Osteoporosis International

, Volume 21, Issue 5, pp 797–803

Fracture incidence in nursing homes in Japan

Authors

    • Department of Community Preventive MedicineNiigata University Graduate School of Medical and Dental Sciences
  • M. Oyama
    • Department of Community Preventive MedicineNiigata University Graduate School of Medical and Dental Sciences
  • S. Takahashi
    • Department of Community Preventive MedicineNiigata University Graduate School of Medical and Dental Sciences
  • Y. Yoshizawa
    • Niigata Council of Institutions for the Elderly
  • R. Kobayashi
    • Department of Physical TherapyNiigata University of Health and Welfare
  • R. Oshiki
    • Department of Physical TherapyNiigata University of Health and Welfare
  • T. Saito
    • Department of Health and NutritionNiigata University of Health and Welfare
  • Y. Tsuchiya
    • Department of Community Preventive MedicineNiigata University Graduate School of Medical and Dental Sciences
Original Article

DOI: 10.1007/s00198-009-1015-x

Cite this article as:
Nakamura, K., Oyama, M., Takahashi, S. et al. Osteoporos Int (2010) 21: 797. doi:10.1007/s00198-009-1015-x

Abstract

Summary

We conducted the first study to determine fracture incidence in institutionalized elderly people in East Asia. Fracture incidence was generally higher than in the general population but was not for all fracture types. Specific fracture prevention strategies for institutionalized people are needed. Ethnic differences in fracture incidence are also discussed.

Introduction

The aim of this study was to determine the incidence of fractures in nursing homes in Japan, where fracture is becoming an increasingly problematic health issue.

Methods

We conducted a cohort study with a 1-year follow-up. Subjects were 8,905 residents (average age, 84.3 years) in nursing homes. Caregivers assessed activities of daily living levels using Barthel’s index, and we collected demographic data of each resident. Fracture diagnosis was based on medical records from the hospitals where patients were admitted.

Results

Incidence of hip fracture was 14.9 per 1,000 person-years for women and 9.7 for men. Incidence of forearm and upper-arm fractures was 1.9 and 5.1 for women and 0.5 and 2.1 for men, respectively. Hip and upper-arm fracture rates were higher than in the general population, but forearm fracture rates were lower. Hazard ratio of fracture incidence in totally dependent to partially dependent or independent subjects was 0.14 (95%CI 0.05–0.33) for hip fracture and 0.34 (95%CI 0.23–0.50) for all fractures.

Conclusions

Fracture incidence in institutionalized elderly people is generally higher but is not higher for all types of fractures than the general population. In addition, totally dependent persons had a much lower risk of hip fracture.

Keywords

Cohort studiesFracturesFrail elderlyNursing homesOsteoporosis

Introduction

Fractures significantly impact many aspects of health in elderly people. For example, hip fracture dramatically decreases the activities of daily living (ADL) level [1], especially in terms of mobility, and is a major cause of hospitalization and admittance to long-term care facilities [2]. The cost to the medical and long-term care system is also exceedingly high. In Japan, cost of treatment and aftercare for hip fractures alone has been estimated as 600 billion yen (five billion dollars) annually [3]. As such, fracture prevention among elderly is a critical public health priority in Japan.

People living in nursing homes are considered to have a high fracture risk. Several researchers in Europe and North America have reported a much higher hip fracture incidence in institutionalized elderly persons than in the general population [4-6]. This finding suggests that institutionalized elderly persons may be more susceptible to falls than those not living in nursing homes. For this reason, it is important to determine epidemiologic features of fracture risk in institutionalized elderly persons. However, to date, relatively few epidemiologic studies have focused on fracture risk in institutionalized persons [7], and very few such studies were conducted in East Asians.

Compared to Caucasians, East Asians generally have a lower risk of osteoporotic fracture [8], so a comparison of fracture incidence in institutionalized and noninstitutionalized elderly Japanese people is of interest. The present study aimed to determine fracture incidence in nursing homes in Japan.

Subjects and methods

Subjects comprised residents living in all 201 nursing homes registered with the Niigata Council of Institutions for the Elderly in Niigata Prefecture, Japan. Japan has two types of nursing homes: nursing homes for frail elderly people who require extensive care (“special nursing homes”) and nursing homes for elderly people who require some or little care (“other nursing homes”). We requested participation in the present study from the 201 homes (including 137 special nursing homes). However, as we received no response from a total of 61 homes, participation was obtained from 140 homes (including 95 special nursing homes). The 8,905 elderly people from the 140 homes as of April 1, 2007 constituted the cohort for this study. The Ethics Committee of Niigata University School of Medicine approved the present study protocol.

The present study was initiated on April 1, 2007 and terminated on March 31, 2008. Age, sex, height, weight, and fracture histories were obtained from medical records. Body mass index was calculated as subjects’ weights (kg) divided by their heights squared (m2). Nurses and caregivers from each home provided assessments of subjects’ ADL characteristics and levels of dementia at baseline. ADL scores were determined using Barthel index [9], which consists of the following ten items: feeding, transfers (bed to chair and back), grooming, toilet use, bathing, mobility (on level surfaces), stairs, dressing, bowels, and bladder. Level of assistance (ranging from complete assistance to independence) required for each item in the index was scored on a 2- to 4-point scale in which a maximum level of assistance needed was scored as 0 points and a minimum level of assistance needed was scored as 100 points (i.e., a higher score indicates lower dependence). Those with a Barthel index score of 0 were determined to be totally dependent in ADL. Levels of dementia were assessed using the dementia scale provided by the Ministry of Health, Labor, and Welfare of Japan [10], which is an observer rating scale and is widely used in the long-term care insurance system for individual care service needed.

During the 1-year follow-up period, each home reported any fractures, and X-ray-based fracture diagnosis was based on medical records from the hospitals where the patients underwent medical examination. Information on dates of discharge from homes was also obtained for the calculation of person-years for each subject.

Fracture incidence was calculated as the number of fractures divided by total person-years. Cox’s proportional-hazards model was used to determine an adjusted hazard ratio. Statistical Analysis Software (SAS; release 9.13, SAS Institute Inc., Cary, NC, USA) was used for data analysis. A p value of less than 0.05 was considered to be statistically significant.

Results

Characteristics of the subjects at baseline are shown in Table 1. Mean age of women in the study was 5.2 years younger than that of men.
Table 1

Subject characteristics at baseline

 

Women

Men

Mean or proportion

Mean or proportion

Age (years)

n = 6,841

85.5 (SD 7.5)

n = 2,064

80.3 (SD 8.6)

Height (cm)

n = 6,544

142.7 (SD 7.5)

n = 1,973

156.2 (SD 7.8)

Weight (kg)

n = 6,828

41.9 (SD 8.2)

n = 2,058

49.9 (SD 9.2)

Body mass index (kg/m2)

n = 6,541

20.6 (SD 3.8)

n = 1,972

20.5 (SD 3.5)

Special nursing home residents

 

5,256/6,841 (76.8%)

 

1,418/2,064 (68.7%)

Total Barthel index score of 0 (totally dependent subjects)

 

1,969/6,838 (28.8%)

 

404/2,064 (19.8%)

Immobile subjectsa

 

3,849/6,841 (56.3%)

 

916/2,064 (44.4%)

Level of dementiab

 Normal

 

1,001/6,828 (14.7%)

 

467/2,062 (22.7%)

 I (slight)

 

482/6,828 (7.1%)

 

188/2,062 (9.1%)

 II (light)

 

1,226/6,828 (18.0%)

 

382/2,062 (18.5%)

 III (moderate)

 

2,116/6,828 (31.0%)

 

578/2,062 (28.0%)

 IV or V (severe)

 

2,003/6,828 (29.3%)

 

447/2,062 (21.7%)

Normal vision

 

5,755/6,817 (84.4%)

 

1,818/2,063 (88.1%)

Normal hearing

 

5,385/6,817 (79.0%)

 

1,693/2,064 (82.0%)

aMobility was assessed according to the “mobility” criteria in the Barthel index as immobile: immobile or <45 m; and mobile: wheelchair independent or walks with help >45 m [9]

bEvaluated by using the dementia scale provided by the Ministry of Health, Labor, and Welfare of Japan [10]

Number of subjects with a limb fracture history separated by sex, age group, and bone site is shown in Table 2. Proportion of subjects with a fracture history tended to increase with age in women. Proportion of subjects with a fracture history was twofold to threefold higher in women than in men for each type of fracture analyzed.
Table 2

Number of subjects with a limb fracture history by sex, age group, and fracture site

 

Women

Men

Age (years)

<70

70–79

80–89

90–99

≥100

Total

<70

70–79

80–89

90–99

≥100

Total

n = 166

n = 1,203

n = 3,336

n = 2,028

n = 108

n = 6,841

n = 230

n = 710

n = 823

n = 286

n = 15

n = 2,064

Upper limbs

3 (1.8%)

69 (5.7%)

176 (5.3%)

138 (6.8%)

9 (8.3%)

395 (5.8%)

9 (3.9%)

15 (2.1%)

20 (2.4%)

5 (1.8%)

0 (0%)

49 (2.4%)

Lower limbs

 

 

 

 

 

 

 

 

 

 

 

 

Femoral neck

9 (5.4%)

121 (10.1%)

516 (15.5%)

460 (22.7%)

29 (26.9%)

1,135 (16.6%)

10 (4.4%)

33 (4.7%)

40 (4.9%)

30 (10.5%)

3 (20.0%)

116 (5.6%)

Other lower limb sites

18 (10.8%)

158 (13.1%)

643 (19.3%)

554 (27.3%)

34 (31.5%)

1,407 (20.6%)

19 (8.3%)

52 (7.3%)

65 (7.9%)

39 (13.6%)

3 (20.0%)

178 (8.6%)

Overall, the 1-year incidence of fracture was 1.6 per 1,000 person-years for forearm (radius or ulna), 4.4 for upper arm (humerus), 13.7 for proximal femur (hip), 3.4 for mid or distal femur, 1.1 for lower leg (tibia or fibula), 3.1 for vertebrae, 7.4 for other bone sites, and 34.7 for total fractures combined. Sixty-one fractures occurred at other bone sites, including two in the head or face, ten in the hand, ten in the clavicle, one in the scapula, 17 in the rib, 11 in the pelvis, two in the patella, and eight in the foot. Table 3 shows the number of fractures and the incidence rates during the 1-year follow-up period by sex, age group, and bone site. Incidence rates were higher in women than in men for all types of fractures. Hip fracture had the highest incidence rate of all fractures, followed by upper-arm fracture in both sexes. Fracture of the lower limbs other than the hip rarely occurred in men. Unadjusted female-to-male ratios of fracture incidence were 3.5 for forearm, 2.4 for upper arm, 1.5 for hip, 7.9 for mid or distal femur, infinite for lower leg, 1.5 for vertebrae, 1.2 for other sites, and 1.8 for all types of fracture combined.
Table 3

Number of fractures occurring during the 1-year follow-up by sex, age group, and fracture site (incidence per 1,000 person-years in parentheses)

Bone site

 

Women

Men

Age (years)

<70

70–79

80–89

90–99

≥100

Total

<70

70–79

80–89

90–99

≥100

Total

Person-years

162

1,154

3,102

1,823

87

6,328

221

654

728

249

9

1,861

Forearm (radius or ulna)

0 (0)

0 (0)

10 (3.2)

2 (1.1)

0 (0)

12 (1.9)

0 (0)

1 (1.5)

0 (0)

0 (0)

0 (0)

1 (0.5)

Upper arm (humerus)

0 (0)

4 (3.5)

21 (6.8)

7 (3.8)

0 (0)

32 (5.1)

1 (4.5)

0 (0)

3 (4.1)

0 (0)

0 (0)

4 (2.1)

Upper leg

 Proximal femura (hip)

3 (18.5)

12 (10.4)

55 (17.7)

24 (13.2)

0 (0)

94 (14.9)

1 (4.5)

3 (4.6)

8 (11.0)

5 (20.1)

1 (108.3)

18 (9.7)

 Mid or distal femur

1 (6.2)

2 (1.7)

14 (4.5)

8 (4.4)

2 (23.0)

27 (4.3)

0 (0)

0 (0)

1 (1.4)

0 (0)

0 (0)

1 (0.5)

Lower leg (tibia or fibula)

0 (0)

1 (0.9)

5 (1.6)

3 (1.6)

0 (0)

9 (1.4)

0 (0)

0 (0)

0 (0)

0 (0)

0 (0)

0 (0)

Vertebrae

0 (0)

4 (3.5)

13 (4.2)

4 (2.2)

0 (0)

21 (3.3)

0 (0)

1 (1.5)

3 (4.1)

0 (0)

0 (0)

4 (2.1)

Othersb

2 (12.3)

17 (14.7)

13 (4.2)

17 (9.3)

0 (0)

49 (7.7)

2 (9.1)

2 (3.1)

5 (6.9)

3 (12.0)

0 (0)

12 (6.4)

Total

6 (37.0)

40 (34.6)

131 (42.2)

65 (35.7)

2 (23.0)

244 (38.6)

4 (18.1)

7 (10.7)

20 (27.5)

8 (32.1)

1 (108.3)

40 (21.5)

aProximal femur includes femoral neck, femoral head, and trochanter

bOthers including head, face, hand, clavicle, scapula, ribs, pelvis, patella, and foot

Totally dependent persons, who also commonly live in nursing homes, rarely fall, and thus the incidence of fracture among this population is expected to differ greatly from that of other elderly persons [11]. Accordingly, we compared fracture incidence among totally dependent vs. partially dependent or independent subjects (Table 4). Adjusted hazard ratios of fracture incidence in totally dependent vs. partially dependent or independent subjects were significantly lower for fractures of the hip and other sites (head, face, hand, clavicle, scapula, ribs, pelvis, patella, or foot) and for all fractures combined. Totally dependent people therefore have approximately one seventh the incidence of hip fracture, one ninth the incidence of fracture at other sites (head, face, hand, clavicle, scapula, ribs, pelvis, patella, or foot), and one third the incidence of any type of fracture, compared with partially dependent or independent subjects. Calculation of the adjusted female-to-male ratios of fracture incidence using the proportional-hazards model revealed that these were 2.7 for forearm, 1.8 for upper arm, 1.3 for hip, 5.1 for mid or distal femur, 14.5 for lower leg, 1.7 for vertebrae, 1.0 for other sites, and 1.4 for all types of fracture combined.
Table 4

Comparison of fracture incidences in totally dependent vs. partially dependent or independent subjects

Site

Totally dependent (Barthel index = 0)

Partially dependent or independent (Barthel index >0)

Adjusted hazard ratioa (95%CI)

Number of subjects with incident fracture

Incidence rate (per 1,000 person-year)

Number of subjects with incident fracture

Incidence rate (per 1,000 person-years)

Forearm (radius or ulna)

0

0

13

2.1

Upper arm (humerus)

8

3.9

28

4.5

0.75 (0.33–1.69)

Upper leg

 Proximal femurb (Hip)

5

2.5

107

17.4

0.14 (0.05–0.33)

 Mid or distal femur

12

5.9

16

2.6

2.06 (0.94–4.51)

Lower leg (tibia or fibula)

2

1.0

7

1.1

0.60 (0.12–2.97)

Vertebrae

0

0

25

4.1

Other sitesc

2

1.0

57

9.3

0.11 (0.03–0.43)

Any fracture

29

14.3

240

39.0

0.34 (0.23–0.50)

aHazard ratio of fracture incidence in totally dependent to partially dependent or independent subjects, adjusted by sex, age, and weight

bProximal femur includes femoral neck, femoral head, and trochanter

cOther sites including head, face, hand, clavicle, scapula, ribs, pelvis, patella, and foot

We also compared fracture incidence in immobile vs. mobile subjects (Table 5). Adjusted hazard ratio of fracture incidence in immobile vs. mobile subjects was significantly lower for fractures of the forearm, hip, vertebrae, other sites (head, face, hand, clavicle, scapula, ribs, pelvis, patella, or foot) and for all fractures combined but was significantly higher for fractures of the mid or distal femur. When fractures of “mid or distal femur” and “lower leg” were combined, the adjusted hazard ratio was 5.03 (95%CI 1.9–13.3).
Table 5

Comparison of fracture incidences in immobile vs. mobile subjects

Site

Immobile subjects

Mobile subjects

Adjusted hazard ratioa (95%CI)

Number of subjects with incident fracture

Incidence rate (per 1,000 person-year)

Number of subjects with incident fracture

Incidence rate (per 1,000 person-years)

Forearm (radius or ulna)

1

0.2

12

2.8

0.06 (0.01–0.44)

Upper arm (humerus)

18

4.3

18

4.3

0.78 (0.39–1.56)

Upper leg

 Proximal femurb (hip)

33

7.8

79

18.7

0.34 (0.22–0.51)

Mid or distal femur

24

5.7

4

0.9

5.03 (1.69–15.01)

Lower leg (tibia or fibula)

8

1.9

1

0.2

5.12 (0.61–42.85)

Vertebrae

4

0.9

21

5.0

0.17 (0.06–0.51)

Other sitesc

16

3.8

41

9.7

0.34 (0.18–0.61)

Any fracture

102

24.1

167

39.5

0.49 (0.38–0.64)

Mobility was assessed according to the “mobility” criteria in the Barthel index as immobile: immobile or < 45 m; and mobile: wheelchair independent or walks with help >45 m [9]

aHazard ratio of fracture incidence in totally dependent to partially dependent or independent subjects, adjusted by sex, age, and weight

bProximal femur includes femoral neck, femoral head, and trochanter

cOther sites including head, face, hand, clavicle, scapula, ribs, pelvis, patella, and foot

Discussion

A comparison of fracture incidence in an institutionalized population to that of the general population may help to determine risk associated with institutional living. Two epidemiologic studies were recently published on fractures in the general population in Niigata. Morita et al. [12] studied the prefecture-wide hip fracture incidence in Niigata (population, 2,486,999), and Sakuma et al. [13] studied hip fracture incidence in Sado City (population, 70,011) in Niigata Prefecture. When their data for women were age-adjusted to our study population, hip fracture incidence was calculated to be 9.7 and 9.9 per 1,000 person-years, in studies by Morita et al. [12] and Sakuma et al. [13], respectively. In comparison, hip fracture incidence in women in our study (14.9 per 1,000 person-years) was about 1.5-fold higher. However, it should be noted that the “general population” in these studies included institutionalized people, and thus the fracture incidence in these “general populations” may be slightly higher than that in noninstitutionalized populations. Higher hip fracture incidences in institutionalized people may signify an increased risk for hip fractures as the level of disability increases, perhaps due to a high risk of falling and decreased bone strength due to inactivity [11, 14].

Other comparative studies on fracture incidence in institutionalized and noninstitutionalized elderly populations found a much higher incidence of hip fracture in institutionalized people than in the general population in Caucasians. Butler et al. [15] reported that hip fracture incidence in institutionalized New Zealanders was 3.2-fold higher than in the noninstitutionalized population. Sugarman et al. [6] reported that the incidence of hip fracture in U.S. nursing home residents is approximately fourfold higher than that of noninstitutionalized elderly people. The present study, however, found that the relative fracture incidence was not as high (1.5-fold, as discussed in the previous paragraph) compared to the general population. Discrepancies between our data and previously reported data [6, 15] are not easy to explain. One possible explanation may be ethnic differences, in that risk of osteoporotic fracture is generally lower in Asians compared to Caucasians [8]. Another reason may be the differences in ADL levels.

ADL is a key factor in determining fracture risk in elderly people. As mentioned above, disabled elderly persons have a high fracture risk. However, fracture risk appears to decrease once a person ceases to act independently. Walter et al. [11] showed that elderly persons who are totally dependent in locomotion have one third the hip fracture risk of those who are independent or partially dependent, and our results were consistent with this finding. We also observed a similar finding in the present study (the hazard ratio of hip fracture was 0.34: see Table 5). Furthermore, we found that totally dependent elderly people have about one seventh the hip fracture incidence and one third the incidence of all types of fracture combined. For hip fracture, in particular, total dependence is a stronger predictive factor of fracture risk than any other established factor, including age, sex, height, and low bone mass [11, 14, 16]. In this context, the relatively low ratio (1.5) of hip fractures in the institutional population in our study compared to the general elderly population may be partly explained by the fact that patients in our study included a significant number of totally dependent elderly persons. Our study population comprised 28.8% of women and 19.8% men who were totally dependent, as opposed to that in another study population, in which only 7% were totally dependent [11].

For limb fractures, Sakuma et al. [13] determined the incidence of fracture of the distal radius and proximal humerus in the general population, and the incidence in women, age-adjusted to our study population, was calculated to be 2.4 per 1,000 person-years for distal radius fracture and 2.5 per 1,000 person-years for proximal humerus fracture. Hagino et al. [17] also investigated arm fractures in the female general population in Japan, and the incidence rate age-adjusted to our population was 4.0 distal radius fractures per 1,000 person-years and 2.2 proximal humerus fractures per 1,000 person-years. In comparison with these data, forearm fracture incidence in our study (1.9 per 1,000 person-years) was lower, but upper-arm fracture incidence (5.1 per 1,000 person-years) was twofold higher than that observed in the general population. This contradictory finding has interesting implications. It is generally believed that when a frail elderly person falls, he or she is more likely to support the body using the shoulders rather than the hands, possibly due to decreased reflex acuity.

Interestingly, immobile persons had a higher incidence of leg fractures (excluding hip fracture) than mobile persons, even though immobile persons had a lower incidence of all other fractures, including hip fracture. This indicates that leg bones in immobile persons may be much more fragile than those in mobile persons. Reasons for this result are unclear but should be investigated further.

To understand how totally dependent people have fracture incidents, we reviewed reports of 29 fracture cases which occurred in totally dependent persons. We found that fractures occurred during care activity for ten cases (34.5%) and occurrence of fracture during care activity was suspected for 16 cases (55.2%). Bones of totally dependent people are thought to be weak enough that even care activity can lead to fracture.

Fracture rate in women versus men varied across different bone sites. Women had a higher rate of lower limb fractures except in the case of hip fracture, and upper limb fracture rate was approximately twice as high in women. In contrast, women and men had an almost identical hip fracture rate (1.3). This suggested that, in institutions, hip fracture prevention strategies should be targeted more at men.

Among women in our study population, hip fracture incidence among subjects in their 90s was lower than that of subjects in their 80s. This was not consistent with the findings of many studies showing that age is one of the best predictors of hip fracture and may be confounded by the proportions of totally dependent persons in those age groups. The proportion of the totally dependent (a group with very low risk for hip fracture) was higher among women in their 90s (31.7%) than among women in their 80s (28.0%; data not shown). In addition, there were many women (22.7%) in their 90s who had a history of hip fracture, and thus the other women (77.3%) in their 90s who avoided prior hip fracture may have done so and survived because they had relatively strong bones.

From nationwide data pertaining to the type of nursing home and age distribution [18], we found that the proportion of special nursing homes in Niigata (the present study population) comprised 67.9% of all nursing homes, whereas that for all of Japan was 68.6%. Age distribution (%) of nursing homes in Niigata and all of Japan was 3.4 vs. 3.7 in their 60s and younger, 18.7 vs. 19.8 in their 70s, 46.5 vs. 45.3 in their 80s, and 31.4 vs. 31.2 in their 90s and older in special nursing homes; and 7.6 vs. 7.4 in their 60s and younger, 29.9 vs. 31.2 in their 70s, 47.3 vs. 46.7 in their 80s, and 15.2 vs. 14.7 in their 90s and older in other nursing homes, respectively. From the strong similarity of the two populations, we surmise that our results can be applied to the general Japanese population in nursing homes.

One limitation of the present study was the low number of incident fractures, especially for men, which made it difficult to compare fracture incidence across genders. In addition, while we presented the incidence of symptomatic vertebral fractures, there may have been additional subjects with undetected asymptomatic vertebral fractures, and therefore our data on vertebral fracture cannot be compared to X-ray-based population studies on vertebral fracture.

This is the first epidemiologic study documenting fracture incidence among institutionalized frail elderly persons in East Asia and demonstrated that fracture incidence in institutionalized people was generally higher than in the general population but that fracture incidence was not higher for all types of fractures in institutionalized people. We found that the hip fracture risk of institutionalized elderly people was approximately 1.5-fold higher than that in the general elderly population; however, this risk ratio (1.5) is lower than that reported (3 or higher) in previous studies of Caucasian subjects [6, 15]. This may be due to ethnic differences and the proportion of totally dependent persons in the study sample, who have a much lower risk of fracture than partially dependent or independent persons.

Acknowledgements

We wish to thank the Niigata Council of Institutions for the Elderly for help in data collection. We also thank Drs. M. Sakuma and N. Endo for providing us with their data and Ms A. Kagoshima for help with data analysis. This study was partly supported by a grant from the Chiyoda Mutual Life Foundation, 2006.

Conflicts of interest

None.

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2009