Factors associated with individual patients (n=16 studies)

Notwithstanding the complexities of determining ‘appropriateness’ of ambulance use from a variety of care-giving and care-receiving standpoints, much of the literature explores how urgent and emergency clinical presentations are recognised and defined by the users of ambulance services, and how this shapes risk management strategies. Studies report wide variation in what both patients and carers deem to be clinical ‘emergencies’, and what would be appropriate self-management thresholds.

Category of clinical problem

One cross-sectional US study16 suggests overall poor physical health status is associated with ambulance use for low-acuity conditions rather than specific symptoms or conditions. This is supported by a similar study showing increased inappropriate use with increasing number of comorbidities, regardless of the actual presenting symptom.17 However, another similarly designed study reports chest pain, respiratory conditions and cardiac symptoms as the most likely clinical problems associated with a true need for ambulance attendance,18 while a Swedish paper reports that nearly half of abdominal pain and urinary presentations to emergency services unnecessarily used ambulances.19

Psychiatric conditions, behavioural disorders and drug and alcohol misuse present particular challenges for ambulance services across the full range of global settings. US studies analysing routinely collected national data sets20 and undertaking case note reviews in emergency departments21 both report high levels of excess ambulance usage among these groups.

Despite these specific symptom-related or diagnosis-related associations, other papers show unclear relationships between the clinical symptom and need for ambulance transport. One comprehensive review of the appropriateness of ambulance transport concludes that there is relatively little variation across study methodologies of ambulance contacts determined as inappropriate on the basis of clinical diagnosis (30–55%).5 These authors highlight the pitfalls of making this assignment retrospectively after clinical assessment, suggesting that basing judgement of appropriateness on information available after the contact (ie, a firm diagnosis) has many limitations in seeking to explain variation by clinical symptoms. Authors of another comprehensive review22 suggest that medical classification of urgency based on physiological measures contrasts with patient classification based on psychosocial factors, rendering distinctions by illness type less informative.

In addition, there is no clear relationship in the literature between whether the ambulance request was deemed to be ‘appropriate’ or not and whether the treatment was for a traumatic injury or a medical symptom. A UK case note review and patient interview study suggests 50% of calls for trauma-related conditions were unnecessary, compared with 20% for medical conditions.23 A Swedish study using similar methodology assesses the figure for trauma to be just 17%.19

Personal anxiety and risk management strategy

Two interview studies24 ,25 with patients from the UK highlight the importance of an individual's risk management strategies when choosing ambulance care. One thematic analysis identifies patient and carer anxieties surrounding the relatively small risks associated with choosing alternative care options.24 Another identifies the process of recognising the need for help, overemphasising the urgency, and therefore accepting that only ambulance-based care will meet this need.25 Feeling isolated or alone during this decision-making process is the criterion in choosing the immediacy of an ambulance response in this analysis.25 Studies that included surveys of patients at the time of ambulance care report a genuine fear of a life-threatening condition in 60% of patients,26 but frequently acknowledge the role of lay bystanders in influencing this assessment.27

One UK study28 set in a rural area explored pre-emptive strategies used by patients to mitigate the perceived risks, suggesting that a common practice in this cohort was to arrange routine primary care appointments early in an illness just in case they became necessary. A different UK study29 that explored ambulance paramedic views through focus groups reports that some ambulance clinicians feel unsupported by their employers in leaving patients at home after a call, suggesting that appropriate risk management is a challenge for both providers and consumers of ambulance care.

Health knowledge

Only one UK study30 attempted to formally account for patient medical knowledge or training via online hypothetical case-vignettes to explore decision-making. This found a negative association between medically unnecessary ambulance use and formal first-aid training, and suggested that national first-aid training programmes may help the population discriminate between routine and emergency problems.

Actions of care-givers and bystanders (n=5 studies)

Of those studies that formally included the perspectives or actions of carers or bystanders (n=5), all were qualitative. All studies reported that the influence of carers and bystanders was towards the use of ambulances, rather than against.

Formal carer responsibilities

The actions of those with formal caring responsibilities, including those in loco parentis of children are often recognised in the literature as important components of the decision to seek help from ambulances. Several studies highlight the lower threshold of medical ‘risk’ tolerated by formal care-givers in these situations,23 ,24 ,27 ,29 with the default action during illness leaning towards ambulance care. Often, the voices of the relatives can be louder than those of the patient, who may be led down a less appropriate decision-making pathway by well meaning but misinformed relations.27 Two thematic studies make reference to the notion that the sick relative is perceived to be less able to make appropriate decisions about their care by virtue of their condition,27 ,29 resulting in de facto vicarious decision-making. Indeed, one interview study illustrates that the roles of relatives and formal carers continue to shape negotiations about non-conveyance even after assessment by the ambulance crew.29 Relatives and formal carers appear to be powerful negotiators who often support the course of action that carries least risk, even though this may not be the most clinically appropriate, or even in line with the patient's preferred priorities.

Bystander actions

Two studies also note that unrelated bystanders often initiate calls for low acuity situations that occur in public places.23 ,31 The initiator of the ambulance call is important in determining how likely it is to be medically necessary, with some relationship between bystander knowledge of basic first aid and the ability to determine what is likely to be a relatively minor illness or injury.23 Members of the public appear to be able to appropriately determine the need for an emergency ambulance when they rate their condition as causing ‘severe pain’ or being ‘potentially life-threatening’.23 However, they can be influenced by bystanders towards requesting ambulance help if their condition is more minor.27

Population-level factors (n=11 studies)

A number of studies (n=11 papers) directly analyse the impact of socioeconomic status and demographic variations in access to and use of ambulance services for non-serious conditions, using either self-reported questionnaire data or routinely collected data at the time of care episodes. Comparisons are complex, as these studies encompass a variety of global health systems and health economies, with varying levels of correcting for confounding. Some comparative studies use the regional level as the unit of comparison, some use national, and some international perspectives. Broadly, markers traditionally associated with deprivation were associated with increased ambulance use for non-urgent conditions. No UK-based studies exploring socioeconomic variation in detail met the inclusion criteria.

Demographics

Globally, ambulance usage for non-serious conditions shows considerable variation according to demographic factors. In Japan, the two studies that met the eligibility criteria used self-reported postal questionnaire methodologies to identify patterns of ambulance usage. Men and the elderly are more likely to use emergency ambulances for non-urgent problems.32 In addition, those who lived alone or without the family support network of younger relatives were more likely to call for an ambulance when questioned about hypothetical situations that did not require an emergency response.33 The authors conclude that access to support from younger family members with positive attitudes to assisting the elderly appears to be associated with seeking treatment from primary care services rather than emergency services.

Similar findings were reported in several US studies, demonstrating old age16 ,34 and male gender17 to be associated with increased inappropriate ambulance use. A separate US study approached the issue from a different perspective, exploring willingness to consider alternatives to emergency ambulances for acute care,35 by interviewing patients at the time of presentation to emergency services. This study supported the above findings, by showing a willingness to consider that non-ambulance care was associated with adults of working age (18–65 years) but, in contrast with other studies, unemployment.

Minority ethnicity was positively associated with ambulance use in two US studies17 ,34 both of which were retrospective analyses of routinely collected national data sets.

Being located in a rural area showed mixed relationships with ambulance usage. A focus group and patient interview study on urgent care-seeking behaviour in Scotland indicated those in the more rural areas would delay seeking urgent ambulance assistance due to their remoteness, preferring a more ‘wait and see’ approach.28 However, a study in the USA that used a substantial national dataset of over 16 million ambulance journeys to emergency departments found that urban location had a corrected OR of 1.46 (1.2 to 1.7) of ambulance use.34 This was not the case in a study of children set in South Carolina (USA), reporting a higher odds ratio of 1.247 (1.041 to 1.492) of unnecessary ambulance transport in rural areas.21 Other studies have shown that patterns of consulting for perceived paediatric emergencies may be different,24 so the significance of this variation would require more detailed exploration.

Socioeconomic status and deprivation

One Japanese study found that those without their own transport were more likely to call ambulances for medically unnecessary situations.32 While increasing household income was associated with reducing rates of ambulance use, the demand for ambulances did not decrease consistently with a theoretical price paid by the user, suggesting the relationship is more complex than monetary cost alone. Socioeconomic factors associated with ambulance use in US-based studies include living in non-private (ie, social) housing,20 unemployment35 and homelessness17—particularly in the elderly, where one retrospective analysis of demographic factors recorded by Fire Department paramedics concluded an eightfold increase in emergency ambulance use compared with matched controls. A self-administered cross-sectional survey of health-seeking behaviour in the USA demonstrated that higher educational level was associated with reduced use of ambulance services,16 and poverty (as defined by total household income) was associated with increased use.

Health insurance status

Several (n=4) US-based studies included health insurance status in their analysis of ambulance usage. For non-urgent conditions, ambulance usage was associated with social healthcare insurance in all three studies. Three studies found that being insured within the Medicaid programme (typically associated with low-income individuals and families) was associated with ambulance use,16 ,21 ,36 whereas one study found the same for Medicare (typically associated with the elderly and disabled).34 As such, all four showed an indirect association with lower socioeconomic and health quality status.

Healthcare infrastructure factors (n=10)

A mixture of qualitative and quantitative studies evaluated access to and satisfaction with the urgent care health infrastructure, either regionally or nationally.

Experience, satisfaction and misconceptions of health infrastructure

Five qualitative papers directly explored patient-reported perceptions of the capacity and capability of the health infrastructure to deal with their perceived or actual problems.24 ,27 ,31 ,37 ,38 Several of these reported themes of misconception about the level or type of care that could be delivered by the ambulance crews24 ,27 or receiving emergency units.31 ,37 One study identified reassurance as a key element of the ambulance response that patients valued,38 which links with another study that identified the amount of time spent and thoroughness of the clinical assessment as features that differentiate an ambulance response from an alternative urgent care avenue.24 Satisfaction scores reduced sequentially with the number of different services a patient contacted prior to definitive care.37

Presence of a primary care model

The presence (or absence) of a primary care model of health within the wider infrastructure was identified in varying degrees as related to ambulance use. One Swedish study estimates that urgent primary care services handle 42 500 potentially life-threatening presentations annually.39 In the Netherlands, a cross-sectional study concluded that 88% of all out-of-hours urgent care requests were handled by primary care. Japanese data indicates that the relatively limited existence of primary care and social isolation are important themes in driving inappropriate and expensive ambulance use.33 Those with access to urgent primary care services in Japan were less likely to use an ambulance inappropriately in hypothetical scenarios.32

Role of other services in unmet needs

Several UK studies explored, qualitatively and quantitatively, patients’ use of services prior to contact with the ambulance service. Ambulances called after a triage-contact with NHS 24 (a telephone triage service) were more likely to be classified as inappropriate.40 One paper reports that 68% of service users had contact with more than one service during their urgent care need, with an average of two contacts per episode.37 Several studies specifically explored patient actions in the lead-up to an emergency call. In general, where primary care services existed, the majority of patients who ultimately were categorised as inappropriately receiving ambulance care had attempted to contact their GP beforehand.24 ,37 ,41 One study found that 1 in 20 calls required either ‘general assistance’, or were not regarding any illness or injury at all.42 Social care was usually the unmet need.

Challenges faced by health professionals (n=21)

Of the included evidence, 21 papers referred primarily or substantially to health professionals’ perspectives in the assessment of need for an ambulance.

Inappropriateness as a concept

The assessment of ‘inappropriateness’ of an ambulance contact is complex and varied in the included evidence; a result of methodological limitations and conceptual variation. The majority of studies sought to determine ‘inappropriateness’ retrospectively from case notes, using semiobjective scoring or coding systems. Assessments were performed by emergency department clinicians20 ,21 ,23 ,26 ,27 ,40 ,43 and pre-hospital staff.17–19 ,21 ,26 ,27 ,29 ,33 Two included assessments of severity of illness undertaken by primary care staff.39 ,41 The professional background and seniority of staff varied. One meta-analysis of US paramedic decision-making reports an aggregate negative predictive value of 0.912 (0.707 to 0.978) for ambulance paramedic determination of necessity for transport at the scene, when compared with hospital physician assessment.44

Only one US study involving children utilised medical necessity criteria agreed at a consensus conference to make these assessments.21 Other studies used a mixture of one or more professional opinions,18 ,19 ,23 ,26–28 coding systems, or scores based on physiological parameters or clinical conditions,17 ,20 ,39 standardised diagnosis codes such as the International Classification of Primary Care,41 or custom designed instruments or topic guides to explore provider perceptions of necessity.29 ,33

Two comprehensive reviews explored the concept of ‘inappropriateness’ specifically and qualitatively.5 ,22 One reported the theme of inappropriateness as divided into two cohorts: those not experiencing a health emergency, and those experiencing an emergency but who do not seek ambulance care when they should.22 The other review concludes that assessment of appropriateness based on information available after clinical assessment will overestimate ‘inappropriate’ use, and neglects the complex psychosocial context of the request for help.5

Fitness for purpose of the triage process

Several studies directly or indirectly address the role of telephone triage systems in supporting or confounding both patient and clinical decision-making. A UK study reports that 26% of ambulance calls coded as the highest clinical priority at triage resulted in no patient being conveyed to hospital.45 Another study highlighted the need for good real-time links between ambulance and primary care triage systems, due to the cross-over of calls.39 Other qualitative studies report the theme of validation of the urgent need for care which can be inferred from triage.24 ,25