Alternatives to direct emergency department conveyance of ambulance patients: a scoping review of the evidence

We used initial pilot searching in two databases to aid in defining effective search terms (JB, DR, additional librarian support). This iterative process provided confidence that the search strategy was adequately capturing all relevant studies. The resulting search strategy remained purposely broad and thus the numbers of studies captured by the search was expected to be large. The initial search strategy was conducted in April 2016 but was updated in April 2017 and February 2020. We searched PubMed, CINAHL, Web of Science and ProQuest Health & Medicine databases. Two separate arms were combined using the Boolean operator “AND” to search for the “who” (paramedic or ambulance) and the “what” (referral or non-conveyance). The full search strategy can be found in Appendix; it was adapted as necessary for each additional database. To ensure all relevant studies were adequately captured, no exclusion filters were applied for language, year (e.g. inception to February 2020) or article type. We also searched grey literature on the following sites: NHS Evidence, CORE, BL.UK, OpenGrey, and HMIC. Additional articles were found by a snowball search of the reference lists of relevant systematic reviews and of articles that met the inclusion criteria.

In the first stage of screening, two researchers independently screened the title and abstracts of each article for relevancy (JB, DR, EP). The second stage of screening involved retrieval and evaluation of the full text to identify if it met the criteria. Again, two of these three researchers independently screened the full text of all potential studies. Reviewers discussed any discrepancies and where necessary, a third reviewer made the final decision. Reason for exclusion in the full text stage was noted.

Inclusion criteria purposely remained broad to capture relevant studies. Articles were included if they considered emergency service callers (e.g. 999, 911, etc.) who were triaged by the on-scene clinician as non-urgent “intermediate” patients and if there was an indication of an alternate route of care to conveyance to ED. “Intermediate” patients were those with non-urgent medical illnesses or injuries who did not require conveyance to the ED, yet could not be safely left at home without medical support [13]. Emergency department is defined as a medical facility responsible for treatment of patients who arrive at the hospital and require immediate medical care. Undifferentiated “intermediate” patients who did not fit within an existing evidence-based pathway were included, while samples of specific clinical patients with pre-existing evidence-based pathways were excluded (e.g. falls, resolved hypoglycaemia, psychiatric and resolved epilepsy pathways). Commentaries, protocols and policy statements were eligible for inclusion, while literature reviews, conference abstract, non-English and non-peer reviewed articles were excluded from final article selection. Due to the expected heterogeneity of studies, there were no limitations on study outcomes. The following software were used in study selection and screening: Mendeley, Microsoft Excel and Rayyan.

JB independently extracted data from all studies using a standardised form that was agreed upon and piloted by the research team (JB, DR, EP). This included data on country, study type, sample size, description of alternate route of care, triage protocol, study outcomes and study findings. A meta-analysis was not possible due to expected heterogeneity of included studies, and a narrative synthesis was conducted instead. The heterogeneity in design, methods and outcomes of studies also facilitated three groups for synthesis: 1) quantitative patient-focused, evidence-based studies; 2) qualitative, evidence-based studies; and 3) consensus-based articles.

In line with the second objective, two authors (JB, EP) independently appraised the quality of evidence following a modified 7-level rating system for the hierarchy of evidence [24]. This tool provides a hierarchy of the likely best evidence (e.g. Levels 1 to 7) and is specifically designed to aid clinicians (and patients) with a rapid appraisal to avoid the need to resort to original sources. Differences in levels of evidence between authors were discussed and agreed upon in a consultative process.

Eighteen studies with patient-focused, quantitative evidence are outlined in Table 1. Seven studies were based in the UK, four in both the USA and Sweden and one in each of Canada, Australia and the Netherlands. Only five studies compared an intervention group (e.g. option to refer to alternate care route) and a control group (e.g. normal practice), with two of these studies randomising into either arm [37, 41]. Retrospective cohorts, including data audits, were the most common study design (n = 8), followed by prospective cohorts (n = 7) and randomised control trials (RCTs) (n = 2). One study used a comprehensive mixed-methods approach of interviews, qualitative telephone data and linked retrospective data [32]. Most studies (n = 14) outlined a triage protocol to guide the ambulance clinician’s decision making; this included six studies with triage tools that led directly to alternate care route outcomes, five studies with a series of protocols for specific incidents and three studies that allowed subjective referral of patients triaged as low acuity using traditional triage tools. Studies commonly detailed multiple alternative care routes that could be accessed through either ambulance transport or referral. Referrals to primary care, including general practitioners (GPs) or nurses, were most frequent (n = 16), although options such as urgent care centres (n = 6), psychiatric or social teams (n = 4) and minor injury units (n = 2) were also common.

We identified two main themes: patient safety and impact on operational efficiency. Most studies suggested that paramedics were able to accurately triage patients to the correct pathway of care [31, 32, 35‐37, 40, 41]. For example, several studies compared decisions between paramedics, first responders (e.g. fire departments) and senior medical experts reporting high sensitivity (> 94%) and lower specificity (< 58%) [31, 35, 36]. Rates of secondary contact with emergency medical services (e.g. calls, transport, ED presentation) were reported in six studies (Fig. 2). Secondary presentation at ED occurred in 5.4–22.4% of all patients who were referred to an alternate care route (range: 48 h to 30 days) [25, 28‐30, 38, 41]; approximately half were subsequently admitted [25, 29, 30]. Only two studies compared rates in intervention and control group, reporting no difference in secondary ED presentation (7.9% vs 8.0%) [28] or admission rates (5.4% vs 6.2%) [38]. Not all studies reported that paramedics could correctly triage patients. For example, one study reported that 53.6% of patients identified by paramedics as eligible for alternative routes of care were subsequently admitted to hospital and were also more likely to experience adverse events than those who conveyed directly to ED [39]. Other studies highlighted concerns of under triage as shown by high secondary recontact rates, overturned decisions by medical experts (19.5–22%) [31, 42] and occurrence of critical events in those incorrectly triaged away from ED (3–11%) [35, 36]. Psychiatric presentations and patients living with dementia were considered at particular risk of under triage [36].

Studies reported that alternative care routes improved operational efficiency by decreasing ambulance job cycle times [28, 29, 37], decreasing ED conveyance rates [26, 28, 32, 34, 40, 41], improving patient documentation [38], increasing clinic care destinations [34], decreasing hospital admissions [25, 28, 33] and improving or maintaining patient satisfaction [31, 34, 37, 38]. Nevertheless, there were conflicting findings that reported longer job cycle times [38] or showed that alternative care schemes had no impact on decreasing ED conveyance rates [37, 38].

Characteristics of the 12 qualitative studies are provided in Table 2. Study characteristics of the qualitative studies were similar to above, with a third of studies (n = 4) in each of the UK and Sweden, two studies in Ireland and one in each of Canada and the USA. As in the quantitative studies, the main alternate care routes comprised of primary care referrals, referral or transport to urgent or community care centres and transport to minor injury units. Data was primarily collected from interviews with paramedics (n = 6) or ambulance nurses (n = 2), patients (n = 3), emergency medical physicians (n = 2) or senior managers/commissioners of ambulance trusts (n = 2). There was unanimous support from all studies that schemes providing alternate routes of care than direct ED-conveyance can deliver vast benefits, although several barriers were recurrently identified.

The main barriers identified by paramedics to successful implementation were training, organisational support and process. First, the absence of training and knowledge on referral triage and processes [13, 43, 45, 47, 48, 52] often manifested as a lack of confidence in both themselves and the system [13, 48], with some paramedics expressing concerns about being held accountable or facing legal action for under triaging patients [47, 49]. Next, there were concerns that the absence of a mandate from their organisation led to a lack of responsibility to refer and, thus, underuse of the scheme [43, 45]. Organisational support from ambulance services and partnerships with primary or community care pathways were identified as crucial components of an effective scheme [45, 47, 48, 52]. Paramedics did not receive feedback on patient outcomes from their own organisation or the referral destination; they reflected that this was demotivating and a lost learning opportunity [43, 48]. Finally, although uncommon, paramedics expressed minor frustrations about the referral process itself [13], perceived time spent to refer [13, 43] and misconceptions from patients who wanted direct conveyance to ED [45, 52].

Consistent with views from paramedics, interviews with stakeholders identified the importance of established structures between organisations [44, 47] and sufficient training and knowledge [50] of paramedics. Power et al. [50] was the sole paper to contrast viewpoints between emergency medical consultants and paramedic towards “treat and referral” schemes. Emergency medical consultants more commonly reported that referral/transport to ED alternatives should be limited to advanced paramedics (57.2% vs 22.6%; p < 0.001), that paramedics should inform the patient’s GP after each referral/alternate transport (88.9% vs 47.5%) and were less optimistic that the scheme would improve ambulance availability (55.6% vs 83.9%) [50]. Finally, patients expressed both a willingness [46] and a preference [53] to be directed to an alternate scheme rather than conveyed directly to ED. They emphasised their desire to contribute to the decision-making process [53], and felt that their views were taken seriously [51, 53].

Six of the 10 consensus-based articles were based on the American EMS systems, three concerning UK systems and one on Australian systems. These articles were primarily opinion-based (n = 2 editorials, n = 2 viewpoints, n = 3 commentaries), but also included two policy statements, a scheme overview and one promising RCT study protocol. Details of all studies are provided in Table 3. Note that results from the study protocol are not publishable due to “unresolvable inconsistencies in data” [57]. The opinion-based pieces and the policy statements both highlighted the promising potential for such schemes, with particular emphasis on projected cost savings [54, 58, 62, 63]. A policy statement, written in 2001 by the Emergency Medical Services Committee and reaffirmed in 2008 by the American College of Emergency Physicians, highlighted seven key elements that should be considered when implementing alternative routes of care [59, 60] (see Table 3). Consistent with above, patient safety and accuracy of paramedic triage decisions were raised as main areas of concern [16, 58, 63, 64]. Sawyer et al. [64] was the sole article to recommend against implementation of such schemes citing concerns of insufficient supporting evidence, under triage having an adverse effect on patient safety and vulnerable patients being disproportionally affected.

Analysis of patient data and interviews with those involved in the care pathway (e.g. patient, ambulance clinician, stakeholder) suggested that successful schemes share four key features. First, clear triage tools are crucial in guiding accurate decision making of ambulance clinicians. Formal triage pathways can ensure that patients are referred or conveyed to the appropriate destination. For example, tools such as the UK-based Paramedic Pathfinder [31] or the Swedish-based Rapid Emergency Triage and Treatment System (RETTS) [28‐30] allow triage outcomes that direct patients to a specific alternative care route (e.g. GP referral, community care centre). Without a guiding framework, non-specific triage tools can lead to incorrect triage decisions [34, 39]. There is no universally accepted triage tool that could be applied everywhere as the availability of schemes differs by health care system, both within and across country. Additionally, geographical considerations play an important role as densely populated urban areas may be able to facilitate non-conveyance of patients better than rural areas with fewer resources. As such, formal triage tools must be specific to individual healthcare systems and provide clear support to guide the decision making of ambulance clinicians.

Second, additional training on correct use of alternative care schemes including triage tools, overview of processes and learned examples is necessary. Although higher skilled ambulance clinicians (e.g. advanced paramedics, ambulance nurses) may have improved decision-making processes, there was a persistent belief that paramedics of all skill levels should still be trained to appropriately use such schemes. Inadequate training likely explains why paramedics have perceived low confidence and hesitation to use these schemes [43, 48, 52].

Third, formal liaisons and partnerships between ambulance services, primary care, urgent care centres, minor injury units or psychiatric and social teams are crucial in facilitating referral or alternate transportation of patients. Without well-established pathways of care, ambulance clinicians are forced to rely on ad hoc decisions and, as a result, are often unsuccessful in finding an appropriate alternative source of care [32]. This must be a consideration when considering implementation of schemes in different countries or regionalised ambulance services, which may limit availability of alternate facilities. Finally, the most important and persistent recommendation in quantitative, qualitative and consensus-based studies was the need for adequate evidence demonstrating patient safety. It was commonly suggested that the current evidence was not sufficient to justify implementation of such schemes; this is described further below.

Research in this area is increasing, contributing to and reflecting positive developments in the paramedicine profession [65, 66]. Half of the studies were published within the last 5 years, with only one study coming before the turn of the century [61]. Expert opinions and viewpoints identify the importance of providing alternatives to direct ED conveyance, while qualitative studies are fundamental in describing the views of ambulance clinicians to understand how human factors can ensure optimal use of these schemes [67]. However, the quality of existing evidence is poor, particularly as it pertains to the most important outcome of patient safety.

The overall level of evidence is low, with only two studies using an RCT design (Level 2) [37, 41]. RCTs are essential to assess the impact of an intervention on patient outcomes [68]. Here, they enable alternative care schemes to be compared to ambulance services without alternative options and may provide a holistic overview of patient outcome differences. Conversely, nearly all relevant studies failed to formally assess whether the schemes are safe and instead, considered accuracy of conveyance decisions (e.g. between ambulance clinicians and expert medical consultants) [31, 35, 36, 49] or rates of secondary contact [25, 28‐30, 38, 39, 41] as proxies for patient safety. While accuracy of decision making may inform future outcomes, it is important to consider both short and long-term patient safety. Secondary contact rates may not provide a reliable indication of individual outcomes. For example, patients may contact emergency services for a reason unrelated to the original incident or recontact rates may be inaccurate if data from all possible health services (e.g. GP, ambulance, ED, etc.) are not obtained. Additional secondary contact with emergency services is common regardless of patient destination; only one study considered how recontact rates in a clear intervention group (e.g. with non-ED conveyance alternatives) compared to a control group, reporting there was no difference in secondary transport within 48 h (7.9% vs 8.0%) [28]. Failure of studies to appropriately consider the impact on the whole system has previously been highlighted [49].

It can be challenging to generalise findings across countries or between services in the same country, as the structure of the EMS system poses unique challenges. For example, there may be higher potential for alternative schemes in the publicly funded UK system which may allow better linkage within healthcare facilities, while privately funded and delivered health care systems, such as the USA, may encounter different challenges [19, 69, 70]. Similarly, Swedish ambulances are staffed with qualified nurses, which may provide them with a larger scope of practice in their decision making than emergency medical technicians in other countries [71, 72]. Most articles proposing cost efficiency of these schemes are US-based and consider specific regional EMS organisations and privatized primary care; this makes it difficult to draw conclusions for a federally funded healthcare system.

Finally, it is important to identify an acceptable under triage rate in non-ED conveyance situations [10]. High sensitivity of tools suggest that ambulance clinicians are able to accurately assess who needs to go to ED, while the lower specificity indicates that they are less able to identify who may benefit from an alternative care route [31, 35, 49]. Over conveyance to ED is preferable to a model that regularly under triages individuals to lower levels of care than is required, and as such, triage tools for non-ED alternatives may necessitate a high-sensitivity, low specificity approach [10]. Further evaluation and assessment of the safest level of under triage requires further investigation.

Due to the heterogeneity of the literature, the identification of relevant articles was challenging; despite a wide and inclusive search strategy, it is possible that relevant articles were not identified. Furthermore, due to ongoing innovation of the paramedicine field, we believe that there is a significant amount of discourse on alternate routes of care that has not been formally researched or published. It is important to disseminate results that have undergone formal peer-review to assure the highest quality of evidence and to help establish the evidence base. Many ambulance services work in isolation as they navigate and identify these alternative care pathways; communication between services both within and between countries is crucial for a collaborative effort to confront these issues. Details of triage tools and non-ED conveyance routes of care are insufficiently provided in study articles. Newton et al. [31] provide a detailed overview of referral routes, the accompanying triage tool and the accuracy of decision making in one UK ambulance trust; future studies should provide similar overviews, whilst expanding to assessment of patient outcomes.

As we aimed to identify non-ED conveyance alternatives that could be adapted to a wide range of intermediate patients, we excluded studies that provided specific clinical pathways; for example, community paramedicine pathways that provide proactive home visits, specific fall pathways or alcohol detoxification centres [73‐75]. However, synthesis of the effectiveness of these targeted schemes is needed. A combination of specific clinical alternatives along with general pathways for intermediate patients can help ensure the highest number of patients can be referred to the optimal level of care.

Finally, the variability in study design, scheme, protocol, outcome and sample population rendered it difficult to synthesize and summarize the evidence. While several studies reported similar outcomes including decision accuracy matrices and recontact rates, heterogeneity of these outcomes made us unable to consider a formal meta-analysis. Additionally, several studies grouped patients who were not immediately conveyed to ED together; thus, it was not possible to assess if referrals or transfers to alternate care routes provide a better safety net than self-care at home.

Given the widespread use of alternative schemes in the UK and Sweden, and increasing implementation in North American countries, it is crucial to commission large scale studies evaluate patient outcomes for those conveyed directly to ED, those left at home and those referred to alternate routes of care. Notwithstanding the impact of decreasing emergency department burden and cost effectiveness, patient safety must remain the most important outcome. Studies consider the entire patient journey, which may involve linkage of data from several emergency or primary care services. There must be improved collaboration between ambulance services within a single country and shared opportunities to learn from other countries. Given that schemes are linked to paramedic skill, training and education (e.g. research nurse, degree paramedic, emergency medical technician), the growing positive advancements in paramedic education in recent years may provide new opportunities and additional scope for these new patient pathways [65, 66].