BACKGROUND

Over 2.8 million older adults (aged 65 years and older) in the United States visit the emergency department (ED) annually after a fall,1 defined as the involuntary and abrupt contact with the ground or lower level.2 Falls in older adults are associated with high injury severity, reduced mobility, functional decline, and death3, 4; the rate of deaths from falls among persons aged ≥ 65 years increased 31% from 2007 to 2016.5 Many patients experience recurrent falls after ED discharge: approximately 10% of older adults experience another fall within 20 days,6 and 2.4% experience another fall-related ED visit within 30 days, which sharply increases to 25% at 1 year.7 Furthermore, mortality rates for older adults who visit the ED for a fall increase from 2% at 7 days to 4.4% at 30 days and significantly increase to 15% at 1 year following the initial ED visit.8 Although falls are common and costly,9 and recent evidence suggest that falls may be preventable in the ED setting,4, 7, 10 ED-initiated fall screening and prevention efforts are not widespread,11 perhaps due to the ED environment; limited resources in time and staffing are barriers to implementing impactful change in ED management of falls.12-14

Given the mortality and morbidity associated with falls, there has been interest within emergency medicine to address falls and initiate ED-based fall prevention efforts. The Society for Academic Emergency Medicine (SAEM) Geriatric Emergency Medicine Task Force recognized fall prevention as one of three priority areas over 10 years ago.15 SAEM’s Task Force recommended five falls research priorities: (1) identification of patients at high-risk for falls, (2) ED-delivered fall prevention, (3) examining acute care versus rapid response services for injurious falls reduction, (4) determining feasibility of hospital-at-home models for managing high-risk patients with falls, and (5) electronic systems to facilitate care communication between emergency and primary physicians about high-risk individuals.15 Furthermore, the 2010 guidelines for emergency medicine resident core competencies specifically addressed the need for understanding how to recognize fall risk factors and prevent falls.16 In addition, the American College of Emergency Physicians’ geriatric emergency department (GED) accreditation process cites fall prevention as a potential quality improvement project for accreditation, highlighting falls as a critical topic in the ED care of older adults.17 Finally, the 2013 multiorganizational GED guidelines recommended initiation of multifactorial fall evaluations for ED patients with falls.18 However, despite these recommendations to prioritize efforts targeting falls, there is a lack of consensus about which ED fall assessment, management tools, and/or interventions are feasible and efficacious in the ED.19 Additionally, a recent large multicenter non-ED study, Strategies to Reduce Injuries and Develop Confidence in Elders (STRIDE), failed to reduce fall injuries despite complex individualized interventions and has left clinicians and policymakers skeptical that fall prevention can be adequately achieved.20, 21

In response to such gaps in the geriatric emergency medicine literature, we formed the Geriatric Emergency care Applied Research (GEAR) Network, an interdisciplinary group of clinicians, researchers, and patient advocates, to identify and examine clinical questions relevant to five geriatric emergency medicine domains: falls, cognitive impairment, medication safety, elder abuse, and care transitions. These domains were identified based on the recommendations of SAEM and the GED guidelines as well as priorities established by GEAR stakeholders. The GEAR-Cognitive Impairment22 and GEAR-Elder Abuse23 recommendations have already been published. The Geriatric Emergency Care Applied Research-Falls (GEAR-Falls) subgroup’s primary objective was to systematically identify, synthesize, and prioritize high-yield research priorities for older adult falls screening and prevention in the ED and prehospital setting. A secondary objective was to report ED best practices for fall risk recognition and prevention based on a scoping review of pertinent emergency medicine research.

METHODS Study design

Experts from geriatric interest groups, such as the SAEM, the American Geriatrics Society (AGS), and the Gerontological Society of America, with an interest in emergency and aging research, were recruited to serve on the GEAR Task Force. Additional researchers were selected based on their publication history in these domains. GEAR-Falls consisted of 12 members: a geriatrician (WH), six emergency physician scientists (UH, CC, SWL, SF, CG, EG), an epidemiologist (EC), a health services researcher (AL), nursing scientists (KS, AM), and a graduate student (NH).

The GEAR-Falls subgroup performed a scoping review adhering to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Extension for Scoping Reviews reporting guidelines.24 GEAR-Falls held monthly meetings to discuss research priorities and develop clinical questions relevant to falls based on clinical expertise and knowledge of preexisting research.

Patient–intervention–control–outcome (PICO) questions GEAR-Falls derived and refined the following PICO questions: PICO-1 Population: ≥60-year-old individuals presenting with a fall for ED care. Intervention: Fall prevention interventions (may include multifactorial risk reduction, medication review exercise training, models of care (hospital-at-home and others) in ED or peri-ED period). We defined peri-ED as care provided during the ED visit or by emergency medical services (EMS) before or after the ED visit. Comparison: Standard of care. Outcome: Quality of ED care metrics, ED operations outcomes (e.g., ED length of stay); patient-centered outcomes (e.g., ED returns for falls, future falls, fear of falling, functional decline, long-term supports, or institutionalization). PICO-2 Population: ≥60-year-old individuals from any pre-ED and ED setting (e.g., EMS). Intervention: Risk stratification and falls care plan. Comparison: No risk stratification and no falls care plan. Outcome: ED referral (if from pre-ED setting), quality of ED care metrics (e.g., preventable ED recidivism, 30-day hospitalization rates), ED operations outcomes (e.g., ED length of stay, cost), patient-centered outcomes (e.g., ED returns for falls, future falls, fear of falling, functional decline, long-term supports, or institutionalization). PICO-3 Population: ≥60-year-old individuals presenting with a fall for ED care. Intervention: Specific risk factors (such as polypharmacy, frailty etc.). Comparison: None. Outcome: Recurrent falls, fall injury, functional decline, acute care utilization (ED, hospital visit), long-term institutionalization, mortality (prognosis question).

GEAR-Falls identified exemplar articles for each PICO question and surveyed 33 investigators from all five GEAR cores to prioritize the PICO questions via teleconference discussions and online voting. PICO questions 1 and 2 were selected as highest priority.

Search strategy

PICO-1 and PICO-2 and the exemplar articles for both formed the basis of an electronic search strategy devised by a medical librarian (MD) for Ovid Medline 1946–, Embase 1947–, CINAHL 1937–, and Cochrane Central Register of Controlled Trials. All search strategies are listed in Data Supplement S1, Appendix S1 (available as supporting information in the online version of this paper, which is available at http://onlinelibrary.wiley.com/doi/10.1111/acem.14279/full).

Study selection and data abstraction

For each PICO question, two authors independently reviewed the titles and abstracts for the search results of PICO-1 and PICO-2 to identify studies that met the inclusion criteria. Inclusion criteria for PICO-1 were fall interventions for adults over age 59 intended to improve quality of care, ED operations, or patient-centered outcomes. Inclusion criteria for PICO-2 were adults over age 59 with fall risk stratification and a falls care plan.

Because PICO-1 had a large number of studies to filter through and adjudicate, PICO-1 studies were divided into three segments using the alphabetical order of the manuscript titles; each segment was reviewed and adjudicated by two independent reviewers (PICO-1 = Segment A—SF, KS; Segment B—CG, AM; Segment C—EC, AL; PICO-2 = WH, EG). Although the authors of this manuscript also authored literature that is included in this review, including AL and CC, neither was assigned to review abstracts they had written. After filtering based on study title and abstract was completed, and before starting adjudication, we calculated the unweighted Cohen's kappa to quantify inter-rater agreement between all reviewers of PICO-1 and the two reviewers of PICO-2. During adjudication, we resolved disagreements on which studies to include or exclude via email or telephone discussions between reviewers. A preplanned third adjudicator (UH) resolved remaining discrepancies. Final articles were abstracted independently by six investigators (PICO-1 = SF, KS, CG, AM; PICO-2 = WH, EG) and key elements were collected using a predesigned template, which included the study setting, inclusion/exclusion criteria, study design, comparator reference standard or control group, and primary/secondary outcomes.

Due to the dearth of literature for both PICO questions, we included conference abstracts, foreign papers with English translation, and nonrandomized studies. We preplanned a separate synthesis for abstract results because the data they contain are limited. We excluded studies if they were abstracts of studies that were already included, studies that reported outcomes that did not include our predefined PICO outcomes, or studies that were not original research. We included systematic reviews if they pertained to our study objectives and included individual studies we were not already analyzing. Our rationale for inclusion of these studies was that they would enhance our scoping review by adding relevant studies and meta-analyses–derived statistical heterogeneity and study quality assessments that we did not conduct.

Synthesis of best practices and building consensus

Summary tables were presented to all GEAR members at the GEAR consensus conference which occurred in October 2019. GEAR-Falls synthesized current-state knowledge and research gaps and then provided preliminary falls research prioritization rationale. GEAR attendees were then divided into smaller discussion groups to consider the scoping review findings, the preliminary research prioritization questions and debate and further consider research gaps. Discussion groups reconvened, shared their perspectives on research priorities, and streamlined them into common themes. Finally, all GEAR members voted and ranked these research priorities as recommendations for future research. Two non–emergency medicine fall prevention experts subsequently created a statement regarding the impact of the scoping review by GEAR-Falls (see Appendix S4). Two patient advocates (LH, LN) reviewed this scoping review and the consensus conference PowerPoint presentations before providing their perspectives via email and teleconference calls (see Appendix S5).

RESULTS Evidence synthesis PICO-1 Fall prevention interventions

We completed the search in February 2019. For PICO-1, after duplicates were removed, there were a total of 3,181 unique citations with a focus on interventions for geriatric patients presenting to the ED for a fall. Of the 3,181 studies focusing on interventions, 3,049 were excluded, because they were not ED-based, geriatric, or interventional studies, leaving 32 studies for inclusion in this review (Figure 1). Cohen's unweighted kappa coefficient for inter-rater reliability for selecting articles demonstrated fair agreement between reviewers (κ = 0.24, 95% confidence interval [CI] = 0.10 to 0.38).25 Discrepancies for initial disagreement included decisions regarding the inclusion of different manuscripts using the same study participants or secondary literature, as well as differing definitions of interventions.

Summary of scoping review quantitative synthesis for PICO-1. PICO, patient–intervention–control–outcome

Participant characteristics

The 32 included studies contained over 500,000 participants from 11 countries worldwide. Studies were published from 1999 to 2019. Mean age of participants ranged from 74.2 to 84.6 years old. All but two studies26, 27 included patients presenting to the ED for a fall or fall-related injury. Of the remaining two studies, Harper et al.26 included patients presenting to the ED with any diagnosis, including falls, provided that they were scheduled for discharge, and a systematic review by Carpenter27 encompassed community-dwelling adults with falls, including those who presented to the ED for any reason and had a subsequent fall within 6 months. Five studies included ED patients with falls if they sustained at least one fall in the year prior to the ED visit.28-32

Study characteristics

Of the 32 studies, three were conference abstracts33-35 and three were commentaries on previously published studies.27, 30, 36 Studies were randomized controlled trials (23),2, 26, 28, 30-33, 35-48 prospective observational (two),33, 49 retrospective observational (four),7, 50-52 and systematic reviews with or without meta-analyses (three).27, 34, 53

Interventions

Twenty-one of the 32 (66%) interventions consisted of a falls risk assessment—screening for future fall risk2, 26, 28-30, 33, 35-38, 40-48, 54, 55—and 15 of 21 (71%) were combined with a personalized exercise or physical rehabilitation session led by physical or occupational therapists (home-based or outpatient geriatric settings).2, 26, 28, 30-33, 36-38, 40, 42, 43, 45, 46 Fifteen studies were multifactorial, targeting the assessment and amelioration of individual risk factors for the fall. They included a falls risk assessment, educational guidelines, and referral to a nurse practitioner focused on falls or physical therapists.2, 26, 27, 30-32, 36, 38, 40, 42, 44, 45, 47, 50, 54 Seven interventions were education focused.6, 26, 32, 44, 49-51 Two studies included medical alert devices.50, 54 Table 1 summarizes the common study, patient, intervention, and outcome characteristics we identified for PICO-1.

TABLE 1. Summary of study, patient, intervention, and outcome characteristics for PICO-1 and PICO-2 PICO-1 PICO-2 Summary

Population: ≥60-year-old individuals presenting with a fall for ED care.

Intervention: Fall prevention interventions (may include multifactorial risk reduction, medication review, exercise training, models of care (hospital-at-home and others) in ED or peri-ED period).

Comparison: Standard of care.

Outcome: Quality of ED care metrics, ED operations outcomes (e.g., ED length of stay); patient-centered outcomes (e.g., ED returns for falls, future falls, fear of falling, functional decline, long-term support, or institutionalization).

Population: ≥60-year-old individuals from any pre-ED and ED setting (prehospital, paramedic, EMS, ED).

Intervention: Risk stratification and falls care plan.

Comparison: No risk stratification and no falls care plan.

Outcome: ED referral (if from pre-ED settings), quality of ED care metrics, ED operations outcomes (e.g., ED length of stay, cost), patient-centered outcomes (e.g., ED returns for falls, future falls, fear of falling, functional decline, long-term support, or institutionalization).

Number of included studies 32 17 Meta-analysis or systemic reviews, n (%) 3 (13) 4 (24) Randomized controlled trials, n (%) 23 (72) 8 (47) Interventions/instruments validated in the ED (n) 9 multifactorial interventions, ranging from: falls risk assessment and physical rehabilitation sessions, preventive education (7); falls risk assessment, educational guidelines, and follow-up with nurse practitioner or physical therapist (15); medical alert devices (2) 12 screening instruments (11 identified) CDC STEADI Paramedic baseline assessment FROP Com FRIDs using clinical pathways Two-item screening tool EGS FIM/FAM Bedside functional tests Timed Up & Go Clinical pathways only SPPB Unidentified (2 studies) Combined with interventions ranging from Education Discharge planning ED-based physical therapy Preventive education Follow-up calls At-home visits Total number of patients recruited 571,071 At least 17,232 (Samuel 2017,58 Beales 201687 did not report sample sizes) Recruitment period 1999–2019 2011–2018 Geographical locations 11 countries 9 countries Mean age range (years) 74.2–84.6 74.3–83.9 Common inclusion criteria (n) Age 65 and older Age 65 and older Presenting to the ED for a fall or fall-related injury Admitted to the ED with a falls-related injury At least one fall in the year prior to the ED visit (5) History of falls within the past year (1) Patient eligible for ED discharge Patient-centered primary outcomes (n) Proportion of participants experiencing recurrent falls (29) Incidence of recurrent falls (8) Frequency of falls per subject (11) Patient participation in fall prevention plans (1) Time to first fall post–ED discharge Proportion of positive screens (1) Change in anxiety from fear of falling Reduction in falls health care costs (1) QALYs Completion of screening (1) Functional ability pre–post fall intervention Clinical relevance (1) Life-threatening complications (1) ED-centered primary outcomes Improvement in documenting a fall-related history or examination Feasibility of screening 9-month ICER Secondary outcomes Physical disability Death Follow-up period post–ED discharge 1–18 months 6–12 months Reference standard (n) Usual care Usual care FROP com with FIM/FAM (1) Criterion standard available None None Risk of bias in meta-analysis/systematic review (n) High (2) Abbreviations: CDC STEADI, Centers for Disease Control and Prevention's Stop Elderly Accidents, Deaths, and Injuries toolkit; EGS, Emergency Geriatric Screening; FIM/FAM, Modified Functional Independence Measure/Functional Assessment Measure; FRIDs, Fall Risk Increasing Medications; FROP Com, Falls Risk for Older People in the Community; ICER, incremental cost-effectiveness ratio; PICO, patient–intervention–control–outcome; QALY, quality-adjusted life-years; SPPB, Short Physical Performance Battery; TUG, Timed Up & Go.

Interventions varied and were often bundled, not only with fall risk assessments but also with other interventions. For example, in one study patient/clinician education was combined with referral to physical therapy.26 This combination of strategies made it difficult to analyze the efficacy of individual intervention components. For example, the PROFET trial showed a statistically significant reduction in subsequent falls for a bundled intervention of medical assessment, occupational therapy assessment, and referral services, but the efficacy of individual components is unknown.2

Study outcomes

Twenty-one of the 32 studies reported their primary outcome as the proportion of participants experiencing recurrent falls.2, 6, 7, 26-30, 32-38, 40-42, 45, 53, 55 The duration of follow-up varied greatly, ranging from 1 to 18 months post–ED discharge.7, 29, 50 Two studies did not report a specific timeline.26, 27 In addition to number of people with falls per group, 11 studies also reported the frequency of falls per subject as their primary outcome.26-30, 33, 37, 38, 40, 41, 53 Two studies reported on process measures including improvement in documenting a fall-related history or examination51 or 9-month incremental cost-effectiveness ratio (ICER).43 Patient-centered outcomes included reduced fear of falling,54 quality-adjusted life-years (QALYs),43 and functional ability.46 Secondary outcomes varied greatly with physical disability being the most common,6, 26-29, 31, 34, 36-38, 40, 41 followed by death.26, 28, 30, 39, 45

Eighteen studies focused solely or included exercise in their intervention bundles.2, 6, 7, 26-34, 36, 38, 40, 42, 45, 56 In seven studies, physical therapy was associated with reduced risk of falls (7/18 studies, 39%).2, 7, 29,