Thirty-seven studies included in this review were identified through database searching and nine identified through searching the bibliographies of included articles.
Included articles were categorised according to the relevant public health domains and subdomains, as shown in Table 2. Most included studies clustered according to the Healthcare Public Health Domain with some articles also clustering around the Health Improvement Domain, primarily the Health Promotion sub-domain. Of note, some domains (e.g. health protection) had nil returns.
Studies were conducted in 7 countries with 72% undertaken in either the United States of America (USA) or Canada. The target population for most of the interventions evaluated in the studies were adults aged > 18 years.
In all studies evaluating public health education and advice interventions (n = 6), the effectiveness of the intervention was assessed using participant self-report measures (e.g., surveys). Four studies used pre-test/post-test measures to assess whether knowledge had increased because of the intervention [9,10,11,12]. Two studies collected only post intervention data [13, 14], but one of these studies compared the results to a control group who did not receive the intervention .
One study used an observational design to assess the feasibility of emergency medical services providing vaccines .
Fourteen studies included in the review described and evaluated paramedicine programmes [16,17,18,19,20,21,22,23,24,25,26,27,28,29]. Seven studies had qualitative designs [17, 24,25,26,27,28,29], two studies were randomised controlled trials (three papers reported on the same randomised controlled trial) [19,20,21,22], one study reported a prospective pre-post intervention , and two were relatively small scale (a pilot study and case report) [16, 23].
Thirteen studies described and evaluated screening tools and referral pathways used by the ambulance sector [30,31,32,33,34,35,36,37,38,39,40,41,42]. Eleven used observational study designs [30,31,32,33,34,35,36,37, 40,41,42], and one study (based on two papers) reported a randomised controlled trial [38, 39].
Twelve studies reported on how ambulance sector data can be used for population health monitoring and intervention [43,44,45,46,47,48,49,50,51,52,53,54]. 2 studies used geocoding methods, 6 studies described data linkage methods, and 2 studies used syndromic surveillance.
Table 3 describes the study characteristics of included studies.
Six UK based pilot initiatives were identified through grey literature searching but these had not been rigorously evaluated [55,56,57,58,59,60,61]. Table 4 describes the studies identified through grey literature searching.
Public health education and advice
Two studies from the USA and Japan reported public awareness of preventative health behaviours increased following public health education interventions delivered by emergency medical technicians [12, 14]. However, neither study measured whether participants changed their behaviour in practice nor looked at health care usage post intervention. A UK study  followed up their participants to document whether members of the public had acted on advice provided by a paramedic regarding their blood pressure. They found 56% of participants who were advised to contact their GP by the paramedic did so and 42% of participants identified as having high blood pressure by the paramedic reported taking actions to reduce their blood pressure (e.g. by increasing exercise levels). However, the follow-up period was short, making it unclear whether the changes were sustained in the long term and what impact this had on future healthcare usage.
Three studies reported delivering public health training to staff working in the ambulance sector [9,10,11]. One Australian study  reported that after completing the population health component of the Graduate Certificate in Rural and Remote Paramedic Practice, 73% of paramedics said they had changed their practice, 20% had concrete plans to change, and 7% were considering making changes. A European study by Lygnugaryte-Griksiene et al. (2017)  delivered suicide-intervention training to emergency medical service providers and found assessment of suicidal risk factors improved six months post training but suicide intervention skills, attitudes towards suicide prevention and strategies for coping with stress remained unchanged. In another study from the USA , participants knowledge, attitudes and situational problem solving about domestic violence improved after attending a training session on domestic violence. However, due to a lack of follow-up data in all three studies it is not possible to say whether changes in clinical practice were sustained in the long-term or had a significant impact on the communities in which the paramedics were based.
Emergency medical services providing vaccines
An observational study, based in the USA, assessed the feasibility of emergency medical service agencies providing influenza vaccines to members of the public . 48% of those who were vaccinated, reported not receiving the influenza vaccine in the previous year, while 34.5% reported that they probably would not have received the vaccine elsewhere if it had not been for the vaccination programme, suggesting emergency medical service agencies were able to reach out to members of the community who may not ordinarily have received the influenza vaccine. However, three-quarters of those vaccinated were 60 years or younger, suggesting further targeting would be needed in the future to reach those most in need of vaccination.
Paramedicine programmes aim to increase access to basic primary care and public health services using specially trained emergency medical service providers.
Three studies reported on the CP@clinic [19,20,21], and three reported on the CHAP-EMS programme [16,17,18], which were delivered by paramedics to older adults living in low-income housing buildings in Canada. Types of activity included weekly risk assessment, disease prevention, and health promotion sessions. The studies reported emergency service call use was significantly lower in the intervention group and QALY, blood pressure and diabetes risk significantly improved for programme attendees [18,19,20]. A cost effectiveness analysis of the CP@Clinic found the reduction in emergency service calls as a result of the programme avoided an estimated C$256,583 (~£147.740), which was almost double the cost of implementing the programme in five communities (C$128,462; ~£73,968) . However, the programmes used ‘accommodated’ paramedics who were unable to undertake traditional paramedic duties due to personal limitations (e.g. injury). This approach may not be feasible in other locations.
A randomised controlled trial in Canada was used to evaluate an ‘ageing at home programme’ where paramedics conducted regular home visits and carried out health monitoring of people with chronic conditions who frequently used emergency medical services . Whilst QALY life scores decreased in the intervention and control groups, this effect was lessened in the intervention group. However, a cost-effective analysis based on cost to realise QALY found it was not cost-effective in the long term.
Two North American studies [27, 29], reported positive outcomes related to paramedicine programmes such as reductions in emergency department visits, and hospital admissions and re-admissions but these findings were not based on rigorous evaluation methods.
Key themes from three qualitative studies aiming to understand the experiences of patients and families involved in paramedicine programmes included:  Paramedics became a trusted and essential member of the patient’s healthcare team;  positive relationships with the paramedics encouraged patients to keep up with positive lifestyle changes and to become more proactive about disease management;  paramedics were described as a ‘safety net’, providing a sense of security and support that patients did not have before;  long-term improvements in the patient’s health status were reported by paramedics, patients and the patients families [17, 24, 25].
Key themes from two qualitative studies exploring the experiences of paramedics and service managers involved in the delivery of paramedicine programmes included:  Community paramedics work in ways that use very different skill sets to the emergency services norm, which can create transitional barriers and role boundary conflicts;  Building trust and long term relationships with patients is key to the success of paramedicine programmes;  Traditional paramedic training is focused on the emergency medical response, with limited education on health promotion, aged care, and chronic disease management [26, 28].
Screening tools and referral pathways used by the ambulance sector
Snooks et al. (2017), conducted a cluster randomised controlled trial in the UK evaluating a falls referral pathway . Paramedics in the intervention group only referred 8% of eligible patients but left fewer patients at the scene without any ongoing care compared to paramedics in the control group. Low referral rates were also reported in another study where paramedics only referred 3% of eligible patients to an 8-week fall prevention programme . One reason cited for the low referral rate was that paramedic training is overly focused on acute care and so paramedics may not recognise a non-injurious fall as a clinical incident requiring follow-up care. Another explanation is that patients may be unwilling to be referred because they feel they do not require assistance .
Two studies, based in the USA, evaluated an outreach referral service run by paramedics and peer recovery coaches for people who had recently survived an opioid overdose [33, 34]. Of the individuals contacted by the outreach team, 33% engaged in same day treatment and 56% of those were still engaged at 90 days. However, more than half (59%) of potential participants proved unable to be located due to homelessness or living in temporary accommodation.
Two US articles described an intervention where paramedics provided education to new and expectant parents on childhood injury prevention [30, 31]. However, no information was provided about whether the intervention changed behaviour in the long-term or had any impact on subsequent injuries sustained in the home.
Three studies reported on the development and validation of screening tools that were not linked to follow-on services [35, 41, 42]. Whilst the studies reported that screening tools were feasible, the lack of follow-on services means conclusions about whether they impacted on public health or ambulance sector outcomes could not be assessed. Shah et al. (2010), evaluated a screening tool used by paramedics where the results of the screening were sent to the patients GP by the research team . At follow-up the authors reported that infrequent discussions had taken place between the patient and their GP about the risks identified during the screening. However, the 2-week follow-up period may not have been long enough for participants to have received any interventions designed to meet their unmet needs.
A UK based study evaluated the implementation of “treat and refer” protocols which allowed ambulance crews to leave patients at the scene with a referral to community-based services or self-care advice . The median job cycle time was 8 min longer for non-conveyed patients in the intervention group compared to the control group. Whilst the increased time on scene may indicate improved quality of care, this would have a considerable impact on operational performance of an ambulance service.
Health intelligence using ambulance sector data
Some studies reported using geocoding type methods to identify hotspots within communities where public health interventions may have the greatest impact. For example, Byun et al. (2019) overlaid emergency medical service call data on census tract rates and Google Maps to explore neighbourhoods in Utah (USA) with high fall counts . This information was used by health planners to identify optimal locations for falls prevention programmes in the community. Another USA based data registry reported collecting addresses of cardiac arrest events from 911 call centre data, emergency medical services data and hospital data, enabling the identification of community level disparities related to bystander cardiopulmonary resuscitation and automated external defibrillator use [49,50,51].
Some studies used data linkage methods to provide a comprehensive picture of population health for use within both regional and country-wide public health initiatives [43, 49, 50, 52]. However, data linkage is highly time consuming as often there is a lack of consistency in reporting methods, and it requires engagement from multiple health agencies [46, 52].
Syndromic surveillance was described in three studies. Syndromic surveillance is the real time collection, analysis, interpretation, and dissemination of health-related data to enable the early identification of the impact of public health threats which requires public health action. Whilst not described as syndromic surveillance, Do et al. (2018), report using ambulance service data to identify opioid-related events in Canada in near real-time . However, one study reported that whilst ambulance sector data had reasonable sensitivity (i.e. it captured most influenza-related events), the data had low specificity (i.e. misidentified some events as influenza-related) . This issue may lead to inaccurate conclusions that could potentially drive strategic decision making.
Data quality and comprehensiveness is frequently cited as a limitation of using ambulance sector data [43,44,45,46, 48,49,50,51,52]. For example, ambulance sector data only captures health events where emergency medical services have been notified, resulting in lower rates of detection. Comprehensive data about patient circumstances and system factors is often unavailable. In the future better data linkage between ambulance service data and other data such as primary care may improve the accuracy and quality of data. In turn, this may lead to greater and more effective use of such data for public health.
Three initiatives involved ambulance sector staff attending the scene of an incident with other health and social care staff such as occupational therapists and physiotherapists, or mental health nurses [55, 56, 59]. All three initiatives reported a reduction in conveyances to the Emergency Department as result of the intervention. Two articles described paramedics who saw patients, identified by their general practitioner (GP) as requiring a home visit, to help reduce GP workload [58, 60]. Reductions were reported in ambulance conveyances, hospital attendances, NHS 111 (urgent) and 999 (emergency) calls, and GP appointments . Furthermore, the proportion of appropriate non-conveyances increased by 35% when paramedics were given the opportunity to work in primary care . Finally, a frailty response line staffed by a consultant geriatrician, accessible to ambulance staff and community care staff to provide care for frail residents reported high proportions of patients being able to remain at home through joint care planning . Whilst the initiatives identified through grey literature searching reported positive outcomes, none of the initiatives had been rigorously evaluated.