Background

The past decades, ambulance care has evolved from a health care facility that conveys patients to the hospital, into emergency medical services (EMS) that provide advanced out-of-hospital care for (non-) life-threatening conditions [1, 2]. At the same time, the utilization of ambulance care has increased throughout the developed world, with various underlying reasons such as ageing of the population, changes in social support, accessibility and costs [3]. Together, these developments put a growing demand on ambulance systems and ambulance capacity, the emergency departments (ED) and the wider healthcare system, and this may compromise patient safety, healthcare quality, and access [3]. In addition to this growing demand, frequent overcrowding of the ED occurs [4, 5].

The ambulance process is situated within this context. This process often results in patient conveyance to an ED or other healthcare facility, but ambulance care can also result in patients not being conveyed. The NHS Litigation Authority (2012) defines conveyance as “the transfer of patients, medical and clinical personnel, equipment and associated records, as appropriate including from one healthcare facility to another as well as the initial journey from the scene.” [6]. Non-conveyance is defined as “an ambulance deployment as appropriate, where the patient after examination and/or treatment on-scene does not require conveyance with medical personnel and equipment to the healthcare facility” [7]. Non-conveyed patients can be treated and ‘discharged’ on-scene, or may be referred to other (primary) healthcare facilities such as the general practitioner. According to the literature, non-conveyance can be divided in two categories: the patient-initiated refusal and the ambulance professional decision [8]. Often, non-conveyance is a combination of these two categories.

Non-conveyance rates of patients who received on-scene emergency care from an ambulance emergency crew, have been reported up to 30% [9, 10]. On the other hand, it has been estimated that 11%–61% of the conveyances is medically not necessary [11]. Factors influencing these non-conveyance rates are patients with low-acuity problems or primary care problems who call an ambulance [12, 13], accuracy of triage systems at the EMS dispatch centre [14], and professional competencies [15].

The priority to conduct research on non-conveyance is reflected on the Dutch National Pre-hospital Research Agenda for EMS 2014–2018 [16]. From patient-safety and professional perspective, little is known about non-conveyance. Insight into characteristics and outcomes of the non-conveyance patients is lacking. Furthermore, it is unknown how often non-conveyance exactly occurs, which complaints non-conveyed patients have, what care is provided after non-conveyance, and how often these patients have adverse events. Conversely from the professional perspective, little is known about the on-scene non-conveyance decision-making process. As ambulance care has become a more complex environment, ambulance professionals are faced with decision-making over multiple care options as conveyance to an emergency department, or another non-emergency service, treat-and-release or referral to another healthcare professional [17]. Literature described that this decision-making process requires adequate competencies, skills and clinical reasoning of ambulance professionals [18], although ambulance professionals curricula include a little on conveyance decision making [19]. Also, few ambulance services developed non-conveyance protocols and policies [20]. However, the question is whether the literature describes guidelines, protocols or triage criteria to support the ambulance professionals in the decision making process for non-conveyance, how competent are they to decide and apply for non-conveyance, and how are they influenced during the decision making process for non-conveyance? These aspects of patient safety and ambulance professional perspectives related to non-conveyance in ambulance EMS have not yet been synthesized in an overview.

Aim

The aim of this systematic review is twofold. The first aim is safety orientated, as we want to describe non-conveyance rates, characteristics of patients, and follow-up care after non-conveyance. The second aim is formulated from the perspective of the ambulance professional, as we want to describe available guidelines or protocols and triage criteria, competencies needed by ambulance professionals to make appropriate (non-) conveyance decisions, and also to describe which factors influence ambulance professionals during the decision-making process.

Methods

Design

A systematic review of the literature was performed according to the steps of the Cochrane Handbook for Systematic Reviews of Interventions [21]. This review is reported in concordance with the Preferred Reporting Items for Systematic reviews and Meta-Analyses (Additional file 1: PRISMA) statement [22].

Search strategy

Firstly, the Cochrane database for systematic reviews and the DARE database were checked for a similar review (protocol). No review was identified, therefore systematic searches were performed in MEDLINE (EBSCO), PubMed, CINAHL (EBSCO), and EMBASE (OVID) in June 2016. Search strategies were developed to represent ‘terms for non-conveyance’ AND ‘terms for pre-hospital ambulance care’. Full search strategies per database are given in Additional file 2: Appendix 1. Searches were not restricted by year of publication. In addition to the electronic searches, after full-text inclusion we hand-searched reference lists to identify relevant studies.

Study selection procedure

We included all types of peer-reviewed systematic reviews, and quantitative or qualitative designs in real clinical practice or simulation situations, on non-conveyance. We defined non-conveyance as ‘the situation where an ambulance was dispatched and where the patient received on-scene diagnostics and/or treatment, followed by professional and/or patient initiated non-conveyance to the ED or another emergency care facility’. Studies were included when reporting on one or more of the following criteria:

  • Non-conveyance rates;

  • Characteristics of non-conveyed patients;

  • Follow-up care after non-conveyance;

  • Non-conveyance guidelines, protocols, or on-scene triage criteria;

  • Professional competencies needed to initiate non-conveyance;

  • Factors influencing the non-conveyance decision-making process.

Conference abstracts, narrative reviews, editorials, personal communications, or unpublished studies were excluded. All articles were screened on title and abstract by two independent reviewers (RE, SB, RS, NT, LV). In case of doubt, a third reviewer (SB, LV) was asked to make a final decision. The remaining articles were screened full text by two independent reviewers (RE, SB, RS, NT, JL, LV). In addition, reference lists of included articles were screened (RE, JL) and potentially relevant publications were screened in a similar way (RE, RS, NT, JL).

Quality assessment

To assess the risk of bias of (pre-, or quasi-) experimental studies we used the ‘risk of bias assessment tool’ [21]. This tool is a domain-based evaluation to assess selection bias, performance bias, attrition bias, detection bias and reporting bias. For non-randomized studies, the Cochrane collaboration recommends to add additional domains. Therefore, we added two domains to the tool: (1) randomization (yes/no), and (2) control group (yes/no). To assess the quality of systematic reviews we used AMSTAR, as recommended by Cochrane [23]. To assess the quality of observational studies (retrospective, cross-sectional, prospective) and qualitative studies we used tools developed for evaluating primary research papers in a variety of fields [24]. From the 14-criteria quantitative tool, we deleted three criteria (criteria five, six, and seven) on experimental research as we assessed quality of experimental studies with the tool described above. For qualitative studies we used the 10-criteria tool. The quality assessment was performed by couples of two independent researchers (RE, RS, NT, JL). In case of doubt, a third reviewer from these four researchers was asked to make a final decision.

Data extraction

Data were extracted by two independent researchers (RE, RS, NT, JL). Outcomes extracted were non-conveyance rates, characteristics of non-conveyed patients, existing guidelines, protocols or triage criteria for non-conveyance, follow-up care by patients after non-conveyance, ambulance professionals competencies needed to perform non-conveyance, and factors influencing ambulance professionals during the non-conveyance decision-making process.

Data synthesis and presentation

Due to heterogeneity of the studies with regard to patient populations, interventions and outcomes, a meta-analysis was not possible. Instead, we extensively analysed and synthesized the studies, by scrutinizing and categorizing data and formulating (sub)themes. To report non-conveyance rates, percentages were extracted or calculated. When patients died or left the scene before ambulance arrival, these were not taken into account for non-conveyance rates. To compare patients’ initial complaints or conditions across studies, we classified these according to the ICD-10 classification [25]. The ICD-10 classification is an international standard to classify diseases or other health problems, and is widely accepted and used. For each ICD-10 category we described the proportions of the patients who had a certain classification.

Results

Review statistics

The initial search identified 2989 unique records, after the selection procedure 67 studies were included (see Fig. 1). A list of excluded articles (n = 67) is provided in Additional file 3: Appendix 2.

Fig. 1
figure 1

study selection process

Full size image

Study Characteristics

The designs of the included studies concerned two systematic reviews [10, 26], four experimental designs: one cluster-randomized controlled trial [27], one quasi-experimental [28], and two pre-test post-test [29, 30], 52 observational designs: 27 retrospective [8, 9, 31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55], 23 prospective [56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78], and two cross-sectional [79, 80], one mixed method design [81], and eight qualitative designs [82,83,84,85,86,87,88,89] (Table 1 and Table 2).

Table 1 Characteristics of quantitative and qualitative included studies (n = 65)
Full size table
Table 2 Characteristics systematic reviews (n = 2)
Full size table

The two systematic review were performed in Australia and the UK. The empiric studies were conducted in North America (n = 36), Europe (n = 17), Australia (n = 6), Asia (n = 5), and Africa (n = 1), and concerned general patient populations or specific patient populations, including patients with hypoglycaemia, patients who refused conveyance, paediatric and/or older patients, patients with supraventricular tachycardia, patient with acute opioid overdose, post-ictal patients, and patients who had fallen. The ambulance professionals in these studies were ambulance nurses, basic and advanced life support paramedics, emergency medical technicians, (specialized) physicians, general practitioners, and first responder fire fighters. For this review we will use the term ‘ambulance professional’ to cover all these types of professionals.

Quality assessment (Additional file 4: Appendix 3, Additional file-5: Appendix 4, Additional file 6: Appendix 5, Additional file 7: Appendix 6)

The two included systematic reviews had moderate [26] and low quality [10] (Additional file 4: Appendix 3). The four experimental designs included one CRCT of moderate quality [27], one quasi-experimental study [28] and two pre-test post-test [29, 30] of poor quality (Additional file 5: Appendix 4). The quality of the quantitative studies (n = 53) varied from good [76] to poor [42] (Additional file 6: Appendix 5), and the quality of the qualitative studies (n = 8) varied from good [83] to poor [88] (Additional file 7: Appendix 6).

Outcomes

Non-conveyance rates (Additional file 8: Appendix 7)

Non-conveyance was initiated by the ambulance professional, the patient and/or his relatives, or a joint decision. Non-conveyance rates for general patient populations ranged from 3.7% up to 93.7% [28, 30, 31, 33,34,35, 37, 38, 40,41,42,43, 45, 46, 49, 51, 52, 57, 60, 61, 64, 68, 77, 81]. Seventeen studies reported non-conveyance rates for specific patient populations. For patients with hypoglycaemia non-conveyance rates ranged from 12.2% up to 84.3% [32, 48, 53, 58, 59, 75]. Non-conveyance rates for people who had fallen ranged from 11%–56% [26, 27, 73, 74]. For paediatric patients non-conveyance rates ranged from 13.2%–27.7% [36, 62, 79]. Two studies reported non-conveyance rates for patients with an opioid overdose, ranging from 6.0%–77.0% [44, 54]. Non-conveyance rates for other specific patient groups were 14.0% for post-ictal patients [53], 33.2% for patients with supraventricular tachycardia [39], 10.7%–11.5% for elder people [69], and 8.6% for patients with injuries [50].

Characteristics of non-conveyed patients (Additional file 8 Appendix 7)

The demographic characteristics were age, gender, ethnicity, and geographic area. For general patient populations, the age ranges from 14 up to 90 years [9, 29, 31, 33, 38,39,40, 45, 48, 50, 52, 54, 56, 62, 63, 65,66,67, 73, 74, 76, 78, 79, 85]. Twenty studies reported on patient gender: in ten studies the gender is predominantly male, in the other studies the population is predominantly female [9, 33, 38,39,40, 45, 48, 50, 52, 54, 62, 63, 65,66,67, 73, 74, 76, 79, 85]. Three studies described the geographic location of non-conveyed patients [33, 65, 74]. Two of these show that most non-conveyed people stay in a metropolitan/urban area. The third study showed that 58.6% of the patient are in their residence. Two studies described the patient’s ethnicity [45, 79], with one study reporting 90.6% of the non-conveyed patient as white, the other study reported 48.3% of the patient as African-American.

The clinical characteristics of the patient were initial complaints and conditions, vital signs, and patient history. A variety of initial complaints and conditions was described [9, 29, 34, 38, 40, 45, 52, 56, 57, 61,62,63, 65, 74, 77,78,79, 85]. Most often, we found initial complaints and conditions classified as VI- diseases of the nervous system (n = 16) or category XX - External causes of morbidity and mortality (n = 16). For category VI the proportion of patients with these complaints and conditions ranged from 1.0%–29.0% [9, 29, 34, 38, 40, 45, 52, 56, 57, 61, 63, 65, 77,78,79, 85], for category XX the proportion ranged from 11.0%–68.5% [9, 29, 38, 40, 45, 52, 56, 57, 61,62,63, 65, 77,78,79, 85].

Three studies described the vital signs of non-conveyed patients [50, 52, 63]. One study on a general population reported that 14.9% of the non-conveyed patients had abnormal vital signs (blood pressure, O2-saturation, Glasgow Coma Scale, and body temperature) [52]. A second study in a non-conveyed general patient population reported that 70.0% had a blood pressure within normal limits, 72.2% had a heart rate within normal limits, and 63.2% had a respiratory rate within normal limits [63]. The last study on vital signs with injured people not conveyed reported a mean systolic blood pressure of 134.7 mmHg (±21.1), a mean pulse rate of 91.8 (±15.9), and a mean Glasgow Coma Scale of 15.0 (±0.3) [50].

Five studies described the patient’s history by describing the medical history and/or current medication use [48, 63, 73, 74, 76]. Two studies [63, 76] described the medical history, for general patient populations 68.7% had no medical history [63], for people aged ≥70 years who had fallen 43.8% had urinary incontinence and 39.0% had a central nervous system disorder.

Follow-up of patients after non-conveyance (Table 3)

Follow-up was reported as (a) repeated access to healthcare and (b) patient outcomes. Sixteen studies combined these outcome categories, the other studies used outcomes within one category [8, 26, 28, 32, 37,38,39,40, 43,44,45, 48, 50, 52, 55,56,57,58,59, 62, 64,65,66,67, 69, 75,76,77,78, 90]. Repeated access to healthcare was specified as repeated access to (1) emergency department (2) EMS-system (call or ambulance run), (3) the general practitioner, and (4) walk-in clinic. For all outcomes, a variety of follow-up periods was used. In every study that reported on repeated access to healthcare a proportion of patients re-entered the (emergency) healthcare system.

Table 3 Follow-up care after non-conveyance
Full size table

Repeated access to the ED is measured in seventeen studies [8, 26, 37, 38, 40, 48, 50, 52, 56, 57, 62, 65,66,67, 75, 78, 90]. For general patient populations, the follow-up periods ranged from <24 h up to <7 days, and repeated access percentages varied from 4.6–7.0% (<24 h), 19.0% (<48 h), 6.4–25.8% (72 h) up to 8.1–80.1% (<7 days). For specific patient populations (hypoglycaemia, people who had fallen, people aged >65 years, children and people with minor injuries), the follow-up periods ranged from <48 h up to <12 months, and repeated access percentages varied from 5.0–26.4% (<48 h), 65.9% (<72 h), up to 12.0–49.0% (12 months).

Repeated access to the EMS-system is measured in ten studies [26, 37, 39, 40, 52, 56, 58, 64, 67, 78]. For general patient populations, the follow-up periods ranged from <24 h up to <7 days, and repeated access percentages varied from 6.1% (<24 h), 2.3–2.5% (<48 h) up to 7.4–13.5% (<7 days). For specific patient populations (hypoglycaemia, people who had fallen, supraventricular tachycardia, and people aged >65 years), the follow-up periods ranged from <48 h up to <12 months, and repeated access percentages varied from 0.0–2.5% (<48 h), 1.3–8.7% (<72 h), 8.3–31.4% (10 months) up to 12.0–49.0% (12 months).

Repeated access to the GP is measured in thirteen studies [8, 26, 38, 55,56,57, 62, 65,66,67, 75, 77, 78]. For general patient populations, the follow-up periods ranged from <48 h up to <7 days, and repeated access percentages varied from 13.0% (<24 h), 36.8–50.0% (<72 h) up to 46.2% (<7 days). For specific patient populations (hypoglycaemia, people who had fallen, children, and people aged >65 years), the follow-up periods ranged from <24 h up to 12 months, and repeated access percentages varied from 54.7% (24 h), 7.4–40.0% (<48 h), 34.1% (72 h) up to 12.0–49.0% (12 months).

Repeated access to walk-in clinic is measured in three studies for specific patient populations (children, people who had fallen, and patients aged >65 years) [26, 65, 78]. The follow-up periods used for this outcome ranged from <48 h up to 12 months, and repeated access percentages varied from 3.8% (<48 h) up to 12.0–49.0% (12 months).

The patient outcomes measured are mortality, hospitalization and recurrence of symptoms. For general patient populations, the follow-up periods for mortality ranged from <24 h up to <30 days, and mortality rates ranged from 0.2–3.5% (<24 h), 0.3% (<48 h), 0.3–6.1% (<72 h), 0.3%–0.7% (<7 days) up to 1.6% (<30 days) [26, 34, 37, 40, 43, 45, 50, 52, 57]. The one study reporting on a specific patient population (opioid overdose) reported a 0.8% mortality rate < 48 h [44].

The hospitalization follow-up period for general patient populations ranged from <24 h up to <14 days, and hospitalization rates ranged from 3.3% (<24 h), 1.0% (<48 h), 4.5–12.1% (<72 h), 5.0–8.1% (<7 days) up to 5.4–6.2% (<14 days) [8, 28, 34, 37, 38, 40, 43, 52, 55, 57]. For specific patient populations (hypoglycaemia, people who had fallen, children, people with minor injuries, and people aged >65 years) the follow-up periods ranged from <48 h up to 12 months, and hospitalization rates ranged from 1.2–7.9% (<48 h), 2.5–5.1% (<72 h) up to 12.0–49.0% (<12 months) [26, 32, 48, 50, 62, 67, 69, 75].

Recurrence of symptoms for specific patient populations (hypoglycaemia and people who had fallen) varied from 6.1% (48 h), 7.9% (<72 h), 4.9% (<21 days) up to 24.7% (<6 months) [32, 48, 59, 76].

Existing guidelines, protocols and triage criteria for non-conveyance (Additional file 9: Appendix 8)

Criteria to guide the (non-) conveyance decision described mostly are abnormal vital functions related to ‘breathing’ (respiration rate, respiratory distress, dyspnea), abnormal vital functions related to ‘circulation’ (systolic/diastolic blood pressure, pulse), suspected or confirmed ingestion of alcohol or drugs, and an altered level of consciousness (Glasgow coma scale) [28, 29, 39, 40, 43, 46, 51,52,53,54, 59, 70, 72, 73, 75, 79, 84, 88]. Ten of these studies described more specific flowcharts, tools, checklist or standard operating procedures for non-conveyance in general [43, 51, 72], patients who refuse conveyance [29, 40, 46], and patients who had fallen [84], with supraventricular tachycardia [39], with social problems [28], with hypoglycaemia [53], and post-ictal patients [53].

Professionals competencies and other factors influencing the non-conveyance decision-making process (Table 4)

Factors influencing the non-conveyance decision-making process are related to the professional, the patient and his relatives, the healthcare process/system, or supportive tools [26, 29, 49, 51, 57, 64, 65, 74, 77, 78, 80, 83,84,85,86,87,88,89] (Table 4). These factors can be present at (a) pre-arrival, when the professional forms an early opinion based on information from the emergency call, during (b) initial patient contact where the ambulance professional gets a first impression of the patient, during (c) patient assessment of vital signs and other parameters, and (d) during the actual (non-) conveyance decision moment [84].

Table 4 Competences and influencing factors (n = 18)
Full size table

As for professional related factors, two studies described professional competencies needed to perform non-conveyance. These studies showed that additional training for ambulance professionals led to higher non-conveyance rates compared to ambulance professionals who received regular training [64, 77]. Besides competencies, other professional related factors are weighing of patient risks and personal litigation risk in case of a wrong non-conveyance decision [87, 88], experience and intuition of the ambulance professional [89], and pragmatism as conveyance being an easy option compared to non-conveyance [89].

For patient related factors, firstly the health status of the patient influenced the non-conveyance decision of the professional [26, 49, 65, 74, 78, 83, 85, 88]. Only three studies specified these physical conditions: the sufficiency of on-scene treatment [26], if problems/injuries have resolved pre-arrival or were only minor [26, 85], patient physiology [74], the absence of new pain or injury [74], and possible changes in usual level of functioning [74]. A second patient related factor is refusal. Refusal might be related to relatives thinking conveyance is not necessary [65], but also by patients concerns about costs of conveyance or ED care [65, 78], or the refusal reasons were not further specified [26, 85]. Thirdly, patient wishes and the patients’ best interest are factors that influence a conveyance decision [86].

Influencing factors related to the healthcare system are access/referral to GP or alternative healthcare facility in case of non-conveyance [26, 87]. To make appropriate conveyance or referral decisions, access to patient information is essential. One study [80] showed that 90.7% of the ambulance professionals felt that a lack of patient information leads to less appropriate care being selected. To make appropriate decisions, ambulance professionals gave high priority to previous medical history, patient’s usual vital signs and resuscitation status as patient information.

Finally, three studies showed that implementing online medical control as supportive tool, where a physician can be contacted by the pre-hospital professional, solely or in combination with a high risk card, increased conveyance rates for patients with high risk criteria or patients who refused conveyance [29, 51, 57].

Discussion

This systematic review includes 67 articles that describe non-conveyance in ambulance care from patient safety and ambulance professional perspectives. Our results show that non-conveyance occurs in all types of EMS systems across the world, and that there is a wide variation in non-conveyance rates for general and specific patient populations. These variations might be caused by differences in patient populations (medical acuity and medical necessity to convey), and differences between EMS-systems in terms of triage systems, types of services, educational levels of ambulance professionals, and type of vehicles (conveying and non-conveying) [91,92,93]. Although non-conveyance in itself is a valid outcome of ambulance care [17], our results do not distinct between justified or unjustified non-conveyance. This can be a focus of future research.

Our review provides a first insight in characteristics of non-conveyed patients. Our results show that patients of all ages and both men and women are represented in the non-conveyance population. Non-conveyed patients most often had a neurological or trauma related complaint or condition. Vulnerable patients as children and elderly, and specific patient groups of people who had fallen or people with hypoglycaemia are relatively high represented in the non-conveyance population. Another subpopulation is patients who refuse care and/or conveyance. From our results it remains unknown what kinds of complaints or conditions these patients have from ICD-10 perspective, and what consequences their refusal has from patient-safety perspective.

Although the assessment of vital signs is an important aspect of the primary survey in ambulance care to make appropriate treatment and triage decisions [94], we found only three studies describing vital signs of non-conveyed patients. These studies show that roughly 15% of the non-conveyed patients have vital signs that deviate from limits. We do not know whether vital signs differ between conveyed and non-conveyed patients. Therefore future research should focus on a comparison of vital signs and follow-up outcomes between conveyed and non-conveyed patient groups. Furthermore, it remains unclear if abnormal vital signs were present in the medical history due to illness or medication use. Poor access to healthcare information systems by ambulance professionals is reported [80], this underlines the possible advantage of access to healthcare information systems in the chain of emergency care, and the accessibility of the general practitioner.

Results show that a significant amount of non-conveyed patient re-enters the (emergency) healthcare system. For instance, 6.1% of the patients re-enters the EMS-system <24 h after non-conveyance, and up to 19.0% of the patient visits an ED within 48 h after non-conveyance. From the patient-safety perspective it remains unclear whether these repeated EMS calls and ED visits are based on medical necessity, as it remained unclear in the data which complaints or conditions these patients had during this repeated access to emergency healthcare, and whether it was similar to the initial EMS contact. Furthermore, the studies did not describe whether the re-entry is based on professional referral or self-referral. Clinical practice could benefit from the development of valid quality indicators for patient safety in the chain of emergency care. These could measure systematically (un)justified re-entry of the emergency healthcare system and quality of care provided.

From the professional perspective, our results indicate that the non-conveyance decision-making process is multifactorial, with influences from the professional, the patient and his relatives, the healthcare system, and supportive tools. Our results do not give clear direction which additional competencies ambulance professionals need to make safe non-conveyance decisions, as only two studies describe positive effects of additional training. Studies not included in our review suggest that pre-hospital professionals with additional training on the conveyance decision, and on management of minor illness and injuries, are less likely to convey patients compared to regular ambulance staff [15, 95]. Initiatives to implement new competencies of pre-hospital professionals in EMS or possibly new professionals with additional competencies in clinical reasoning and conveyance decision-making should be explored and tested regarding patient safety.

As for supportive tools, our results show that there is a limited number of flowcharts, checklists or protocols available to guide non-conveyance decisions for general and specific patient populations. However, it remains unclear how these tools were developed and to what degree they are evidence-based. This urges the need to develop evidence-based supportive tools to guide non-conveyance decision-making for different patient groups. In order to do so, future research should be aimed at identifying factors to guide accurate non-conveyance decision making, to predict non-conveyance in the EMS dispatch phase through tailored triage criteria, or to predict follow-up outcomes such as mortality and re-enters in the emergency healthcare system. This with the aim to support professionals in their decision making and to enhance quality and safety in pre-hospital care.

Limitations of included studies

As described in the result section, the quality of included studies varied. For the quantitative studies (Additional file 4: Appendix 3, Additional file 5: Appendix 4, Additional file 6: Appendix 5), the quality assessment criteria objective/aim, design, methods of subject/group selection, appropriateness of sample size, description analytical methods, and detailed reporting of results scored good quality. The moderate assessment criteria were description of subject characteristics, outcome definition, and the relationship between results and conclusion. The reporting of estimate of variance was poor, and due to design most studies could not be controlled for confounding. Within the qualitative studies (Additional file 7: Appendix 6) the quality assessment criteria objective/aim, design, connection to theoretical framework, data-collection and data-analysis scored good quality. The moderate assessment criteria were description of context, sampling strategy, and conclusion supported by results. Use of verification procedures and reflexivity of account were the two poor assessment criteria. Another limitation concerned the studies describing initial complaints and conditions. These studies used different types of classification systems, or systems were lacking. Therefore, we recommend to use one classification system, such as the ICD-10, in future research to enhance generizability and comparability of results.

Study strengths and limitations

Despite the fact that this systematic review is the most complete and systematic analysis to date of non-conveyance in ambulance care, there are some limitations. A possible limitation is that our review did not cover the entire ambulance care process, as we focused on the phases after ambulance dispatch. Additional research should focus on the accuracy and predictive value of current EMS dispatch systems for non-conveyance decisions. Secondly, a meta-analyses was not feasible due to heterogeneity amongst studies. Another limitation concerns the quality assessment tools for quantitative and qualitative designs. A variety of these tools exist without a clear evidence-base. Strengths of our study concern the usage of Cochrane and PRISMA methods and tools to perform and report our research.

Conclusion

This systematic review shows that non-conveyance occurs in all types of EMS systems across the world, and that a wide variation in non-conveyance rates for general and specific patient populations exists. Patients in the non-conveyance population present themselves with a variety of initial complaints and conditions, although initial complaints or conditions related to trauma and neurology, and vulnerable patients groups such as children, elderly and patients with hypoglycaemia, are well represented. Nevertheless, further insight in characteristics of the non-conveyance population is needed. From patient safety perspective it turns out that a proportion of non-conveyed patients re-enters the emergency healthcare system within one or 2 days after non-conveyance. Why these patients re-enter the emergency healthcare system, and what outcomes these patients have remains unclear. For ambulance professionals the non-conveyance decision-making process is complex and multifactorial, with influences from the professional, the patient and his relatives, the healthcare system (referral or access to general practitioner) and supportive tools. Competencies needed to perform non-conveyance are marginally described, this should be priority in future research. Despite the fact that a limited amount of supportive tools is available for general and specific non-conveyance populations, there is a need to develop evidence-based guidelines and protocols to guide non-conveyance decision-making.