Osteoporosis International

, Volume 20, Issue 9, pp 1613–1620

External hip protectors are effective for the elderly with higher-than-average risk factors for hip fractures

Authors

    • Department of RheumatosurgeryOsaka City University Medical School
  • Y. Orito
    • Department of Orthopaedic SurgeryOsaka City University Medical School
  • H. Toyoda
    • Department of Orthopaedic SurgeryOsaka City University Medical School
  • M. Tada
    • Department of Orthopaedic SurgeryOsaka City University Medical School
  • R. Sugama
    • Department of Orthopaedic SurgeryOsaka City University Medical School
  • M. Hoshino
    • Department of Orthopaedic SurgeryOsaka City University Medical School
  • Y. Nakao
    • Department of Orthopaedic SurgeryOsaka City University Medical School
  • S. Kobayashi
    • Department of Orthopaedic SurgeryShinshu University School of Medicine
  • K. Kondo
    • Department of Public HealthOsaka City University Medical School
  • Y. Hirota
    • Department of Public HealthOsaka City University Medical School
  • K. Takaoka
    • Department of Orthopaedic SurgeryOsaka City University Medical School
Original Article

DOI: 10.1007/s00198-008-0824-7

Cite this article as:
Koike, T., Orito, Y., Toyoda, H. et al. Osteoporos Int (2009) 20: 1613. doi:10.1007/s00198-008-0824-7
  • 524 Views

Abstract

Summary

In our cluster randomised controlled trial for efficacy of hip protector with 672 ambulatory elderly women, a hip protector was more effective for prevention of hip fractures in residents with fall history (n = 202; hazard ratio (HR), 0.375; 95%CI, 0.14–0.98; p = 0.05) and body-mass index (BMI) ≤ 19.0 (n = 206; HR, 0.37; 95%CI, 0.14–0.95; p = 0.04) by a Cox proportional hazards regression model.

Introduction

Hip fractures result from both osteoporosis and falling. A potentially cost-effective method of preventing hip fractures involves the use of hip protectors but recent studies have revealed the uncertain effectiveness of hip protectors even in institutional settings.

Methods

This study was a cluster randomised controlled trial with nursing homes. We randomly assigned 76 homes with 672 ambulatory but frail elderly women. Several risk factors were assessed at baseline and incorporated into a Cox proportional hazards regression model. UMIN Clinical Trials Registry number is UMIN000000467. Research period was between January 2004 and March 2006.

Results

In the intervention group, 19 hip fractures occurred (54.0/1,000 person-years), whereas 39 hip fractures occurred in the control group (78.8/1,000 person-years). Hazard ratio of hip fracture in the intervention group was 0.56 (95%CI, 0.31–1.03; p = 0.06) after adjusting for risk factors. In subgroup analysis, hip protectors were more effective for prevention of hip fractures in residents with fall history (n = 202; HR, 0.375; 95%CI, 0.14–0.98; p = 0.05) and BMI ≤ 19.0 (n = 206; HR, 0.37; 95%CI, 0.14–0.95; p = 0.04). Overall compliance with use of hip protectors was 79.7%.

Conclusion

Risk of hip fracture can be reduced by hip protectors among elderly women with fall history and low BMI.

Keywords

FemaleHip fractureHip protectorOsteoporosisRCTRisk factor

Introduction

Hip fracture is one of the most devastating consequences of falls in the elderly, due not only to the resultant advanced disability and deteriorating quality of life, but also the higher morbidity and mortality rates [1]. Hip fractures are the result of both osteoporosis and falling. Efforts aimed at preventing hip fractures include targeting the elderly with lifestyle advice and multi-faceted interventions to reduce the risks of falling, including exercise and medication [2, 3]. The majority of hip fractures among older adults are caused by a sideways fall onto the hip and hip protectors have thus been developed as an alternative method for preventing hip fractures [4]. Hip protectors use protective pads to cover the greater trochanter and attenuate and/or disperse the force of a fall sufficiently to prevent fracture. A potentially cost-effective method with immediate effects on preventing hip fractures involves the use of external hip protector pants [5, 6]. The recently updated Cochrane review of hip protectors has found no preventive effect of hip protectors among ambulant community-dwelling individuals and the effectiveness of hip protectors in an institutional setting remains uncertain [7]. To date, at least 15 randomised controlled trials (RCTs) examining the effect of external hip protectors on incidence of hip fractures have been published. Hip protectors led to a significant reduction in the incidence of hip fractures in five studies [812] while results were borderline significant in two studies [13, 14] and no significant reduction in hip fracture incidence was seen in eight studies [1522]. These discrepancies among RCTs might be attributable to differences in the methods of randomisation, biomechanical properties of the actual hip protectors used and compliance rates for use of hip protectors. In the review article, Parker et al. concluded that there is no evidence of effectiveness of hip protectors from studies in which randomisation was by individual subject within an institution or for those living in their own homes [23]. At the same time, they also concluded that the pooled data from six cluster-randomised trials performed in institutions showed evidence of a significant reduction in incidence of hip fractures in the groups allocated hip protectors [9, 1114, 21, 23]. These two conclusions appear conflicting, particularly with respect to the institutionalised elderly populations examined. Although the same energy-shunting type of hip protector (Safehip; Tytex A/S, Ikast, Denmark) was used in eight RCTs [9, 12, 1416, 18, 21, 22], the efficacy of hip protectors was inconsistent across studies. Furthermore, compliance with required hip protector use was not well monitored and was relatively low in most previous RCTs [24]. Another possible explanation for the discrepancy between RCTs might be the heterogeneity of risk factors for hip fractures at baseline among subjects recruited in trials. We hypothesised that use of hip protectors by residents of nursing homes who are at higher risk of hip fracture would reduce the incidence of hip fractures if adequate compliance with hip protector use could be achieved. This large RCT used cluster randomisation in an institutional setting with adequate assessment of risk factors for hip fractures to address some of the limitations of previous studies.

Materials and methods

Study design and subjects

The study was a cluster-randomised controlled trial, with nursing and residential homes acting as the clusters. We asked 250 nursing and residential homes in Osaka and surrounding areas in Japan to participate in the study through Osaka City Council of Social Welfare and 76 nursing homes and residential care facilities expressed interest. In previous studies, low compliance rates with protector use have been problematic. Compliance is largely determined by the motivation and competence of staff in nursing homes, so to reduce burdens on staff in the intervention group and increase compliance rates for the use of hip protectors, the nursing staff and researchers selected five residents from each home in the intervention group and 15 residents from each home in the control group to be subjects, according to predefined inclusion criteria after cluster randomisation. Facilities were thus randomly allocated in a 1:3 ratio to control and intervention groups (Fig. 1). Randomisation was generated from a table of random numbers. A numbered and sealed opaque envelope containing allocation details was opened by the author (TK) after enrolling homes.
https://static-content.springer.com/image/art%3A10.1007%2Fs00198-008-0824-7/MediaObjects/198_2008_824_Fig1_HTML.gif
Fig. 1

Flow of nursing home clusters and participants

We originally intended to recruit women displaying hip fracture risk factors in aged-care facilities in Japan. The populations targeted for study were older women at higher-than-average risk for falls and hip fracture. Women were eligible to take part in the trial if age was ≥65 years and at least one of the following risk factors was present: history of any prior fracture, low body-mass index (BMI), family or individual history of hip fracture, frequent faller status, current smoker, or other frail residents. These risk factors were taken from large prospective studies reported from the United States and Japan [25, 26]. Women were ineligible if they had experienced bilateral hip fractures or hip replacement surgery. Current use of any anti-fracture medication was not an exclusion criterion, nor did we exclude women with any form of illness unless they were bedridden. The UMIN Clinical Trials Registry number for this study is UMIN000000467 [http://www.umin.ac.jp/ctr/].

Outcome measures

The primary outcome of interest was fracture of the proximal femur. Secondary outcome variables were non-hip fractures including pelvic fractures, falls and, in the intervention group, compliance with wearing hip protectors (defined as being observed to wear the hip protectors for 24 h or at least during the daytime). Data on hip and other fractures in all facilities were obtained through information collected by the clinical research nurse or researchers during monthly visits to the homes and crosschecked against local hospital records and roentgenograms.

Data collection

The study required the use of various data collection tools, particularly for the assessment of risk factors. For the comparison of intervention and control groups regarding demographic and health-related circumstances at baseline, personal details, information on medication and existing fall risks were gathered by researchers and caregivers in each home. Bone density was assessed by an ultrasonographic measurement (speed of sound [SOS]) at the calcaneus using a CM-100 bone densitometer (Furuno Electric, Nishinomiya, Japan) [27]. Mental status was scored using the Mini-Mental State Examination (range, 0–30) [28]. A trained clinical research nurse and researchers made monthly regular visits to all intervention and control homes throughout the trial to promote implementation and monitor progress. Residents and staff in all homes were provided with posters and monthly newsletters to remind about the prevention of hip fracture.

In the study, 76 facilities were enrolled and the heterogeneity of facilities needed to be assessed. Characteristics of homes were evaluated (Table 1).
Table 1

Baseline characteristics of nursing homes

Characteristic

Intervention group (n = 54)

Control group (n = 22)

Ownership of homes (%)

 Public

8 (15)

6 (27)

 Non-profit

8 (15)

3 (14)

 Private

38 (70)

13 (59)

Number of residents per home

 Total

82.2 ± 43.1

70.0 ± 21.2

 Women

66.1 ± 33.8

56.5 ± 18.3

Participants in this study (%)

12.4 ± 13.6

29.4 ± 12.0

Number of residents per caregiver

 During day time

6.5 ± 3.7

6.8 ± 2.7

 During night time

23.0 ± 7.9

26.2 ± 17.6

Number of falls per home (per 100 occupied beds)

 2001a

41.1 ± 38.6

55.2 ± 63.9

 2002

79.6 ± 88.2

88.3 ± 92.8

 2003

101.5 ± 149.8

104.7 ± 95.9

 2004

117.8 ± 93.7

112.6 ± 97.8

Number of hip fractures per home (per 100 occupied beds)

 2001

2.6 ± 2.4

2.2 ± 2.1

 2002

2.6 ± 2.4

2.5 ± 2.4

 2003

3.0 ± 2.4

2.7 ± 2.6

 2004

2.6 ± 2.6

2.2 ± 2.1

Number of other fractures per home (per 100 occupied beds)

 2001

1.6 ± 1.7

1.3 ± 2.7

 2002

1.9 ± 2.2

2.1 ± 2.5

 2003

1.8 ± 1.8

3.3 ± 3.8

 2004

2.5 ± 2.7

2.3 ± 2.0

Values represent mean ± SD unless stated otherwise

aThe term “2001” means the period from April 1, 2001 to March 31, 2002

Hip protector

Intervention group participants were issued with three pairs of hip protectors (Safehip; Teijin Pharma, Tokyo, Japan, compatible with the product by Tytex) at the beginning of the study and general advice on how to reduce fracture was provided in the form of a leaflet. Hip protectors could be replaced as required during the project in case of loss, damage or shortage. Residents in the intervention group were thus encouraged to wear the protectors day and night or as much as possible. Conversely, residents in the control group received only the leaflet and standard care.

Statistical analysis

The required sample size was calculated on the basis of a reduction in the rate of hip fracture from 9.7% to 1.2% over 12 months (based on results from an earlier study in Japan) [9], using a ratio of control to intervention participants of 1:1, 80% power and a 5% level of significance. This yielded target sample sizes of 374 participants in each group. All principal analyses were intention-to-treat analyses, using data from all 672 residents entered into the study. Hazard ratio (HR) and 95% confidence interval (CI) were estimated using the Cox proportional hazards regression model. In the case of group comparisons, Wilcoxon’s rank sum test for continuous variables and the χ2 test or Mantel–Haenszel test for dichotomous or non-continuous variables. The period of observation used in calculating hazard ratio began at the date of enrolment and ended at the date of receiving hip fracture as primary outcome. In the analysis, we adjusted for age, BMI, bone status, previous falls, previous fractures and use of diapers. We also performed analyses stratified by fall history and BMI at baseline, exploring an a priori hypothesis that the effects of the hip protector are affected by the grades of risk factors. All statistical analyses were performed using SAS version 9.1 (SAS Institute, Cary, NC, USA) by the authors (TK, KK and YH).

Ethical approval

All subject protocols were obtained from the Institutional Review Board of Shinshu University School of Medicine. Written informed consent was obtained from all residents in the intervention and control homes. If a resident was not mentally competent to provide consent, their proxies were asked to consent to their relative participating in the study.

Role of the funding source

The sponsors of the study had no role in the study design, data collection, data analysis, data interpretation or the writing of the report.

Results

During the study, 250 nursing or residential homes were asked to enter the study. A total of 76 homes agreed to participate and were randomly allocated to achieve a 3:1 ratio of intervention to control homes. In total, 345 elderly women from 54 homes were allocated to the intervention group and 327 residents from 22 homes to the control group between January 2004 and March 2005, with follow-up through to March 2006 (Fig. 1). Baseline characteristics of homes and individuals of both groups are presented in Tables 1 and 2, respectively. As residents with higher risk factors for hip fracture were enrolled, residents in the intervention group displayed greater risk for osteoporotic fractures than the control group in several characteristics: BMI, medical condition, cognitive status, ability to move, previous falls and previous hip fractures (Table 2).
Table 2

Baseline characteristics of subjects

Characteristic

Intervention group (n = 345)

Control group (n = 327)

p value

Age (years)

85.3 ± 6.6

85.2 ± 7.0

0.87a

Height (cm)

143.1 ± 7.5

144.6 ± 7.4

0.02a

Weight (kg)

41.9 ± 7.8

45.1 ± 8.5

<0.001a

Body mass index (kg/m2)

20.5 ± 3.6

21.6 ± 4.0

<0.001a

Medical conditions (n (%))

 Heart disease

68 (20)

67 (20.5)

0.80b

 Hypertension

106 (31)

125 (38)

0.04b

 Previous stroke

56 (17)

56 (18)

0.81b

 Diabetes mellitus

13 (4)

6 (2)

0.13b

 Osteoarthritis (knee or hip pain)

101 (31)

87 (25)

0.12b

 Eye disease

28 (8)

40 (12)

0.08b

 Parkinsonism

26 (8)

9 (3)

0.005b

Cognitive status (MMSE score)d

11.4 ± 8.7e

14.6 ± 9.8e

<0.001a

Ability to move (n (%))

  

0.008c

 Entirely independently

82 (24)

86 (26)

 

 With cane

43 (12)

62 (19)

 

 With walker

75 (22)

83 (25)

 

 With wheelchair

145 (42)

96 (30)

 

Use of diaper (n (%))

215 (62)

143 (44)

<0.001b

Falls during the preceding 2 months (n (%))

  

<0.001c

 0

207 (60)

250 (76)

 

 1

51 (15)

40 (12)

 

 2 or 3

55 (16)

26 (8)

 

 ≥4

32 (9)

11 (3)

 

Fracture history

 Hip fracture

98 (28)

45 (14)

<0.001b

 Other fractures

63 (18)

53 (16)

0.48b

SOSf (m/s)

1461.1 ± 30.2g

1457.4 ± 28.2g

0.12a

Values represent mean ± SD unless stated otherwise.

ap values were calculated using Wilcoxon’s rank sum test for continuous variables

bp values were calculated using the X2 test for dichotomous variables

cp values were calculated using Mantel Haenszel test for non-continuous variables

dCognitive status was scored using the Mini Mental State Examination (MMSE)

eNot recorded for three people in each group due to poor communication with participants

fSOS: speed of sound, ultrasonographic assessment of calcaneal bone

gNot recorded for 15 people in the intervention group and 17 in the control group due to poor communication with participants. T-scores of SOS (mean ± S.D.) in control and intervention groups were −2.4 ± 0.85 and −2.3 ± 0.91, respectively.

Hip fractures

Total follow-up time was 352 person-years for the intervention group and 495 person-years for the control group for the analysis of hip fractures. In the intervention group, 19 hip fractures occurred in 19 persons (54.0/1,000 person-years) (Table 3). Of the 19 fractures in this group, seven fractures occurred in falls while wearing hip protectors, seven fractures in falls without hip protectors, two fractures without falls and three fractures before intervention. In the control group, 39 hip fractures occurred in 39 residents (78.8/1,000 person-years). There was no hip fracture without falls in the control group. HR of hip fracture in the intervention group was 0.635 (95%CI, 0.37–1.10; p = 0.11) in univariate analysis and 0.56 (95%CI, 0.31–1.03; p = 0.06) after adjustment for age, BMI, bone status, previous falls, previous fractures and use of diapers (Table 3).
Table 3

Fractures during follow-up in the intervention and control groups

Type of fracture

Intervention group (n = 345)

Control group (n = 327)

Hazard ratioa (95% CI), p value

Number of fractures

Incidencebn/1,000 person-years

Number of fractures

Incidencebn/1,000 person-years

Before adjustment

After adjustmentc (n = 639)

Hip fracture

19

54.0

39

78.8

0.635 (0.37–1.10) p = 0.11

0.56 (0.31–1.03) p = 0.06

Other fracture

17

48.4

28

56.6

0.81 (0.44–1.49) p = 0.50

0.65 (0.33–1.27) p = 0.20

aHazard ratio was calculated by Cox proportional hazards analysis

bTotal duration of follow-up was 352 person-years in the intervention group and 495 person-years in the control group for hip fracture analysis. Total duration of follow-up was 343 person-years in the intervention group and 475 person-years in the control group for non-hip fracture analysis

cRelative hazards were adjusted for age, body mass index, speed of sound at calcaneal bone, previous falls, previous fractures, use of diapers

In subgroup analysis, HR of hip fracture in the intervention group was 0.62 (95%CI, 0.28–1.35; p = 0.23) among subjects who had not experienced falls during the 2 months before enrolment (n = 437) (Table 4). Conversely, the hip protector was effective for preventing hip fractures in residents with a history of falls (n = 202; HR, 0.375; 95%CI, 0.14–0.98; p = 0.05). Furthermore, another subgroup analysis was performed according to BMI classification. Among participants with BMI ≤ 19.0 (n = 206), HR of hip fracture in the intervention group was significantly low (0.37; 95%CI, 0.14–0.95; p = 0.04). However, no significant difference was identified between intervention and control groups with regard to time to first hip fracture for residents with BMI > 19.0 but ≤22.3 (n = 217) or >22.3 (n = 214) (Table 4).
Table 4

Hip fractures during follow-up in subgroups

Subgroups

Number of events

Hazard ratio

p value

Control

Intervention

(95% CI)a

No previous falls (n = 437)

24

11

0.62 (0.28–1.35)b

0.23

Previous history of falls (≥1, n = 202)

15

8

0.375 (0.14–0.98)b

0.05

 

 

 

 

 

Low BMI (≤19.0, n = 206)

17

7

0.37 (0.14–0.95)c

0.04

Mid BMI (>19.0, ≤22.3, n = 218)

12

8

0.63 (0.24–1.69)c

0.36

High BMI (>22.3, n = 215)

10

4

0.73 (0.19–2.82)c

0.65

aRelative hazard of hip fracture was calculated by Cox proportional hazards analysis

bRelative hazards were adjusted for age, body mass index, speed of sound at calcaneal bone, previous fractures, use of diapers

cRelative hazards were adjusted for age, speed of sound at calcaneal bone, previous falls, previous fractures, use of diapers

Pelvic and other fractures

Risk of pelvic and other fractures was similar between groups. Adjustment of the results for potentially confounding variables did not alter these findings (Table 3). The Cox proportional hazards model was used for this secondary endpoint and the outcome measure was the time to first pelvic or other fracture. Total follow-up time for this analysis was 343 person-years for the intervention group and 475 person-years for the control group.

Falls

No difference was seen in number of falls per individual between groups (Table 5). However, the percentage of residents without falls during the observation was greater in the intervention group.
Table 5

Falls and use of hip protector

 

Intervention group (n = 345)

Control group (n = 327)

p value

Falls (n)

825

758

 

Falls per person-years (n)

3.4 ± 7.4

3.5 ± 13.7

0.64d

In person with

  

0.03e

 No fall (%)

158 (46)

99 (30)

 

 1 fall

52 (15)

84 (26)

 

 2 or 3 falls

65 (19)

78 (24)

 

 ≥4 falls

70 (20)

66 (20)

 

% of falls protected with hip protector (n (%))

 100%

121 (65)

  

 51–99%

26 (14)

  

 1–50%

20 (11)

  

 0%

20 (11)

  

Falls in which hip protector was used

 Totala

85.3%

  

 Meanb

79.8 ± 33.7%

  

% of durationc

 24-h wearing

54.1 ± 36.7%

  

 During the daytime

25.6 ± 31.3%

  

 No wearing

13.5±2.2%

  

Values represent mean ± SD unless stated otherwise

aCalculated using the formula: total number of falls in which hip protector was used/total number of falls observed by all participants in intervention group

bCalculated using the formula: mean of (total number of falls in which hip protector was used/total number of falls observed by each participants in intervention group)

cCalculated using the formula: (corresponding days/duration in each participant) × 100

dWilcoxon’s rank sum test

eMantel–Haenszel test

Compliance and adverse events of hip protectors

Compliance with hip protector use (for at least part of the day) was 79.7% throughout the study period. In the intervention group, 65% (121/187) of fallers experienced falls always with hip protector (Table 5). The percentage of falls that occurred while hip protectors were being worn was 85.3% in total, 79.8% in mean (Table 5).

Six residents in the intervention group reported skin-related adverse events and refused to wear hip protectors from that time on.

Discussion

Hip protectors were effective in preventing hip fractures in this study, with a large subject cohort (n = 672) and high compliance rate for use of hip protectors (79.7%) by the intervention group according to intention-to-treat analysis in an institutional setting. In particular, hip protectors were effective among residents with higher risk for hip fractures, that is, previous history of falls and lower BMI. Cummings and Nevitt hypothesise that four conditions must be satisfied for a fall to cause hip fracture: (1) impact of the fall near the hip, (2) failure of the normal protective responses of the individual, (3) poor energy absorption/distribution of soft tissues around the hip and (4) failure of reduced bone strength to withstand energy transmitted to the proximal femur [29]. When a person falls on the hip, the hip protector is designed to absorb and/or shunt away the impact toward the soft tissues, to keep force on the trochanter below the fracture threshold. Frequent fallers and lean residents are at highest risk for hip fractures and hip protectors should theoretically be effective for such individuals.

In previous RCTs, assessment of risk factors has been relatively poor. Of 15 RCTs, only two have included BMI as a risk factor for the estimation of the relative hazard, and those trials showed significantly reductions in hip fractures with the use of hip protectors [9, 11]. The most recent report as to the effect of hip protectors failed to demonstrate any protective effect on hip fracture incidence in nursing home residents, despite high compliance [20]. That RCT was unique, with 37 nursing homes randomly assigned to having residents wear a one-sided hip protector on the left or right hip. Using this method by which each subject served as their own internal controls, many problems inherent in the previous designs of cluster or individually randomised trials were avoided. Conversely, the effects of risk factors that should be the same for both sides of the hip could not be evaluated. Cognitive impairment represents a known risk factor for fall and hip fracture. The present trial identified few effects of cognitive status (data not shown), probably due to relatively low scores on the MMSE in both groups (Table 2). In the multivariate analysis, we adjusted for age, BMI, bone status, previous falls, previous fractures and use of diapers. These risk factors except use of diapers were identified in previous studies [25, 26]. We presupposed that the use of diapers affected the motion of hip joints and walking abilities. By Cox proportional hazards analysis, HR of hip fracture in the residents with use of diapers was 1.48 (95%CI, 0.87–2.50; p = 0.15) in univariate analysis and the numbers of falls among subjects with use of diapers were greater than non-users of diaper in both control and intervention groups (data not shown). Our study has several limitations. Cluster randomisation may introduce a methodological bias. Nursing homes designated for intervention may differ from control facilities with respect to hip fracture rate, awareness of falls or aggressiveness of fall-prevention programs [30]. However, cluster randomisation was essential, as the intervention program relied on changes to nursing techniques. In our study, no significant difference was seen in the number of falls between groups (Table 5). This suggests no difference in the nursing techniques for preventing falls between groups. Furthermore, we selected five residents from each facility allocated to the intervention group and 15 residents from control homes. Only 12.4% of residents in intervention homes and 29.4% in control homes participated in our trial (Table 1). Consequently, significant differences existed between groups in some variables at baseline (Table 2). The intervention group had lower BMI, greater numbers of previous falls and hip fractures, greater use of diapers, and lower cognitive condition than the control group, as characteristics that may increase fracture risk (Table 2). However, other differences in unknown fracture risk characteristics may have been present between groups. Hip protectors clearly do not always prevent hip fracture in the event of a fall, as 36.8% (7/19) of hip fractures in the intervention group occurred while protectors were being worn. Cameron et al. reported that hip protectors improve fall self-efficacy [31]. In our trial, no difference was identified in the number of falls between groups, but the percentage of non-fallers was higher in the intervention group (Table 5). The use of hip protectors might lessened fear of falling and improved falls efficacy. Making definitive comparisons between studies is difficult, as compliance has been measured in many different ways [24]. However, seven randomised controlled studies provided the rate of wearing hip protectors at the time of fall (24–74%) [1114, 1618]. The level of compliance achieved in this study was superior to previous studies, particularly in terms of the percentage of falls protected with hip protectors (85.3%, Table 5). This high rate of hip protector use in the intervention group might be one of the reasons a positive effect was found for hip protectors in our study. O’Halloran et al. found that the policy of making hip protectors available free-of-charge to residents of nursing and residential homes was ineffective in reducing hip fracture rates [21]. We agree with this finding and propose selective provision of hip protectors for specific subgroups at higher risk of hip fractures. Future studies or re-analyses of previous RCTs should examine the effects of risk factors on the results of interventions using hip protectors in detail.

Acknowledgements

This study was supported by grants from the Health and Labour Sciences Research Grants for Comprehensive Research on Aging and Health (TK and KT), Japan and the Research Society for Metabolic Bone Diseases (TK), Japan. We are indebted to the local research coordinators and caregivers at participating facilities for their help, recognition and dedication in conducting the trial. We also wish to thank Tadao Tsuboyama, Kazuki Sakamoto, Motohiro Fujii and Hideaki Ishibashi for their help in collecting data and Tomoko Watanabe for her special efforts as a research nurse in recruiting subjects, conducting interviews and communicating with caregivers.

Conflicts of interest

None.

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2008