Table 1 Summary of quantitative, evidence based studies (n = 18)

Blodgett 2020 [25]

UK

n = 5283

Pilot data linkage of retrospective patient data

To determine feasibility linking data to assess differences in patients conveyed directly to ED and those referred to a GP referral scheme

Paramedic Pathfinder (protocol tool outlining alternate routes of care)

Referral to partner GP providers

– Patients were more likely to be referred to GP if they were:

 i. women

 ii. older

 iii. Lower priority at dispatch.

– 22% of referred patients presented and 13% were admitted to ED w/i 30 days.

– There was no difference in hospital outcomes between GP-referred and directly conveyed groups.

Positive support

– GP referral scheme provides a safe alternative path of care and does not increase risk of poor outcomes.

– Recommendation for a large-scale study to provide evidence-based recommendations for changes to EMS care pathways.

3

Ebben 2019 [26]

Netherlands

n = 426

Retrospective observational study

To describe characteristics of non-conveyance ambulance incidents

National protocol for EMS decisions

Referral to GP/medical specialist

– 31.1% of patients in a 12-month period were not conveyed.

– 36.6% of non-conveyed patients were referred to GP and 6.1% to medical specialist.

Inconclusive support

– A significant number of ambulance visits end in non-conveyance.

– Note, differences between those conveyed, referred and left at home were not examined.

6

Krumperman 2005 [27]

USA

n = 2143

Retrospective cohort

To compare patient satisfaction and referral adherence in two systems:

 i. “evaluate, treat and refer”

 ii. “telephone triage and referral”

No description of triage process

Referral to:

 i. primary care provider

 ii. urgent care centre

– Patients evaluated and referred by paramedic were less likely to follow instructions than those referred by telephone [odds ratio: 0.31 (0.14–0.69)].

– Patients were highly satisfied with the alternate route of care.

Positive support

– Systems that use both pre-ambulance telephone triage and on-scene referral pathways can help avoid unnecessary ED visits.

4

Larsson 2017 [28]

Sweden

n = 394

Prospective cohort study compared with a matched retrospective control group

To examine pre-hospital assessment of non-urgent patients, and investigate outcomes of different levels of care

Rapid Emergency Triage and Treatment System (RETTS)

- Consulted with GP to decide alternate route:

 i. primary home healthcare supervision

 ii. transportation to primary healthcare unit

– Intervention group resulted in:

 i. decreased ED conveyance (17.4%; 53.1%)

 ii. no difference in transport to primary care unit (8.7%; 10.4%)

 iii. Reduced on-scene ambulance time (87 min; 94 min)

 iv. decreased hospital admissions

(11.4%; 25.6%)

 v. no additional secondary transport w/i 48 h (7.9%; 8.0%).

Positive support

– Collaboration between ambulance nurses and GPs can improve appropriate level of care for non-urgent patients and safely decrease unnecessary ED conveyance.

3

Magnusson 2016 [29]

Sweden

n = 529

Retrospective observational study

To describe characteristics, assessments, and routes of care of low priority patients (as assessed by dispatcher)

RETTS

Referral to:

 i. primary care appointment

 ii. community nurse

 iii. Mobile psychiatric or social care team.

– Compared to ED-conveyed patients, patients who were referred or given self-care advice:

 i. were younger

 ii. required a shorter job time.

– Of those referred or left at home, 19% (visited ED within 72 h; half of these were admitted and a further half of those admitted required intervention/treatment).

Mixed support

– Single-responder nurse can safely triage to the appropriate level of care, providing more effective use of emergency services.

– Note that the study did not solely consider a group of referred patients (e.g. combined with self-care patients) so conclusions specific to referrals cannot be made.

4

Magnusson 2020 [30]

Sweden

n = 6712

Prospective cohort

To assess patient characteristics and evaluate appropriateness of:

 i. initial triage and;

 ii. non-transport decisions

RETTS- Adults

Referral to:

 i. primary care;

 ii. social or home care

– Compared to ED-conveyed patients, non-conveyed patients were more likely:

i. to be younger

ii. to be women

iii. have no medical history

iv. have better vital signs

v. to have been lower priority at initial dispatch.

– 10% of non-conveyed patients were admitted to ED within 72 h (1% considered time-critical).

Mixed support

– Defining patient characteristics that may help initial assessment.

– Improved assessment tools, appropriate use of full triage and better education is necessary.

4

Newton 2013 [31]

UK

n = 481

Prospective cohort

To evaluate if paramedics can safely use Paramedic Pathfinder to direct patients into alternate routes of care

Paramedic Pathfinder (protocol tool outlining alternate routes of care)

– Two alternate routes:

 i. community care pathway (referral to ambulance GP)

 ii. transport to urgent care centre

– There was high agreement in decision-making between expert senior medical practitioners and ambulance clinicians.

– Sensitivity (95%) and specificity (58%) of the tool were sufficient.

Positive support

– Ambulance clinicians can successfully use Paramedic Pathfinder to identify patients that do not require ED care.

– The potential benefits of using the tool fully depend on provision of suitable community alternatives.

4

O’Cathain 2018 [32]

UK

i. n = 49 interviews

ii. n = 615,815 calls

iii. n = 20 interviews

iv. n = 42,796 non-conveyed incidents

Mixed methods including:

 i. paramedic, manager, commissioner interviews

 ii. ambulance dispatch data

 iii. Qualitative telephone advice data

 iv. linked ambulance, hospital and mortality data

To understand differences in non-conveyance between ambulance services

Different triage systems in different services; no description of on-scene triage process

Alternative routes of care include referrals to:

 i. GP out-of-hours service (face to face or via telephone)

 ii. pharmacy

 iii. MIU

 iv. rgent care centre

 v. social worker

 vi. psychiatric pathways

 vii. Community services (home attendance)

– Non-conveyance to ED was facilitated by:

 i. formal referral pathways

 ii. informal relationships with local services

 iii. Organisational facilitation of connectivity between ambulance service and other emergency and urgent care services.

– Ambulance trusts with higher rates of non-conveyance:

 i. had higher skilled paramedics

 ii. better valued training/skill of these skilled paramedics

 iii. Better organizational support

 iv. lower ED rates within 3 days of non-conveyed incident.

Positive support

– Non-conveyance variation between ambulance services is due to:

 i. staff skill (e.g. advanced paramedics)

 ii. perceived value of advanced paramedics

 iii. Perceived risk adverse views of senior management

 iv. commissioning of services.

– Standardisation of successful processes between ambulance services could reduce unwarranted differences in non-conveyance rates.

4

Pickstone 2019 [33]

UK

n = 1084

Retrospective audit of referral services

To determine if referral service reduces ED attendances

No description of triage process

– Referral to @home team (which offers 25 acute in-home clinical care pathways)

– 755 (72%) referrals (including ambulance, community services and acute settings) over a 3-month period were accepted, with an estimated 397 ED attendances prevented.

– This reduced total number of ED attendances by 0.3%.

Low support

– The @home referral service reduces ED attendances by a small amount.

– Investment of local health services does not have a sufficient impact on service delivery.

4

Schaefer 2002 [34]

USA

n = 1016 in intervention

n = 2617 in control

Prospective cohort study compared with a matched retrospective control cohort

To evaluate if EMTs can correctly triage patients alternate care destinations

Two criteria:

 i. non-urgent severity code

 ii. one of 24 diagnosis codes

Referral to:

 i. urgent care centres

 ii. walk-in clinics

 iii. GP practices accepting walk-in patients

– Intervention group resulted in:

 i. increased clinic care (8.0%;4.5%)

 ii. decreased ED conveyance (44.6%; 51.8%).

– Patients reported high satisfaction.

Positive support

– Alternate care destinations can safely reduce ED visits and provide satisfactory care.

– Further investigation of ways to ensure appropriate care of non-urgent patients is needed.

3

Schmidt 2000 [35]

USA

n = 1300

Prospective cohort study with linked retrospective EMS chart review

To evaluate if emergency medical technicians can safely apply protocols to assign transport options

Series of triage protocols for categories of complaints (e.g. musculoskeletal injuries)

Referral to primary care provider

– There was no difference in classification of transport decision between EMTs and first responders (e.g. fire departments).

– 3-11% of patients that were determined not to need ambulance had a critical medical event.

– Based on occurrence of critical events, protocol sensitivity was high (95%) and specificity low (33%).

Low support

– A better triage tool or improved triage adherence is required for EMTs to appropriately triage patients to alternate care routes.

4

Schmidt 2001 [36]

USA

n = 1300 (same sample as above)

Prospective cohort study with linked retrospective hospital chart review

To evaluate if emergency medical technicians can safely apply protocols to assign transport options

Series of triage protocols for categories of complaints (e.g. musculoskeletal injuries)

Referral to primary care provider

– 9% (13/140) of patients who were diverted away from ED were under triaged.

– Patients with psychiatric complaints and dementia are at higher risk of under triage.

Mixed support

– Protocols must be created and refined to minimise undertriage rates and ensure correct care pathways for patients.

4

Snooks 2004 [37]

UK

n = 409 in intervention

n = 425 in control

Cluster randomised controlled trial and semi-structured interviews

To

 i. evaluate effectiveness of direct transport of patients to Minor Injury Unit (MIU)

 ii. describe factors that impact MIU use through interviews with ambulance crews

Protocol outlining 23 minor injuries eligible for transportation to MIU

Transportation to an MIU

– Alternate transportation scheme:

 i. did not increase non-ED conveyance in intervention group (25.9%; 23.1%)

 ii. decreased job cycle time, time to treatment and time in unit compared to ED

 iii. Improved patient’s rating of care.

– Ambulance crews reported that location, patient needs, job times, improved service delivery and handover encouraged use of MIU.

Positive support

– Despite underuse of MIUs, there are no adverse consequences for correct use and many potential benefits.

2

Snooks 2004 [38]

UK

n = 251 in intervention

n = 537 in control

Phase 1:

Protocol development

Phase 2: Prospective cohort with matched control group

To develop and evaluate ‘Treat and Refer’ protocols

Treat & Refer protocols; training delivered to intervention crews (2-day course)

Referral to community based services (GP, district nurse, etc.)

– Referral scheme:

 i. did not increase non-conveyance in intervention group (37.1%; 36.3%)

 ii. improved documentation

 iii. Increased patient satisfaction

 iv. increased job times

 v. yielded safety concerns (5.4% of non-conveyed patients were admitted to ED w/i 14 days).

Mixed support

– Referral scheme did not reduce unnecessary ED conveyance, although patient satisfaction was improved.

– There were some concerns with the safety of referral protocols and further research is needed.

3

Tohira 2016 [39]

Australia

n = 67,387

Assessment of past patient care records

To evaluate if paramedics can safely identify patients who can be managed in the community

Ordinal triage scale to determine acuity; no clinical guidelines to determine transport

Referral to health services in the community

– 4.8% of ED-transported patients were identified as potentially suitable for community-care.

– 53.6% of these were admitted to hospital after direct ED conveyance.

– Patients identified as suitable for community care were more likely to require subsequent ambulance request, ED visit and hospitalisation within 24 h than those who were not.

Low support

– Paramedics were unable to accurately and safely triage patients to non-ED alternatives; this approach is high risk and requires further evaluation.

4

Verma 2018 [40]

Canada

n = 1851

Retrospective cohort study

To examine associations between paramedic home care referrals and use of services (911 emergency call, ED, home care)

No formal triage criteria

Referral to community services via Community Care Access Centres

- Referrals reduced 911 emergency calls by 10% and ambulance transport to ED by 7%.

Positive support

– Paramedics can successfully refer patients to community care access centres.

– This has promising benefits for reducing future emergency care access including reduced emergency calls and ED conveyances by ambulance

4

Vicente 2014 [41]

Sweden

n = 410 in intervention

n = 396 in control

Randomized controlled trial

To evaluate feasibility and safety of alternate transport to geriatric care

Decision support tools for 11 predefined conditions

Transportation to:

 i. geriatric care

 iv. community emergency care centre

– 20% of patients were transported to alternate route of care.

– 6.7% of non-conveyed patients required transfer to ED w/i 72 h.

Positive support

– Ambulance nurses can appropriately triage to alternate routes of care.

– Such schemes can prevent inappropriate use of the ED and improve care of older adults.

2

Villarreal 2017 [42]

UK

n = 23,395

Audit of routine ambulance data

Semi-structured paramedic interviews

Triage criteria covers 19 specific incident types

Referral via GP to:

 i. intermediate care teams

 ii. social services

 iii. Community hospitals

 iv. referral to patient’s own registered GP

– 78% of those who were referred to GP via telephone were not transported to hospital.

Patients were more likely to be referred to GP if they were:

 i. women

 ii. older

– assessed by GP face-to-face.

Positive support

– Collaboration between paramedics and GP can reduce unnecessary ED transfers.

– Recommendation for follow-up of hospital outcome and use of services in subsequent days to assess overall impact and safety of scheme.

4