHS England and local initiatives, could inform the development of a coherent national plan for network information-system development.

Summary of main findings

We have conducted five separate rapid reviews spanning a range of themes relevant to the NHS England review of emergency and urgent care. 6 A comprehensive discussion and conclusions are provided at the end of each review chapter. The main findings from each review are briefly summarised here.

Common themes across subject areas

We have conducted five separate reviews focused on specific areas of service delivery, and the synthesis of evidence for each individual review revealed a number of common themes that spanned one or more topic areas. This fits with the view that delivery of emergency and urgent care needs to be viewed from a whole-system perspective and highlights the clear interdependencies between the different system components that contribute to efficiency and effectiveness.

Although the theme on understanding demand for urgent and emergency care is not a key theme in the NHS England review, we included it in this rapid review to provide context for the underlying issues that are contributing to pressure in the emergency and urgent care system and hence the need for service change. The issues and evidence gaps identified from the analysis of evidence on trends in and characteristics of demand were relevant to several subject areas. The most highly relevant was emergency and urgent care networks. The envisaged primary function of a network is to design and plan an emergency and urgent care system that can meet the needs of a heterogeneous and changing population with a range of complex health problems ranging from acute and life-threatening to minor and chronic illness and injury. To do this, a detailed understanding of the profile of the population accessing care, the different types of care they will need and the relative proportion of people requiring different types of service within the system is needed. The identified need to develop more sophisticated models that map demand and the needs of local populations and can begin to estimate how this will change as population characteristics change has direct relevance to the potential successful functioning of an emergency and urgent care network.

At an individual service level this also has implications for the success of the expected changes in delivery of care. For example, clear estimates and thresholds for the likely proportion of patients who could be safely managed in the community by the ambulance service would inform ambulance service resource and workforce planning. Similarly, initiatives in the ED to better manage specific patient groups require information on the likely volumes of patients who, for example, will be suitable for GP care in a co-located primary care unit or a specialist major trauma or frail elderly care service.

Limitations

The scope of this rapid review and the volume of evidence included mean we have not been able to conduct a detailed analysis of the quality of the evidence base; however, we have provided a broad assessment and commentary of relevant issues in each of the separate review chapters. We have included a substantial number of systematic reviews within this review and have drawn on the findings of these studies to identify significant methodological issues. In addition to the limitations regarding lack of scale and measurement of system impact already discussed, some key themes have been identified that are common across the reviews:

• Overall, the evidence base on clinical effectiveness and cost-effectiveness of different models of care for delivering emergency and urgent care is weak. Only a small number of interventions have been evaluated using RCT designs. It is acknowledged, however, that this is a difficult environment in which to conduct this type of study.

• There is a high reliance on uncontrolled before-and-after studies within the overall evidence base, with consequent serious limitations on interpretation of findings. The majority of studies were single-site and small in scale, with small sample sizes. This makes generalisability difficult, particularly in terms of relevance to different health systems. There is a lack of population-based studies to inform development of emergency and urgent care systems and services.

• The majority of studies have measured processes, such as times, attendance or contact rates, re-admission and recontact rates, to assess effectiveness. With a few exceptions, patient outcome measures have been confined to patient satisfaction, and there has been little consideration of important patient outcome measures concerned with safety, such as adverse events. The reviews on telephone triage and management of patients by paramedics in the community did identify studies that included patient safety.

• Little attention has been paid to the costs and cost-effectiveness associated with interventions.

There are some limitations to the rapid review method we have used. First, we have not been able to exhaustively search for and synthesise all the relevant literature for each of the five themes. We have drawn extensively on existing systematic reviews and only extracted detailed information on primary studies not included in these reviews. This has allowed us to provide a comprehensive overview of the existing evidence base for each theme. However, this was a particular limitation for the review of models of care for delivering ED services. For this theme we only included systematic reviews and therefore have not been able to confidently identify the related evidence gaps. This theme warrants additional, more focused reviews.

Second, emergency and urgent care encompasses a broad patient population, range of services and related factors. Given the potential scope and scale, we have used a framework to guide the reviews, based on the key policy objectives for emergency and urgent care set out in the related NHS England review,6 in order to provide focus and relevance. This framework also guided the systematic search strategies tailored to each individual review but, given the scope, we have further restricted the focus of two themes (helping people get the right advice and managing more urgent care outside hospital) to the specific areas of telephone-based triage and advice and management of patients by paramedics in the community, respectively. This does mean that the broader literature on community-based urgent care services has not been addressed.

Third, although using the framework provided structure that enabled us to synthesise a large body of evidence in a short time frame, it does mean that there are related themes and subject areas that have been excluded from this review. The most obvious gap is a separate review of models of urgent care within primary care. There is significant overlap within the existing reviews, for example, telephone triage, management of patients by paramedics within the community setting and provision of primary care services within the ED. However, there may be more specific evidence we have not considered. There is scope to conduct an additional review within this area utilising the search libraries we have generated for this review. We have also not considered the specific but important issues related to the emergency and urgent care workforce, such as education, training, skills, recruitment and retention. This may also warrant an additional review, although it would be large in scope as there is a wide range of professional groups to consider.

Finally, the short time frame and volume of included material meant we were not able to include any patient and public involvement input in this review. However, this perspective will be important when considering future research commissioning. We have identified a substantial number of evidence gaps which will need to be prioritised, and patient and public involvement assessment of those gaps and questions will be of value in understanding which ones are most important to patients and the public and which need most urgently addressing.

Conclusions

We have conducted five separate rapid evidence reviews on themes related to the delivery of emergency and urgent care in the NHS. These themes were (1) trends in and characteristics of demand; (2) telephone triage and advice; (3) management of patients in the community by ambulance clinicians; (4) models of service delivery in the ED; and (5) emergency and urgent care networks.

We have found there is a paucity of evidence to explain the complex reasons that have driven the increases in demand for emergency and urgent care and the detailed characteristics and consequent service needs of the population. There exists a considerable evidence base on the effectiveness of some interventions for improving service delivery, but overall the evidence base is weak and based in small single-site studies, with no assessment of impact at scale or on the wider emergency and urgent care system. A small number of interventions have been robustly evaluated. The delivery of ED care is complex and multidimensional and it was beyond the scope of this review to identify clear evidence gaps. Furthermore, more focused evidence reviews are needed in this area. There is no evidence to support the development of emergency and urgent care networks.

We have identified a number of evidence gaps across the reviews. The gaps that appear to be in most immediate need of addressing are detailed below:

• Research to characterise and map demand at a population level and link this to service need so that emergency and urgent care systems can be designed that can effectively, efficiently and safely respond to patient needs.

• An assessment of the current state of play in the development of emergency and urgent care networks and longer-term evaluation of the effectiveness of different network models to identify how best networks can deliver NHS England objectives.

• Expanding the current evidence base on existing interventions that are viewed as central to delivery of the NHS England plan, such as increasing the number of patients managed by paramedics in the community, by developing integrated care pathways and assessing the implications of increasing the intervention at scale. Measurement of the costs and system impact will be critical to the assessment of effectiveness and resource requirements needed to support expansion.

Although not an evidence gap, a clear theme that emerged across the reviews was the need for robust, high-quality and linked patient data to support service planning, measure whole-system benefits and costs and support research. This may need to be addressed alongside any newly commissioned research studies.

There is also a need to plan and design better-quality research studies. Some of the problems associated with existing research studies have been highlighted above. Future research will need to address these problems. The broad range of questions evident from the research gaps means a range of designs will be needed to adequately answer these questions. It is not possible to be prescriptive, as each question will need a different approach, but there is a clear need to move away from uncontrolled studies toward better use of quasi-experimental observational studies that do allow comparison against current practice. Evidence gaps concerned with scaling-up interventions would benefit from mixed-methods approaches that can consider a range of associated issues. Clinical effectiveness may have been demonstrated, but translating this into common practice needs consideration of both the population (quantitative) effects across multiple sites or whole systems and, as importantly, the ‘real-world’ processes that effect change and mediate successful adoption of new ways of delivering services. Valuable knowledge can be gained from more qualitative studies that can identify the factors and processes that contribute to success or failure of service change. This is linked to the clear lack of consideration of workforce implications apparent in the existing evidence. Future studies, particularly those concerned with scaling up, cannot ignore this component, and specific studies providing a primary assessment of workforce and skill mix requirements may need to be a precursor to any large scale expansion of interventions. Existing studies have also been limited by only measuring short-term process and outcome changes and future studies could make better use of, for example, interrupted time series designs to provide a more robust assessment of the impact of service changes in the longer term. Robust economic evaluations should also be conducted alongside any changes to assess cost-effectiveness, as well as the broader cost shifting that occurs around an emergency care system when large-scale service delivery changes are implemented.

There are significant problems within the emergency and urgent care system in the NHS. In consequence, there is pressure to instigate change and services are operating in a rapidly changing environment. Specific programmes, such as the NHS England review delivery plan,8 refer to a number of existing pilot schemes and interventions under development. There are also broader programmes across the NHS that may also have an impact on emergency and urgent care, such as the move to 7-day working, the Prime Ministers Challenge Fund to improve access to primary care (URL: www.england.nhs.uk/ourwork/qual-clin-lead/calltoaction/pm-ext-access/) and vanguard sites for new models of care deliver (URL: www.england.nhs.uk/ourwork/futurenhs/5yfv-ch3/new-care-models/ and local initiatives). These programmes mean that there are already potentially large numbers of relevant initiatives already under way or in development, but it is difficult to identify the range and scope of existing projects. Before new research to address evidence gaps is commissioned, it may be worth considering commissioning a preliminary project to identify and map existing plans and initiatives to distinguish if any evaluation research is already under way, whether or not the planned evaluation is robust and whether or not there are potentially suitable initiatives that are at development stage but have no evaluation plan in place. This could reduce duplication of effort and help ensure new models of care are rigorously assessed before they are adopted on a large scale.

Contribution of authors

Ms Janette Turner (Senior Research Fellow) undertook reference sifting and evidence synthesis for the review on demand for urgent care; undertook reference sifting, data extraction into summary tables and evidence synthesis for the review on emergency and urgent care networks; contributed to writing the report, including the discussion and conclusions; and managed the project.

Mrs Joanne Coster (Research Fellow) undertook reference sifting, data extraction into summary tables, evidence synthesis and contributed to writing the report for the review on telephone triage.

Dr Duncan Chambers (Research Fellow) undertook reference sifting, data extraction into summary tables, evidence synthesis and contributed to writing the report for the review on management of patients by ambulance clinicians in the community.

Mrs Anna Cantrell (Information Specialist) contributed to the proposal writing, designed and ran the literature searches, and contributed to writing the report.

Mr Viet-Hai Phung (Research Assistant) undertook reference sifting, data extraction into summary tables and evidence synthesis, and contributed to writing the report for the review on models of care in the ED.

Dr Emma Knowles (Research Fellow) undertook reference sifting for the review on management of patients by ambulance clinicians in the community.

Mr Daniel Bradbury (Medical Student) undertook data extraction into summary tables for the review on demand for urgent care.

Professor Elizabeth Goyder (Professor of Public Health) contributed to the proposal writing.

Data sharing statement

No new data have been created.

Disclaimers

This report presents independent research funded by the National Institute for Health 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, NETSCC, the HS&DR programme or the Department of Health. 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, NETSCC, the HS&DR programme or the Department of Health.

General search strategy

Targeted search strategies