Falls Among Community-Dwelling Elderly in Japan
Japanese have a lower incidence of hip fracture than Caucasians despite having lower bone mass. Hip fractures usually occur after a fall, and differing incidence rates of falls might explain the observed differences in hip fracture rates. To explore this hypothesis, we studied falls and related conditions among 1534 (624 men, 910 women) community-dwelling people aged 65 years and over in Japan and compared the prevalence of falls to Japanese-Americans living in Hawaii and to published studies of Caucasians. In Japan, 9% of the men and 19% of the women reported one or more falls during the past year. The prevalence of falls increased with age in both genders and was greater among women compared with men. In logistic regression models, having musculoskeletal disease, physical disability or limited activity increased the risk of falls by two to four times in both genders. Most fallers (92%) reported fear of future falls, and about one third of fallers reported that they went out less often as a result of their falls. Compared with native Japanese, the age-standardized prevalence of falls among Japanese-Americans was similar but about twice as high for Caucasians, which may explain the lower hip fracture risk of Japanese.
JAPANESE (both in Japan and in Hawaii) have a lower incidence of hip and other nonspine fractures than Caucasians in North America.1,2 The lower rates might have resulted from a lower prevalence of risk factors for fractures among the Japanese. Low bone mass is one important risk factor for fractures.3 However, several studies have reported that Japanese have similar or lower bone mass than Caucasians.4–7 Thus, the lower fracture rate among Japanese is not due to greater bone mass. Hip fractures usually occur after a fall, and differing incidence rates of falls might explain the observed differences in hip fracture rates. In support of this hypothesis, the incidence of falls among Japanese in Hawaii is approximately half that of Caucasians in North America and Britain.8 However, there have been few if any population-based studies of falls in Japan. Therefore, we compared the prevalence of falls among community-dwelling elderly people in Japan to Japanese-Americans living in Hawaii and to published studies of Caucasians. In addition, we examined the circumstances and consequences of falls among people in Japan. Analyses of the circumstances and consequences of falls among Japanese-Americans have been published separately8 and are not compared here because these data were collected prospectively for Japanese-Americans and cross-sectionally for Japanese in Japan.
MATERIALS AND METHODS
The study of native Japanese was carried out in July 1994 on community-dwelling people ages 65 and over in a rural area (Mitsugi-gun, Hiroshima, Japan). The target population consisted of 1827 people who were sent questionnaires relating to falls, which were designed for self-completion. The surveys were standardized to those of the Hawaii Osteoporosis Study (HOS; below) by a meeting of the investigators prior to beginning the Mitsugi study. When the subjects could not answer the question by themselves, their families, relatives, or friends were asked to answer for them. This study area has 23 Minsei-iin, a Japanese term for community volunteers who promote social welfare. These Minsei-iin also checked the responses to improve reliability and collected the questionnaire. One thousand six hundred and twenty (88.8%) people returned questionnaires. Subjects were asked: Do you have any musculoskeletal disease, for example, osteoarthritis, rheumatoid arthritis? (MUSCULOSKELETAL DISEASE, answered: yes/no); Do you have any physical disabilities, for example, Parkinson's disease, hemiplegia due to stroke? (PHYSICAL DISABILITY, answered: yes/no); Are you bedridden, require partial bed rest, or house bound? (LIMITED ACTIVITY, answered: yes/no); and Have you experienced any falls during the previous 12 months? (Answer: number of falls). Subjects reporting one or more falls were asked about circumstances and consequences separately for each fall (time, date, reason and place of falls, presence/absence and nature of injury, going to a doctor, fear of future falls, and going out less often after falls). Eighty people who failed to answer the question about the presence/absence of falls during the previous year, and 9 people who were completely bedridden were excluded from analyses, leaving 1534 people (624 men and 910 women, 84.0% of the target population). The average age was 73.4 years for men and 74.5 for women.
The Japanese-Americans were participants at the eighth examination of the HOS conducted from January, 1992 through September, 1994. The HOS is an extension of the Honolulu Heart Program (HHP), which originally invited all noninstitutionalized men of Japanese ancestry to participate if they were born from 1900 to 1919 and living on the island of Oahu, Hawaii. A 30% random sample of the men attending the third examination of the HHP were asked to join the HOS in 1981. Wives were also invited if of pure Japanese ancestry. Published details are available concerning both the HOS and the HHP.9,10
We compared the prevalence of falls among Japanese-Americans living in Hawaii and published studies of Caucasians to the prevalence among Japanese in Japan. Subjects in both Japan and Hawaii were community-dwelling people. Since rates of falls are often higher among institutionalized than among community-dwelling elderly,11 we selected published studies of community-dwelling Caucasian elderly. In addition, to reduce the effect of recall bias,12 we limited comparisons to published data selected in the same way, that is falls experienced during the previous year.
Age-standardized risk ratios (stratified by 5-year age categories) were used to compare the women to the men within our native Japanese study and to compare the prevalence of falls among these men and women to Japanese-Americans living in Hawaii and published studies of Caucasians. These risk ratios and 95% confidence intervals (CIs) were calculated using dEPID (public-domain epidemiology software created by Kevin Sullivan, Centers for Disease Control, Atlanta, GA, U.S.A. and David Foster, Department of Epidemiology, University of Michigan, Ann Arbor, MI, U.S.A.). Associations of falls with predictor variables (musculoskeletal disease, physical disability, and limited activity) were evaluated using logistic regression models to estimate odds ratios and 95% CIs, adjusting for age. The logistic regression analysis was conducted using SAS software (SAS Institute, Cary, NC, U.S.A.).
The age and gender distribution of fallers during the past year among native Japanese and Japanese-Americans is provided in Table 1. Nine percent of the men and 19% of the women in Japan, and 11% of men and 17% of women among the Japanese-Americans reported one or more falls during the previous year. Five percent of the men and 9% of the women in Japan, and 5% of the Japanese-American men and women reported two or more falls during the previous year. The prevalence of one or more falls among native Japanese increased with age in both genders (p = 0.002 for men and p = 0.001 for women) and was greater among women compared with men. Among people in Japan, the risk among women was approximately twice as great for one or more falls, for only one fall, and for two or more falls (Table 2).
Table TABLE 1. AGE AND GENDER DISTRIBUTION OF FALLERS DURING THE PAST YEAR AMONG NATIVE JAPANESE AND JAPANESE-AMERICANS
Table TABLE 2. AGE-STANDARDIZED RISK RATIOS COMPARING THE PREVALENCE OF FALLS IN WOMEN TO MEN IN JAPAN
Table 3 provides the results of age-adjusted logistic regression analyses for men and women in Japan. Having a musculoskeletal disease, a physical disability, or limited activity were associated with increased risk of having only one fall, except for musculoskeletal disease and limited activity for men. These conditions were significantly associated with the risk of having two or more falls in both genders. Associations with two or more falls were consistently higher than those for having only one fall.
Table TABLE 3. AGE-ADJUSTED ODDS RATIOS (95% CIS) OF RISK FACTORS FOR FALLS AMONG NATIVE JAPANESE MEN AND WOMEN
Data were missing or uncertain concerning the circumstances and consequences of falls for some participants. The percentage of subjects who answered completely ranged from 64% for the time of falls to 89% for fear of future falls. There were no significant differences in age between respondents who filled out the survey completely and those who did not. Table 4 provides circumstances of falls among native Japanese. Most falls (83% for men, 85% for women) occurred during the day (between 06:01 and 18:00). Falls were most common in spring (March-May), followed by winter, were more common outdoors than indoors, and the proportion of outdoor falls was greater for men. The most common causes of falls were unstable legs and tripping. Table 5 provides consequences of falls in native Japanese. Among fallers, 62% of the men and 64% of the women reported injuries, of which bruises were the most common. Of those with injuries, 40% of the men and 57% of the women went to a doctor. Most fallers (92%) reported fear of future falls, and about 30% of fallers reported that they went out less often as a result.
Table TABLE 4. CIRCUMSTANCES OF FALLS AMONG NATIVE JAPANESE
Table TABLE 5. CONSEQUENCES OF FALLS AMONG NATIVE JAPANESE
Age-standardized risk ratios of falls in Japanese-Americans and published studies of Caucasians compared with native Japanese are provided in Table 6.13–16 The prevalence of falls in native Japanese was similar to that of Japanese-Americans but was only about half that of Caucasians for both genders.
Table TABLE 6. AGE-STANDARDIZED RISK RATIOS OF FALLS COMPARED WITH NATIVE JAPANESE
We repeated analyses for the subset (64%) of native Japanese fallers who answered the circumstances and consequences completely. The trends in prevalence of falls with age were similar to the entire sample. Furthermore, age-adjusted logistic regression analyses with musculoskeletal disease, physical disability, and limited activity yielded findings similar to those in Table 3, except that the association of physical disability with only one fall was no longer significant for men, possibly because of the reduced sample size and statistical power.
Our study revealed that the proportion of people who reported falls in the previous year increased with age, and falls were more common in women than in men. These findings are consistent with the results of other published community studies of falls among the elderly.13–17 After age-standardization, the proportions of falls in the previous year both for native Japanese and for Japanese-Americans living in Hawaii was about half that of published studies of Caucasians,13–16 suggesting that the difference in prevalence of falls may be related to genetic factors rather than environmental factors. However, the hip fracture incidence rates among both Caucasians and Japanese in Hawaii were much lower than those reported for Caucasians in North America and Northern Europe,18 suggesting that environmental factors may be responsible, or that both ethnicity (genetics) and lifestyle (environmental) factors may be involved. Two factors that might account for the reduction of risk of falling among Japanese, compared with Caucasians, are better neuromuscular function from sitting directly on the floor (a traditional Japanese custom) plus the shorter legs of Japanese people (at least partly related to genetics).19
Environmental conditions, such as the difference between rural and urban community, may affect falling. This study (Mitsugi) represents a rural farming district, where most residents younger than 70 years of age continue to grow rice and vegetables by manual labor, sometimes using machinery. Although the Hawaii study was primarily in an urban setting, many of the men worked as carpenters or laborers when they were younger, and some individuals continued to do light gardening at home and often walk for exercise. Although people in a farming community are expected to have better strength and balance than an urban population, Yasumura et al.20,21 recently reported a lack of difference in prevalence of falls between rural and urban Japanese communities. Thus, environmental differences between urban and rural Japanese communities do not appear to have caused major differences in the rates of falls.
Musculoskeletal disease and physical disability contribute to the risk of falling in the elderly14,22 and are more likely to be associated with recurrent falling.15,23 These conditions would lead to limited activity. In our results, musculoskeletal disease, physical disability, and limited activity were associated with falls, especially with two or more falls.
Lach et al.24 divided reasons for falls into extrinsic and intrinsic factors. The studies among community-dwelling elderly8,11,13,22,24 found that half of the falls resulted from extrinsic factors such as tripping or slipping. This finding was consistent with our results, although there was a high proportion of missing data.
Even if a fall does not result in physical injury, it can lead to other serious consequences. The shock of falling can generate a fear of falling again, which often leads to anxiety, loss of confidence, social withdrawal, and restriction in daily activities.25 One recent study has shown that the prevalence of fear of falling increased with age, and fear of falling was associated with decreased satisfaction with life, increased frailty, depressed mood, and a recent falling experience in community-dwelling elderly.26 Most fallers (92%) in the current study reported a fear of future falls, and about 30% of fallers went out less often as a result.
Our study has several limitations. First, there may be inaccuracies in subject recall about falls in the previous 12 months.15,27 Cummings et al.12 found that surveying falls during the past 12 months at a single point in time underestimated the true number of falls by 13%. Especially among the cognitively impaired, falls could have been forgotten or reported more than once. Second, there may be methodologic differences among studies, including participation rates and survey methods (mail vs. interview). Participation among the native Japanese was high (84% of all community-dwelling elderly), although there was a high proportion of missing data for some variables.
Japanese have lower bone mass than Caucasians in all skeletal sites,4–6 except possibly the arm.7 Low bone mass in the femoral neck is strongly associated with higher hip fracture incidence.3 However, hip fracture incidence rates among Japanese (both in Japan and in Hawaii) were about half that of Caucasians.2 Although Japanese have shorter hip axis lengths than Caucasians,28,29 which may partly explain the lower hip fracture risk among Japanese,30 hip fractures usually occur after a fall. In our results, the prevalence of falls in native Japanese was similar to that in Japanese-Americans living in Hawaii but about half that of Caucasians in other communities, which may help explain the lower hip fracture risk of Japanese. Additional studies are needed to elucidate which factors are responsible for the observed differences in falls and fractures between Japanese and Caucasians.
The authors thank Dr. Noboru Yamaguchi (Mitsugi Public General Hospital) for his support of the investigation and Dr. Kazuhiko Moji (Department of Public Health, Nagasaki University, School of Medicine) and Dr. Sumihisa Honda (Department of Radiation Epidemiology, Atomic Bomb Disease Institute, Nagasaki University, School of Medicine) for their insightful thoughts and helpful suggestions. This study was funded in part by the Hawaii Osteoporosis Foundation, and by grant AG10412 from the National Institutes of Health, National Institute on Aging.