NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Collaborating Centre for Nursing and Supportive Care (UK). Clinical Practice Guideline for the Assessment and Prevention of Falls in Older People. London: Royal College of Nursing (UK); 2004 Nov. (NICE Clinical Guidelines, No. 21.)

  • This publication is provided for historical reference only and the information may be out of date.

This publication is provided for historical reference only and the information may be out of date.

Cover of Clinical Practice Guideline for the Assessment and Prevention of Falls in Older People

Clinical Practice Guideline for the Assessment and Prevention of Falls in Older People.

Show details

6Guideline recommendations with supporting evidence reviews

Below are the recommendations agreed by the GDG, with associated evidence statements, evidence summaries and, where relevant, GDG commentaries on the consideration and interpretation of the evidence.

1.1. Case/risk identification

(please see Sections 5.2 and Sections 5.3 for evidence review methods)

1.1.1. Recommendation

Older people in contact with health care professionals should be asked routinely whether they have fallen in the past year and asked about the frequency, context and characteristics of the fall/s. [C]

Level of evidenceEvidence statement
Level IIILevel III
A previous fall is the most frequently reported risk factor in prospective cohort studies, suggesting that an older person with a history of falling would be at high risk of a subsequent fall.

Evidence summary

Falls history

Falls history is a frequently reported significant risk factor and predictor of potential further falls. Ten studies reported falls history as statistical significant, among community-dwelling older people (Northridge 1996; Covinsky 2001; Tromp 2001; Friedman 2002; Stenbacka 2002; Wood 2002), and among residents of extended care facilities (Thapa 1996; Cavanillas 2000; Kallin 2002). For older people in community-dwelling settings, the range of summary statistics (OR/RR) reported was: 1.5-4.0. Three studies were of high quality; three of medium quality and one was low quality, with a reported OR of 4.0.

Studies conducted in extended care settings reported significant results, of which one high quality study reported a incident density ratio of 2.23 (1.4-4.37). Two other studies, of low quality, reported an odds ratio range of 1.9-4.65. Seven studies reported falls history as significant in bivariate analysis but not in multivariate. Heterogeneity between these studies hinders interpretation of the clinical relevance of this finding.

GDG commentary

There is good evidence from cohort studies that an older person who has had a previous fall would be at risk of a subsequent fall. The group was keen to recommend that an older person be asked about their falls history based on this evidence. The purpose of obtaining this history would be to establish where possible, the frequency of falling; context and circumstances of the fall; and severity or injuries sustained from the fall. There was debate within the group of the best approach to identifying older people at risk, based on their previous falls history. Some were in favour of an annual review based on screening. Others considered that a case finding approach was more appropriate, asking an older person if they had fallen in the last year when seen by a health care professional. The group was in support of this being done yearly but did not want to reflect this in the recommendation.

1.1.2. Recommendation

Older people reporting a fall or considered at risk of falling should be observed for balance and gait deficits and considered for their ability to benefit from interventions to improve strength and balance. (Tests of balance and gait commonly used in the UK are detailed in Section 5.) [C]

Level of evidenceEvidence statements
Level III
  • Mobility impairment, gait disorders and balance deficits have frequently been reported as significant risk factors in prospective cohort studies.
  • Many tests for the assessment of balance and gait are available to support clinical skill and the choice of such a tool should be determined at a local level.
Level I
  • Intervention trials focusing on gait and balance have shown a reduction in falls.

Evidence summary

Mobility impairment, gait disorders and balance deficits have frequently been reported as significant predictors of future falling in prospective cohort studies (Bueno-Cavanillas 2000; Cesari 2002; Covinsky 2001; Northridge 1996; O'Loughlin 1993; Stalenhoef 2002).

Tests are available for the assessment of an older person's balance and gait that can inform clinical judgement. A detailed list of such tests is provided in Appendix E, Evidence table 3. These range from simple, pragmatic tests that require no special equipment, to those that require a trained health care professional with skill to administer.

GDG commentary

The group felt that assessment of older people who have fallen at least once should include observation for balance and gait deficits. This could be done on first contact by an appropriately trained health care professional in any setting. Clinical judgement should support the use of any test referred to in the clinical evidence and many other tests, developed by different disciplines, are likely to be available in trusts. However, a simple observation of a patient's ability to stand, turn and sit is considered adequate as a first level assessment.

Older people with observed gait or balance problems should be referred for targeted interventions. Identifying those most likely to benefit should also be considered.

The group was unable to recommend specific tests for use in practice, as there was a lack of robust validation studies. A profile of tools and tests identified in the assessment review is provided in Appendix E, Evidence table 3. The choice of tests should be determined at local level.

1.2. Multifactorial falls risk assessment

(please see Section 5.2 and Section 5.3 for evidence review methods)

1.2.1. Recommendation

Older people who present for medical attention because of a fall, or report recurrent falls in the past year, or demonstrate abnormalities of gait and/or balance should be offered a multifactorial falls risk assessment. This assessment should be performed by a health care professional with appropriate skills and experience, normally in the setting of a specialist falls service. This assessment should be part of an individualised, multifactorial intervention. [C]

1.2.2. Recommendation

Multifactorial assessment may include the following: [C]

  • identification of falls history
  • assessment of gait, balance and mobility, and muscle weakness
  • assessment of osteoporosis risk
  • assessment of the older person's perceived functional ability and fear relating to falling
  • assessment of visual impairment
  • assessment of cognitive impairment and neurological examination
  • assessment of urinary incontinence
  • assessment of home hazards
  • cardiovascular examination and medication review.
Level of evidenceEvidence statements
Level IIIMany individual risk factors have been proven to be predictive of a subsequent fall; therefore presence of more than one of the factors listed below increases the risk of falling:
  • falls history
  • gait deficit
  • balance deficit
  • mobility impairment
  • fear of falling
  • visual impairment
  • cognitive impairment
  • urinary incontinence
  • home hazards
  • number of medications
  • psychotropic and cardiovascular medications
  • muscle weakness.

Evidence summary

Gait deficit

Three community-dwelling studies reported this risk factor as statistically significant with a range of OR: 1.96-2.2 (Koski 1998; Cesari 2002; Northridge 1996). Four studies in community-dwelling settings reported non-significance in multivariate analysis (Northridge 1996; Stalenhoef 2002; Wood 2002; Tinetti 1995).

No studies in extended care settings reported gait deficit as significant (Cavanillas 2000; Kallin 2002) although one study carried out detailed gait analysis and found ‘sitting down incorrectly’ as significant in multivariate analysis significant (Cavanillas 2000).

In all of the above studies, the method of measuring gait and aspects of gait analysis differed between studies.

Balance deficit

Three studies conducted among community-dwelling participants, reported balance as statistically significant with a range of summary statistics of 1.83-3.9 (O'Loughlin 1993; Stalenhoef 2002; Covinsky 2001). However, each study measured different aspects of balance including dizziness, unbalanced and postural sway.

Eight studies did not find aspects of balance significant in multivariate analysis, two of which were conducted in extended care settings (Bueno-Cavanillas 2000; O'Loughlin 1993; Tinetti 1995; Northridge 1996; Koski 1998; Wood 2001; Stalenhoef 2002; Kallin 2002). Again, different aspects of balance were analysed.

Mobility impairment

Two community-dwelling studies reported statistical significance: In study one: trouble walking 400m: IRR=1.6(1.2-2.4); trouble bending down: IRR=1.4(1.0-2.0) (O'Loughlin 1993). Study two conducted statistical modelling adjusting for different variables and reported the range for both multivariate models: OR=2.64-3.06 for mobility impairment (Covinsky 2001).

Four studies reported non-significance but as discussed earlier, different methods and aspects of mobility were measured (Bueno-Cavanillas 2000; Kallin 2002; Cesari 2002; Stalenhoef 2002).

Fear of falling

Three community-dwelling studies reported statistical significance of this factor, with a range of summary statistics 1.5 – 3.2, although different methods of measuring fear were used (Arfken 1994; Cumming 2000; Friedman 2002). This included use of the falls efficacy scale (FES) to explore different cut-off values for determining risk and verbal rating scales to identify the degree of fear present. One study measured fear at baseline and reported non-significance in the results (Tromp 2001).

Friedman et al. (2002) carried out a prospective cohort study to examine the temporal relationship between falls and the fear of falling with n=2212 community-dwelling participants aged between 65 and 84 years. Fear was measured at baseline and at one-year follow-up with a simple yes/no answers to whether they were worried or afraid of falling, with a further question relating to their activity limitation when afraid of falling. This study was of high quality with a large sample and detailed baseline measurement. Logistical regression with adjustment for other variables in the model was performed on the data and results as follows. Fear of falling at baseline was significantly predictive of falling at follow-up with OR=1.78 (1.41-2.24), as well as fear at baseline predictive of fear at follow-up OR=5.40 (4.23-6.91). In addition to this, a fall at baseline was predictive of fear at follow-up 1.58 (1.24-2.01). Shared predictors of both falls and fear at follow-up include female gender and history of stroke.

Cumming et al. (2000) carried out a prospective study to assess the impact of fear of falling with n=418 community-dwelling aged 65 and over. This study was of medium quality with a smaller sample size than others. The FES was administered at baseline with a total score of 100 indicating high fall related self-efficacy and 0 low fall related self-efficacy. Cut-off points were tested for predictive ability of falling in the analysis. Adjusted hazard ratio for all study participants with a FES score of <75 = 2.09 (1.31-3.33).

Tromp et al. (2001) conducted a prospective study to examine all predictors for falls with n=1285 community-dwelling participants aged 65 years and more. This was a high quality study, with detailed baseline measurement and fall events measured with falls calendars. Fear was determined using a modified FES where answers were rated on a scale 0 (no confidence) to 3 (completely confident). Odds ratio for 1 fall and recurrent falls were significant in bivariate analysis but non-significant in logistic regression analysis.

Arfken et al. (1994) recruited patients from a prospective cohort study in which the purpose was to develop a screening tool for predicting falls in older people. The sample was 890 community-dwelling participants stratified in age groups ranging from 66 to 81+years. Baseline data were collected as part of the parent study and falls surveillance was conducted with participants reporting falls to a hotline plus monthly postcards reporting the incidence of falls. At one-year follow-up, the participants received a structured in-home assessment including demographics, health status, activity level, satisfaction with life, depressed mood and a brief physical assessment. Fear was determined with a three point verbal rating scale and dichotomised to summarise outcome as odds ratios: A= moderately fearful and not fearful, B= very fearful. Logistic regression models adjusted for gender and age. Results indicated that those who were moderately or not fearful predicted falling at least once: A=1.52 (1.06-2.17) and very fearful participants: (B= 2.49 (1.48-4.20). Those experiencing frequent falls were more likely to be very fearful of falling: B=3.12 (1.61-6.06) than those moderately or not fearful A=1.71 (1.01-2.89).

Visual impairment

Two community-dwelling studies found that older people with a visual impairment were significantly at risk of falling, OR range=1.18-2.3 (Northridge 1996; Koski 1998). One extended care study of low quality reported OR =5.85 (Kallin 2002).

Eight studies reported non-significance in multivariate analysis, two of which were extended care setting studies (Tinetti 1995; Northridge 1996; Tromp 1998; Cesari 2002; Stalenhoef 2002; Wood 2002; Thapa 1996; Bueno-Cavanillas 2000).

Different aspects of vision were measured in these studies and included: visual impairment, visual acuity, depth perception and others.

Furthermore, there are a number of prospective cohort studies which we have been alerted to by stakeholders that demonstrate that visual impairment is an independent risk factor for falls and hip fractures (Felson et al. 1989; Cummings et al. 2003; Ivers et al. 2000 and 2004).

Cognitive impairment

Two studies in community-dwelling settings reported that older people with cognitive impairment were significantly at risk of falling OR=2.2-2.4 (Tinetti 1995; Van Schoor 2002). One low quality study in an extended care setting reported OR 6.2 (1.7-23.3) (Bueno-Cavanillas 2000).

However, nine studies did not find older people with cognitive impairment significantly at risk of falling in both settings (Tinetti 1995; Northridge 1996; Tromp 1998; Cesari 2002; Stalenhoef 2002; Van Schoor 2002; Thapa 1996; Kallin 2002; Wood 2002).

Urinary incontinence, including stress and urge incontinence

Two studies reported that older people suffering from urinary incontinence were at risk of falling with OR range=1.26-1.8 (Tromp 1998, (2001; Brown 2000). Additional studies that support incontinence as a significant risk factor include Luukinen 1996 and Tinetti 1995.

Five studies did not find incontinence a significant predictor of falling (Tinetti 1995; Koski 1998; Brown 2000; Cesari 2002; Thapa 1996).

Home hazards

Two studies reported that the presence of home hazards increased an older persons risk of falling, One study reported OR=1.51 (95% CI1.43-1.69) (Cesari 2002). The other study (Gill 2000) carried out detailed analysis and reported that the following contributes to the risk of falls:

Loose rugs and mats: hazard ratio=5.87(95% CI 1.42-24.2)

Carpet fold or tripping hazard: hazard ratio=3.45(95% CI 1.29-9.27).

Multiple medications

Seven studies were included in a systematic review and meta-analysis of cardiac and analgesic drugs (Leipzig et al. 1999a). All report that patients taking more than three to four medications were at risk of recurrent falls compared with patients taking fewer medications (range of results: OR 1.61 to 3.16). The studies included in this review were cohort, case control and cross sectional in design.

Anti-arrhythmic medications

In a meta-analysis of cohort, case control and cross-sectional studies (Leipzig 1999a), the following pooled results of 14 studies indicated that taking type 1A anti-arrhythmic drugs increase the risk of falling (OR 1.22, 95% CI 1.05 to 1.42).

Psychotropic medications

In a systematic review and meta analysis of cohort, case control and cross-sectional studies examining psychotropic drugs and falls (Leipzig 1999b) the pooled results for the association between taking any psychotropic drug and risk of falling was 1.73 (1.52 to 1.97).

Muscle weakness

Muscle weakness has been reported as a significant risk factor (Perell 2001). Our updated review did not identify any studies reporting statistical significance of this factor. One study conducted in extended care and one in community-dwelling setting reported non-significance in multivariate analysis (Bueno-Cavanillas 2000; Koski 1998).

Discussion

We have reported here risk factors that are associated with falling. These results were statistically significant in multivariable analyses. The evidence suggests that although each factor can be a predictor of falls, in some population groups or settings some risk factors may be more important than others. This is illustrated by studies that have carried out multivariate analysis and reported non-significance for each factor. However, there was substantial heterogeneity between studies and within each risk factor. Many different methods of measurement of risk factors are reported and no one study replicates another. An important example of this is those studies examining gait, balance and mobility problems. There is substantial overlap between each study's definition of each domain and method of measurement. The possible synergism between different risk factors should also be considered.

GDG commentary

Assessment of older people with a history of falling and the presence of other risk factors should be undertaken. The identification of older people at risk will enable practitioners to refer older people for effective interventions targeted at specific factors. Multifactorial assessment is an important process but must be linked to interventions. The grading of this recommendation reflects both the evidence on risk factors and level I evidence of assessment linked to intervention(s).

This multifactorial assessment should be done in the context of a comprehensive geriatric assessment where indicated.

1.3. Multifactorial interventions

(please see Sections Section 5.5, Section 5.6, Section 5.9, Section 5.11 for evidence review methods)

1.3.1. Recommendation

All older people with recurrent falls, or assessed as being at increased risk of falling, should be considered for an individualised multifactorial intervention. [A]

In successful multifactorial intervention programmes the following specific components are common (against a background of the general diagnosis and management of causes and recognised risk factors): [A]

  • strength and balance training
  • home hazard assessment and intervention
  • vision assessment and referral
  • medication review with modification/withdrawal.

1.3.2. Recommendation

Following treatment for an injurious fall, older people should be offered a multidisciplinary assessment to identify and address future risk and individualised intervention, aimed at promoting independence and improving physical and psychological function. [A]

Level of evidenceEvidence statement
Level IMultidisciplinary, multifactorial, tailored interventions are effective in reducing falls in the following population groups and settings:
  • community-dwelling older people
  • older people in extended care settings
  • older people presenting at A
  • E following a fall.
Level IThree trials suggest that multifactorial, multidisciplinary rehabilitation programmes are effective in reducing the incidence of further falling in older people who have suffered an injurious fall. This evidence is supported by key documents, in particular the expected standards of care outlined in the NSF for older people (standard six).

Evidence summary – multifactorial interventions

Community-dwelling, unselected (fallers and non-fallers in the population studied)

Of the eight studies that evaluated a multifactorial screening and intervention programme in community-dwelling older people, who were recruited on the grounds of age and domestic circumstances, without a requirement for the presence of known risk factors, data were pooled from four (Fabacher 1994; Jitapunkul 1998; Newbury 2001; Wagner 1994) involving 1,651 participants. The pooled data are homogeneous and show that the interventions are effective in reducing the proportion of fallers in the intervention group (pooled RR 0.73, 95%CI 0.63 to 0.85).

Data were not pooled from the other four studies in this category. In Carpenter (1990) (539 participants), which was cluster randomised by household, the intervention involved an assessment by trained lay volunteers using a disability rating scale; an increase in disability score at a repeat visit was reported to the family medical practitioner. Only the total number of falls in each group in the month before the final interview was reported. The trialists reported significantly fewer falls in the experimental group during that period, but insufficient data were available to calculate an effect size. The fourth of the incremental interventions in Steinberg (2000) also cluster randomised, had a medical screen, home hazard assessment, and exercise. There was no significant difference in the incidence of falling between this group (59 participants) and the control group (63 participants) who received an information package alone. Van Rossum (580 participants) found no difference in the incidence of falls between the intervention and control groups, but no data were provided. Vetter (1992) (674 participants) was cluster randomised (by household). There were 95 of 350 fallers in the intervention group and 65 of 324 in the control group.

Community-dwelling, targeted (population studied are known fallers or have identified risk factors prior to enrolment)

Data from two studies in this category were not pooled as cluster randomisation was employed. Coleman (1999) (169 participants) reported that screening and intervention in a chronic care clinic provided no significant improvement in the incidence of falls at 12 or 24 months. Tinetti (1994) (301 participants) reported a significant reduction in the number of fallers in the intervention group, adjusting for age, sex, previous falls, and number of risk factors (adjusted incidence rate ratio 0.69, 95%CI 0.52 to 0.90). Data were pooled from the other five studies (Close 1999; Hogan 2001; Kingston 2001; Lightbody 2002; van Haastregt 2000). The pooled data show a significant reduction in the proportion of fallers in the intervention groups (pooled RR 0.86, 95%CI 0.76 to 0.98).

Exercise, visual correction, and home safety intervention

Day (2002), in a study of factorial design, examined the effect of exercise, visual correction and a home safety intervention. The impact of these three interventions combined was a significant reduction in the number of participants falling (RR 0.76, 95%CI 0.61 to 0.94). Further analysis was carried out for the data for exercise plus vision correction (RR 0.76, 95%CI 0.62 to 0.95), and for exercise plus home hazard management (RR 0.84, 95%CI 0.69 to 1.03). These analyses are somewhat less favourable than the adjusted analyses presented by the authors in their original report.

Extended care

In Jensen (2002), a cluster randomised trial of an 11-week multidisciplinary programme, including general and resident-specific tailored strategies, reported a reduced incidence of falls in the intervention group (adjusted incidence rate ratio 0.60, 95%CI 0.50 to 0.73).

McMurdo (2000) (133 participants), also a cluster randomised study in an institutional setting, reported no significant difference between intervention and control groups in the percentage of participants falling in the six-month period after completion of the intervention. Ray (1997) (482 participants) was also cluster randomised. Data were reported on recurrent falls and injurious falls. The reporting of the data provides insufficient detail to confirm whether the reduction in recurrent falls experienced in the intervention group was significant. Rubenstein (1990) (160 participants) found no benefit from nurse practitioner assessment and physician referral within seven days of a fall (RR 0.97, 95%CI 0.84 to 1.11). Vassallo (2001) evaluated a multidisciplinary fall assessment in a cluster randomised trial in a geriatric rehabilitation setting, and reported fewer fallers (39/275) in the intervention group, compared with 111/550 in the control group.

Becker (2003), in a cluster randomised trial (N = 6) involving 981 long stay residents of community nursing homes, reported that the number of fallers was less in the intervention group (RR 0.75, 95%CI 0.57 to 0.98, trialists' analysis). The incidence density rate of falls per 1,000 resident years was also reduced in the intervention group (RR 0.55, 95%CI 0.41 to 0.73, trialists' analysis).

Cognitively impaired (any residence)

Shaw (2003), in a comparison of multifactorial assessment and intervention in 274 older people with cognitive impairment or dementia recruited from an A&E department following a fall, could not confirm the effectiveness of this intervention (RR 0.92, 95%CI 0.81 to 1.05). There is a lack of evidence of effective interventions for this group of older people. Many trials specifically excluded older people with a cognitive impairment.

Economic evidence

Tinetti et al. (1994) and Rizzo et al. (1996) both report on the same study. Tinetti reported on the clinical effectiveness. Rizzo undertook the cost effectiveness analysis. This study reported that the intervention package was cost effective in the high risk individuals. The high cost of the intervention was offset against the treatment costs of the high risk individuals. However, in this study not many of the control group had costly hospital admissions and the data was skewed. They undertook sensitivity analysis. There still remained a number of individuals in the intervention group who required costly treatment. The overall effect of this was to reduce the expected benefit in the intervention group. The analysis presented in the cost effectiveness analyses chapter (see Section 5.11) assumes that the at risk population can be reliably identified. Clearly the specificity and the sensitivity of the assessment tools will impact upon the cost effectiveness of the interventions.

The systematic review of assessment tools did not identify any information on the sensitivity and specificity of the existing assessment tools. In this context, there is even greater uncertainty about the true cost effectiveness of these interventions. The greater the ability of assessment tools to differentiate between those who are likely to fall without the intervention and the rest of the elderly population, the more cost effective the interventions will be. The figures presented in this chapter represent a best case, where the assessment is completely accurate. Nandy et al. (2004) report a high specificity (0.92) but a relatively low positive predictive value (0.57). Using this assessment tool, slightly more than 40 per cent of patients identified as being at high risk using this tool would not be expected to fall. This would have a significant upward impact upon the cost effectiveness results presented above. This evidence became available too late in the process for it to be incorporated directly in cost effectiveness modelling.

The existing evidence base for judging the cost effectiveness of these interventions is poor. If the at risk population can be identified, our analysis indicates that the multifactorial intervention is likely to be cost effective compared to conventional thresholds, although there is a large degree of uncertainty around the actual incremental cost effectiveness ratio.

GDG commentary

The evidence above suggests that multifactorial interventions targeted to risk factors are effective in reducing falls in older people. However, it is difficult to make a definite recommendation of the key effective components for specific settings and populations. It is sensible therefore to refer a patient for intervention(s) that target known risk factors. Illustrative examples of good practice were nominated from trials by Close (1999), Jensen (2002) and Tinetti (1994).

Close (1999) identified older people living in the community who presented at A&E following a fall. The intervention included a detailed medical and occupational therapist assessment, with referral to relevant services for targeted interventions either by modification of risk factors where possible; referral to multidisciplinary team for further interventions; and drug medication review by the GP. Advice and education was given by the occupational therapist (OT) about safety in the home and modifications were made where appropriate. The OT supplied minor equipment or referral was made to social or hospital services as required.

Jensen (2002) recruited older people from extended care settings who received assessment by a physician and physiotherapist. This assessment included a full clinical examination and medication review. Targeted interventions included staff education, environmental modifications, exercise, supply or repair of aids, medication review and hip protectors.

Tinetti (1994) recruited older people living in the community with the presence of one of the following risk factors: postural hypotension; use of sedatives; use of at least four medications; impairment in arm or leg strength or range of motion, balance, and ability to move safely from chair to bed. Assessment was conducted by a study nurse practitioner and physiotherapist. The intervention group was given either a combination of adjustment of their medications; or behavioural instructions and exercise programmes aimed at modifying their risk factors in the form of decision rules and intervention protocols for each risk factor.

These trials provide an example of approaches to providing effective multifactorial interventions, but the fundamental element is to prescribe or refer for targeted interventions.

Evidence summary: rehabilitation

Two trials reported a significant reduction in the incidence of further falling in those who had received attention for a previous fall (Close 1999; Crotty 2002). The intervention in the trial by Tinetti (1999) did not show an effect on the risk of falling, but there was a significant reduction in the incidence of individuals hospitalised. The key components of these studies included medical, physiotherapy and occupational therapy assessments with follow-up interventions, medical assessment to identify primary cause of fall and other risk factors with intervention or referral as required. Interventions may involve individually tailored exercise programmes aimed at improving strength, gait, balance, transfers and stair climbing. Social care and support were also part of some programmes.

The safety and efficacy of an exercise protocol designed to improve strength, mobility, and balance and to reduce subsequent falls in older patients with a history of injurious falls was examined in Hauer (2001). This RCT was a three-month intervention trial, with an additional three-month follow-up in an outpatient geriatric rehabilitation unit. The participants included 57 female patients, above the age of 75 years, admitted to acute care or inpatient rehabilitation, with a history of recurrent or injurious falls, including patients with acute fall-related fracture. Fall incidence was reduced non-significantly by 25 per cent in the intervention group, compared with the control group (RR: 0.753 CI: 0.455-1.245).

The studies reporting significant results suggest that a multifactorial approach, including multidisciplinary assessment and targeted interventions, could have some impact on reducing the incidence of falling as part of a rehabilitation programme following a fall resulting in medical attention. It is less clear from this evidence of the impact of these complex interventions on other factors – such as confidence; quality of life and acceptability – as limited data were available. Perhaps there also needs to be consideration of the planned withdrawal of such programmes and the ability of these individuals to sustain the improvement shown.

The evidence from geriatric hip fracture (GHFP) and early supported discharge (ESD) programmes suggest that they decrease the total length of hospital stay for older people who have suffered a hip fracture and inpatient treatment. In addition, these structured programmes of care achieve higher rates of return to previous residential status (Cameron et al. 2002). However, it is unclear what the effect these programmes have on reducing the incidence of a further fall. Furthermore, less is known about the impact on function, morbidity and quality of life for older people participating.

Two trials suggest that a multidisciplinary, multifactorial approach to management of older people, who have suffered an injurious fall and who have received treatment in a primary care or acute care setting, is an effective intervention package. Important components include assessment and a targeted intervention(s), underpinned by detailed discharge planning.

It is less clear which specific mechanisms of this multifactorial approach to rehabilitation are effective, but the fundamental key to success may be through comprehensive discharge planning.

In addition, the overall aim of these programmes for older people should be to regain confidence and subsequently prevent further falling. However, practitioners need to assess the extent to which the older person is likely to cooperate with the intervention programme and the usefulness of the overall prevention strategies in the Cochrane review.

GDG commentary

There was substantial overlap between secondary prevention interventions and rehabilitation strategies. It was not possible to review the evidence of the effect of these interventions on important rehabilitation outcomes – such as improvement in function, mobility and psychosocial health – as these outcomes were outside the scope of the guideline. In this guideline, rehabilitation is considered as part of the secondary prevention of falls, but users of the guideline need to be aware of the potential for improvement in outcomes other than falls prevention.

1.4. Strength and balance training

(please see Sections Section 5.5, Section 5.6, Section 5.9, Section 5.11 for evidence review methods)

1.4.1. Recommendation

Strength and balance training is recommended. Those most likely to benefit are older community-dwelling people with a history of recurrent falls and/or balance and gait deficit. A muscle strengthening and balance programme should be offered. This should be individually prescribed and monitored by an appropriately trained professional. [A]

Evidence summary

Exercise and/or physical therapy

Level of evidenceEvidence statement
Level IA programme of muscle strengthening and balance training, individually prescribed at home by a trained health care professional is effective in reducing falls (pooled results from three trials).
Community-dwelling: targeted interventions

Pooled data from three studies from New Zealand, with a total of 566 participants (Campbell 1997; Campbell 1999; Robertson 2001a), using the same individually tailored programme of progressive muscle strengthening, balance retraining exercises and a walking plan, indicated that this intervention significantly reduced the number of individuals sustaining a fall over a one-year period (pooled RR 0.80, 95% CI 0.66 to 0.98). The number of people sustaining a fall resulting in injury was also significantly reduced (pooled RR 0.67, 95%CI 0.51 to 0.89). Seventy-four per cent of participants in the control group and 69 per cent in the exercise group in Campbell (1997) continued for a second year. After two years, the rate of falls remained significantly lower in the exercise group (Campbell 1999a). The relative hazard for all falls in the exercise group was reported to be 0.69 (95%CI 0.47 to 0.97); the relative hazard for a fall resulting in a moderate or severe injury was 0.63 (95%CI 0.42 to 0.95).

These three studies involved older participants, but the components of the successful intervention suggest that balance retraining may be an important component of successful exercise programmes.

Economic evidence

The two papers by Robertson et al. (2001a, 2001b) report on the trials of the same home-based exercise programme in different centres carried out by different health care professionals. The programme delivered by the practice nurse was less costly than that delivered by the physiotherapist. However, the study undertaken with the practice nurse did not reduce hospital costs overall between the control and intervention groups.

The cost effectiveness analyses of exercise programmes for older people at risk of falling are reported. The exercise programme is likely to be cost effective but less cost effective than the multifactorial intervention. This said, exercise may produce other health benefits that have not been incorporated into the analysis presented in Section 5. However, in the absence of a sensitive and specific method for identifying those older people at high risk of falling, the cost effectiveness of exercise falls prevention strategies cannot be confirmed and any recommendation to implement such programmes should be treated with caution.

GDG commentary

The group agreed that strength and balance training should be administered by an appropriately trained professional. Although the evidence is relevant to community-dwelling older people with either a history of falls and/or a balance and gait deficit, this evidence could be generalised to other settings. At present, individually prescribed exercise has been shown to be effective in falls prevention. Evidence of effectiveness of group exercise interventions is emerging and will be considered in the update of the guideline. In addition, the health benefits of exercise should be considered.

1.5. Exercise in extended care settings

(please see Sections Section 5.5, Section 5.6, Section 5.11 for evidence review methods)

1.5.1. Recommendation

Multifactorial interventions with an exercise component are recommended for older people in extended care settings who are at risk of falling. [A]

Level of evidenceEvidence statement
Level IThe evidence suggests individually prescribed or group approached exercise interventions in extended care settings are not effective in reducing falls (pooled results from two trials and one single trial, non-significant). However, three trials report effectiveness of exercise as a component in multifactorial programmes.

Evidence summary

Nowalk (2001) in a study in long-term care facilities – ranging from independent living to skilled nursing care – reported no significant difference in number of falls between a control group and two untargeted exercise groups (resistance endurance training or Tai Chi).

Data were pooled from two studies. Donald (2000) studied the effect of a targeted physiotherapy programme in 54 patients in an elderly care rehabilitation ward. Mulrow (1994) studied elderly nursing home residents (194 participants), comparing a three times weekly exercise programme with a friendly visit of the same duration. The pooled data showed no evidence of effectiveness in this context (RR 1.02, 95%CI 0.74 to 1.41).

Schnelle (2003) compared a low intensity functionally oriented exercise and incontinence care programme with usual care in 190 incontinent nursing home residents. There was a non-significant trend towards a reduction in the number of fallers in this study, which may have been underpowered (RR 0.62, 95%CI 0.37 to 1.06).

GDG commentary

Whilst there is insufficient evidence to recommend exercise as a single intervention in extended care settings, multifactorial interventions in this setting with an exercise component have been shown to be effective. Please refer to page 66 for further details.

1.6. Home hazard and safety intervention

(please see Sections Section 5.5, Section 5.6, Section 5.11 for evidence review methods)

1.6.1. Recommendation

Older people who have received treatment in hospital following a fall should be offered a home hazard assessment and safety intervention/modifications by a suitably trained health care professional. Normally this should be part of discharge planning and carried out within a timescale agreed by the patient or carer, and appropriate members of the health care team. [A]

1.6.2. Recommendation

Home hazard assessment is shown to be effective only in conjunction with follow-up and intervention, not in isolation. [A]

Level of evidenceEvidence statements
Level IHome safety interventions/home hazard modifications have been shown to reduce the incidence of falls, especially in older people with a history of falling (pooled results from four trials).
Level IThere is no evidence for the effectiveness of home hazard modification in those without a history of falls in the previous year before enrolment (one trial, non-significant).

Evidence summary

Evidence for the effectiveness of home hazard management in people with a history of falling is somewhat strengthened by new data from the updated Cochrane review.

The association of domestic hazards with falls in the home has been controversial, despite its face validity (Clemson 1996; Gill 2000; McLean 1996; Northridge 1995; Parker 1996; Sattin 1998). However, six trials with a substantial home hazard modification component (Carter 1997; Cumming 1999; Day 2002; Hornbrook 1994; Nikolaus 2003; Pardessus 2002) have reported data that supports its effectiveness, particularly in those with a history of previous falls. Cumming (1999) cautioned that ‘this effect is unlikely to be caused by home modifications alone’ since the reduction in falls was not confined to falls inside the home. This is true also of the reduction in the number of participants reporting two or more falls in Carter (1997), where falls in the yard/garden associated with the dwelling were also eligible, and in the study reported by Stevens (2001). Hornbrook (1994) also used a complex intervention. While the evidence supports interventions designed to reduce home hazards, the exact mechanism of the effect remains unclear.

Five studies evaluated home safety interventions alone (Cumming 1999; Day 2002; Nikolaus 2003; Pardessus 2002; Stevens 2001). Data for number of participants falling are available from four, (Cumming 1999; Day 2002; Nikolaus 2003; Pardessus 2002). Amongst those participants with a history of falling in the year prior to randomisation, there was a significant reduction in the number of participants sustaining two or more falls in the study period (RR 0.66, 95%CI 0.54 to 0.81). An overall analysis including all participants, fallers and non-fallers prior to randomisation, showed a significant, but smaller, effect (RR 0.85, 95%CI 0.74 to 0.96).

In those without a history of falls in the previous year (Cumming 1999) there was no evidence for the effectiveness of home hazard modification (RR 1.03, 95% CI 0.75 to 1.41). In Cumming (1999) the rate of falls away from home was reduced by a similar extent to the reduction in falls at home.

Stevens (2001), in a population with mixed fall status, reported results of a cluster randomised study in which the individual household was the unit of randomisation. After one year there was no significant difference in the rate of falls (overall, and falls at home), the rate of fall injuries, or the proportion of fallers in the intervention group, compared with the control group.

Economic evidence

In a well-conducted cost effectiveness analysis, Salkeld et al. (2000) recruited patients during hospital admission, a number of whom had a history of falls. The intervention was implemented by an experienced occupational therapist. There was little improvement in the falls in the intervention group as a whole, but there was a statistically significant reduction in the number of falls in those with a previous history. The cost effectiveness relates to the high risk groups of older people.

Smith and Widiatmoko (1998) modelled the costs of fall with the costs of a home hazard intervention. Over the 10-year period of the model, they demonstrated a cost saving of A$92 per person. However the various sources of the data used, and assumptions made, indicate that although useful, it is not necessarily a substitute for empirical evidence.

GDG commentary

It is clear from the evidence that providing a home hazard assessment with an intervention aimed at modification for older people with a history of falling is effective. It is not clear which component of this intervention has the most impact on preventing further falls. However, a combination of advice, education interventions aimed at increasing confidence, risk awareness and home modifications are effective. Cumming (1999) reported a significant reduction in two or more falls in older people with a history of falls. Assessment was carried out by an occupational therapist and recommendations for prevention supervised as necessary. This intervention not only reduced the incidence of falls within the home but also falls outside the home.

There was debate about who should carry out home hazard assessments. The GDG acknowledged that in practice this may not always be carried out by a health care professional, but by a suitably trained member of the health care team. The personnel involved in assessment within the studies reviewed were trained health care professionals – including a doctor, and occupational therapist (Pardessus 2002); nurses, physiotherapists, occupational therapists and social workers (Nikolaus 2003); occupational therapist (Cumming 1999); and a trained assessor (Day 2002).

1.7. Psychotropic medications

(please see Sections Section 5.5, Section 5.6, for evidence review methods)

1.7.1. Recommendation

Older people on psychotropic medications should have their medication reviewed, with specialist input if appropriate, and discontinued if possible to reduce their risk of falling. [B]

Level of evidenceEvidence statement
Level IIOne trial of older people above 65 years suggests that a psychotropic medication withdrawal programme, involving a gradual withdrawal of psychotropic medication over a 14-week period, is effective in reducing the risk of falls.

Evidence summary

Psychotropic drugs include neuroleptics, sedatives/hypnotics, antidepressants, and benzodiazepines. These can increase an older person's risk of falling, as can the use of multiple medications. Results of a systematic review and meta analysis to identify particular medications that may increase an older person's falls risk suggest that older people taking more than three to four medications were at risk of recurrent falls; and those taking psychotropic medications were also at risk of falling.

Campbell (1999) reported the results of a study of factorial design, in which the interventions were an individually tailored exercise programme of progressive muscle strengthening and balance retraining; a walking plan (also used in Campbell 1997 and Robertson 2001a); and a placebo-controlled psychotropic medication withdrawal programme. This was gradual withdrawal of psychotropic medication over a 14-week period. Inclusion criteria included those above the age of 65 years who were currently taking benzodiazepine, any hypnotic, antidepressant or major tranquilliser.

The analysis reported by the investigators, using a Cox proportional hazard regression model, showed that the overall risk of falls was lower for the medication withdrawal group (relative hazard 0.34, 95%CI 0.16 to 0.74).

Economic evidence

One Australian and one US study (Andrews et al. 2001 and Coleman & Fox 2002) looking at the contribution of medication use were also assessed. These involved pharmacy reviews of medication, which may have resulted in falls. Neither of these studies provides strong economic evidence, but they highlight the importance of assessment following a fall. The costs detailed in the paper by Andrews (2001) show the relationship between medication and the outcomes for patients. The study by Coleman illustrates some potential cost savings in reviewing medications.

GDG commentary

In addition to the evidence for psychotropic medication review, polypharmacy was identified as a risk factor for falling and medication review should be part of a multifactorial assessment, as described in recommendation 3.

1.8. Cardiac pacing

(please see Sections Section 5.5, Section 5.6, for evidence review methods)

1.8.1. Recommendation

Cardiac pacing should be considered for older people with cardioinhibitory carotid sinus hypersensitivity who have experienced unexplained falls. [B]

Level of evidenceEvidence statement
Level IICardiac pacing in fallers with cardioinhibitory carotid sinus hypersensitivity is effective in reducing falls and syncope (one trial).

Evidence summary

Cardiac pacing in fallers with cardioinhibitory carotid sinus hypersensitivity (Kenny 2001) was associated with a statistically significant reduction in the number of participants who were not cognitively impaired, sustaining syncope (RR 0.48, 95%CI 0.32 to 0.73). In addition, the mean number of falls in 12 months in the intervention group was significantly reduced (WMD -5.2, 95%CI -1.0 to -9.4).

GDG commentary

This recommendation reflected the evidence for a standalone intervention for older people who have cardioinhibitory carotid sinus hypersensitivity. The evidence is also reflected in recommendation 3, which indicates that the GDG considered it necessary that a cardiovascular assessment should be carried out as part of a multifactorial assessment, where appropriate.

1.9. Encouraging the participation of older people in falls prevention

(please see Sections Section 5.7, Section 5.8, for evidence review methods)

1.9.1. Recommendation

To promote the participation of older people in falls prevention programmes the following should be considered. [D]

  • Health care professionals involved in the assessment and prevention of falls should discuss which changes a person is willing to make to prevent falls.
  • Information should be relevant and available in languages other than English.
  • Falls prevention programmes should also address potential barriers such as low self-efficacy and fear of falling, and encourage activity change, as negotiated with the participant.

1.9.2. Recommendation

Practitioners who are involved in developing falls prevention programmes should ensure that such programmes are flexible enough to accommodate participants' different needs and preferences, promoting the social value of such programmes. [D]

Level of evidenceEvidence statements
Level III-IVPeople may be reluctant to participate in falls prevention programmes if they have not previously exercised, do not perceive a risk of falling, have a fear of falling or perceived poor functional ability or have not been adequately consulted about what changes they are willing to make.
Level IVMuch of the current information provision on falls prevention programmes may alienate rather than encourage participation by stereotyping older people, not being available in languages other than English, not emphasizing that many falls may be preventable and not promoting the social value of falls prevention programmes.

Evidence summary

The review of the quantitative and qualitative evidence on older people's views and experiences enabled the identification of factors that may promote the idea of falls prevention. Multiple barriers to participation in falls programmes were identified, the most significant of which are summarised in Table 16.

Some studies indicate that much of the information on falls prevention alienates rather than encourages participation by stereotyping older people (Aminzedah & Edwards 1998; Ballinger & Payne 2000); and by not producing information in languages other than English (Aminzedah & Edwards 1998; Kong et al. 2002). Other information needs include giving special advice to older people about the benefits of physical activity and falls prevention and how to stay motivated in the face of multiple barriers (Commonwealth of Australia 2000; King 1995).

Some studies also reported a mismatch between the strategies willingly accepted by older people – for example, walking aids, home modification, low intensity exercise – and those that are most effective (balance and strengthening training) (Commonwealth of Australia 2000; Health Education Board 1999; Stead et al. 1997; King 1998). Two studies pointed out that imposition of strategies thought most optimal by health professionals may alienate the target group (Simpson et al. 2003; Porter 1999) and that health professionals need to find out which characteristics people are willing to modify and what changes they are prepared to make (Porter 1999) before suggesting strategies. This should be an ongoing process (Grossman et al. 2003).

Some of the individual factors that were shown to increase participation in falls prevention programmes or specific components of these programmes were: high exercise self-efficacy, past exercise history and general good health and functional ability King 1995; Oman 1998; Resnick 2000). Aspects of the format of falls prevention programmes that appeared to improve participation and maintenance included: home-based, telephone supervised, peer role models, low intensity exercise – for example, walking – moderate frequency – for example, two to three times per week – and be perceived as relevant, beneficial and fun for the participants. The social aspects of falls prevention programmes are probably their strongest selling point (Health Education Board 1999; Kong et al. 2002), particularly to older people without a history of physical activity.

Factors that appeared to be barriers to either initial participation or long-term maintenance of falls prevention programmes were mainly personal, rather than programme format issues. These included: low self-efficacy or lack of perceived ability to undertake components of the programme; fear of falling; fear of exertion; illness; denial or under-estimating personal risk of falling; embarrassment or increased inconvenience regarding use of assistive devices (Bruce 2003; King 1998; Yardley 2002).

In addition, the economic systematic review identified two studies that used quality of life measures (SF36) to look at the impact of fear of falling. The paper by Cumming et al. (2000) showed a link between fear of falling, SF36 measures and the admission to a long-term care institution. The study by Suzuki et al. 2002 showed that those subjects who expressed a great deal of fear of falling had SF36 scores, reflecting their increased anxiety and depression.

The most commonly occurring and consistent themes across all studies (observational and qualitative) were as follows:

Preferred strategies
  • People may be reluctant to participate in falls prevention programmes that have an exercise-based component (including balance training), if they have not previously regularly exercised and in which the social value of participation is not promoted. This requires consideration in light of the Cochrane review findings that a) a programme of muscle strengthening and balance retraining, individually prescribed at home by a trained health professional and b) a 15-week Tai Chi group exercise intervention are likely to be beneficial (Gillespie et al. 2003).
  • Interventions not involving behaviour change, such as home modification and assistive aids, appear to be more readily accepted among potential participants. There was a fairly consistent finding across the reviewed studies that prevention programmes that were home-based, moderate or low intensity exercise with frequent professional contact were most acceptable and showed higher participation rates (Hillsdon 1995; King 1998; Oman 1998). Other single interventions reported as being beneficial in the Cochrane review (Gillespie et al. 2003) – such as cardiac pacing and withdrawal of medicines – similarly may be more acceptable to some people.
Individual factors
  • Although trials of multifactorial packages have reported beneficial results (Gillespie et al. 2003), in clinical practice there may need to be more emphasis on finding out what characteristics a person is willing to modify and what changes are they prepared to make at what stage in their lives. This somewhat concurs with the finding that individually tailored interventions delivered by a health professional are more effective than standard or group delivered programmes (Gillespie et al. 2003).
  • There was also evidence that the following factors are associated with activity avoidance: increasing age, being female, increasing anticipation of loss of function (Yardley 2002), not facing up to the risk of falling, (Simpson 1995) lack of perceived ability (King 1998), fear of falling (Bruce 2003) and fear of exertion (Grossman et al. 2003). However, fall prevention programmes that address self-efficacy and encourage activity change may result in increased uptake of falls prevention programmes (Cheal 2001; Resnick 2002). This suggests that consideration of these factors is important when devising falls prevention programmes to ensure practical and appealing interventions are developed.
  • Barriers need to be addressed prior to participation in a falls prevention programme to ensure commitment to the strategies.
Health promotion and information needs
  • There is a need to inform and educate older people that many falls are preventable.
  • Perceived relevance of falls prevention may be low until a fall has been experienced.
  • The social value of falls prevention programmes, as well as the physical benefits, needs to be promoted to make them attractive to intended participants.
  • Those from non-English speaking backgrounds may require targeted health promotion.

1.10. Education and information giving

(please see Sections Section 5.8 for evidence review methods)

1.10.1. Recommendations

All health care professionals dealing with patients known to be at risk of falling should develop and maintain basic professional competence in falls assessment and prevention. [D]

1.10.2. Recommendations

Individuals at risk of falling, and their carers, should be offered information orally and in writing about: [D]

  • what measures they can take to prevent further falls
  • how to stay motivated if referred for falls prevention strategies that include exercise or strength and balancing components
  • the preventable nature of some falls
  • the physical and psychological benefits of modifying falls risk
  • where they can seek further advice and assistance
  • how to cope if they have a fall, including how to summon help and how to avoid a long lie.

Evidence summary

See evidence summary above associated with ‘encouraging participation in falls prevention programmes’.

1.11. Interventions that cannot be recommended

(please see Sections Section 5.5, Section 5.6 for evidence review methods)

1.11.1. Brisk walking

Level of evidenceEvidence statement
Level IIThere is no evidence that brisk walking reduces the risk of falling.

One trial showed that an unsupervised brisk walking programme increased the risk of falling in post-menopausal women with an upper limb fracture in the previous year. However, there may be other health benefits of brisk walking by older people.

Evidence summary

In one study (Ebrahim 1997), brisk walking in n=165 women with an upper limb fracture in the previous two years, reported RR 0.69,95% CI 0.12-4.03. This UK study included postmenopausal women identified from A&E and orthopaedic fracture clinic records, with a history of an upper limb fracture in the last two years. The intervention group received initial advice on general health/diet and then encouraged to build up to brisk walking 40 minutes, three times per week. The control group received initial advice on general health/diet and encouraged to perform upper limb exercises to improve post-fracture function. Falls events were greater in the intervention group.

GDG commentary

The group had reservations about this trial. It was a small trial with a specific group of older women. Although there was a significant increase of falls (I=52/81 vs. C=50/84 and fractures (I=2/81 vs. C=3/84) in the intervention group, the GDG recognise the limitations of the generalisability of these findings. For some other groups of older people, walking may have health benefits and should not be discouraged.

1.12. Interventions that cannot be recommended because of insufficient evidence

(please see Sections 5.5,Section 5.6 for evidence review methods)

We do not recommend implementation of the following interventions at present. This is not because there is strong evidence against them, but because there is insufficient or conflicting evidence supporting them.

1.12.1. Low intensity exercise combined with incontinence programmes

Level of evidenceEvidence statement
Level IThere is no evidence that low intensity exercise interventions, combined with continence promotion programmes, reduces the incidence of falls in older people in extended care settings (one trial, non-significant).

Evidence summary

Schnelle (2003) compared a low intensity functionally oriented exercise and incontinence care programme with usual care in 190 incontinent nursing home residents. There was a non-significant trend towards a reduction in the number of fallers in this study, which may have been underpowered (RR 0.62, 95%CI 0.37 to 1.06).

Economic evidence

The study by Schnelle et al. (2003) made a number of assumptions that were not all reported in the paper. They acknowledge that this was an expensive and labour intensive intervention. They do not detail the costs but refer to them in the discussion. This intervention resulted in no significant difference between the control and intervention groups in the costs of assessing and treating acute conditions. The only statistically significant result was the stable fall rate in the intervention group. However, the authors recommend caution when interpreting these results, as this was a post hoc decision to analyse the data in this way.

1.12.2. Group exercise (untargeted)

Level of evidenceEvidence statements
Level IExercise in groups should not be discouraged as a means of health promotion, but there is little evidence that exercise interventions that were not individually prescribed for community-dwelling older people are effective in falls prevention.

Evidence summary

Community-dwelling: untargeted interventions

Using the FICSIT definition of falling, participants (n=200) exposed to the 15-week Tai Chi intervention had a lower rate of falling than controls in one trial (risk ratio 0.51, 95%CI 0.36 to 0.73) (Wolf 1996). Local advertisements and direct contact recruited the participants in this study. Inclusion criteria included ambulatory older people, above the age of 70 years, living in unsupervised environments.

Eleven studies, involving a total of 1,480 participants, reported the results of exercise interventions offered to groups of older community-dwelling people, where exercise interventions were not individually prescribed. Pooled data from nine studies (Buchner 1997a; Cerny 1998; Cornillon 2002; Day 2002; Ebrahim 1997; Lord 1995; McMurdo 1997; Pereira 1998; Rubenstein 2000) does not confirm the effectiveness of untargeted exercise interventions in community-dwelling older people based on number of fallers (pooled RR 0.89, 95%CI 0.78 to 1.01). Data from Wolf (1996) were reported as adjusted estimates from a Cox proportional hazards analysis, and raw data to allow pooling were unavailable.

Carter (2002), in a comparison of a twice-weekly exercise class with no intervention, reported no difference between groups in the number of people falling. Means (1996) recruited 65 participants, with a history of falling, who all underwent a six-week supervised low to moderate intensity programme designed to improve balance and mobility. Thirty-one participants practised on an obstacle course, in addition to the exercise intervention, while 34 did not. No statistically significant difference in the mean number of falls was reported.

There were three complex intervention studies that included exercise. In a factorial design, Day (2002) compared group-based exercise, home hazard modification and management of reduced vision. Although group based exercise alone was the most potent single intervention in this study RR 0.82 (0.70-0.97), falls were also reduced when exercise was combined with home hazard management, or reduced vision management, or both.

The remaining two trials were cluster randomised; their data could not be pooled. One (Reinsch 1992) evaluated the effectiveness of classes teaching exercise, relaxation and health and safety topics relating to fall prevention, and classes without the exercise component. Results did not demonstrate a statistically significant reduction in number of fallers for either intervention. The other (Steinberg 2000), using a cumulative intervention in which three out of four groups received a monthly one-hour exercise class and encouragement to exercise between classes, reported that the intervention strategies could achieve an 18 to 40 per cent reduction in the incidence of falling, but the hazard ratios were not significant.

Conclusion

The evidence for effectiveness of group exercise interventions remains limited, apart from the Tai Chi intervention of Wolf (1996) and Day (2002). However, the three trials from New Zealand (Campbell 1997; Campbell 1999; Robertson 2001a), which used an individually tailored exercise programme of progressive muscle strengthening, balance retraining and a walking plan, demonstrated effectiveness. These three studies involved older participants, but the components of the successful intervention suggest that balance retraining may be an important component of successful exercise programmes. However, there is no evidence of clinical effectiveness of other exercise interventions that was untargeted to specific older people at risk of falling.

GDG commentary

The GDG recognises the emerging positive evidence for group exercise with two studies published beyond the date of the literature review underpinning these guidelines (Lord et al. 2003; Barnett et al. 2003). This new evidence will need to be included in the guideline update. In addition the global health benefit of exercise needs to be emphasised.

Economic evidence

The study by Buchner et al. (1997) reported a relative risk for falls in the control group of 0.61. This study also measured quality of life using the SF36. They note that the hospital use between the two groups was very similar and the length of stay for the control group was likely to be longer resulting in additional costs.

1.12.3. Cognitive/behavioural interventions

Level of evidenceEvidence statements
Level IThere is no evidence of effect that cognitive/behavioural interventions alone reduce the incidence of falls in community-dwelling older people of unknown risk status (two single trials, non-significant). Such interventions have included risk assessment with feedback and counselling and individual education discussions.

There is no evidence that complex interventions – in which group activities included education, behaviour modification programme aimed at modifying risk, advice and exercise – are effective in falls prevention with community-dwelling older people (four single trials, non-significant).

Evidence summary

Cognitive/behavioural therapy alone

In Gallagher (1996) (100 participants), comparison of the two risk assessment interviews and a feedback/counselling interview, with a single baseline assessment interview, showed that the intervention had no statistically significant impact on the main outcome measures. In Ryan (1996) (45 participants), analysis of the number of fallers at three months showed no evidence that individual education sessions provided by a trained nurse were more effective than the one-hour group discussion of intrinsic and environmental risk factors.

Complex interventions including cognitive/behavioural intervention

Carter (1997) (658 participants) and Hornbrook (1994) (3182 participants) used a behavioural approach after carrying out an environmental safety assessment. Data have not been pooled from these studies, as Hornbrook (1994) is cluster randomised (by household). Both had co-interventions. Hornbrook (1994) included group sessions designed to modify risk taking behaviour and an exercise component, and reported survival analyses for sustaining any fall, injury fall, medical care fall, fracture fall, and fall causing hospitalisation. Unadjusted rates for all falls were significantly lower among intervention participants; for other categories of fall (injury falls, medical care falls) there were no statistically significant differences between groups. In Carter (1997) advice on optimising medication was given to the two intervention groups; a low intensity intervention in which advice alone was given on home safety, and a high intensity intervention that included professional formulation of an action plan. There was no evidence of a difference in the number of individuals falling between the control group and either intervention group. However, both interventions were associated with a significant reduction in the number sustaining two or more falls (low intensity intervention RR 0.27, 95%CI 0.08 to 0.95; high intensity intervention RR 0.22, 95%CI 0.05 to 0.98). In a cluster randomised trial, Reinsch (1992) evaluated the effectiveness of classes teaching exercise, relaxation and health and safety topics relating to fall prevention, and classes without the exercise component. The trial did not identify a statistically significant reduction in number of fallers. In another cluster randomised trial (Steinberg 2000), a cumulative intervention in which three out of four groups received encouragement to exercise and a monthly one-hour exercise class, the intervention strategies achieved an 18 to 40 per cent reduction in the incidence of falling, but the hazard ratios were not significant in any group.

1.12.4. Referral for correction of visual impairment

Level of evidenceEvidence statements
Level IExercise in groups should not be discouraged as a means of health promotion, but there is little evidence that exercise interventions that were not individually prescribed for community-dwelling older people are effective in falls prevention.

Evidence summary

In Day (2002) there was no evidence that referral for correction of vision in community-dwelling older people was effective in reducing the number of people falling (RR 0.88, 95%CI 0.54 to 1.43). This study, using a factorial design, compared a control group with groups receiving exercise, correction of visual impairment, and home hazard modification, each alone, and in combination. Results above reflect analysis for the visual correction alone group.

GDG commentary

Whilst there is insufficient evidence that single interventions targeting vision impairment are effective in reducing falls, referral for visual correction as part of a multifactorial intervention has a significant impact on falls reduction.

Identifying older people with visual impairment and referral for intervention should be considered within a multifactorial intervention.

1.12.5. Vitamin D and oral supplementation

Level of evidenceEvidence statement
Level IThere is evidence that vitamin D deficiency and insufficiency are common amongst older people and that when present they impair muscle strength and possibly also neuromuscular function via CNS-mediated pathways. In addition, the use of combined calcium and vitamin D3 supplementation has been found to reduce fracture rates in older people in residential/nursing homes and sheltered accommodation. Although there is emerging evidence that correction of vitamin D deficiency or insufficiency may reduce the propensity for falling, there is uncertainty about the relative contribution to fracture reduction via this mechanism (as against bone mass) and on the dose and route of administration required. No firm recommendation therefore can currently be made on its use for this indication. Guidance on the use of vitamin D for fracture prevention will be contained in the forthcoming NICE clinical practice guideline on osteoporosis that is currently under development.

Evidence summary

There is no evidence from one small trial involving 50 participants (Gray-Donald 1995), for the effectiveness of a programme of oral nutritional supplementation – in this case, a high energy, nutrient-dense supplement – in preventing falls in a group of frail elderly women RR 0.10 (0.01 to 1.69).

Five studies (Bischoff 2003; Dawson-Hughes 1997; Latham 2003; Pfeifer 2000; Sato 1999) evaluated the effect of vitamin D on falling. Data were pooled from Bischoff (2003); Pfeifer (2000) and Latham (2003) (461 participants). In these studies both intervention and control groups received calcium supplementation; the intervention group in each received oral vitamin D supplementation. Within this group of pooled studies, no evidence was produced of the effectiveness of vitamin D supplementation in reducing the number of people who fall amongst community-dwelling or hospitalised older people (RR 0.87, 95%CI 0.70 to 1.08). In Pfeifer (2000), the reduction in the number of falls resulting in fracture was not statistically significant (RR 0.48, 95%CI 0.02 to 11.84).

In Sato (1999) (86 participants), the administration of 1-alpha-hydroxyvitamin D alone to people with Parkinson's disease (Hoehn and Yahr Stage <5) significantly reduced the number of fracture falls (RR 0.12, 95%CI 0.02 to 0.98), but did not reduce the mean number of falls in the intervention group (WMD 0.10, 95%CI -0.71 to 0.91).

In a placebo-controlled trial of administration of vitamin D and calcium supplementation to community-dwelling men and women over 65 years, Dawson-Hughes (1997) (445 participants) reported that the number of participants falling did not differ significantly between intervention and control groups. Data were not presented.

Vellas (1991) (95 participants) reported that administration of the vaso-active medication raubasine-dihydroergocristine to older people presenting to their medical practitioner with a history of a recent fall, significantly reduced the numbers of the intervention group who reported falls in the six months of therapy (RR 0.48, 95%CI 0.29 to 0.78).

A recent published meta-analysis of vitamin D supplementation suggests there is a reduction in falls (Bischoff-Ferrari, 2004). There results showed that vitamin D supplementation appears to reduce the risk of falls among ambulatory or institutionalised individuals with stable health by 20 per cent.

However, although there is emerging evidence that correction of vitamin D deficiency or insufficiency may reduce the propensity for falling, there is uncertainty about the relative contribution to fracture reduction via this mechanism (as against bone mass) and on the dose and route of administration required. No firm recommendation therefore can currently be made on its use for this indication. Guidance on the use of vitamin D for fracture prevention will be contained in the forthcoming NICE clinical practice guideline on osteoporosis that is currently under development.

1.12.6. Hip protectors (please see Section 5.10 and Section 5.11 for evidence review methods)

Level of evidenceEvidence statements
Level IReported trials that have used individual patient randomisation have provided no evidence for the effectiveness of hip protectors for the prevention of hip fractures when offered to older people living in extended care settings or in their own homes.

Data from cluster randomised trials provides some evidence that hip protectors are effective in the prevention of hip fractures in older people living in extended care settings who are considered at high risk.

Evidence summary

Incidence of hip fractures

Data from the five cluster randomised studies were not pooled with data from the individually randomised studies. Cluster randomisation methods were used in five studies (Ekman 1997; Harada 2001; Kannus 2000; Lauritzen 1993; Meyer 2003). However, an uncorrected exploratory analysis of the five cluster randomised studies was conducted by the trialists. In Kannus (2000), the exploratory analysis (RR 0.34, 95% CI 0.19 to 0.61) that uses the raw numbers of participants sustaining fracture in each group differs slightly from that in the primary report (relative hazard 0.4, 95% CI 0.2 to 0.8), which used Cox proportional hazards analysis adjusted for age, sex, body mass index, mental status, ability to walk, previous falls and previous fractures.

The cluster randomised trial by Ekman (1997) reports RR 0.34 (0.12-1.01) for the incidence of hip fractures, randomised by unit or nursing home. Harada (2001) reported the number of hip and other fractures, number of falls and compliance with hip protectors. Results for the incidence of hip fractures was RR 0.11 (0.01-0.84) and the incidence of other fractures RR 4.33 (0.21-88.74).

In the trial by Lauritzen (1993), the incidence of hip fractures, randomised by unit or nursing home, was RR 0.44 (0.20-0.93) and the incidence of pelvic fractures was RR 0.34(0.02-7.01).

The incidence of other fractures was RR 1.02 (0.55-1.89) in this trial.

Meyer (2003) reported the number of hip fractures, and other fractures; falls; mortality; compliance of wearing the hip protectors and the reasons for non-compliance. The incidence of hip fractures, randomised by unit or nursing home, was RR 0.53 (0.32-0.87) and the incidence of other fractures RR 1.14 (0.74-1.78).

Pooling of data from the seven trials in which randomisation was by individual showed no significant reduction in the incidence of hip fracture in those allocation to wearing hip pads (64/1306 (4.9%) versus 64/1086 (5.9%), RR 0.94, 95% CI 0.67 to 1.31).

Pooling of data from five individually randomised trials conducted in nursing/residential care settings (1,426 participants) (Cameron 2001; Chan 2000; Jantti 1996; Hubacher 2001; Van Schoor 2003) showed no statistically significant reduction in hip fracture incidence (hip protectors 37/822 (4.5%), controls 40/604 (6.6%), RR 0.83, 95% CI 0.54 to 1.24). The reviewers note that by the end of the one-year observation period, nearly half (16/36 versus 17/36) of the individuals in Jantti (1996) had been lost to follow-up through death or permanent hospitalisation.

Two individually randomised studies recruited community-dwelling older people (Birks 2003; Cameron 2003). These studies did not achieve a statistically significant reduction in the incidence of hip fractures (27/484 (5.6%) versus 24/482 (5.0%), RR 1.11, 95% CI 0.65 to 1.90).

Villar (1998) studied compliance with wearing hip pads in a study with a follow-up period of 12 weeks. As this study excluded mentally incapacitated patients, participants were at lower risk of hip fracture. No hip fractures occurred in either the 101 participants allocated to receive protectors or the 40 participants in the control group. Thus this study contributed no data to the meta-analysis.

Incidence of pubic ramus and other pelvic fractures

There is insufficient evidence to confirm whether the use of hip protectors significantly reduces the incidence of pelvic fractures. Data on the incidence of pubic ramus and other pelvic fracture were available in 10 studies. In the six studies that used individual randomisation there were 16/1266 (1.3%) in the protector group and 13/1055 (1.2%) in the control group (RR 1.15, 95% CI 0.58 to 2.31).

Incidence of other fractures/injuries

The use of hip protectors appears to have no effect on the incidence of other fall associated fractures. Data on the incidence of other fractures that occurred over the study periods were reported in 10 studies. Pooling of results from the individual randomised studies showed that 63/1266 (5.0%) occurred in the protector group and 56/1055 (5.3%) in the control group (RR 1.06, 95% CI 0.75-1.50).

Compliance

Amongst those who were assigned to their use, compliance with wearing of hip protectors was limited. It is not clear in some trials how compliance was measured, but for those that stated the method of measurement, the length of time wearing them was calculated.

Chan (2000) reported a compliance of 50.3 per cent, with dementia given as a reason for non-compliance. Ekman (1997) reported an average compliance of 44 per cent, although it is not clear how this was calculated. Harada (2001) reported that 17/88 (19 per cent) of those allocated to the protectors refused to wear them. Complete compliance estimated by hours worn was 70 per cent and partial compliance 17 per cent. Jantti (1996) stated that, of the 19 participants available at one year, 13 (68 per cent) were still using hip protectors. Of the subgroup of 45 individuals allocated to hip pads monitored in Lauritzen (1993), only 11 (24 per cent) wore the protectors regularly. In Kannus (2000), 31 per cent of those eligible declined to participate in the study, while a further 71 of 446 patients discontinued use during the study. Compliance in those who agreed to participate in the study – assessed as the number of days the protector was worn as a percentage of all available follow-up days – was 48 per cent (±29%, range <1 to 100%). Van Schoor (2003) used random visits to assess compliance and found that, at one month, 39 per cent were not compliant with wearing the protectors. This figure had risen to 55 per cent at six months and 63 per cent at one year. Hubacher (2001) reported that for 384 allocated to the protector group, 138 were regular wearers, 124 discontinued wearing them and 122 refused to wear them. Even the 138 ‘regular wearers’ only wore the pads 49.1 per cent of the time. Birks (2003) gave an overall compliance figure of 34 per cent. Cameron (2001) stated total compliance was 57 per cent. At the end of the study only 37 per cent were still regular wearers of the protectors. Meyer (2003) reported that the hip protectors were worn by 34 per cent of the intervention group participants. Cameron (2003) approached 1,807 potential subjects living in their own homes and 34 per cent of these agreed to participate. By two years, the end of this study, only 33-38 per cent of participants were wearing the protectors all the time. In Villar (1998), of the 288 individuals approached only 141 consented to participate. Of the 101 who received the protectors only 27 (27 per cent) wore them throughout the 12-week study period. In a breakdown of the reasons for non-compliance presented by Villar (1998), discomfort and poor fit were the most common reasons for discontinued use.

Other evidence reporting compliance problems is also worth summarising, as these sources of evidence also confirm many of the Cochrane review findings reported above.

A systematic review of the literature reported that the acceptance of, and compliance with, of hip protectors (Van Schoor 2002) ranged from 37 per cent to 72 per cent (median 68 per cent) for acceptance and 20 per cent and 92 per cent (median 56 per cent) for compliance. No details were given of specific settings or populations.

In a randomised controlled trial (Cameron 2000), the effect of hip protectors on fear of falling was examined in 131 women aged 75 and above who had two or more falls in the previous year. The results of this study report that hip protector users had greater improvement in falls self efficacy at follow-up.

In a prevalence study (Villar 1998), which aimed to assess compliance with the use of hip protectors in a residential setting, only 27 per cent wore the hip protectors for the full 12-week study period and half of the women wore them for less than one week. The reasons for non-compliance were poor fit or discomfort.

Pakkari (1998) conducted a before and after study assessing the acceptability and compliance with hip protectors in 19 ambulatory residents in a nursing home. The small sample size for this study prevents generalisability, but results indicated that the tight fit of the hip protectors reduced the ability for independent toileting.

Complications (including skin damage/breakdown)

Ekman (1997) mentioned that the occurrence of skin irritation was used as a reason for non-compliance. Villar (1998) reported three individuals who were unable to tolerate the special undergarments during a heatwave and also mentioned discomfort as the prime reason for non-compliance. Kannus (2000) reported skin irritation or abrasion in 15 cases. In addition, one person reported the protector caused swelling of the legs and another that it caused bowel irritation. Hubacher (2001) reported that aches and pains and an uncomfortable feeling with wearing the protectors was given as a reason for non-compliance. Minor skin irritation was reported in Cameron (2001), and Cameron (2003) reported minor skin irritation or infection caused by hip protectors in 16 users (5 per cent). Meyer (2003) reported five cases of skin irritation. In addition some of the care homes reported increased dependency of some of the residents at toileting, more difficulty in dressing and discomfort from wearing the protectors.

For the results of other outcomes measured in this review, see Evidence table 11 (Appendix E).

Summary

The cluster randomised studies, which formed the bulk of the evidence from the previous review (2001), supported a significant beneficial effect of hip protectors in reducing the incidence of hip fracture (Parker et al. 2003). However, this significant protective effect was not confirmed by pooling of data from studies using individual randomisation in the updated version (Parker et al. 2003). For those living in their own homes, the review authors suggest there is insufficient evidence from randomised trials to support any benefit of hip protectors. The authors note that in a number of the cluster randomised studies, although allocation was by institution, analysis was by individual, without allowing for the effect of clustering. This leads to an estimation of the treatment effect in which the confidence intervals are inappropriately narrow Thus there is a risk that a statistically significant effect appears to exist, when in fact it may not. This may have encouraged inappropriate interpretation of the strength of the evidence.

The authors of the Cochrane review also noted other shortcomings – such as evidence of heterogeneity amongst the populations studied in respect of baseline risk of fracture; that most of the individually randomised studies were underpowered; that the use of protectors appears to have varied between trials and within trials; and that initial acceptance of, and later compliance with, wearing the hip protectors were reported as problems in all of the studies.

The reader is referred to the Cochrane report for full details.

Finally, the studies included in the Cochrane review (Parker et al. 2003) and additional studies on hip protectors involve the use of a number of different designs of hip protector. It is not possible to be sure that the different types of hip protector used had equal effectiveness. A variety of different types of hip protectors have now been produced and clinical studies will be required to see if these new designs of protector are equally effective in reducing the risk of hip fracture. In addition, the compliance may vary for the different types of hip protector.

Economic evidence

Two studies were identified as being relevant to the use of hip protectors. The first paper by Kumar and Parker (2000) looked at the cost effectiveness of hip protectors using the audit data from an English hospital and the Cochrane review of musculoskeletal injuries (Parker et al. 2001). The intervention was the wearing of hip protectors and the control was no intervention. The outcome measure was the number of hip fractures prevented. As the cost and benefit period was calculated over one year, discounting was not necessary. Direct costs only were used in the analysis and the number of protectors needed per person was obtained from previous studies and communication with the authors. The cost per item was obtained from the manufacturer. The authors use a previously published paper to estimate the average cost of a hip fracture the data updated to their cost year (1998).

The cost results showed that the three hip protectors required for each person cost £113 per year. The average cost of treating a hip fracture was £7,200. The results were presented by age group. The cost of fracture prevented in the 50 to 59 age group was £508,500. The cost per fracture prevented in the above 85 age group was £2,485. The authors conclude that the use of hip protectors in the above 85 age group appears to be cost effective.

However, there are a number of assumptions made in this study that may influence the results shown. The costs were calculated for those people who complied. They did not cost the supply of protectors to people who did not comply. They report a compliance rate of 36 per cent, which suggests that there is a problem. In addition, no sensitivity analysis was carried out on the price of the protectors. No indirect costs for hip fracture were included. The results of this study should be treated with caution.

The second study by Segui-Gomez, Keuffel and Frick (2002) was a state transition model. This models the movement of patients through the probability of sustaining a fall resulting in a hip fracture, not falling or dying from any cause. That is to say the patient is in one of three states: well, hip fracture or dead. Data for models are obtained from published literature, epidemiological data, quality of life data or utility data. The data driving the model was obtained from published literature of trials. The authors state that they made some assumptions concerning the effectiveness of the protectors, which is normal when modelling. However, these assumptions need to be explicit in order to give validity to the model.

This model was populated by two hypothetical groups of 500,000 65-year-old men and women in the USA. The model was run for 35 years.

As with other studies it is difficult to generalise between health care systems. However, they did include a cost utility analysis. They obtained QALY data from expert opinion (a sample of gerontologists) and a sample of older people using a VAS scale. The authors do not give information about the sample other than it being one of convenience. There have been recent concerns about the use of VAS scales in deriving QALY data and this does raise some questions about their results (Brazier et al. 2003).

The authors showed that hip protectors are cost effective in the above 85 age group. The QALY data they collected showed that women gained QALYs overall, but with men there was a decrement. This is attributed to the inconvenience for men of wearing the protector.

There is considerable uncertainty about some of the sensitivity analysis. Compliance is an issue, as the authors state that hip protectors only result in cost savings when compliance is 70 per cent. The literature illustrates that there are problems with compliance and achieving 70 per cent would be difficult.

There are methodological questions with this model that make it difficult to use the results to inform practice.

Both of these studies have no intervention – that is doing nothing is the comparator. It is likely that this may not be the case in some areas where prescribed vitamin D and calcium or bisphosphonates may occur as part of a fracture prevention programme.

In view of recent effectiveness data, which show fewer benefits than previously anticipated, these two flawed cost effectiveness studies demand that their results be treated with caution.

GDG commentary

The GDG acknowledged that the evidence is less convincing of the effectiveness of hip protectors in the prevention of falls, following the update of the Cochrane systematic review on hip protectors. There was discussion about the benefit of hip protectors for high risk groups of older people. Older people at high risk might include those with the presence of multiple risk factors. However, the GDG felt that it was not possible, on the basis of the current clinical effectiveness evidence, to make a potentially expensive recommendation about their use until there are trials evaluating the newer types of hip protectors and national standards for their manufacture and safety are made.

Copyright © 2005, Royal College of Nursing.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means electronic,mechanical, photocopying, recording or otherwise,without prior permission of the Publishers or a licence permitting restricted copying issued by the Copyright Licensing Agency, 90 Tottenham Court Road, London W1T 4LP.This publication may not be lent, resold, hired out or otherwise disposed of by ways of trade in any form of binding or cover other than that in which it is published,without the prior consent of the Publishers.

Bookshelf ID: NBK55844

Views

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...