[A comparison between three methods to investigate falls among the elderly living in the community]

Interview survey is the often selected method to investigate falls among the elderly. However, the reliability of data obtained by that method has not be confirmed. The aim of this study is to clarify the number of intervals of recollection that should be specified to ensure accurate data when we investigate falls by using interview surveys among the elderly.

We carried out a total health survey among the elderly in Shigenobu town, and reported about the frequency of falls and its relating factors. Sex, age, walking ability over a distance of 1 km and hospitalization experience for past 1 year were related to falls significantly. In this study, we set 3 groups matching for the above factors in addition to history of falls during the preceding a year, and for one year we asked about falls every month for the first group (116 participants), every 3 months for the second group (116 participants) respectively. The third group (118 participants) was asked once at the end of this study. These three groups totaled 350 persons who were participants of physical fitness measurements that we carried out in Shigenobu town. The existence of falls was investigated by mail, and the nonrespondents were questioned by direct telephone calls.

For reasons such as death, moving out, hospitalization and answer denial, persons whose answer was not obtained were excluded from the analysis. Finally, 87.1% in the 1st group, and 89.7% in the 2nd group and 96.6% in the 3rd group were analyzed. Statistically significant difference for the above factors related to falls did not exist among the 3 groups. Annual incidence for falls tended to be more frequent in the 1st group that in the 3rd group (20.5% in the 1st group, 15.9% in the 2nd group and 6.4% in the 3rd group) in males. In females, such a difference was not observed. (26.3% in the 1st group, 18.3% in the 2nd group and 20.9% in the 3rd group).

There were no significant differences for the factors related to falls in 3 groups, therefore, the groups had almost similar backgrounds. In men, the difference of annual reported incidence between the 1st group and the 3rd group may be due to differences in the method of recollection.