Support and Assessment for Fall Emergency Referrals (SAFER) 2: a cluster randomised trial and systematic review of clinical effectiveness and cost-effectiveness of new protocols for emergency ambulance paramedics to assess older people following a fall with referral to community-based care when appropriate

Falls in older people

Falls in older people are recognised as an important issue internationally,1,2 with high human costs, for the individual and their carers, and high organisational costs. It is estimated that around 30% of home-dwelling people aged ≥ 65 years fall every year,3–8 and the rate of falls is even higher among care home residents. 9 Studies suggest that falls in older people are the cause of 20–30% of mild to severe injuries and 10–15% of emergency department (ED) attendances in this group. 10,11 Falls in older people are often due to more than one underlying cause or risk factor; as the number of risk factors rises, so does the risk of falling. 12 Risk factors include muscle weakness, balance and gait, low blood pressure (or drop in blood pressure on assuming upright posture) and cognitive decline. The severity of fall-related complications increases with age. 13,14 Falls are a cause of substantial rates of mortality and morbidity, as well as considerable contributors to immobility. 8 Reduction in quality of life and physical activity leads to social isolation and functional deterioration with a high risk of resultant dependency and nursing home institutionalisation. 15–18 Recovery from fall injury is often delayed in older people, which increases the risk of subsequent falls. 8 Recurrent falls are associated with greater physician contact and functional decline. 19–21 An additional complication is post-fall anxiety, leading to a fear of falling, which in turn can further contribute to deconditioning and weakness and in the long run may actually increase risk of falls. 8

The About the National Health Interview Survey22 indicates that falls are the largest single cause of restricted activity days among older adults. Most people who fall do not seek any medical advice, and those who do are more likely to report to their general practitioner (GP). 23,24 The combination of high falls with a high susceptibility to injury makes a relatively mild fall dangerous. This is because of a high prevalence of clinical diseases such as osteoporosis and age-related changes, that is slowed protective reflexes. 8 In the UK, it has been estimated that falls account for 3% of total NHS service expenditure,10 with additional costs incurred by social care providers. 25 The prevention of falls in older people has been highlighted as a priority. 11,26

Although prevention appears effective,11 reducing falls and associated morbidity depends on early identification of people at high risk and delivery of interventions across traditional service boundaries;27 these priorities are now reflected in national and international guidelines. 28–30

Emergency department overcrowding and potentially avoidable attendances

Population growth, the increasing burden of chronic disease and population ageing, and the shortage of health-care workers are affecting health-care systems in many countries. 31 EDs are under considerable pressure and overcrowding is a significant international problem, with a negative impact on both patient care and providers. 32 Department of Health data33 show that from 2003/4 to 2012/3 the number of attendances in English NHS ED units increased by nearly 32% (from 16.5 million to 21.7 million), although the majority of the increase was in minor injuries units and walk-in centres. Many of the older people presenting to EDs have had a fall. 34 One study showed that older people presenting to EDs after a fall represented 20% of all attendances and 14% of all hospitalisations in people aged > 65 years,35 and the size of the problem is likely to be underestimated as falls are poorly defined, coded and recorded. 36

Older people can be kept for a longer time in the ED than younger adults because of difficulties in assessment and management, that is, they undergo more tests than younger adults, they are more likely to have cognitive limitations37 and potential underlying diseases or, if they have common diseases, atypical symptoms,38 and they are also more likely to be admitted than younger adults. 39 Pressures on acute services to discharge patients as soon as possible may mean that there is little focus on fall prevention interventions. 36 Older people taken to EDs following a fall are at high risk of falling again in the next year, with a 30% chance of sustaining fracture or dislocation. 35

It has been argued that many admissions of older people are avoidable, and that the demands placed on the hospital system may be relieved if alternatives can be found. 40 The 2006 White Paper Our Health, Our Care, Our Say set the direction on ‘improving patient experience and significantly reducing unnecessary admissions to hospital’,41 with paramedics and enhanced/senior paramedics to play a role in providing care at the scene. The National Medical Director of NHS England recently proposed key areas of change in the 2013 Keogh report Transforming Urgent and Emergency Care Services in England 2013. 42 The reports look at developing the ambulance service into a mobile urgent treatment service capable of treating more patients at the scene so they do not need to be conveyed to hospital to initiate care. The report highlights opportunities to shift care closer to home, stating that 40% of ED patients are discharged requiring no treatment, up to 1 million emergency admissions were avoidable in 2012 and up to 50% of calls could be managed at the scene.

Conveyance rate of older adults who fall

Although demand on UK ambulance services continues to increase steadily, a recent study found that only 10% of calls are life-threatening, and an estimated half of all emergency calls relate to patients who could be treated at home. 43

People aged ≥ 65 years commonly call an emergency ambulance (through the emergency services) following a fall. In London (UK), this group accounts for about 60,000 attendances (8% of emergency ambulance attendances) each year. 44 This is very similar to the proportion reported in an urban emergency medical services (EMS) system in the USA. 43

Non-conveyance to EDs is high in this group, at close to 40% in London,44 elsewhere in the UK45,46 and the USA. 43 In most cases (90%), patients who fall and are not conveyed to an ED have fallen in the home. 46 Non-conveyance of patients is recognised internationally as a safety and litigation risk. 47 In some UK ambulance services, performance targets are set for individual enhanced paramedics and paramedics regarding the percentage of patients they should leave at home, and the enhanced paramedic or paramedic decides who can be safely left at home. In 2011/12, the Department of Health, in conjunction with the College of Emergency Medicine, introduced eight accident and emergency (A&E) clinical quality indicators as part of the NHS Outcomes Framework. These became mandatory from 1 April 2011. Indicator 1 (ambulatory care) measures the proportion of patients who are able to be treated at home by improved pathways of care for patients to avoid hospital admission. 48

Little is known about how, in the absence of specific guidelines or training to leave older people who fall at home, paramedics make these decisions. A study acknowledged the pragmatic nature of negotiation with patients whether or not to go to hospital. 49 A UK study identified that the non-clinical factors affecting these decisions include experience and confidence of ambulance staff, time into the shift, presence of carers, quality of the accommodation, waiting times at the local ED and prior knowledge of the patient. 50

Ambulance services, with key stakeholders, have been required to develop alternative care pathways, some of which involve the direct referral of older patients who have fallen to primary51 or community services. 52 This is particularly relevant at a time when pressures on EDs give rise to safety concerns and challenges to demand management;53 general practice and wider community services are being required to support the ED directly in order to avoid secondary care, in a time of major system reorganisation. 54 Alternative care pathways (avoiding use of the ED) for ambulance clinicians have been incentivised through local targets in the Commissioning for Quality and Innovation payment framework. 55 This emphasis is likely to continue given the pressure on EDs.

The National Service Framework for Older People26 advocates that ambulance staff/clinicians refer to community-based care older people who have fallen, although this reflects consensus rather than research evidence. Previous studies in this setting have found that change in practice is difficult to achieve and new pathways of care are difficult to exploit. 56

Service models: community falls services

A range of different types of trial performed in community-dwelling individuals (someone who lives among the general population and is not living in an institution),57 care home residents and hospital patients has established the evidence base for falls interventions. 11 Recent clinical guidelines from the American Geriatrics Society, British Geriatrics Society and American Academy of Orthopaedic Surgeons panel on falls prevention29 have strongly advocated preventative approaches based on multidimensional risk factor assessment, exercise programmes (which include balance, strength and endurance training) and environmental assessment and modification. A 2004 meta-analysis demonstrated a 37% risk reduction in the monthly rate of falling for community-dwelling individuals when multifactorial intervention was implemented. 58 Falls clinics are one approach by which falls and injuries can be managed multifactorially. 59 A falls clinic is a space where older adults prone to falling can receive holistic support and treatment from nurses, physiotherapists and other health professionals. A recent randomised controlled trial (RCT) showed that the Chaos Falls Clinic in Finland is effective in preventing falls in home-dwelling people over 70 years of age at high risk of falling. 60 Patients in this trial were guided towards the falls clinic by regional health-care professionals (physicians, nurses or GPs), but relatives of patients could also contact the clinic for an assessment for eligibility.

A recent study found that referral to a community-based falls prevention of older people who had fallen and been left at home by their attending ambulance clinicians, service reduced further falls and improved clinical outcomes,61 and achieved cost-effectiveness. 62 Recent policy changes in the UK have encouraged the development and implementation of alternative models of care for the delivery by the ambulance service. The literature points to the development of extended skills paramedics in line with government policies and supported by protocols. The use of enhanced paramedics has the potential to reduce running costs of the service considerably, while improving quality and focus, although significant ‘one-off’ costs associated with training are also necessary. However, Newman63 noted, the ‘emphasis on the national standards plus local flexibility appeared to reflect supposed shifts in the role of state’ but more particularly presented a range of models which resulted in ‘confusing messages about the relationship between government, the professionals and the public’.

Aim

To assess the benefits and costs to patients and the NHS of a complex intervention comprising education, clinical protocols and pathways enabling paramedics to assess older people who have fallen and refer them to community-based falls services when appropriate.

Objectives

• To compare outcomes, processes and costs of care between intervention and control groups:

  • patient outcomes: rate and pattern of subsequent emergency health-care contacts or deaths, for any reason; health-related quality of life; psychological status, especially fear of falling; and change in place of residence

  • processes of care: pathway of care at index incident; subsequent health-care contacts; ambulance service operational indicators; and protocol compliance including clinical documentation

  • costs of care: provided by NHS and personal social services (PSS); incurred by patients or carers in seeking care.

• To estimate wider system effects of the introduction of the intervention on ambulance service performance and costs.

• To understand how patients experience the new health technology.

• To identify factors which facilitate or hinder the use of the intervention.

• To inform the development of methods for falls research especially outcome measures recommended for trials of interventions for older people who fall. 59

• To carry out a systematic review on the implementation and impact of interventions by emergency medical staff who care for older people who fall.

Background

Since the publication of the National Service Framework for Older People 2001,26 the NHS has prioritised older people who fall. As the overall use of EDs increases, policy is shifting to encourage ambulances to convey fewer patients and refer them to other services, when appropriate.

Objective

To review evidence about effects of interventions within EMSs for older people who fall, which aim to reduce demand for EDs.

Method

We undertook a systematic search of 18 electronic databases. Studies were eligible for inclusion if they included empirical data on novel interventions by EMS in the community for older people who fall. We extracted outcomes, assessed studies for methodological quality and used narrative synthesis to analyse results.

Results

Of the 5355 articles identified, we selected 12 studies reported in 13 papers. Identified studies were excluded for the following main reasons: not English language, not EMS, not falls, no or entirely insufficient empirical data, no intervention or the outcomes of falls were not differentiated. The 13 papers reported data from 7411 participants: 7399 patients and 12 paramedics. Interventions fell into two groups: prospective screening on-scene or at the time of the emergency call and retrospective screening and referral to falls prevention.

Up to half of patients screened on-scene were conveyed to hospital, and referrals to falls services were low except when automatic (retrospectively). The majority of studies were judged to be of poor quality and data were heterogeneous, limiting opportunities for comparison and meta-analysis. Of the higher-quality studies, one trial reported a reduction in ED attendance and hospital admission and increased patient satisfaction; the other reported a reduction in ambulance calls over 12 months, with patients reporting fewer falls and increased confidence.

Discussion

Despite national policy to prevent falls and reduce use of emergency services, studies were few and evidence was mainly weak on effective ways to provide emergency care to older people who fall. When high-quality trial data were available, positive impact on patient and service outcomes was reported.

Study registration

This systematic review is registered on PROSPERO number CRD42013006418.

Background and rationale for the review

Since the publication of the National Service Framework for Older People 2001,26 the NHS has prioritised older people who fall. These people have reduced quality of life and levels of physical activity, leading to social isolation and functional deterioration with high risk of increased dependency and institutionalisation. 15–17 Treatment of falls is a major and rising cost for health systems internationally. In the UK, falls cost the NHS more than £2.3B per annum,67 about 3% of total expenditure. 10 In the USA, costs will increase by almost 75% between 2010 and 202068 and in Australia by more than double between 2004 and 2021. 69 Older people who fall and need emergency attention make up a substantial part of the EMS workload. In London, for example, they generate over 60,000 attendances, about 8% of the workload. 44,51 International proportions are similar. 70 Standard paramedic practice is to assess injury and immediate care needs, move patients from where they have fallen and convey them to an ED unless they refuse. 71 In the face of rising ED attendances, policy is shifting to encourage emergency ambulance services to convey fewer patients to EDs in the UK and internationally. 40,72

Research has highlighted opportunities for alternative treatment for older people who fall that reflect their health status and risk. 43,46,73,74 Some older people who fall and call the emergency services can be identified in the ambulance call centre, allowing rapid and targeted alternative responses. 75 The cost of EMS attendance to older people who fall is high. 76 Innovative approaches are necessary and feasible,77,78 but evidence about cost-effectiveness of alternative treatment is needed. 76

A Cochrane review11 of 159 randomised trials of interventions to reduce incidence of falls among older people living in the community reported that exercise interventions reduce risk and rate of falls. However, very few studies reported interventions by EMS staff. 61,62,71,79 Gates et al. 65 found limited evidence supporting fall prevention programmes in EDs, primary care or the community but did not evaluate of any initiatives delivered by EMS. More recently a systematic review by Mikolaizak et al. 74 reported that one study found that 49% of non-transported patients who had fallen had unplanned health contact within 14 days;44 another study reported that 33% of such patients were admitted to hospital within 28 days;80 and two studies that found that attendance by specially trained paramedics reduced subsequent falls, hospital attendance and other AEs. 61,66,74 However, this review did not assess the quality of included studies or observe Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines,81 thus limiting interpretation of findings. None of these reviews considered issues affecting implementation of interventions for older people who fall, notably the views of EMS staff delivering care.

Protocol and registration

We registered this review with PROSPERO and the All Wales Systematic Review Register. We adhered to the guidance for reviews of health care by the Centre for Reviews and Dissemination82 and PRISMA. 81

Eligibility criteria

We defined inclusion and exclusion criteria in terms of the population, intervention, comparator, outcomes, context and study design (PICOCS) of included studies, which is an adaption of population, intervention, comparator, outcomes (PICO). The additional ‘C’ in the acronym recognises the importance of the context within which the intervention is delivered. 82,83

We included studies that evaluated interventions by EMS staff in the community to older people who had fallen and generated an emergency service call. We considered any treatment above the care routinely provided by a paramedic or other EMS staff. Table 1 shows our inclusion and exclusion criteria.

Sources and search strategy

We searched published and grey literature and reference lists of included studies. We searched the electronic databases listed in Table 2 during the first week of July 2013 using a search strategy focusing mainly on the target population and the setting of the intervention. We used medical subject headings (MeSHs) and keywords when available. Table 3 shows the search strategy for MEDLINE, the Allied and Complementary Medicine Database and Cumulative Index to Nursing and Allied Health Literature Plus.

Our search strategy focused on two elements of PICOCS:

1. population, using terms such as ‘aged’, ‘health services for the aged’, ‘frail elderly’, ‘age’, ‘accidental falls’, ‘trip’ and ‘slip’

2. context, using terms such as ‘emergency medical services’, ‘ambulatory care’, ‘ambulances’, ‘emergency medical technicians’, ‘urgent care’, ‘emergency care’ and ‘prehospital’.

The terms for each database varied slightly according to MeSH or other indexed terms. The key terms were included, as well as keywords in title, abstract or text.

The search strategy used in three of the databases (MEDLINE, EMBASE and PubMed) can be summarised as follows:

• population, using the MeSH terms of ‘Aged’, ‘Health Services for the Aged’, Frail Elderly’, ‘Accidental Falls’ and the keywords of ‘age*’, ‘trip’ and ‘slip’

• context, using the MeSH terms of ‘Emergency Medical Services’, ‘Ambulatory care’, ‘Ambulances’, ‘Emergency Medical Technicians’ and the keywords of ‘urgent care’, ‘emergency care’, ‘prehospital’, ‘pre-hospital’ and ‘emergency medical service*’

• combination of elements, combining each of the population terms and each of the context terms with ‘OR’; and combing the population and the context terms with ‘AND’.

Study selection

We downloaded electronic search results into the EndNote bibliographic software (Thomson Reuters, CA, USA). One researcher (RC, MH or MK) screened the title, abstract and keywords of each study using the inclusion and exclusion criteria in Table 1. A second researcher, blind to the first researcher’s decision, screened 1 in 10 references, to ensure consistent decision-making. We resolved differences in discussion with the third researcher. We retrieved the full text of papers whose abstract met the inclusion criteria. We compared final recommendations and again achieved consensus on eligible studies in discussion with the third researcher.

Data extraction

We developed and piloted a data extraction framework reflecting the objectives of the review in line with guidance. 82 Two researchers independently extracted the following items:

• year of publication, country of origin, setting, study design

• aim or research question, outcome measure(s), method(s), study population, sample size

• description of intervention

• key findings.

When possible, we used the taxonomy developed by the Prevention of Falls Network Europe (ProFaNE),84 from the work of Hauer et al. ,85 to describe these data, thus increasing consistency of extraction across the included studies and opportunity for synthesis. The full data extraction table can be found in Appendix 2. We compared completed forms and agreed on content in discussion with the third researcher.

Assessment of the risk of bias (quality)

We assessed quality of included studies using checklists recommended by Centre for Reviews and Dissemination82 to inform comparisons and guide interpretation. 85 We identified these checklists as potentially suitable in the Centre for Reviews and Dissemination guidance. We used the Scottish Intercollegiate Guidelines Network checklist86 for all study methods except qualitative reports, for which we used the summary criteria of Walsh and Downe. 87 We assessed general methodological quality as high, acceptable or low, defined by Scottish Intercollegiate Guidelines Network86 thus:

• high: almost all criteria met; low risk of bias; conclusions unlikely to change after further research

• acceptable: most criteria met; some flaws with an associated risk of bias; conclusions may change after further research

• low: either most criteria not met or clear flaws in study design; conclusions likely to change after further research.

Data synthesis

We assessed quantitative results for their potential to contribute to a meta-analysis. We used narrative synthesis to synthesise heterogeneous data. 88 We explored relationships within and between studies, seeking to explain similarities and differences between study findings.

Inclusion

We identified a total of 5355 references. Our review of titles and abstracts suggested that 138 articles might be eligible. After studying full texts, we confirmed that 13 papers61,62,66,80,89–97 reporting 12 studies met our inclusion criteria, as detailed in Figure 1. Table 4 summarises the 12 included studies; Tables 5 and 6 describe them in more detail and Table 7 assesses their quality. Appendix 3 lists excluded studies (at full-text screening stage) with reasons for exclusion.

FIGURE 1.

The PRISMA flow chart.

Study characteristics

Of the 12 studies, six61,62,66,80,89,94 came from the UK, four90–93 came from the USA, one97 came from Australia and one95 came from New Zealand. Study designs comprised two RCTs61,66 (one of which61 published a separate paper evaluating cost-effectiveness element of the study),62 nine cohort studies (of which six did not include a comparator arm89,91–93,95,97 and three included intervention and control arms80,90,96) and one qualitative study. 94 Of the nine cohort studies, three were retrospective,92,95,96 five were prospective80,89–91,97 and one had both prospective (screening) and retrospective (referral) elements. 93 These study characteristics are summarised in Table 4.

The one-armed nature of six89,91–93,95,97 of the studies restricted them to reporting descriptive outcomes. All but Logan et al. 61 reported outcomes at the time of the intervention or soon after. The most common outcome measures were referrals made to a receiving unit or service,66,89,91–93,97 acceptance of referral by the patient66,89,91,92,97 and patient conveyance following attendance by EMSs. 66,80,95 Five studies reported follow-up outcomes for patients receiving a novel intervention (see Table 6). 61,66,80,90,97 Three66,80,90 used short follow-up periods (14 and 28 days) with limited inclusion of outcome measures in two,80,90 whereas another collected physical health, quality-of-life and mortality outcomes. 66 One written as two research papers reported outcomes over 12 months;61,62 but one did not report the period of follow-up. 97

Ten61,66,80,89,90,92,93,95–97 of these studies included a total of 7399 patients and the qualitative study interviewed 12 paramedics. 94 However, one study did not specify its sample size. 91

Methodological quality

The methodological quality of the studies was variable, but was generally weak (see Table 7). The controlled trials,61,66,90 economic evaluation62 and qualitative study94 were considered acceptable or high quality and scored highly for internal validity. The cohort studies were mostly of weak quality and did not report sufficient information to determine if their methods minimised risk of bias or confounding. Seven out of eight of the cohort studies80,89,90,92,93,96,97 failed to identify potential confounders in study design and analysis. Shah et al. 93 and Metcalfe89 did not provide a focused research question. Gray and Walker80 scored fairly low using the quality assessment tool, but reviewers defined this study as of acceptable quality as, compared with the other studies included in this review, it included a control arm, a clearly focused aim and clearly written methods and results sections. Kue et al. 92 did not define outcomes or reliably assess exposure. However, reviewers considered all studies that reported data that were directly relevant to this review, and all were included. 98 The full quality assessment for each included study is presented in Appendix 2, with a brief summary of the internal validity score, risk of bias and noted limitations given in Table 7.

Study findings

Strengths and limitations of this review

We used a systematic process in line with good practice82,99 to identify international research and review reported data. In undertaking this review, we were unable either to consider papers in languages other than English or to request original data from study authors because of time constraints.

Heterogeneity of interventions and outcomes and low study quality mean data should be interpreted with caution. It also limits ability to generalise findings.

The quality assessment tools that were utilised were not tailored to single-group cohort studies. Therefore, many of the questions were not relevant to the majority of studies that were included in this review. Nevertheless, the inability to answer a question was regarded as a sign of low quality.

Review findings as context for Support and Assessment for Fall Emergency Referrals 2

The SAFER 2 trial selected the age of 65 years and older as its cut-off point for inclusion, as both the National Service Framework for Older People 200126 and the National Institute for Health and Care Excellence (NICE) guidelines for falls67 use this cut-off point, although they recognise this as an arbitrary point that varies in relevance to individual health and disability.

In the systematic review, however, we chose to also include studies that stated their lower age for inclusion to be 60 years, as we were aware from our reading on the topic that two61,66 UK trials had used this slightly lower age breakpoint.

Although we recognise that this might suggest that the cut-off point of ≥ 60 years of age could or should have been used for the SAFER 2 trial, we are reassured that our results are comparable when we look at the mean age of participants in the 10 of the 12 studies that provide us with this demographic information about participants:61,66,89–93,95–97 7 of these 10 studies report mean ages of participants in the ‘frail older people’ group, ranging from 77 to 83 years61,66,90–93,97 and one of the most common age group of participants as 80–89 years. 89 The two studies of exception were those where fall was one condition only amongst a range studied, and these reported average ages of 4796 and 62 years. 95 We therefore feel confident that the different lower age parameter for the systematic review and the trial did not impact on relevance of one to the other.

We suggest that the fact that this review found limited evidence on the topic of support and assessment for older people who fall and are attended by the emergency ambulance service, and even more limited evidence of high quality, is the context and justification for the SAFER 2 trial, the findings of which are now reported.

Aim

To assess the benefits and costs to patients and the NHS of a complex intervention comprising education, clinical protocols and pathways enabling paramedics to assess older people who have fallen and refer them to community-based falls services when appropriate.

Objectives

• To compare outcomes, processes and costs of care between intervention and control groups:

  • patient outcomes: rate and pattern of subsequent emergency health-care contacts or deaths; health-related quality of life; falls efficacy (fear of falling); and change in place of residence

  • processes of care: pathway of care at index incident; subsequent health-care contacts; ambulance service operational indicators and protocol compliance including clinical documentation

  • costs of care.

• To understand how patients experience the new health technology.

• To identify factors which facilitate or hinder the use of the intervention.

• To inform the development of methods for falls research.

Trial design

The trial was conducted in geographically defined sites within three UK ambulance services. The unit of clustering at each site was an ambulance station. The clusters comprised the paramedics based at eligible stations who volunteered to participate in the trial. Once stations had been randomly allocated to intervention or control groups, participating paramedics based at those stations were asked to deliver care according to their group allocation, irrespective of their location on any particular shift.

We chose a cluster design, as opposed to a patient-level RCT, as the intervention included training and new assessment skills, and it would be impossible to switch on and off as would be required in a patient-level RCT. Other units of randomisation have been used in trials based in prehospital emergency care, with limited success. 100 We chose to randomly allocate stations in order to allow the hosting ambulance services to support change in practice for paramedics based at intervention stations while minimising contamination to practice by paramedics based at control stations.

Trial management

Following the Medical Research Council guidelines for good practice in clinical trials,101 the management structure for the trial (see Appendix 4) included an independent Trial Steering Committee (TSC) and Data Monitoring and Ethics Committee (DMEC), with an internal Trial Management Group (TMG), local implementation team (LIT) in each area, core research team, and task and finish groups for specific aspects of the trial such as data management. The TSC provided external oversight and advice to the chief investigator, the Health Technology Assessment (HTA) programme and the sponsor on all aspects. The DMEC had access to unblinded comparative data to monitor the data and make recommendations to the TSC if there were ethical or safety reasons why the trial should not continue. The TMG managed the trial at an overarching level. The LITs dealt with operational issues at each site and provided liaison opportunities between participating services. The core team that managed the trial at a day-to-day operational level was smaller and included the chief investigator, site principal investigators and researchers.

Setting

We undertook the trial in prehospital emergency care, with paramedics delivering the intervention in partnership with community-based falls services (known as falls prevention services). We selected sites within three ambulance services in England and Wales in which a falls service was available, but no process was in place for paramedics to make direct referrals from the scene of emergency service call attendances. Appendix 5 maps stations and hospitals with ED departments at each site.

Participants

Ambulance stations were eligible for inclusion in the study as a cluster if they were situated within a participating ambulance service and were a base for paramedics who regularly attended patients within a participating falls prevention service’s catchment area.

We invited paramedics based at selected ambulance stations to participate in the trial before allocating those stations randomly between intervention and control groups. Paramedics were approached via e-mail, letter and station posters and were asked to return a reply slip if they were interested in participating. Paramedics who volunteered to participate were given a £50.00 voucher.

Patients received either intervention or control group care, depending on the group allocation of the attending paramedic. During training, intervention paramedics were instructed to lead on care if there was also a control group paramedic on-scene, and these patients were included in the intervention group for analysis.

Inclusion criteria:

• aged ≥ 65 years

• resident in the catchment area of participating falls services

• attended by a study paramedic following an emergency call to the ambulance service which was coded by a dispatcher as a fall without priority symptoms (Advanced Medical Priority Dispatch System Code 17)

• first eligible call during the trial period.

Randomisation

Randomisation of stations to groups was carried out in accordance with principles related to cluster randomisation and blinding of analysis team, as outlined in the West Wales Organisation for Rigorous Trials in Health and social care (WWORTH) standard operating procedure (SOP) on randomisation. 102 The trial statistician (WYC) and manager (JP) undertook randomisation after recruitment of paramedics had closed, to avoid selection bias, using trial site, volume of calls, falls service and the number of participating paramedics as stratification variables.

Recruitment

Patients were identified as being potentially eligible for study inclusion from routine emergency service dispatch records through standardised queries written for each site. Site researchers confirmed eligibility of individual patients by retrieving corresponding patient report forms (PRFs), which are routinely completed by paramedics when they attend a patient. PRFs include patient identifiers and demographics; and operational, clinical assessment and treatment information.

In this cluster trial we did not ask paramedics to approach patients for consent to participate in the trial during the emergency episode. Instead, following identification of eligible patients as detailed in Participants, we sought consent from patients for follow-up through routine medical records and by postal questionnaire. Following considerable discussion with the Research Ethics Committee, our consent process included contacting patients by post and then, if necessary, by telephone or a home visit. Patients then had the opportunity to actively ‘consent’ or ‘decline consent’ (dissent). However, we were unable to contact some patients at all (no response by post or telephone). We had Health Research Authority Confidentiality Advisory Group (HRA CAG) permission to follow up this group anonymously (but they neither ‘consented’ nor ‘declined consent’). The group we analysed consisted of those who had consented and those who we could not contact (follow-up was through an anonymised route). Only those who had actively declined consent (dissented) were excluded from this group. The postal consent pack (see Appendix 6) included:

• covering letter

• consent form

• patient information sheet

• 1-month questionnaire (see Appendix 7)

• Freepost return envelope

• £5.00 voucher, to thank participants for their time.

Interventions

Outcomes

Outcome measures at 1 month and 6 months after the patient’s index incident were consistent with recommendations of ProFaNE. 103

Data collection

• Ambulance service: routine clinical and operational management information about index and subsequent contacts from despatch records (completed by emergency service call takers) and PRFs completed by paramedics during or immediately after patient attendance.

• Falls services: information about referrals and falls service response.

• Participants (or their carers): questionnaire data, sent by post to eligible patients at 1 month (see Appendix 7) and those who consented to a further questionnaire at 6 months (see Appendix 10).

• Health and Social Care Information Centre (HSCIC) (England) and National Welsh Informatics Service (Wales): routine data about ED attendances, hospital admissions and deaths for all matched study patients over the study period.

At all three sites, patients who dissented by any means were excluded from all data collection. For all other patients, we collected anonymised data on index and subsequent health-care contacts. Identifying information collected at ambulance and falls services was split from clinical information by our trial partners within the three ambulance services. In order to improve the matching rate achieved by National Welsh Informatics Service and HSCIC, our ambulance service partners used imperfect emergency service call information such as name, address and date of birth to look up NHS numbers. Once patients were matched, clinical outcome data were retrieved and transferred for analysis into the Secure Anonymised Information Linkage (SAIL) gateway when it was joined up with questionnaire data.

Data entry

The method of data entry varied by source and site. Ambulance service despatch data were downloaded electronically to study databases. There were two types of PRF: paper and electronic. Sites 2 and 3 used only paper forms and site 1 used a mix of paper and electronic forms. Paper forms were input into study databases manually by the study team, whereas electronic forms were input automatically after they were downloaded from a central server. Questionnaires were input into study databases using Teleform software version 10 (Cardiff Software, Sunnyvale, CA, USA). A quality assurance check was carried out on all manually entered data.

Blinding

Blinding was carried out when possible. Paramedics, trial managers and site researchers were not blinded, as they needed to know the allocation of participating ambulance stations for operational reasons. The trial statistician and all other TMG members were kept blind to allocations in order to eliminate reporting bias. This was undertaken by coding the allocated groups in order to prevent group identification. Blinding statisticians is a technique recognised, both by statisticians in general and by the WWORTH SOP on statistics,107 as best practice in statistical analysis. Even though the WWORTH SOP also requires that trials agree statistical analysis plans before starting analysis, these cannot cover all eventualities, with the result that unblinded statisticians could unconsciously take analytical decisions that favour intervention over control or vice versa. Unblinding occurred after completion of primary analyses.

Sample size and power

We estimated our trial sample size from our principal outcome: the proportion of participants who, within 6 months, die or contact emergency services. From previous trials of interventions for older people who have sustained a fall and presented for emergency treatment, summarised in a recent systematic review,108 we made the conservative estimate that trial patients had about a 50% chance of making another emergency contact or dying within 6 months. We judged that a change of 5% in this proportion could be clinically and economically important. In the absence of clustering, a sample size of 4190 evaluable participants would have yielded 90% power to detect a change of at least 5% (from 50% to ≤ 45% or ≥ 55%) when using a two-sided 5% significance level. As participants came from 25 clusters, we needed to adjust this sample size to allow for intracluster correlation (ICC). We estimated this ICC from the findings of the SAFER 1 trial,79 which evaluated the clinical effectiveness and cost-effectiveness of computerised clinical decision support software for use by paramedics when attending older adults who had a fall. The SAFER 1 trial estimated the ICC for the same outcome, but over 1 month rather than 6 months, as 0 when clustering participants by station (as in the SAFER 2 trial), but 0.005 when clustering participants by paramedic (as in the SAFER 1 trial). 79 To be conservative, we allowed for an ICC of 0.002 by increasing the target evaluable sample to 6290, namely 25 clusters × 251.6 participants per cluster. This sample would have more than 90% power to detect a change of 0.18 in the number of emergency contacts of 1.8 over 6 months, given a standard deviation (SD) of 1.5. Hence, the SAFER 2 trial was able to detect a difference of 1 emergency contact in 10 avoided (or induced) by the intervention.

We had originally postulated that patients recruited to the study would have a 40% chance of making an emergency contact or dying within 6 months and that the ICC could be as high as 0.03. Under those assumptions our target sample of 6290 would have yielded 80% power to detect a change of at least 10% (i.e. from 40% to ≤ 30% or ≥ 50%) when using a two-sided 5% significance level. When the SAFER 1 trial showed that the assumed ICC was unduly pessimistic, recruitment was progressing well. Therefore, rather than finish the trial early, we decided, with the approval of both TSC and DMEC, to be less conservative in assuming a worst ICC of 0.02, thus yielding enough power to detect a change of only 5% in the emergency contact rate, still a clinically important difference in the view of our advisers.

Statistical methods

The full analysis plan can be found in Appendix 11. Primary analysis was by ‘treatment allocated’. Analyses included logistic regression for binary outcomes; cross-tabulations and risk ratios for categorical outcomes; and survival analysis including Cox’s proportional hazards models for times to events. We used multilevel modelling to estimate (random) station effects and (fixed) group effects and analysed repeated observations as such.

Our principal outcome is made up of a hierarchy of events. We therefore undertook analysis incrementally: first, deaths; second, emergency admissions plus deaths; third, ED attendances plus admissions and deaths; and, finally, emergency service calls plus attendances, admissions and deaths. Although we originally planned to undertake analysis for all events and for those coded as a fall, in practice data quality across the levels precluded analysis of falls only. We have presented results as follows: the proportion of patients who called the emergency services, attended ED, were admitted or died; survival analysis of the time to the first subsequent emergency contact; the mean number of further emergency contacts adjusted for time at risk, excluding days in hospital or after death; and recurrent event analysis when feasible. We also examined the effect of the intervention on patient satisfaction, health-related quality of life and costs (as described under Economic evaluation).

Potential predictors of triage decisions include the distance between the site of the index event and the ED; patients’ age, sex and history of previous falls; type of presentation [specifically, a seasonality term capturing when, during the calendar year, the call was made, and an indicator of whether or not it was made out of (GP) hours]; and time since recruitment started (recruitment point). The list of adjusting factors (covariates) includes those suggested by the TSC and may be divided into three subsets: (1) group and site interactions, reflecting study design; (2) participant characteristics (age and gender); and (3) characteristics related to the timing and location of the index incident. We used the covariates in the analysis. It is possible that patients in the catchment area of one station may receive care for a subsequent event from a paramedic based at another station participating in the study but allocated to a different group. Nevertheless, analysis was still by treatment allocated. We have identified referrals made to falls services at subsequent emergency service call attendances and are able to use this for censored analysis in order to minimise contamination, but this is not included in the current report.

We proposed to identify any wider system effects, by comparing response times during the trial period across the study catchment area and surrounding areas with pretrial response times and response times elsewhere. This analysis would only be necessary if effects were found on ambulance operational performance indicators – job cycle time within the trial population.

To inform the development of outcome measures for falls research as recommended by ProFaNE,103 we compared SF-12 and derived Short Form questionnaire-6 Dimensions (SF-6D) scores with mFES scores to establish their construct validity. We also assessed their predictive validity by comparing scores with the number of further events and the time to the first subsequent event.

Missing data

We adopt a consistent approach to missing data relating to both clinical effectiveness and cost-effectiveness, except when individual outcome measures require some variation in that approach. For each variable, we consider the frequency of missing data; if there is no reason to suspect that data are not missing completely at random, we use appropriate imputation methods to mitigate the problem of missing data. Specifically, in addition to taking the SF-6D score as 0 when a participant is dead at the data collection point, we imputed missing values by regression (thus adjusting for other covariates) from all available values of that score at other data points.

Adverse events

As the study population had high mortality and morbidity, we did not routinely record or report AEs that were neither serious nor adverse reactions (ARs) in the sense of possibly being caused by the new clinical protocol for referring to falls services. The main potential AR is misdiagnosis, which could lead to an inappropriate pathway of care. As misdiagnosis is reliably identifiable only through patient complaints or coroner’s inquest, we focus on these, and treat them as serious ARs. Any patient complaint or coroner’s inquest at which the ambulance service was asked to supply information related to non-conveyance of a trial participant from the index incident triggered was investigated by the local principal investigator and chief investigator. We also investigated suspected ARs brought to our attention in any other way.

Death or emergency hospital admissions are serious adverse events (SAEs). As these form the primary outcome of this trial, and are not unusual or unexpected in the study population, we report them at the end of the trial. In particular, the imbalance between intervention and control groups in the occurrence of SAEs or serious ARs was the subject of statistical analysis at the end of the trial.

Aim

We undertook an economic evaluation alongside the RCT from the perspective of the UK NHS and PSS, in line with the approach recommended by NICE. Economic analysis estimated the costs of providing the intervention and the consequences of the scheme for the NHS and PSS in terms of inpatient admissions, ED attendances, GP consultations, out-of-hours GP contacts, NHS Direct contacts and use of social services.

Method

We estimated the costs of providing the new intervention by utilising data collected from financial reports and documents, relevant information logged as a part of routine practice and from resource utilisation recording sheets, together with reference to patient records and discussions with relevant finance staff. We collected data on participants’ use of health service and social services resources from paramedic records, routine hospital records and patient-completed questionnaires. We estimated NHS resource use from routine data including duration of ambulance job cycles and episodes of care; records of resource use; and patient records. We also estimated social services resource costs from discussion with relevant social services departments. Costs were then calculated using published unit costs such as the Personal Social Services Research Unit (PSSRU) Unit Costs of Health and Social Care 2013123 and NHS Reference Costs 2011–12. 109

Health-related quality-of-life outcome

The SF-6D scores (derived from SF-12 scores) were used to estimate the QALYs gained from the intervention and an incremental cost per QALY, applying appropriate threshold incremental cost-effectiveness ratios, such as the £20,000.00 and £30,000.00 per QALY thresholds used by NICE. These ratios were presented along with their associated cost-effectiveness acceptability curve.

Lost to follow-up/missing data

If a participant was dead at the data collection point, the SF-6D score was taken as 0. To avoid outliers, SF-6D and mFES scores were taken to be the minimum value observed for that measure in the relevant treatment group. The problem of missing data was addressed using an appropriate regression-based imputation approach, as reported earlier.

Discounting

As the outcomes were assessed at 1 month and 6 months, no discounting was employed.

Uncertainty

We addressed uncertainty by applying bootstrapping for cost-effectiveness acceptability curves and CIs. Sensitivity analyses were undertaken to assess the robustness of results to changes in the configuration of the intervention and other health service costs.

Aim

The qualitative component of the trial was designed to address the following two of the study objectives:

1. Gain an in-depth understanding of how the intervention is experienced by patients.

2. Understand how the intervention is delivered in practice, identifying factors which enable or hinder its use.

Participants

Patient interviews

A schedule (see Appendix 12) for face-to-face, semistructured patient interviews was developed by the research team, drawing on the aims and objectives of the study, in order to explore:

• patients’ experience of the ambulance service

• patients’ experience of those seen by a falls service

• patients’ health since fall

• patients’ satisfaction with treatment.

Face-to-face semistructured interviews were conducted by experienced researchers from the study team (RA, RC, MK, LW). In some cases, a carer or other person who had been present at time of fall also took part in the interview or was present to support the participant. Interviews took place in patients’ own homes, and were recorded and later transcribed in full.

Paramedic focus groups and interviews

Data collection with paramedics took place at two points in the study:

1. at baseline, after paramedics had been trained in the SAFER 2 trial intervention but before they started using it in practice (see Appendix 13)

2. after the end of the patient recruitment period (see Appendix 14).

Separate topic guides were developed by the research team for each phase of data collection, drawing on the aims and objectives of the study. The topic guides were reviewed by members of the LITs, which included ambulance service personnel at different clinical, operational and management levels. The focus group topic guides were adapted for use in one-to-one interviews. The topic guide covered:

• paramedics’ views and attitudes towards the new intervention

• any preconceptions about the new way of working

• factors which enabled the use of the new referral pathway

• factors which hindered the use of the new referral pathway.

Each focus group was led by one researcher, with a second member of the team acting as support and taking notes to enable the linkage of texts to speakers, and noting other details, such as points of consensus or disagreement, issues that drew strong emotional responses such as anger, fear or anxiety. All focus groups and interviews, with the permission of participants, were recorded and transcribed in full.

Stakeholder focus groups and interviews

Topic guides (see Appendix 15) were developed by the research team for use with ambulance service personnel and wider stakeholders, and covered:

• perception of how well the intervention worked

• impact on the stakeholder’s organisation

• cross-organisational working

• views on the process of implementing the intervention.

Stakeholder focus groups took place before the main results of the trial were known. Each was led by one researcher, with a second member of the team acting as support and taking notes to enable the linkage of texts to speakers, and noting other details, such as points of consensus or disagreement, issues that drew strong emotional responses such as anger, fear or anxiety. Each interview was carried out face to face, and was conducted by one member of the research team. All focus groups and interviews, with the permission of participants, were recorded and transcribed in full.

Qualitative data management

As this was a team-based project involving collection of data in multiple sites, maintaining consistency in data collection and transcription was crucial. The co-ordinator set up a data management system consisting of instructions on converting raw data to computer files (in the form of a transcription protocol), on organising data storage and on data archiving steps and a data management checklist. The transcription protocol ensured that standard conventions were adopted throughout the transcription process, and that a standard presentation format was used. All transcripts were anonymised and stored securely. Transcripts of interviews with stakeholders were sent back to the participant for checking, to ensure that they were accurate.

Qualitative data analysis

The data were analysed thematically using a modified version of the framework analysis approach for applied policy research. 111 This is a systematic, dynamic and transparent method of analysis, which generates themes from the original accounts of participants. 112

Two separate analysis tasks were carried out: data from patient interviews were analysed to address objective 3; and data from focus groups and interviews with paramedics and stakeholders were analysed to address objective 4. Each task consisted of the following stages:

• The transcripts were distributed for reading among members of the research team (RA, RC, SG, MK, AP, LW and AT, the patient representative) so that each had at least five to read, with each transcript being read by two people.

• Each member of the research team made notes on themes and ideas, guided by the relevant study objective.

• The research team met to discuss and agree a coding frame.

• Two researchers independently coded up two transcripts each using the coding frame, discussed how well it worked, and then revised the coding frame. The final coding frames are shown in Appendix 16.

• RA, RC, MK and AP used the revised coding frame on one transcript, then compared coding to check consistency.

• Transcripts were then shared among RA, RC, MK and AP for coding. RA then entered all codes into an NVivo database (QSR International, Warrington, UK) (one database relating to objective 3 and one relating to objective 4).

• RC, MK and AP met to discuss overall themes and findings relating to each objective, drawing on the NVivo database and the coding frame.

The qualitative findings in Chapter 6 present anonymised quotations from participants. In line with the guidance from Corden and Sainsbury,113 quotations have been used to present evidence, to support explanation of complex ideas, to provide illustration, and to deepen understanding, particularly when respondents had an emotional response. Although many quotations were selected to illustrate the majority view, some show exceptional viewpoints, as noted. In some cases, quotations were selected to illustrate complexity or ambiguity of perspective, as noted.

Involvement of service users

Involving service users in research is encouraged to improve relevance, quality and accountability of research. 114–117 The SAFER 2 trial followed the principles and procedures outlined in the WWORTH SOP for user inclusion118 to ensure that members of the public were actively involved throughout the research process. This SOP provided a starting point for developing a model of user involvement in the SAFER 2 trial. Moreover, the SAFER 2 trial aimed to address the challenges of involving older people, who are often in poor health because of their falls history or falls risk. The model took into account the multiple layers of a multisite trial and proposed three tiers of involvement in different forums at the strategic, site and local levels where overall or specific input could be made. We followed best practice guidance by involving two service users in each of the strategic trial committees. 114,119,120 We also involved service users at site (n = 3) and local levels (n = 18) of the trial to provide a range of forums and opportunities for involvement by older people with risk of falling. Our involvement model received the Involving People Award in 2014 as an example of best practice.

Trial registration

The trial is registered as Current Controlled Trials ISRCTN60481756 and UKCRN 6801. The trial protocol has been published. 71

Individual patient level

Although we expected recruitment of patients to be similar across ambulance service sites, and similar numbers of ambulance stations participated in the trial at each site, recruitment of patients was much higher at site 1 as a result of several factors including:

• longer recruitment period at site 1

• operational practice which took paramedics out of the trial catchment area, particularly in site 3

• loss of six stations and related paramedics at site 2.

There was also variation in recruitment between trial arms across sites, with a higher proportion of control group patients recruited at site 1 than at either site 2 or 3. There was very little difference between groups in age, overall or at any site. There were more women than men recruited to the trial, although the proportion of men was slightly higher in the intervention group than the control group, with about 3% more at each site and overall. There was little difference between trial arms in the proportion of calls made outside usual GP operating hours was similar between trial arms, distance to ED or time to recruitment (Table 13). There were some differences between sites, for example site 3 had a lower proportion of female patients, a lower proportion of calls made ‘out of hours’ and shorter distances to ED. This was our most urban site, which may have accounted for these differences in study population.

Cluster level

We randomly allocated one more station to the intervention group than the control group at each site. However, we recruited fewer paramedics recruited per station in the intervention arm than in the control arm at each site.

Contamination

Table 22 shows the percentage of control patients subsequently attended within the trial period by a paramedic from the intervention group, < 5% at 1 month rising to about 15% at 6 months, with broadly similar rates at each site.

Patients ineligible for referral to the falls service

A subgroup of participants in site 1 were registered at GP practices without agreement for referral of patients to falls services and were less able to benefit from the full SAFER 2 intervention. Appendix 19 therefore presents the main results after excluding these patients. Results in these tables are consistent with those presented in the preceding sections.

Patterns of referral to the falls service

Sixty per cent of designated intervention group paramedics referred trial eligible patients to a falls service (Table 23). Paramedics referred up to 11 patients, although most made referrals once or twice. Patient age, sex, distance to ED did not appear to influence referral (Table 24); however, patients were more likely to be referred to falls services out of hours and less likely to be referred at site 1 than at the other sites (Table 25).

For most patients left at the scene by intervention group paramedics without referral to a falls service, no reason was identified. At site 1, 231 patients were not eligible for referral because, although they lived in the study catchment area, they were not registered with participating GPs. Other reasons for non-referral included refusal, not appropriate for referral or information that the patient was already under the care of a falls team or other service (Table 26).

Comparison of outcome measures

To meet our methodological objective we compared the performance of the generic SF-12 with the mFES (‘Fear of Falling’), a falls-specific outcome measure, in this population (Table 27 and Figures 3–6). The SF-12 mental component summary (MCS) and physical component summary (PCS) scores at 1 month both showed mean values at the lower end of the scale (lower scores indicate worse health), indicating mental and physical scores below the population norms. Similarly, the mFES at 1 month showed a clustering of scores at the lower end of the scale, indicating worse health. At 6 months, both scales again showed mean score values at the lower end of the scale, indicating worse health. There was, however, a small increase in MCS, PCS and mFES scores, indicating a slight improvement in patient-reported outcomes since the 1-month follow-up.

FIGURE 3.

Short Form questionnaire-12 items PCS and mFES scores at 1 month: correlation = 0.546.

FIGURE 4.

Short Form questionnaire-12 items MCS and mFES scores at 1 month: correlation = 0.553.

FIGURE 5.

Short Form questionnaire-12 items PCS and mFES scores at 6 months: correlation = 0.627.

FIGURE 6.

Short Form questionnaire-12 items MCS and mFES scores at 6 months: correlation = 0.547.

There was moderate correlation between the SF-12 MCS and PCS scores and the mFES scores at both 1 and 6 months (range 0.55–0.63). Unusually, there were significant floor and ceiling effects for the condition-specific measure, mFES, indicating less sensitivity to detect small changes in patient outcomes than the SF-12 disease-specific measure.

The fall in context

Every patient we interviewed described having a fall. Fifty-one respondents talked about the circumstances of the index incident and what they felt caused it. The majority (28) of the falls were described as taking place within the home, many of them trips and slips when moving around or using steps, and eight were cases of respondents sliding off chairs or the bed. A further 11 of the respondents described falls in the garden or outside the house, while hedge-trimming or watering plants, shovelling coal, putting the bins out, and so on. The remainder of the falls happened while the respondent was out and about. A minority of respondents described injuries resulting from the fall, ranging from fractures and being knocked unconscious to cuts and bruises.

Some falls were described as having a clear mechanical cause, such as 2–6H missing his footing on a stepladder while cutting the hedge. However, for many more, both inside and outside the home, the cause seemed unclear, as in the case of one respondent who fell in her kitchen:

All of a sudden, I never felt it, it just . . . just . . . like it just . . . my mind went blank. I just felt myself falling and when I looked around I was on the floor. So I thought to myself ‘silly bugger’.

3–21N

Many of the respondents explicitly stated that they did not know why they fell, while for a striking number the fall was something which ‘just’ happened: ‘my leg just gave way’ (1–1H); ‘I just dropped’ (1–7F); ‘I just slid down’ (1–2N), and so on.

Others seemed to be seeking an explanation for why they fell:

But you know being in the middle of the night, waking up out of a sleep, you’re not very steady, are you? I mean in the best of times.

3–9F

The falls happen like somebody’s switching off the light switch according to the doctor. It’s not a stumble or a trip, it–, this is just a fall which happens [clicks fingers], which is related to probably blood pressure or whatever. They are looking at the fact an epileptic fit may be involved here.

2–7H

In some cases this was expressed in terms of wanting to find an external cause for the fall:

But as I say the fact that I fell was simply something mechanical. I took a step back and I’d got a saucepan in one hand and a spoon in the other so I wasn’t holding onto things as I usually do. It’s the sort of thing that can happen to anybody, I think.

3–20N

Forty-three of the respondents described having had one or more previous falls, before the index incident, although not all of these incidents would have been attended by an ambulance. Some seemed habituated to falling:

Cause I’ve fell loads of times before.

1–11N

I have falls . . . I’m a faller.

3–18F

I can fall over at a drop of a hat, can’t I?

1–5H

However, while some of those who had previously fallen had said that they had managed to do so without damage, and without professional intervention to get back on their feet, for others there was past history of significant damage sustained in falling:

I said, ‘Oh my goodness, not again [wife’s name],’ ‘cause she’d broke the other hip 12 months ago.

1–4Hc

Thirteen of the respondents described having fallen again between the index incident and the interview. For the majority of them, this was a single fall, or a series of minor falls dismissed as ‘little trips’ (3–5H), but three described having had two or even three more major falls for which an ambulance had been called.

Patient experience of the ambulance service

Patient experience: those referred to a falls service

Twelve of the respondents had been identified as having been referred to a falls service by the attending ambulance crew. Respondents described being contacted within 2 weeks of referral at most, some within a week, and receiving a range of interventions, such as 6-week courses in falls avoidance at a day centre, home visits with exercise training, the installation of grab rails in the home, and equipment such as bed levers and perching stools.

Among the respondents who had been referred to the falls service and had received input, satisfaction was generally high:

Oh it’s the best thing that’s ever happened in my life.

1–8F

They’re tops. They really are. Not one of them you could fault. Very, very helpful . . . They’re marvellous people. And so patient with you.

1–12F

I had wonderful care from [local authority]. Absolutely remarkable.

3–4F

Being identified as suitable for a falls service did not necessarily equate to receiving the intervention. One respondent felt that the falls service offered nothing helpful, for another the intervention was stopped for medical reasons and a third was identified as suitable for referral to a falls service, but refused it altogether because:

I thought it was a waste of time really.

3–21N

Three respondents recorded as having been referred could not recall any contact, which may reveal a fault with the recording system or with the referral system.

Although respondents were forthcoming about their views on the falls services themselves, it was hard to disentangle any opinion on the process of being referred to the falls service as part of the SAFER 2 trial intervention. One respondent was unusual in commenting on this, indicating that she felt some confusion about the referral process:

Well, they just said to me when we–, ‘You know we have to refer you now to the falls team’ and so I said, ‘Right’. But I wondered what’s the falls team? What it was all about.

1–12F

More generally, it was not always clear whether the intervention the respondent was talking about was one which was delivered by a falls service associated with the SAFER 2 trial, or was actually a similar service delivered by another team or accessed through a different route. Respondents themselves were in some cases unclear about this:

Well, I feel as though, yes, I mean I got this appointment at the falls, although the doctor, obviously, did it, but I think it was through the SAFER 2 that they did it.

1–7F

Oh, sorry . . . I don’t know who arranged that then.

1–14F

Patient experience: those who attended hospital

Thirty-six of the respondents had been taken to hospital as part of the SAFER 2 trial. The care they received in hospital varied greatly, from being ‘sent home with some paracetamol’ (1–1H), to being admitted for surgery. Others were radiographed, stitched or were put in a cast, and a small number ended up with a lengthy stay in hospital.

The majority of respondents had a positive evaluation of their visit to hospital. Generally, this was expressed in terms of praise for the care provided by individual clinicians:

[The nurse] did fill me with confidence, you know, he was very, very good.

1–10H

The hospital staff, they were all absolutely wonderful, can’t rate them enough.

1–15H

They were very comforting, you know, they gave me confidence in them and they were thoughtful and kind and helpful. I can’t praise them enough, nor the nurse that did the stitching.

2–16H

But, yeah, the doctor was marvellous, the people who done the X-ray, great. No problems.

2–6H

Two respondents stated explicitly that the decision to go in to hospital had been the right one, the wife of 3–6Hc suggesting that it ‘saved your life’.

However, for some the experience of attending hospital proved to be very frustrating. 2–6H was radiographed for suspected broken ribs, although even after the radiograph, ‘they couldn’t say for sure if there was a break or not’. Another respondent was reported by a carer as receiving even less in the way of an intervention:

We went into [ED] in the [hospital], we were there till quarter to three when a doctor eventually came to see her. She had a quick look at her, asked me to get her off the bed and walk her, and because she walked, to take her home . . . not even an X-ray to see whether she had broken her leg.

2–3Hc

Other negative aspects of attending hospital reported by respondents included delays in handover and admission:

We were all lined up, it was like a production line, waiting to be handed over to the doctor.

2–10H

Another respondent mentioned the difficulty of getting only a one-way journey in the ambulance:

But, the only thing I have against all that, they’ll get you to the hospital but they don’t see about you getting home again.

1–1H

Patient health and quality of life since fall

Respondents were asked how their health had been since the fall, leading to extensive discussions not just of health, but also of mobility and quality of life. Some patients made a comparison between their current state and their prefall status, while others compared how they were at the time of the interview with their immediate post-fall status; a few talked through the whole sequence of fall and recovery. A great range of experience was described, across all three outcome groups, and all three study sites. It was not possible to discern a clear pattern of difference between the three outcome groups, except that the group of patients who had been taken to hospital, as might be expected, included some people with more significant injuries.

A minority of respondents had very positive stories of recovery after the fall. These were most striking in the case of people who had sustained fractures or other significant injury in the fall:

Getting better all the time.

2–10H

Mum is amazingly better. She walks around the whole house on her own. When she came back from hospital she was actually moved with a hoist, she had no movement whatsoever and that has changed dramatically.

2–17Hc

Among those who had been referred to a falls service, a minority described very clearly how the intervention they were receiving was supporting an improvement in their well-being:

And now with the exercises . . . they show you to do, which I keep doing. I mean I couldn’t have walked from here to the table at one time, now I can walk to the table . . . Walk round the house with confidence, you know . . . not frightened of falling.

1–12F

Well I’m walking better and I’m feeling better in myself because I’ve gotten over the fall. And I try to do little things. I haven’t quite got enough balance at the moment . . . and that’s what they’re trying to work on.

3–9F

Although there were positive stories of recovery, a larger number of respondents talked about the increased limitations they experienced since the fall. This was a case of adapting what they did, taking more care and keeping clear of risky places such as the garden:

I’ve been very careful . . . No I haven’t fallen since, I haven’t done any gardening though you see and I’ve only done very little housework so I’ve tended not to do anything where I might fall.

1–6N

For some, having to accept these limitations was distressing:

I haven’t been up that road on my own since that happened . . . I used to go out once a week. Yeah, I loved it. One leg wasn’t like 100% but I could do it. I can’t do it now, it breaks my heart . . . I definitely can’t walk like I used to.

2–4H

For many of the respondents, it was a loss of confidence and a fear of falling that prevented them from going out and about:

So I’ll freeze for fear of falling . . . Hoping to tell me legs where I want to go.

2–13H

No but I won’t go out in the street on my own. I can’t, I just can’t, I just freeze. I go to that door, I can’t even go out to the street door . . . Oh yeah. I used to go shopping and everything. I am frightened I am gonna fall.

3–7H

Some expanded on this theme, talking about the fall as something which had pushed them a little further along the inevitable journey into old age:

I have a loss of confidence, I am old but suddenly I feel much older. I feel a great sense of loss of freedom, of a loss of independence . . . Somebody said to me what do I most miss and I said ‘being out in the air’ I do stand at the back door, but I’m frightened of the stairs.

3–9H

I don’t get up around quite so much as I did. And it’s hard ‘cause I now, which I now, which I never felt before, feel old. I am a bit nervous about doing physical things and where I’m going and should I take my walking stick and things like that.

3–15H

I’m walking now and I start crying. Then they say why and I say all this things I don’t feel it before and it’s my age.

3–1N

A large proportion of the respondents described their health, following the fall, as much the same as before. However, many of these then went on to report pain, reduced mobility or lack of confidence since the fall, presumably because they were thinking of these as being distinct from health. For example, 2–15H reported his health as being ‘just the same’ since the fall but went on to describe having to take tablets for pain relief and now finding it hard to get up after sitting on the settee.

The patient in context: other health and social care provision

Nearly half of the respondents described the networks of health and social care provision which supported their lives, much of which had direct relevance to preventing falls or managing their impact.

The most commonly reported form of support was an emergency alarm (also known as a Lifeline or Linkline), mentioned by 12 respondents across all three study sites. These alarms feature a pendant carried around the neck, which the user presses to call for help after a fall. For some respondents, they featured as part of the story of the index incident; for others, they were mentioned more as a source of reassurance.

In all three areas, many respondents reported having previously been provided, through different routes, with the type of support offered by the falls services associated with the SAFER 2 trial intervention. This provision included the fitting of grab rails and other adaptations to the home, equipment such as commodes and walking frames, and physiotherapy. Routes to accessing services varied; for one respondent, the GP functioned as an effective gateway:

Well apparently if you fall more than a certain number of times in a year, it has to be reported which my GP did, so I had a occupational therapist came out to see me and a physiotherapist. And they sorted out all the things, the aids that they thought I would need, namely a trolley and the second hand rail on the stairs, and grip rails everywhere, which I’m thankful for.

1–5H

One respondent described how previous falls had acted as a trigger to push her up the waiting list:

Then when I had that fall, she got a bit worried so it moved up a bit to the next one, I moved up, and that’s how it’s been going on, moving up and moving up to the next one ‘til they found time to come down and assess me.

3–9F

An alternative route to provision was on hospital discharge, which was cited as a way of accessing both practical home care for a limited period, rehabilitation (described by one as ‘walking stick instructions’ 3–13H) and a range of useful equipment such as trolleys which contributes to falls prevention. These routes to access were not always straightforward, with two respondents describing having gone through the assessment process but then being found not eligible for support.

They did send an [occupational therapist] out to me, but she wasn’t very good, I didn’t end up getting anything . . . But my husband has–, well, we’ve been trying to get a rail for the bath for years, each time he’s had an assessment, but we never get it.

1–8F

Other respondents described other forms of health or social care that contributed to their well-being at home, including specialist falls clinic, foot clinic, pain clinic and visits from the district nurse. A small number of respondents described support which they paid for themselves, outside the range of public sector or third-sector provision. In two cases this was people employed to help with cleaning, laundry and gardening; in another three cases it was privately purchased equipment, such as a specialist bed, while two reported paying privately for physiotherapy.

Paramedic and stakeholder expectations of the new intervention (pre trial)

Both paramedics and stakeholders expected significant usage of the new referral pathway. Paramedics thought that the new referral protocol would be utilised widely if it was simple to complete.

I do think it can’t be too difficult to do that referral, that’s going to be the biggy. It’s whether it takes an hour to sort something out. It needs to be a 5/10-minute phone call and ‘yeah, no problem’.

Site 1 paramedic

If you make it a 20-page document people are not going to go with it.

Site 3 paramedic

Paramedics believed that patient care would be improved through the intervention, through avoidance of ED and through referral to a falls prevention service.

Once they can determine why the patient’s having the falls, it could be just poor housing, rugs, things like that, slippers, could be medical, could be down to their medicines what they’re taking, . . . it’s got to build their confidence, it’s got to do.

Site 1 paramedic

Paramedics from all three study sites expected personal gains, including an increase in job satisfaction resulting from offering a more complete care intervention:

You feel like you’re finishing a job. You’re not leaving it half done. You know, it’s ‘Doris has fallen again, we’ll go back to her tomorrow’ [but] ‘Oh Doris is fallen, she’s being referred, someone is looking after her’.

Site 1 paramedic

You’ll feel like you’ve actually achieved something for the patient because you’re not actually just moving them along on a conveyer belt to hospital. You’ll feel that you’re actually benefiting them.

Site 3 paramedic

With the exception of one, paramedics believed that the intervention would increase time on-scene. Paramedics from sites 2 and 3 expressed anxiety on how the control room would respond to an increase in the time on-scene.

If you’re leaving people at home you need to do the checks to make sure they’re safe to stay, that’s the thing isn’t it, you can’t just do a quick blood pressure, a quick pulse . . . then go. You know, in the real world of course it isn’t like that and I think a lot of people up in like control, they think it is like that and it’s not . . .

Site 3 paramedic

The referral bit will take time. I don’t know how control will feel about that.

Site 2 paramedic

Despite paramedics’ worries, ambulance service managers in sites 1 and 3 did not feel that increases in time on-scene would be an issue.

I have no problem with paramedics spending an extra half an hour, three-quarters of an hour on-scene doing a thorough assessment that’s appropriate for the patient. And at the end of the day, the call cycle time will still be quicker than going into ED.

Site 1 stakeholder

I don’t necessarily think that time constraints on-scene would be a problem, I mean most paramedics are quite adamant that time on-scene is my time, I’m not going to get any pressure from, you know, control, if I’m with a patient then that’s the time that I need so I don’t think that that would have any impact on them referring or not referring.

Site 3 stakeholder

Concern was raised by falls team leaders at the number of referrals that would be received the trial, and the capacity of the services to deal with this, although one spoke of plans to bring in locum staff in order to cover increases in demand for their service.

Paramedics’ and other stakeholders’ experiences and views on context and existing practice

The majority of paramedics felt that current practice does not provide appropriate care to elderly patients who fall. Paramedics from all three of the ambulance services stated that current practice involved assessing patients for injuries, and depending on the outcome of the assessment and patient preference, conveying the patient to an ED, leaving the patient at home with no onward referral, or leaving the patient at home and referring to their GP. In two of the three sites (sites 1 and 3), there had been previous attempts to set up alternative pathways for elderly patients who had fallen, referring them to, for example, intermediate care teams, rapid response teams, district nurses and physiotherapy services, although these pathways had fallen out of use.

Overall, paramedics felt that they are able to make decisions regarding leaving patients at home and onward referrals, but they lack the referral pathways to properly undertake this. This was stated explicitly by a paramedic within site 1, with paramedics from sites 2 and 3 stating that they routinely make decisions regarding patient conveyance.

But this has been happening for years hasn’t it. We’ve never had anything put in place. I’d say we‘ve got more access to more places now than we’ve ever had, and it’s still not enough . . . because nine times out of 10 we know exactly what that patient needs, and we just can’t access it.

Site 1 paramedic

The process of deciding whether a patient stays at home or is referred to hospital is not always a straightforward one, and paramedics gave examples from existing practice where patients and paramedics disagreed. These were most striking in cases in which a patient wished to stay at home, but the paramedic felt, based on clinical assessment, that they should go to hospital; in these reported cases the paramedic’s view prevailed, in part because of anxieties about risk:

I had a lady the other day, she clearly needed to go in and she was arguing with us for half an hour. I think we just wore her down in the end, but she had to go.

Site 2 paramedic

You bully them, you do. You bully them to some extent ‘cause you think, I’m not risking it.

Site 2 paramedic

A number of paramedics discussed their anxieties about the risks of leaving a patient at home following a fall, worrying both about the patient, and about any comeback on the paramedic if it were to turn out to be the wrong decision:

I always think back at the end of the shift, oh God, I hope they’re all right.

Site 2 paramedic

At the back of our minds, it’s always gonna be . . . ‘am I gonna be wrong, am I gonna get in trouble’, you know, not that anybody wants to do anything wrong, not that we’d purposely do it. We’re always trying to do the right thing, but if you miss something for whatever reason, and we’re only human, . . . what are the ramifications gonna be?

Site 1 paramedic

Paramedics within sites 1 and 3 discussed the concept of ‘covering their backs’ when attending an elderly patient at home following a fall, in order to mitigate the risk of action against them if something were to go wrong. In some cases, this resulted in the patient being conveyed to hospital even if the paramedic could not see a clear clinical need:

Sometimes we convey the patient when quite clearly they didn’t need to go in. In essence you feel like you need to cover your own backside because nobody else will.

Site 3 paramedic

Referrals to GPs were seen as problematic in all three sites: paramedics were not confident that the GP would respond to a referral (owing to not realising the urgency or simply not receiving the referral) or that they would make an onward referral if appropriate:

They [GPs] just don’t seem to understand they have options as well, that they can refer people on. I just don’t trust the GP to make that choice, to know that there are more choices. It’s alright for us, we know we can refer onto social services . . .

Site 3 paramedic

Paramedics’ professional status was seen as an important factor in the ability of paramedics to consider alternatives to conveying patients to the ED. One paramedic appeared to express frustration that the potential of paramedics to act as the gateway to a range of care services was untapped:

I went to a fall, and there was a district nurse who turned up just after I got there, . . . and the questions she got asked, and the ease of [the district nurse] getting that person respite – I was just gobsmacked at how easy it was. All those questions I could have just easily done . . . I think that paramedics have generally been seen as just take them to hospital, they haven’t been seen as having any clue.

Site 1 paramedic

Pay, and the responsibilities attached to it, were mentioned by paramedics (although not managers) as an issue in sites 1 and 2:

Yeah, people are like, ‘Well I don’t get paid to [undertake additional decision-making to leave patients at home] so why should I do it?’ So then everybody goes to hospital, so then you’ve got an A&E then that is absolutely chocca full of people who could possibly have been treated elsewhere by a different pathway.

Site 2 paramedic

Experience of implementing the Support and Assessment for Fall Emergency Referrals 2 trial intervention

Post trial, paramedics and stakeholders reflected on the process of implementing SAFER 2 trial as part of the research study. Paramedics stated three main reasons for volunteering to take part in the study: persuasion by a colleague; opportunity for personal development and learning new skills; and dissatisfaction with existing practice, in terms of both the lack of referral options available and the frustration of multiple attendances to patients for falls.

And you’re taking somebody in and you know you’re going to be straight back out because, you know, so if there’s something that you can stop from going in so it’s not a . . . waste of resources.

Site 2 paramedic

In sites 2 and 3, parallel developments of other pathways, while the SAFER 2 trial was under development, were felt to have led to confusing messages for paramedics:

It made it quite complicated from a local point of view, so I had 30 paramedics who were trained as intervention paramedics and 60 who weren’t and trying to tell them, ‘No, you can’t choose the [SAFER 2] one yet,’ ‘Oh but we have seen adverts everywhere and we’re getting letters from operations directors saying we must refer patients,’ ‘Well no you can’t.’ So it was quite complicated getting the right message out to the right staff, I think we got there but it was quite challenging to do that.

Site 3 ambulance manager

In all three ambulance services, the majority of managers felt that implementation of the SAFER 2 trial intervention had an impact on their role and workload during the initial phases of the study, when paramedic training was being carried out, but less so in the latter parts of the study:

[My involvement] all seemed to be very front loaded and then I think I lost my way, if I’m perfectly honest with it, during the latter–, I’d still come to the meetings and stuff, but once the training was there and everything was set up, it was almost like, right, sort of sit back and watch what happens.

Site 1 manager

Factors which enabled/encouraged the use of the Support and Assessment for Fall Emergency Referrals 2 trial intervention

According to the paramedics, usage of the new referral pathway was encouraged by its simplicity.

Overall, paramedics were positive regarding the suitability of the training they received. One paramedic compared the benefits of the SAFER 2 trial training to previous training for novel ambulance service interventions, which (as at the other two sites as well) seemed to be limited to e-mail or notice board updates.

Well, there hasn’t been any training apart from this [the SAFER 2 trial training], . . . I mean this has been superb really, you know. You get to sit and relax and get trained on it but the new pathways are reading a piece of paper and acting on it. So, we’re not trained on it [previous pathways].

Site 1 paramedic

Only one stakeholder commented on the value of the SAFER 2 trial training, and in positive terms:

The [paramedics] have now had some training that’s been approved by some specialists so they feel very confident now at making those decisions with the backing of this triage tool.

Site 3 stakeholder

Paramedics in all three trial sites stated that they were aware of the clinical support available to them throughout the trial, although the majority did not actively seek clinical support at any point during the trial. Paramedics in all sites thought newsletters and e-mail updates were important to provide feedback and so maintain the motivation to refer:

[The newsletters were] a kind of confirmation that you are doing the right thing. I can’t remember any specific points but I remember one and I don’t know what it was, I can remember reading it and thinking, oh yeah, OK, I’m doing that so I’m doing the right thing.

Site 3 paramedic

Factors which hindered the use of the Support and Assessment for Fall Emergency Referrals 2 trial intervention

Both stakeholders and paramedics reported aspects of paramedic practice and the organisation of the ambulance service which hindered use of the SAFER 2 trial intervention. Paramedics from all three sites stated that taking a patient to the ED was often the easier option, especially when nearing the end of a shift.

You’ve got a lot of paramedics who, anything for an easy life – ‘I’ll go back take them to [ED] . . . Some of them want to do the best so they get off on time, or don’t have a late finish, or least path of resistance.

Site 1 paramedic

Time of day, coupled with the patient’s social situation, was mentioned by paramedics as affecting their ability to make a referral. Paramedics from all sites stated that they would feel uncomfortable leaving an elderly patient at home late at night without support, even if they were not injured. A small number of paramedics stated that a more immediate intervention than SAFER 2 trial was required by paramedics in order to assure confidence when leaving a patient at home.

I think if we had, as you said earlier, a direct phone number to a person that we knew was going to come out here in 2 or 3 hours, we’d maybe feel a little bit more confident to leave her there.

Site 2 paramedic

Paramedics said they often forgot to refer when on-scene, this being influenced by the time of day and the amount of time they had already been on the road.

At different times of the day . . . you’re more tired, so you do occasionally leave the scene and then think, ah, I haven’t done that [the SAFER 2 trial decision tool].

Site 2 paramedic

Ambulance service managers were aware of this, one suggesting that the use of the SAFER 2 trial declined over time:

I just feel that–, possibly ‘cause it’s over 12 months they did it– and . . . there wasn’t any other trigger mechanism to trigger them to participate actively, participate. And bit by bit, they started just to forget.

Site 1 clinical tutor

In contrast to examples described in Paramedics’ and other stakeholders’ experiences and views on context and existing practice of paramedics persuading patients that they needed to go to hospital, they also described examples of patients, or those who cared for them, putting pressure on paramedics for a hospital transfer, even where paramedics might propose that a patient be left at home with a referral to a falls team. Even though the SAFER 2 trial decision tool had been used, paramedics were frustrated in taking the recommended step to refer to a falls service:

You feel pressure. When you attend elderly fallers, you face pressure from the families, their next door neighbours, . . . generally speaking that patient will take the advice of their next door neighbour and they’ll go to the hospital. So we face a lot of pressures from around us to take them in. The family will say, ‘Oh, I’d prefer her to go in and see a doctor. I’d rather she went in and had an X-ray,’ even if you explain that they don’t X-ray people routinely. So there’s a lot of pressure around taking them in.

Site 2 paramedic

You’re dealing with a frail, elderly group and often with little support in the community . . . added to which I think there is still the ‘you call, we haul’ mentality . . . call for an ambulance, you expect to go to hospital, that’s what we’ll do then.

Site 3 stakeholder

Obtaining consent from patients to refer to the falls team was seen as an issue in all sites:

Patients didn’t want to be referred, we meet a lot of obstinate people who [laughs]–, ‘Get me off the floor, get me bed, I don’t want anything else from you and I’ll be calling you back tomorrow night’.

Site 3 paramedic

Paramedics and managers also reported that it could be difficult to ascertain from patients if they already have involvement from such a service, or from an equivalent service:

If you went to a couple of falls to the same person and in the meanwhile they’d had an intervention of some description from a falls team, they might not immediately recognise that as being the intervention . . . It’s really quite hard, I think.

Site 1 ambulance service manager

And we just feel, are you going on overkill almost, you know, if you start referring to another little team, you know, and they’ve already got this, that and the other coming in? You feel you’re just repeating things.

Site 1 paramedic

Paramedics reported particular issues to do with referrals of patients living in nursing homes, where the referral to the falls service could be seen by nursing home staff as threatening:

If you referred somebody in their own private address, it was welcomed because you explained the package . . . But in the nursing home environment, you felt as though you were actually putting them on the spot . . . they were being scrutinised because they weren’t providing the care or the facilities as what was available.

Site 2 paramedic

Paramedics in sites 1 and 2 saw lack of information regarding the role of the falls teams as an issue, specifically the time frames for initial contact and the interventions provided. Paramedics suggested that increased knowledge of the intervention would have positively influenced the number of referrals.

Evident across all interviews with ambulance service managers were some gaps in communication between them and the study paramedics. Although paramedics spoke of the importance of clinical support in the initial stages of a new pathway, most managers stated that they had not spoken directly to paramedics about the study and had not provided clinical support to study paramedics, with one manager in site 1 describing the difficulty of ‘catching’ a paramedic in order to discuss progress. Paramedics in all sites told a similar story.

[Asked if management were supportive of paramedics being involved in the trial] No one ever said anything.

Site 1 paramedic

I don’t think they [ambulance service managers] knew who was taking part and who wasn’t.

Site 2 paramedic

Impact of the Support and Assessment for Fall Emergency Referrals 2 trial intervention on paramedic practice

The Clinical Decision Flow Chart, used by paramedics on-scene, was felt to be a useful aide-memoire, but its impact on practice was variable. Paramedics from all three sites discussed a process of ‘settling in’ to the new assessment and referral process after their training. Many paramedics stated that it took time to get used to using the Clinical Decision Flow chart, but once they had, it was only used to remind them of specific details on-scene.

The more you used it, the less you needed to refer to it, or only certain points or anything. And if you weren’t sure about something then you’d–, oh, just let me go and check on that. So once you got used to it, as I say, it was very straightforward to use.

Site 2 paramedic

I found it useful as a memory jogger.

Site 3 paramedic

For those based in sites 1 and 3, the flow chart did not contain any new procedures that were not already undertaken on-scene, apart from the referral to the falls team:

I think it [the flow chart] just reinforced what we already know. I guess . . . we’re all used to dealing with falls.

Site 1 paramedic

In site 2, by contrast, the flow chart was felt to bring new assessment techniques into practice, to which paramedics responded positively:

I never did two blood pressures on people, when they were on the floor and when they got up. That didn’t cross my mind. So you know that was definitely a learning curve, so that’s given us extra skills.

Site 2 paramedic

Paramedics from all three sites felt that the SAFER 2 trial did not bring about significant changes in decision-making on-scene; paramedics felt that the decision to convey or leave a patient at home was a decision that they often made with some confidence but the new referral protocol ‘reinforced what they already knew’.

Paramedics in sites 1 and 2 stated that the SAFER 2 trial intervention increased their confidence to leave a patient at home through two mechanisms: the reassurance that the falls team will contact the patient and the confidence in the assessment of a patient who has fallen. Paramedics within site 3 only stated the former. Opinion of ambulance service managers was split between those who thought the confidence had increased, and those that believed that confidence was still an issue.

People are very confident, they are happy making those decisions [not to convey a patient to A&E] I’d say in general.

Site 3 manager

One of the things that we’ve highlighted as a possible reason [for not referring patients] is the lack of confidence, which is meaning that people aren’t leaving patients at home that could be possibly left at home.

Site 1 operational support manager

Paramedics from all three sites thought that the new pathway increased time on-scene, mainly through three mechanisms: increased time undertaking patient assessment, increased time recording patients’ assessment on PRF, and time spent on-scene making the referral to the falls team. Paramedics within site 3 discussed ‘welfare checks’ (a call from the control room to ensure the welfare of the paramedic after a certain period of non-communication), which the paramedics perceived as ‘hurry-up checks’. One paramedic realised a benefit to spending more time on-scene with the patient:

I think for the length of time that we would have been there before SAFER 2–, ‘cause you know, obviously when you actually do the referral it takes up to an hour, so then you get to see the patient sort of in their own environment and possibly doing things that you wouldn’t normally because you’d probably just take them down the hospital.

Site 3 paramedic

Impact of the new pathway on falls services

Falls service representatives from all three sites commented on the appropriateness of the referrals made from paramedics; the majority of patients were thought to have been referred appropriately, although one falls service representative suggested some were not:

. . . there were some who maybe should have gone into hospital and actually not being kept at home.

Site 3 stakeholder

Representatives stated that the referrals they received through the SAFER 2 trial were a mix of patients already known to the falls services, and patients of whom the falls services had not previously been aware. Falls service team leaders from sites 1 and 3 reported that the newly identified patients referred through the SAFER 2 trial were often more frail than those normally receiving treatment from falls teams; a number of patients referred had subsequent A&E attendances and emergency service calls between the referral being made and contact with the falls teams.

The impact of paramedic referrals on the falls service differed between trial sites. In site 2, falls service representatives reported the impact as substantial in terms of paperwork, time and difficulties with onward referral, although the assessment and onward referral process was not thought to be very different from normal practice.

It was a bit of a put off really, and I have to be honest, when the referral come through . . . you were sort of like dreading it a little bit.

Site 2 falls service representative

Representatives from sites 1 and 3, however, felt that the new referral pathway did not impact significantly upon the service, with representatives in site 3 commenting on the positive impact of the new pathway on their falls service.

I think the response time was quite interesting ‘cause ours was a community team, we had a response time and within 1 to 10 days or quicker . . . What SAFER 2 did was . . . it made us firm up a few systems in our single point of access to flag up these fallers, patients, with the SAFER 2 you know, they go on the top, they need to be prioritised so we had to do a bit of work around that which was I think good for the service to kick in a sort of more urgent response, which we hadn’t had before, we weren’t an urgent sort of rapid response team. But after SAFER 2 we had that ability.

Site 3 falls service representative

A number of representatives discussed receiving repeat referrals for the same patient; this was generally felt to be useful, helping to identify gaps in current patient care.

At a personal level, falls service representatives from sites 1 and 3 were positive about their involvement in the study, stating that it was ‘worthwhile’ and ‘enjoyable’. Despite the difficulties, representatives within site 2 were also positive regarding their involvement.

Keeping people well and healthy in their own homes for longer, so I’m glad we did take part. And I think it is something that, you know, the health board will look to develop further in future, but you know, we do need to make sure we get it right, and address some of the issues that you’ve all raised.

Site 2 falls service representative

Suggestions for modifications or developments of the Support and Assessment for Fall Emergency Referrals 2 trial intervention

Participants suggested modifications in four categories: the project set-up, training, patient assessment and referral processes/post referral. Suggested modifications differed by study site, reflecting the variety of issues experienced across the sites.

In sites 1 and 3, ambulance service managers suggested that key ambulance service staff should be involved as soon as possible in the set-up of a study, to make important decisions on behalf of the trust and to have input into the study design and data collection methods.

Falls service representatives from site 1 identified training as an issue, although their equivalents in the other two sites did not, perhaps reflecting the more limited involvement falls service representatives had in training in site 1. A falls service representative from site 1 suggested that training in the SAFER 2 trial should be carried out at regular intervals, in order to include and update more paramedics:

Going back to the training really and maybe just doing some supplementary training and . . . some scenarios with them . . . a quick half hour would have been enough just to say, ‘Thank you for your referrals, thank you for your interest’, . . . they could think of patients that they thought of, ‘oh yeah well I maybe could have referred that lady but she seemed quite fit . . . she seemed quite able but she’d had a couple of falls’.

Site 1 falls service representative

Paramedics in sites 2 and 3 felt that the clinical decision flow chart needed to be simplified, to include less routine information, and more about the causes of falls:

I think on the flow chart if you’re looking at possibly improving it maybe something a bit more . . . shorter version, maybe tick box type of thing.

Site 3 paramedic

On the flow chart, the current one, sometimes there was a bit too much information.

Site 2 paramedic

One site 3 paramedic felt that a smaller A5-sized flow chart would be better as it would to fit into paramedics’ pockets, to allow improved access on-scene. By contrast, other stakeholders across all sites did not comment on the flow chart being an issue.

Paramedics within both sites 1 and 3 felt that the decision to convey a patient would be easier to make if they had access to patients’ records on-scene.

A copy of what their blood pressure is normally like, their normal observations so then . . . we can see it’s . . . abnormal and then it’s going to help our decision-making whether we’re going to leave them or not.

Site 3 paramedic

In site 1, electronic patient report forms (ePRFs) were implemented during the study period, and ambulance service managers suggested that patient assessment could be improved if the ePRF included prompts to remind paramedics to consider a referral to the falls service. An assessment checklist on the ePRF was also considered beneficial by stakeholders.

. . . putting something on to the electronic Toughbook[^® Panasonic, Newark, NJ, USA], create some sort of mandatory check. You know, . . . do not pass go unless you have notified the falls team, etc., whatever it might be, so there is some sort of check there built into the system and–, so it’s programmed and you’ve identified there’s a fall . . . for me, that way there, you will ensure that you’ve actually triggered the response that you want, to trigger it, it’s going to be referred.

Site 1 clinical trainer

Other suggested improvements included increasing the number of people on telephone lines receiving referrals, increasing the capacity to refer (suggested by a site 2 paramedic), and providing an information sheet to referred patients to help them understand ‘what will happen next’ (site 3 paramedic). Improvements to the feedback process were suggested by paramedics from all three study sites, and by stakeholders in site 1. Paramedics from sites 1 and 3 argued for prompt feedback following a falls service referral, while a paramedic from site 2 explained that confirmation that the patient received the intervention would increase their confidence in the referral process:

We weren’t sure what happened after we left. There was a presumption that they were going to be seen. It would have been nice to have some feedback to say that Doris has been seen yesterday and this has been done, so that you know it’s working rather than presuming it’s working.

Site 2 paramedic

Patient experience

We used the qualitative phase of research to assess whether or not patients were satisfied with the new intervention and whether or not, if relevant, the referral pathways by which we had hypothesised that the SAFER 2 trial intervention would make a difference were effective.

Almost all the respondents were very satisfied with the processes of assessment and examination carried out by paramedics and tended to see them as part of a relationship based on empathy and care. The majority of those who were in a position to make a comparison between the SAFER 2 trial encounter with the ambulance service and a previous contact after a fall could not see much difference in the assessment and care delivered. When patients talked about having a say in decision-making, it was generally in terms of whether or not they would travel to hospital, not in terms of whether or not they wished to have contact with the falls service.

Even when patients were referred to the falls service, this did not necessarily lead to them receiving an intervention. Patients referred to falls services may in practice have come by many and various routes. Many of the patients in the SAFER 2 study were already tapping into a complex, and potentially confusing, network of community-based services, delivered by a range of providers (NHS, social services or third sector).

Although some patients reported good recovery after their fall, many had lasting ill-effects, with loss of confidence a major issue. Many worked out their own ways of reducing risk of a future fall, but at some cost to their quality of life.

For the vast majority of patients, the index incident was not the first fall they had experienced, although previous falls may not have resulted in contact with the ambulance service, and many went on to have a subsequent fall. Some seemed quite habituated to falling. Many falls were reported by patients as having an unspecified or unclear cause.

Paramedic and stakeholder experience

Through the qualitative component of the study, we aimed to examine the perspective of paramedics and other clinical stakeholders on the hypothesised mechanisms by which the SAFER 2 trial intervention was meant to work, that is, improving the decision-making of paramedics in terms of safe non-conveyance and referral of older people who have had a fall to appropriate community-based services, and necessary factors to allow this to happen.

The intervention was hypothesised to increase paramedics’ clinical knowledge of how to make appropriate non-conveyance decisions and their knowledge of falls services and pathways for referral. In sites 1 and 2 (but not site 3), paramedics reported that a lack of knowledge around the role of the falls team and the structure of their intervention limited their ability to refer.

It was also hypothesised that it would increase paramedics’ confidence about making a non-conveyance decision and reducing anxiety about risk and would increase the likelihood that paramedics will consider non-conveyance and referral pathways as an option in appropriate cases. There were mixed messages about whether or not confidence had been increased from both paramedics and managers. Paramedics emphasised that the decision-making process remained complex, with patients and carers having a voice which might contradict the paramedic and the protocol.

We also examined the three factors that we felt needed to be in place: effective referral pathways and sufficient capacity in falls teams; effective training and support for paramedics; and motivation on the part of paramedics to use the intervention. Before introduction of the intervention, falls service stakeholders also felt that the new pathway would impact upon their role greatly, although after the intervention stakeholders from two of the three sites reported no significant change to workload. Falls team managers in sites 1 and 2 stated that many referred patients declined the falls service intervention, although this was not mentioned as an issue at site 3. Paramedics in all three sites described reasons for not referring to falls services relating to patients’ social situation and patient autonomy.

After implementation, paramedics reported that support for implementing the new intervention (i.e. paramedic training, flow chart and clinical support) had a positive effect on their ability to refer and make decisions about patient conveyance. The structured training on the SAFER 2 trial was felt to be more helpful than previous, more informal, approaches to introducing innovations, especially in sites 1 and 2. In sites 1 and 2, the SAFER trial processes such as the get up and go test and taking sitting and standing blood pressures were new to paramedics, while paramedics in site 3 already undertook these.

In terms of motivation, before introduction of the intervention, paramedics generally saw it as an effective way to deal with elderly patients who fall, and potentially an improvement on current practice. Current practice was felt to be geared towards conveyance to hospital even when not appropriate for the patient, although paramedics from all sites stated that they routinely make the decision to leave a patient at home following a fall. Paramedics expected to make significant usage of the new referral pathway, resulting in increased time on-scene, but also personal gains in terms of their skills and job satisfaction.

Throughout the transcripts, there was a lack of evidence of paramedics and managers having ‘useful’ contact with each other. Paramedics from all three study sites stated that they had limited contact with their managers during the study period. This is reinforced by the fact that paramedics were ‘worried’ about increased times on-scene, but managers did not see this as an issue.

Suggested modifications differed by study site, reflecting the variety of issues experienced across the sites. Paramedics from all three sites agreed that a ‘simpler’ intervention would have sufficed, with a less detailed flow chart. This was not commented on by ambulance service managers.

Patient experience

The sampling of patients for the interviews was heavily skewed (36 out of 58) towards those who had been transported to hospital. This may have meant that the patients interviewed were more likely, on average, to have had a major fall, and/or were more likely to have wanted to be transported to hospital.

Some patients found it difficult to recall details of the index incident many months after the event. When patients had had a number of falls, it was not always entirely clear which fall they were recalling and describing in response to the interviewer’s questions.

It was hard to disentangle patients’ views on the SAFER 2 trial assessment protocol specifically from their views on the more general process of examination and assessment.

Paramedic and stakeholder experience

It proved difficult to interview an equivalent cross-section of ambulance service managers within each trust, which led to a small variation in the set of questions asked in each site.

Focus groups and interviews were undertaken at different times between May 2011 (pretrial focus groups) and March 2013 (post-trial focus groups). The time of year may have affected the responses provided by both stakeholders and paramedics, as service delivery pressures and priorities might have differed; therefore, it could possibly have affected the issues at the forefront of the interviewees’ minds.

Each site varied in the basic training provided to paramedics, the set-up of the service, service delivery pressures and priorities at the time of the study.