Falls in the elderly are a rising concern in society, as their incidence is expected to increase with the ageing “baby boom” generation. Moller1 anticipates that the total health costs attributable to injurious falls in elderly people will increase threefold in the next 50 years in Australia alongside significant increases in nursing home places and additional hospital bed days. Recent studies in the USA have shown that falls among older people are the reason for 10% of all visits to emergency departments and 6% of urgent hospitalizations.2 The elderly living in long-term care facilities are at a higher risk of falling than community-dwelling elderly. The incidence of falls in residential care facilities is approximately three times higher than that in the community, equating to about 1.4 falls per person per year.3 Injuries such as hip fractures have an in-hospital mortality of 15% and a 1-year survival of only two-thirds.4

There is evidence to suggest that these costly falls are preventable through the design and implementation of multifaceted falls-intervention programs.5 These programs typically entail staff and resident education, environmental modification, exercise programs, medication reviews, and other intervention strategies. Multifaceted falls-prevention programs have mainly been implemented for community-dwelling older people and with more success than programs using single-strategy interventions.6 There is thus a need to see whether importing these multifaceted programs into the residential care setting will yield similar results. Although single interventions such as exercise and home hazard assessments are known to be efficacious by themselves,6 “falls are often the result of a complex, interdependent constellation of factors, resulting in multiple causes acting together to produce a fall.”7 Therefore, falls-prevention programs that are multifaceted, in that they include more than one implemented intervention strategy to attempt to reduce predisposing and situational risk factors for falling caused by both the surrounding environment and the resident’s behavior, are entirely justified.

This systematic review is unique in several regards. It focuses exclusively on multifaceted falls-intervention programs in residential care facilities, unlike previous reviews.348 This restriction is consistent with the growing trend towards nursing homes in developed countries such as the USA where 40% of people aged 65 years or older are expected to use a residential care facility at some point in their lives.9

Also, unlike previous reviews on the efficacy of multifaceted falls-intervention programs, our paper takes into account the number of recurrent fallers because of their frequency and potential contribution to the problem. It is important to study the number of recurrent fallers because they contribute considerably to the total number of falls recorded. Vassallo and colleagues10 found that, compared with non-fallers, recurrent fallers were more likely to have pre-admission falls (p = 0.004), longer stays in hospital (p<0.001), and more nursing home discharges (p<0.001). Recurrent fallers are therefore a high-risk group and contribute significantly to the overall health burden caused by falls in the elderly.

The objective of this review is to evaluate the effectiveness of multifaceted intervention programs in reducing the number of falls, fallers, recurrent fallers, and injurious falls among elderly people living in residential care facilities.

METHODS

Table 1 presents the definitions of terms used in this review.

Inclusion criteria

1. Randomized controlled trials

2. Concurrent comparison group

3. Participants who were all 60 years of age or older living in a residential care setting including nursing homes

4. Multifaceted programs, which included more than one intervention strategy; common strategies include: staff/resident education on falls prevention, environmental modification, exercise programs, medication review, the supply of (free) hip protectors, and maintenance of mobility-related aids such as wheelchairs

5. At least a 6-month follow-up period for evaluation of residents after the intervention program (eg, post-fall assessments, monitoring of falls and fractures)

6. Studies that measure at least one of the following outcomes: the number of residents sustaining a fall, number of falls, number of injuries (eg, fractures, brain injuries) resulting from falls, and number of recurrent fallers

Exclusion criteria

1. No concurrent comparison group

2. Studies where the primary setting is not residential care such as nursing homes

3. Intervention programs that only focus on one strategy hence not multifaceted

4. Studies that only report proxies such as postural hypotension or the risk of falling

5. Non-peer-reviewed reports

Search strategies

We searched the following electronic databases: Medline (to July 2007); PubMed (to July 2007); EMBASE (to July 2007); CINAHL (to July 2007).

Keywords searched were: “falls,” “accidental falls,” “prevention,” “residential care,” “hip fracture,” “fracture,” “brain injury,” “head injury,” “aged,” “multifaceted,” “trauma,” “randomized controlled trial,” “prophylaxis.” Using a combination of medical subject headings and free text, articles were retrieved for preliminary consideration. The searches were not restricted by language or publication status.

The citation lists for all included studies and the internet search engines Dogpile, Yahoo, and Google Scholar were also searched.

Study identification and selection

The articles were screened using the inclusion and exclusion criteria. Of the initial 21 potential studies, five remained. The most common reasons for exclusion were exclusion criterion 1 (26%), criterion 2 (58%), criterion 3 (11%), and criterion 4 (5%). Disagreements were resolved through consensus.

Data extraction

Data were extracted independently by two authors (JK, KS).

Description of search

The full text of 21 studies was retrieved for preliminary consideration; five met the inclusion criteria.

Quality assessment

Methodological quality was assessed using the Downs and Black checklist.13 The five main areas assessed were: reporting, external validity, bias (internal validity), confounding (internal validity), and power.

RESULTS

Description of studies

Jensen et al5

In this study, the residents (n = 402) were from nine residential care homes in a northern city in Sweden. The participants in both the intervention and control groups completed a baseline assessment. More than half of all invited staff in the intervention nursing homes attended a physician-and-physiotherapist-led educational seminar. Another intervention strategy, environmental modification, entailed reducing the environmental hazards in common areas by rearranging furniture that posed a risk for falling, among other things. The exercise program was delivered in the form of resident-tailored training aimed at improving strength, balance, gait, and safe transfer. Finally, the drug regimen review strategy sought to adjust medication for residents because of the presence of side effects that could potentially enhance the risk of falling. The residents in the control group received the usual care.

Becker et al14

In this study, the sample consisted of 981 residents from six community nursing homes in Germany. The control group did not undergo any specific program activities during the intervention period. The participants in both the intervention and control groups completed a baseline assessment. Staff education in the intervention nursing homes comprised a 60-min course delivered to staff members and the provision of written information on falls. Resident education also involved the provision of written information on falls prevention. For the environmental modification component, an environmental hazard check was completed with particular emphasis on appropriate lighting, chair and bed height, floor surfaces, room clutter, additional grab bars for toilets and bathrooms, and proper use and maintenance of walking aids. The group exercise program relied on a blend of balance exercises and progressive resistance training.

Dyer et al15

For this study, the sample consisted of 196 residents living in residential care homes in west Wiltshire, England. Before randomization, baseline assessments of participating residents and homes were conducted. The control group was visited every 3 weeks by the research assistant so that fall records would be complete. For the intervention group, staff education entailed the dissemination of written information pertaining to the study’s objectives by the care home manager to the staff. Another feature, a 40-min group exercise program (thrice weekly for 12–14 weeks), involved offering physiotherapist-supported exercise sessions, which targeted balance and gait, flexibility, strength, and endurance. Environmental modification largely consisted of written reports, completed by an occupational therapist assistant, on specific risk factors. Medication reviews were advised for residents with suspected medical risk factors.

McMurdo et al16

This study reported on 133 residents in nine local authority residential homes in Dundee, Scotland. The control group received 6 months of reminiscence therapy while the intervention group received 6 months of falls risk factor assessment/modification and a seated balance exercise training program. In the former group, the reminiscence sessions entailed participation in a twice-weekly gathering to promote social interaction through the use of quizzes and music among other items. The risk factor assessment/modification incorporated environmental modification, with a specific focus on appropriate lighting, in addition to medication reviews. Medication that could have contributed to hypotension in the residents was reviewed, and dose reduction, discontinuation, or therapeutic substitution was recommended as necessary (this was not formally done for the control group). The exercise feature of the intervention program was a twice-weekly, 30-min seated group exercise session to music.

Ray et al12

This study reported on 482 residents in seven pairs of Tennessee nursing homes. For the intervention group, an individualized treatment plan was developed after preliminary assessments, and each nursing home chose a falls coordinator, who was responsible for implementing the recommendations stated on the treatment plan. The environmental modification strategy entailed safety recommendations on modifying the lighting, flooring, and call lights, and a suggestion to purchase raised toilet seats which would ensure a safer bathroom setting. Another feature of the multifaceted program was drug regimen reviews in which a team reviewed the suitability of psychotropic drug use among regular users who were ambulatory and at high risk of falling. The educational component offered instruction pertaining to safer transferring techniques to residents and staff. The control group did not undergo any program activities.

Table 2 provides a summary of the following components in each of the studies: methods, participants, eligibility, duration of intervention, topics received by intervention group and the delivery method, control group activities, outcomes measured, main results, cluster, statistical analyses, randomization method, allocation concealment, follow-up period, and any additional notes.

The performance of the included studies in each of these areas is discussed below and summarized in table 3.

Reporting

Reporting was adequate for all the included studies. This included reporting of the study hypothesis, the main outcomes and interventions, estimates of random variability, and main findings. However, none of the included studies attempted to report the adverse effects associated with the undertaken intervention strategies such as cardiac events or falls during the exercise component of the programs. None reported on costs or sustainability of the multifaceted interventions.

Internal validity

As the study population and the study investigators could not be blinded with respect to the intervention, concern for internal validity bias is present. However, the absence of concealment was due to the nature of the intervention itself, and the studies were carried out with otherwise sound methods (appropriate statistical tests, reliable compliance with intervention measures, valid outcome measures).

In all the included studies, apart from one, all patients in the residential care facilities were given the option to participate. McMurdo et al16 did not report on this aspect of consent.

Bias and confounding

All the examined studies stated that the control groups were not exposed to any aspect of the intervention strategies. However, in our opinion, it is not clear from the reported studies as to whether there was the potential for contamination of the control groups. For example, staff who received educational sessions may have worked in both intervention and control nursing homes. Although residents had different doctors, it is possible that doctors in the intervention nursing homes could have communicated with doctors in the control nursing homes about reviewing/revising medication. This may have inadvertently contaminated the control groups.

Controls were recruited from the same populations as the intervention groups and over the same period of time, and the total number of potentially eligible patients was always reported. Losses of patients to follow-up were taken into account in all studies (table 4). However, the internal validity for one of the studies may have been threatened by the inadequate adjustment for confounding in the analyses from which the main findings were drawn.5 This study by Jensen and colleagues proposed conclusions that were based on analyses of treatment rather than intention-to-treat.

Statistical power

Three studies reported having sufficient power to detect a clinically important effect where the probability of a difference being due to chance is less than 5%.51214 Jensen et al5 determined that their sample size (167 residents in the control group and 157 residents in the intervention group) was sufficiently powered to detect a 12% difference in falling between the two groups at a significance level of 0.05.

The study of Becker et al14 was underpowered to detect differences in fractures because of the low incidence of hip fractures in the control group and in the total group. However, there was adequate power to observe a significant difference in the rate of falls and recurrent fallers between the control and intervention groups using 95% CI (table 5). Furthermore, the larger sample sizes used in this study pose fewer problems concerning its statistical power.14

Ray et al12 designed their study so that a total sample size of 500 residents would provide power to detect a 20% reduction in recurrent fallers and a 35% reduction in injurious falls. The study was sufficiently powered to detect a significant reduction in recurrent fallers in intervention facilities compared with control facilities. However, it was inadequately powered to detect a significant reduction in injurious falls.

Dyer et al15 noted that the relatively high intracluster correlation coefficient in their study might be linked to its low statistical power. The intracluster correlation coefficient measures how similarly residents within a cluster are behaving and therefore how much of the behavior is attributed to the cluster, in this case the nursing home, as opposed to the intervention program. Higher coefficients give lower power, with a coefficient of zero signifying that each patient was acting independently in response to the intervention. The high coefficients in this study indicate that the nursing home had a major effect on fall rates, therefore undermining its power to detect any statistically significant reduction in falls in the intervention group.

For the study of McMurdo et al,16 the high drop-out rate may have produced a type II error. In addition, see table 4.

Data analysis

Data were available as measures of comparison between control and intervention groups in the number of residents sustaining a fall, number of falls, number of injuries (eg, fractures) resulting from falls, and number of recurrent fallers. One study chose to express the outcomes (eg, falls and fallers) in terms of incidence density rates, defined as incidence of falls per 1000 resident years.14 In the study of Ray et al,12 differences between paired intervention and control homes were expressed as mean facility proportions of recurrent fallers.

A narrative approach was adopted to describe and synthesize the results because of the heterogeneity of the data of samples and uniqueness of methodological design (eg, follow-up periods) for each included study. The narrative approach was also adopted because no single prevention program used exactly the same combination of intervention strategies.

Effect measures

In the study of Becker et al,14 there was an unexpectedly low incidence of hip fractures in the control group, possibly underpowering the study for detecting significant differences in this particular injury. Another possible explanation for the results in this study was the low adherence to hip protectors. Residents who chose to wear hip protectors apparently observed a lack of staff support with toileting, dressing, and the supply of hip protectors during the initial intervention period.

Jensen et al5 and Ray et al12 both defined recurrent fallers in their studies to be residents who sustained two or more falls during the follow-up period. The other three studies defined recurrent fallers to be residents who sustained three or more falls during the follow-up period.141516

Table 5 summarizes the results of the outcome measures for each of the included study areas.

DISCUSSION

This review found evidence that multifaceted falls-intervention programs can significantly reduce the number of recurrent fallers and incidence of falls among the elderly living in residential care facilities. Of the four studies that measured the number of fallers and number of falls, both Jensen et al5 and Becker et al14 reported a significant reduction in both outcomes with the implementation of a multifaceted program. Only Jensen et al5 reported a significant reduction in the number of injurious falls with the implementation of the intervention program. However, the analyses of this study were not based on intent-to-treat, so the results on the number of injurious falls remain inconclusive.

Other confounding factors may also explain the decrease in the reported falls, injurious falls, fallers, and recurrent fallers besides the intervention program. Dyer et al15 commented that marked differences in the philosophy of residential care homes (to provide care or promote autonomy) may have a significant effect on the degree of enforcement of the multifaceted programs and therefore may have influenced outcomes.

Despite the health burden of recurrent fallers, there is still no agreed definition of what constitutes a “recurrent faller”. Sensitivity analyses using receiver operator characteristic analyses of large databases of fallers might help to define this in a more valid way. All five of the included studies attempted to evaluate the efficacy of multifaceted intervention programs on reducing the number of recurrent fallers in residential care. Three of the studies reported a significant reduction in the number of recurrent fallers of 7–11% with the implementation of a multifaceted falls-intervention program.51214 More high-quality studies that measure this outcome are required.

Although the ageing demographic is a global phenomenon, none of the studies reviewed came from South America, Asia, Australia/Pacific or Africa. No study came from a middle- or low-income country. The data represent the findings of high-income nations in the Northern Hemisphere. If any global impact is to be made on this pressing problem, further research is required from an international community.

One of the strengths of this review is the high standard of methodological quality required for the inclusion of studies. All of the included studies were controlled and had random assignments to control and intervention groups. Four of the five included studies performed analyses appropriate for cluster randomization.5121415 Despite the methodological quality, there is still room for improvement, particularly in the powering of studies for injurious falls in the future.

IMPLICATIONS

Practice

The best available evidence, albeit limited in quantity and quality, suggests that a multifaceted intervention program, comprising resident-specific, group-specific, and general intervention strategies designed for residents living in residential care facilities is likely to be effective in reducing falls. Resident-specific intervention strategies include drug regimen reviews for residents with specific side effects believed to increase the risk of falling. Group-specific strategies commonly take the form of group exercise sessions. General intervention strategies include resident and staff education on falls prevention and environmental modification to the nursing home to reduce safety hazards, which may increase the risk of falling. These are general strategies in the sense that they can be applied to a larger number of residents and staff.

In a setting of limited resources, a reasonable approach for practitioners would be to focus on high-risk groups such as recurrent fallers because they contribute substantially to the overall health burden caused by falls in the elderly.10 Studies included in this review showed that multifaceted programs significantly reduce the number of recurrent fallers.51214 These studies shared common general intervention strategies including environmental modification and education on falls prevention and specific strategies tailored to the needs of individual residents, such as medication reviews and the provision of hip protectors for recurrent fallers. Certain aspects of these multifaceted programs may be easier and less costly than others to implement, but further work is required across different countries to confirm that this is the optimal combination that is consistently effective before a recommendation for widespread implementation of such programs.

Research

Within the setting studied, insufficient evidence exists on adverse outcomes (eg, adverse outcomes attributed to a medication change in a previously stable patient, or an adverse event induced by an exercise program in an otherwise frail elderly person), costs, cost-effectiveness, and sustainability of such programs. No data exist on the value of these programs outside of higher-income countries in North America or Europe. Consideration of adverse outcomes, long-term sustainability, and cost-effectiveness of these programs globally should be given priority in future studies with sufficient power to detect differences in injurious falls.

Other outcomes that could be considered in evaluating the effectiveness of such programs for the elderly in residential care in the future include benefits to patients’ quality of life, social interactions, and reductions of adverse events such as cardiac events and drug interactions that could theoretically arise from interventions intended to reduce falls. An unexplored area for consideration is the effect that such programs can have on care givers and family/friends.

Given the growing demographic of the elderly globally, there is an urgent need for well-designed studies in these areas to inform widespread integration of multifaceted programs into residential settings.