What are the psychosocial consequences when fear of falling starts or ends? Evidence from an asymmetric fixed effects analysis based on longitudinal data from the general population

The purpose of this study was to identify whether the onset and the end of fear of falling (FOF) are associated with psychosocial consequences (in terms of depressive symptoms, loneliness, social isolation, autonomy, and subjective well‐being).

increased neuroticism, 3 frailty, 4 and increased depressive symptoms. 5 Furthermore, it can, for example, result in reduced physical activity as well as in the loss of self-esteem. 6 Therefore, knowing the factors associated with FOF is crucial. However, to date, there is a lack of longitudinal studies determining whether changes in FOF are associated with psychosocial consequences. Therefore, the first aim of this study was to investigate the link between changes in FOF and psychosocial factors in terms of depressive symptoms, loneliness, social isolation, autonomy, and subjective well-being. More specifically, and probably the particular benefit of our study was to identify whether the onset Psychosocial factors that appeared both to be theoretically and empirically important were considered as outcomes in our study. Previous research has demonstrated that there is a link between FOF and increased depressive symptoms. 7 Thus, we hypothesize that FOF is associated with increased depressive symptoms. Moreover, it has been demonstrated that FOF is associated with increased loneliness. 8,9 Consequently, we hypothesize that FOF is associated with increased loneliness and social isolation. Oh et al showed that there is an association between FOF and decreased life satisfaction. 10 Furthermore, a recent study showed that FOF is associated with lower positive and higher negative affect. 8 In sum, we hypothesize that FOF is associated with reduced subjective well-being. Furthermore, we hypothesize that FOF is associated with reduced perceived autonomy.
With regard to our second aim, we examined the asymmetric effects of FOF (consequences of the onset of FOF and the end of FOF on psychosocial outcomes) in an exploratory fashion because there is a lack of studies using this approach. In our study, data were used from wave 5 and 6 because FOF was solely measured in these waves. In total, over 10 300 participants took part in wave 5 and more than 6600 participants took part in wave 6. Klaus et al provide additional details with regard to the DEAS study. 11 All participants provided written informed consent. The DEAS study follows the principles of the Declaration of Helsinki. Because the criteria for an ethical statement were not fulfilled (eg, risk for the respondents or use of invasive methods), an ethics committee approval was not required for the DEAS study. It should also be noted that the German Centre of Gerontology (DZA), who is responsible for the DEAS study did not apply for an ethics vote, based on the recommendation of a standing council of the DEAS that decided no ethics vote to be necessary.

| Dependent variables
A tool, developed by Schwarzer, 12 was used to assess the perceived autonomy (consisting of four items). The score ranges from 1 to 4, whereby higher values correspond to higher perceived autonomy.
In our study, Cronbach's alpha was .81.
Bude and Lantermann 13 developed a scale to measure social isolation, consisting of four items (each ranging from 1 [strongly agree] to 4 [strongly disagree]). All items were recorded and subsequently the mean rating across all items was computed, with higher values reflecting higher perceived social isolation. In our study, Cronbach's Alpha was .88.
A short 6-item version of the widely used 11-item De Jong Gierveld Loneliness Scale 14 was used to assess loneliness (from 1 [strongly agree] to 4 [strongly disagree]). The psychometric properties have been shown elsewhere. 15 Higher values correspond to higher levels of loneliness. In our study, Cronbach's alpha was .83.
In this study, the 15-item version of the Center for Epidemiologic Studies Depression Scale (CES-D) 16 was used to quantify depressive symptoms (2 items have been recorded; each item ranges from 1 (rarely/none of the time) to 4 (most/all of the time); value range was recorded from 1 to 4 to 0 to 3). The scale ranges from 0 (no depressive symptoms) to 45 (severe depressive symptoms). This means that the scale reflects the sum of all 15 items. Favorable psychometric properties have been shown. 17 In our study, Cronbach's alpha was .87.
The cognitive evaluation of life (life satisfaction) was measured using the Satisfaction with Life Scale (SWLS), 18   To assess positive (eg, joy) and negative emotions (anxiety or anger), the Positive Affect and Negative Affect Schedule (PANAS) 19 was used (in each case: 10 items). The final score ranges from 1 to 5, with higher values corresponding to higher positive or negative affect, respectively (Cronbach's alpha for the positive affect subscale was .87, and Cronbach's alpha for the negative affect subscale was .86 in our study).

| Independent variables
The key independent variable was fear of falling. Individuals self-rated fear of falling ("Were you afraid that you might fall during the last 12 months?" no; yes). This is a widely used form in large cohort studies to quantify FOF (eg, in the Survey of Health, Ageing and Retire- With regard to covariates, it was controlled for age, familial status (married, living together with spouse; others (divorced; single; widowed; married, and living separated from spouse), and employment status (distinguishing between employed, retired, and other-not employed). Furthermore, it was controlled for several health-related factors, namely for physical functioning by using the subscale physical functioning of the SF-36 20 which ranges from 0 (worst) to 100 (best), for self-rated health (from 1 = very good to 5 = very bad), and for the number of physical illnesses (eg, cardiac and circulatory disorders; ranging from 0 to 11).
In the sensitivity analysis, we examined whether gender or age (younger than 65 years; 65 years and above) moderated the relation between FOF and the psychosocial outcome measures.

| Statistical analysis
In accordance with previous studies analyzing the determinants of psychosocial factors over time, [21][22][23] linear FE regressions were used. FE regressions offer the advantage that they allow for an arbitrary correlation between time-constant unobserved factors (such as genetic factors). 24 Hausman-tests with robust standard errors also substantiated our choice (eg, with depressive symptoms as outcome measure: Sargan-Hansen statistic equaled 67.3, P < .001). FE estimates exclusively use intraindividual variations in the observation period (eg, changes in FOF within individuals from wave 5 to wave 6).
To check how the onset and the end of FOF affect psychosocial factors, asymmetric FE regression analysis was used. 25 The significance level was set at P < .05. Stata 15.1 (StataCorp, College Station, Texas, USA) was used in the present study.

| Description of the analytical sample
Observations of the FE regression analysis (with depressive symptoms as outcome measure; FE regression analysis with other psychosocial outcome measures had almost the same analytical sample) are described in Table 1. In the analytical sample used, mean age was 65.5 years (±10.7 years) and 50.4% were female. Average depressive symptoms score was 6.5 (±5.7), average loneliness score was 1.7 (±0.5), average social isolation score was 1.6 (±0.6), average life satisfaction score was 3.9 (±0.5), average positive affect score was 3.6 (±0.5), average negative affect score was 2.1 (±0.5), and average autonomy score was 3.5 (±0.5). Further details are given in Table 1.

| Regression analysis
Results of FE regressions are described in Table 2 (with changes in FOF as main independent variable),  In sensitivity analysis (results not shown, but available upon request), it was tested whether gender or age (younger than 65 years; 65 years and above) moderated the relation between FOF and the psychosocial outcome measures. However, none of the interaction terms achieved statistical significance.

| DISCUSSION
Using longitudinal data from the nationwide representative DEAS study, the objective of this study was to determine whether the onset  Beta-coefficients were reported; cluster-robust standard errors in parentheses; ***P < .001, **P < .01, *P < .05, +P < .10. and the end of fear of falling (FOF) were associated with psychosocial consequences-in terms of depressive symptoms, loneliness, social isolation, autonomy, and subjective well-being.
negative affect. We assume that the end of FOF has the potential to mark a decisive turning point in life for individuals who scored high in these adverse conditions (severe depressive symptoms, high loneliness, or frequent negative emotions) when they had FOF. Thus, it appears plausible that these individuals are better off when FOF ends.
However, future research is required to elucidate the underlying reasons for the end of FOF.
In sum, the study findings showed that while the onset of FOF   11 Self-reported data were used in this study (in contrast to validated falls efficacy scales used in other studies 4,5 ). Therefore, the possibility of a recall bias cannot be completely dismissed (eg, remembering whether one experienced FOF in the past 12 months).
Furthermore, due to reasons of data availability, frailty, 4 and personality factors, such as neuroticism, 3 were not included in our study.
Future longitudinal studies are required to clarify the link between FOF and these factors.

| CONCLUSION
Findings of the study suggest that future studies should analyze the consequences of FOF differently (onset and end of FOF) which has practical, important implications. More specifically, while strategies to avoid the onset of FOF may help to maintain satisfaction with life and autonomy, strategies to end FOF may contribute to avoid increased loneliness, feelings of negative affect, as well as increased depressive symptoms.

CONFLICT OF INTEREST
The authors declare no conflict of interest.

AUTHOR CONTRIBUTIONS
A

DATA AVAILABILITY STATEMENT
The data used in this study are third-party data. The anonymized data sets of the DEAS (1996,2002,2008,2011,2014, and 2017) are available for secondary analysis. The data have been made available to scientists at universities and research institutes for scientific purposes.
The use of data is subject to written data protection agreements.
Microdata of the German Ageing Survey (DEAS) is available free of charge to scientific researchers for non-profitable purposes. The FDZ-DZA provides access and support to scholars interested in using DEAS for their research. However, for reasons of data protection, signing a data distribution contract is required before data can be obtained.