Our study reveals that compared with older adults without FOF, those with FOF shared several risk predictors of falls in multivariate logistic regression models, such as impaired gait maneuverability, number of comorbidities, and depressive symptoms, implying that experience of FOF might be adopted, in combination with other relevant risk factors, for segmentation of the target population for fall prevention interventions.
The prevalence of falls in the second-wave survey, although similar to that of Beitou study [3], was still lower than reported in Korea [5], posing two possible explanations: First, Chinese older adults had more vigilant attitudes and behaviors in walking stick usage, higher levels of planned activities, and lower levels of incidental activities [27], even though some were reluctant to report fall experiences for a health belief leaning toward fatalism [28]. Second, it might be attributed to the healthy-participant effect. Most of the older adults in Hunei used to make a livelihood in agriculture or fishery, and these blue-collar occupations, involved in more physical activities, might protect them from falls [29]. Moreover, the prevalence of FOF in the second-wave survey, even higher than that in the first wave survey, was also lower than in those preceding studies [3–6]. Further studies should be conducted to elucidate whether it reflected not only cultural differences in reporting FOF but also variations in the definition and measurement of FOF across countries/studies.
Consistent with our previous study [2], female gender, number of comorbidities, and depressive symptoms were identified as significant independent risk factors for falls. In addition to age-related differences in the knee muscle strength in women [30], the study findings reflected the additive effects of chronic diseases on the risk for falling [31] and the substantial influence of depression on fall risk driven by biologically based mechanisms [32, 33]. As these mechanisms contribute to poor concentration and low-energy levels [33], older adults with depression are less capable to perform the dual task and pay less attention to fall-related environmental hazards and participation in their usual ADLs. Hereby, inactivity results in reducing strength, range of motion, and endurance [34] and leads to increased risk for falling [9, 16]. However, our study did not identify whether the depressive symptoms–fall relationship carries more weight than the opposite pathway [35] because it is beyond the scope of the present study to explore an intricate bidirectional and self-perpetuating interaction between falls and depression [32, 33]. Moreover, the fact that gait maneuverability was also an important risk predictor of falls is consistent with several previous study findings. Older adults at risk for falls had a greater gait variability when compared with both older adults who are non-fallers [36]. Some quantitative gait markers are independent predictors of falls in older adults [37], especially for slow gait velocity [38], stride time variability [39], and stride-to-stride variability [40].
Several other risk factors, whether established in univariate analyses or not, did not withstand the multivariate adjustment. For instance, although taking drugs from the central nervous system therapeutic group increases the risk for falls by seven times for patients [41] and by twofold for old home-dwellers [42], our study did not proclaim polypharmacy or use of sleeping pills or sedatives as an independent risk factor. Interestingly, our study reported that both having gouty arthritis and taking anti-arthritic or other specified medications were significantly associated with an increased risk of falling in bivariate analysis. Further study is needed to investigate whether it was due to arthritis-related sarcopenia in patients with rheumatoid arthritis [43] or knee osteoarthritis [44]. In contrast to a previous study [42], self-rated health and FOF, although selected into the multivariate logistic regression models, were not significantly associated with the risk for falls, increasing two possible explanations: First, it might be due to the limited power for a small sample size, and second, the over-presentation of some selected characteristics among non-participants makes it difficult to tell their relative differences in the risk for falls.
This study has two strengths. First, different questionnaires were administered during each survey to collect data with the following validated scales: balance/gait maneuverability, GDS, MMSE, and PASE, among others. Second, the overall fall experience in two surveys provided broader outcome base for identification of correlated risk factors of falls. The overall FOF experience in two surveys as a stratification variable made it possible to obtain an almost equivalent number of older adults with and without FOF for a between-group comparison of risk factors for falls. Our study findings inform twofold policies. On one hand, it implies a likely clustering of risk factors for falls among older adults with FOF. As reported, having more risk factors for falls predisposes to not only a higher risk of falls for older adults living at home or in sheltered housing [42] but also a higher risk of RF for residents of intermediate care facilities [45]. Remarkably, regardless of a fall history, older adults might still have other risk factors of falls, such as depressive symptoms, to become an anxious group with low physiologic fall risk but a high perceived risk for falls [46]. This might also elucidate the phenomenon of a high proportion of older adults without a history of falls but still having FOF in robust community-dwelling older adults [13]. Noteworthily, even though usually precipitated by body and mind incompetence, FOF in older adults can be quantified and interpreted in terms of psychological and physical characteristics, social support, and global functional capabilities [18], and probably reduced to a limited extent immediately after exercise interventions [47]. Hereby, geriatric rehabilitation aims to increase muscle strength, trunk stability, balance and coordination, together with better gait performance. While doing these, those older people with FOF may have the opportunity to rebuild their physical capabilities and self-confidence to FOF. On the other hand, it not only supports the common recommendation for risk stratification of clinical fall prevention guidelines that fall history, FOF, and gait and balance difficulties should be incorporated into “case-finding” self-reported questions [48], but also expands the conceptualization of risk factors for falls from functional outcomes to the psychosocial context of the individual [33]. Accordingly, further study might consider a large sample size whether a simple question of FOF experience, combined with questions related to fall history, and assessment of postural stability, comorbidities, and depressive symptoms, could successfully identify the potential target population of preventability from the risk for falls with multifactorial interventions [49].
Nevertheless, this study has several limitations. First, data collection based on questionnaire interviews is liable to recall bias, and a higher missing rate (5.7%) was found based on balance/gait maneuverability. Second, the study explored the association, instead of causal linkage, between selected covariates and falls. Despite the overall prevalence of falls/FOF standing for summary experiences during the two-wave surveys, almost all explanatory variables were taken from just one wave. The lack of repeated measurements of selected covariates makes it difficult to detect their time-dependent changes associated with falls/FOF and to examine a cascade effect between previous and future falls [42]. Third, in contrast to the likely underestimation of the incidence rate reported in an 8-year follow-up study [33], prevalence odds ratios overestimate the prevalence rate when the proportion of outcome is not rare (≥ 10%) [50]. Fourth, FOF was measured with a “yes/no” question and could not detect variability in FOF degrees [11]. Fifth, more falls/FOF could have been missed because ascertainment of falls/FOF was not performed between the first- and second-wave surveys. Effects of RF or persistent FOF might be underestimated. Sixth, environmental home hazards were not adopted into further analysis in the study due to a lack of discrimination validity between fallers and non-fallers. Notwithstanding, older adults with a higher level of FOF were less likely to have environmental home hazards but were more likely to have functional home hazards compared with those without FOF [4]. Finally, the overall number of medications does not reflect the real added impact of each medication because not all medications affect falls/FOF with a similar magnitude [41].