Objective To determine whether multifactorial falls prevention interventions are effective in preventing falls, fall injuries, emergency department (ED) re-presentations and hospital admissions in older adults presenting to the ED with a fall.
Design Systematic review and meta-analyses of randomised controlled trials (RCTs).
Data sources Four health-related electronic databases (Ovid MEDLINE, CINAHL, EMBASE, PEDro and The Cochrane Central Register of Controlled Trials) were searched (inception to June 2018).
Study selection RCTs of multifactorial falls prevention interventions targeting community-dwelling older adults ( ≥ 60 years) presenting to the ED with a fall with quantitative data on at least one review outcome.
Data extraction Two independent reviewers determined inclusion, assessed study quality and undertook data extraction, discrepancies resolved by a third.
Data synthesis 12 studies involving 3986 participants, from six countries, were eligible for inclusion. Studies were of variable methodological quality. Multifactorial interventions were heterogeneous, though the majority included education, referral to healthcare services, home modifications, exercise and medication changes. Meta-analyses demonstrated no reduction in falls (rate ratio = 0.78; 95% CI: 0.58 to 1.05), number of fallers (risk ratio = 1.02; 95% CI: 0.88 to 1.18), rate of fractured neck of femur (risk ratio = 0.82; 95% CI: 0.53 to 1.25), fall-related ED presentations (rate ratio = 0.99; 95% CI: 0.84 to 1.16) or hospitalisations (rate ratio = 1.14; 95% CI: 0.69 to 1.89) with multifactorial falls prevention programmes.
Conclusions There is insufficient evidence to support the use of multifactorial interventions to prevent falls or hospital utilisation in older people presenting to ED following a fall. Further research targeting this population group is required.
- accidental falls
- systematic review
- emergency department
- fall prevention
Statistics from Altmetric.com
Falls are a leading cause of emergency department (ED) presentations in older adults.1 2 In the USA, an older adult is treated in the ED for a fall every 15 s, and an older adult dies following a fall every 29 min.3 It is estimated that £2.3 billion is spent annually on fall-related injuries in people over the age of 65 years in the UK.4 Age-standardised rates of hospitalised fall-related injury cases in older Australians are also steadily increasing.5 Fall-related presentations to the ED are not isolated events; older patients frequently experience subsequent falls (46%–51%);6 re-hospitalisation (49%)7 and substantial functional decline7 within the 12-month period following ED presentation.
Clinical practice guidelines in the USA, UK and Australia recommend the use of multifactorial interventions that involve an assessment of individual risk factors, followed by speciﬁc interventions targeted to those identified risk factors, to prevent falls in older adults living in the community.4 8 9 A number of systematic reviews have also established evidence for the effectiveness of multifactorial interventions in reducing falls in community-dwelling older adults.10–13 However, there is conflicting evidence regarding the effectiveness of these interventions when applied specifically to those presenting to the ED with a fall, based on findings from a systematic review of available evidence until March 2007, conducted by Gates and colleagues.11 Similarly, a more recent systematic review concluded that falls prevention interventions found to be effective in the general older population did not appear to be transferrable to those recently discharged from hospital.13 These conflicting results are likely because of the different care needs of the populations concerned.14 There is a lack of current evidence on the effectiveness of fall prevention interventions for older people presenting to the ED with a fall, who have different care needs to their community-dwelling peers and those who have been recently discharged from hospital.15
An updated review of the effects of multifactorial interventions in people presenting to the ED is warranted, given the addition of new published trials, the increasing number of older people at risk,16 the major physical and psychological consequences associated with falls17 18 and high associated healthcare costs.19 The purpose of this review was to determine the effects of multifactorial falls prevention interventions on falls, fall injuries, fractures, ED presentations and hospitalisations in older adults presenting to the ED with a fall. The results will be of importance to healthcare services and policymakers considering the high cost of the associated injuries and management, and the costs associated with implementing such interventions.20
A systematic review and meta-analysis were performed according to the criteria of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA).21
Search strategy and selection criteria
Five electronic databases (Ovid MEDLINE, CINAHL, EMBASE, PEDro and The Cochrane Central Register of Controlled Trials) were searched, from inception to June 2018. A sensitive search strategy was developed using medical subject heading search terms and keywords (online supplementary appendix 1 outlines full Medline search strategy), and was customised to each database as needed. References of included studies were reviewed for further relevant literature.
Study inclusion criteria: (1) randomised controlled trials (RCT) published in English; (2) included participants aged 60 years or older who presented to an ED after a fall; (3) the intervention included any multifactorial falls prevention intervention delivered to the target population (multifactorial interventions were defined as including two or more subdomains of interventions provided to participants in any possible combination that addresses two or more individual risk factors for falls from assessment findings22 and (4) at least one fall or hospitalisation outcome was reported. Falls prevention interventions were classified according to the Prevention of Falls Network Europe taxonomy.22 Fall outcomes included the number or rate of falls, fallers (number of people who experienced one or more falls in the follow-up period), fall-related injuries or fractures. Hospitalisation outcomes included ED re-presentations or hospital admissions. Studies that included participants who were recruited from an alternative setting (eg, primary care or community setting) were excluded from this review.
Study selection and data extraction
Initially, two reviewers independently screened and excluded studies based on title and abstracts. For articles not excluded, full-text versions were independently assessed by both reviewers to determine whether they met inclusion criteria. In the event of multiple reports from one trial, only the study with the most complete reporting was retained. Disagreements were resolved by a third reviewer if required. Reviewers independently extracted data from included studies. Demographic information (including country of origin, population specifications, sample size), definitions for fall and hospital outcomes and intervention characteristics (such as falls-risk assessments, healthcare professional undertaking the assessment, the intervention strategies used, timing and intensity of interventions offered, and participation) were extracted using a standardised data extraction form. Data available for all relevant outcomes were extracted and tabulated.
Methodological quality assessment
All included studies were assessed for methodological quality by the same reviewers using the PEDro scale.23 This scale rates 11 aspects of methodological quality of RCTs as being either absent or present. A third reviewer was called on if consensus could not be reached.
Data synthesis and meta-analysis
Study characteristics and demographic data, such as sample size, gender and age of participants were reported using mean (SD), median (IQR) or frequency. Meta-analysis was conducted to assess the effect of multifactorial interventions on review outcomes. The rate ratio and 95% CI (if available) was used to describe the treatment effect for falls, fractures, ED presentations and hospitalisations. For the outcome of faller versus non-faller, we used the risk ratio and 95% CI as the treatment effect. If both adjusted and unadjusted effect estimates were reported, we used the unadjusted estimate unless the adjustment was for clustering. Effect estimates were manually calculated when needed. When a study reported multiple effect estimates for an outcome, the follow-up most comparable with other studies, or with the longest follow-up, or adjusted for the largest number of covariates was selected. We conducted two pre-planned subgroup analyses: (1) based on frequency of interventions (interventions with two or more interactions vs those that included less than two) and (2) type of interventions included (interventions including only referral-based interventions vs those that included direct treatment to address risk factors).
As heterogeneity of data was anticipated, due to differences in study populations, follow-up duration and intervention components, a random effects model was used. An inverse variance method was also used to weight each estimate. Between-study variability was assessed using the I2 statistic,24 where I2 values greater than 50% were considered to have a high degree of statistical heterogeneity.25 Where data were unable to be pooled in a meta-analysis due to heterogeneity of outcome measures, descriptive analysis was performed. All analyses were conducted with the use of Review Manager, V.5.2 (RevMan, The Cochrane Collaboration; Oxford, UK). A sensitivity analysis was performed by removing each individual study from our meta-analyses to assess the robustness of our findings.
The electronic search identified 851 potential studies for screening of eligibility, after duplicate studies were removed. Following screening of titles and abstracts and full-text review, 12 studies were retained for inclusion (figure 1), based on our described inclusion and exclusion criteria. There were a total of 3986 participants in the trials included. The median number of participants randomised per trial was 340 (range 10926 to 7126). Mean age was 78 years and ranged across studies from 7327 to 8428 years. Average gender mix was 69% women (ranging from 55%27 to 8029%).
Study and sample characteristics
The 12 included studies were published between 1999 and 2018 and conducted in six countries (table 1). Population age was specified as equal or greater than 65 years in all studies except two,6 27 which included those 60 years and above. Ten studies excluded participants with cognitive impairment, although definitions varied across studies. Five studies utilised the Mini-Mental State Examination (MMSE) (score ranging from 16 to≥25), three the abbreviated mental test (scores ranging from 4 to ≥7) and two did not use a validated tool (patients were excluded based on being described as having dementia in hospital records). Only one study included older people with cognitive impairment (MMSE<24).29
Methodological quality of included studies
Studies were of variable methodological quality (table 1). Methodological strengths included allocation concealment and between-group comparisons for statistical analysis. Common limitations were the lack of blinding and inadequate follow-up of participants. The complete PEDro assessment of studies has been outlined in online supplementary appendix 2 in the supplementary data on the journal website.
Multifactorial falls assessment and interventions
All studies included an assessment of falls risk factors (online supplementary appendix 3 provides a detailed summary). Assessment tools and risk factors assessed varied considerably across studies, as did the type of healthcare professionals undertaking the assessments. The most common falls risk factors assessed were home environment (10 studies), mobility or gait (nine studies), vision (10 studies) and balance (seven studies). The time from index fall until baseline assessment was described in five studies, and ranged from within 2 weeks to 1 month after the fall-related ED presentation. Assessments were also undertaken in a variety of settings and on occasion by more than one health professional, including the participant’s home (10 studies), an outpatient setting for example, day hospital or clinic (four studies) or as an inpatient (two studies).
Table 2 provides the detailed summary of the interventions of included studies. Interventions were led by a variety of healthcare professionals, including occupational therapists, physiotherapists, registered nurses and medical professionals. The specific interventions delivered were highly variable, including education (11 studies), referral to healthcare services (11 studies), home modifications (eight studies), exercise (six studies) and medication change (five studies). Some studies provided only limited treatment options, such as education and referral to healthcare services, whereas others provided many potential intervention strategies. The time until delivery of intervention was reported in only six studies, and ranged from 2 to 8 weeks after completion of baseline assessment. Frequency of the recommended interventions varied from 1 to up to 16 sessions. Only 6 of the 12 included studies reported on patient uptake of referrals and/or fall prevention recommendations, ranging from 7%6 to 100%.28
A number of fall outcomes were captured across the 12 studies (table 1). Studies used varying definitions for these outcomes, summarised in online supplementary appendix 4. All studies utilised falls calendars or diaries to record information on falls. Of these, 11 studies required participants to return the calendar regularly (weekly, monthly), nine contacted the participants monthly to verify/retrieve falls data from the calendar, while one collected data only on follow-up (12 months).30 All studies, except one31 that included ED presentations and/or hospital admissions as an outcome measure, assessed medical records to confirm details. The length of follow-up for studies was between 6 and 12 months.
Effectiveness of falls prevention interventions
Rate of falls and number of fallers
As shown in figure 2, there was no reduction in the rate of falls (rate ratio: 0.78, 95% CI: 0.58 to 1.05) with the use of multifactorial falls prevention programmes. Nor did the programmes significantly reduce the number of fallers (risk ratio: 1.02, 95% CI: 0.88 to 1.18; figure 2). Substantial statistical heterogeneity was noted between individual studies for both the rate of falls (I2=94%) and the number of fallers (I2=75%). Subgroup analyses demonstrated that studies that included two or more interactions in their multifactorial intervention programme had a significant reduction in the rate of falls (rate ratio: 0.62, 95% CI: 0.45 to 0.86) (online supplementary appendix 5). Similarly, studies which included treatment of risk factors, rather than just referral-based interventions, demonstrated a significant reduction in the rate of falls (rate ratio: 0.78, 95% CI: 0.58 to 0.93). No difference was observed for number of fallers in these subgroups.
Fall injuries and injurious falls
Meta-analyses were not performed for fall injuries and injurious falls, as the definitions were too heterogeneous across included studies (online supplementary appendi 4). However, of the eight studies that reported on either injurious falls or fall injuries, no study observed a statistically significant effect.
Number of fractures
Due to the variability in the type of fractures reported, only studies that reported on the outcome fractured neck of femur (NOF) were included in this meta-analysis (three studies). Multifactorial falls prevention programmes did not significantly reduce the number of fractured NOFs (risk ratio: 0.82, 95% CI: 0.53 to 1.25; figure 2). Of the remaining two studies that reported on other fractures, one27 reported a reduction in the rate of fractures in the intervention group (incidence rate ratio: 0.37, 95% CI: 0.15 to 0.91). The other study6 observed no effect on fractures.
Rate of fall-related ED presentations and hospitalisations
Due to the limited number of studies that reported all-cause ED presentations and hospitalisations, only fall-related ED presentation or hospitalisation outcomes are reported. The pooled data demonstrate no significant effect on the rate of fall-related ED presentations (rate ratio: 0.99, 95% CI: 0.84 to 1.16; figure 2), or fall-related hospitalisations (rate ratio: 1.14, 95% CI: 0.69 to 1.89; figure 2). Statistical heterogeneity was noted between individual studies for hospitalisations (I2=58%). Of the three studies that reported all-cause ED presentations and/or hospitalisations, no study observed a statistically significant effect on either outcome. Subgroup analyses did not identify any difference in the results for fractures, ED presentations or hospitalisations.
Sensitivity analysis was performed whereby each individual study was removed from each of the meta-analyses. The direction of association remained consistent when individual studies were removed.
This review extends on previous reviews on the effect of multifactorial falls prevention programmes10 12 focusing on the specific population of older adults who present to ED with a fall. It includes seven additional studies published since the review by Gates and colleagues in 2007.11 We found that multifactorial intervention programmes did not reduce falls in older people who present to the ED with a fall, unless the programme included two or more interactions, or the treatment of risk factors, rather than referral-based interventions alone. No significant reductions were observed in the number of fallers, fractured NOFs, ED presentations or hospitalisations with the delivery of multifactorial falls prevention programmes. We were unable to synthesise data relating to rates of injuries, due to the variability in how this outcome was defined.
There are three key reasons that may have contributed to the limited findings of effect; heterogeneity of multifactorial interventions, poor tailoring of interventions to the ED population and the inconsistent definition of outcomes between studies. The findings of this review are consistent with findings from the recently updated Cochrane systematic review which examined the efficacy of multifactorial interventions in community-dwelling older adults.12 As observed by Hopewell et al (2018), we noted substantial diversity between the multifactorial interventions delivered within studies included in our review. Some studies managed multiple risk factors with multiple intervention strategies from a variety of healthcare professionals over numerous interactions, whereas others concentrated more on education and home modifications in a single interaction. Studies also used various combinations of multifactorial interventions. Although the majority of studies included referral to relevant healthcare services and education, the location and mode of delivery of the interventions were different. Based on results from our subgroup analyses, programmes should at minimum include more than one interaction or actual treatment of one or more risk factor, and not be purely referral based. Meta-analyses of the number of fallers, rate of falls and rate of ED presentations showed significant statistical heterogeneity (I2 58%–94%). While is it expected that multifactorial interventions will differ based on a person’s individual risk factors, the low number of studies and the heterogeneity observed may have diluted any possible effects.
Many of the included interventions were not tailored specifically to the care needs of this population group but were generic interventions that can be implemented to community-dwelling older adults. People that present to the ED as a result of a fall are generally older, frailer, have higher risk of subsequent falls, have more multi-morbidity, complex social issues and more severe injuries when compared with those who do not attend the ED.15 32 The ED is a challenging environment in which to initiate falls prevention interventions as staff, workflows and processes are focused on managing the acute care needs of a patient (eg, injury assessment and management) as opposed to prevention. The studies included in this review may not have sufficiently addressed these differences.
Finally, our outcomes of interest were falls, fall injuries, fractures, ED presentations and hospitalisations. Similar to the review by Hopewell and colleagues (2018), it was apparent in our review that these outcomes were defined inconsistently across the included studies. Definitions varied for falls and fall injuries. Some studies excluded specific types of falls (eg, those due to an acute medical event or in which the person came to rest on furniture or a wall), whereas others included all falls. The rate of peripheral fractures has been recommended as the only robust and feasible measure of an injury.12 33 However, only NOF fractures were able to be pooled in this review, due to varying definitions. As injuries related to falls contribute to the significant burden on the health of patients, and the healthcare system,19 it is important that these more robust outcome measures are consistently defined and reported to allow pooling of results in future reviews.10 33 34 The lack of effect on hospitalisation outcomes may reflect the broader health concerns in this population.
Adherence to fall prevention recommendations and referrals is an important consideration when assessing the efficacy of a falls prevention programme, particularly when numerous interventions are being provided. However, we found adherence to be inconsistently reported. Previous studies have found that participation and engagement in falls prevention programmes in general community-dwelling older adults is likely to be around 50% at 12 months35. A likely contribution to those that present to ED with a fall having poorer participation and engagement in fall prevention programmes may be their complex health needs. Poor adherence could also be explained by the ‘better for others than for me’ phenomenon as described by Haines et al, where individuals at risk of falls can see benefits in falls prevention interventions generally, but do not adhere to these as they do not believe the interventions are relevant to them.36 When developing new interventions, it is important to consider the full range of factors that may influence the effectiveness of an intervention including factors that may limit engagement and participation.35 Future research would benefit from exploring these factors this population group in more detail.
This review was undertaken using the robust methods recommended in the PRISMA statement. Limitations include the consideration of papers only published in English, potentially excluding other high-quality studies, and the analysis of only objective outcomes (eg, falls, fall injuries). Subjective information regarding fear of falling, falls self-efficacy and health-related quality of life were not considered, which may have provided further insight into meaningful change in psychological consequences of falls that contribute to the overall burden. The majority of studies also excluded older adults with significant cognitive impairment, which restricts generalisability of this review’s findings to those with delirium or dementia. All studies only had a maximum follow-up period of 12 months. It is possible that different conclusions may result if a longer follow-up was applied for these less sensitive measures. Despite the limitations, this review reinforces the fact that falls prevention for this high-risk population is complicated and current guidelines regarding falls prevention practices may need to be targeted specifically for this population.
Falls and their sequelae are a significant health burden worldwide. This systematic review and meta-analysis of RCTs found little evidence to support the use of multifactorial falls prevention programmes for older adults that present to ED with a fall. More studies in this complex population are required. Studies with a reproducible type and dose of intervention, and powered to detect effects on fall injuries and fractures, are needed to resolve the uncertainty of effectiveness of intervening in this population group. Research with consistent definitions of fall outcomes, along with adequate reporting of intervention components, intensity and adherence are crucial.
What is already known on the subject
Falls are a leading reason that older adults present to emergency departments (EDs).
There is systematic review evidence for interventions to reduce falls in older people living in the community.
When similar interventions are applied to those presenting to the ED with a fall, there is a lack of effectiveness.
What this study ads
There remains little evidence to support the use of multifactorial falls prevention programmes for older adults that present to ED with a fall.
More studies in this complex population are required.
We acknowledge Eliza Bradley, Research Assistant, for assistance with final stages of manuscript editing.
Funding NHMRC Career Development Fellowship [APP1143538] to JR and National Institute for Health Research Career Development Fellowship [CDF-2015-08-030] to SRN.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.