Association of fear of falling and low physical activity with the quality of life of community-dwelling older adults

Abstract

Background and purpose

Both fear of falling (FOF) and low physical activity (PA) are prevalent risk factors for falls in an aged society; however, their effects on the quality of life (QoL) remain insufficiently understood. This study aimed to investigate the influence of FOF and low PA on various QoL dimensions.

Methods

Data were collected from 600 older adults aged 70 years and above through structured questionnaire interviews conducted in 1996 and 1999 in the Hunei community of Kaohsiung, Taiwan. Their QoL was assessed using the SF-36 and World Health Organization Quality of Life (WHOQOL)-BREF instruments, which were administered in the 1999 follow-up only. The relationship between the combinations of FOF (vs. no FOF) and low PA (vs. moderate-to-high PA) and each subscale QoL was analyzed using one-way analysis of variance (ANOVA) and further evaluated through multiple linear regression (MLR) models. The models sequentially included the combinations and other explanatory variables—namely, age, sex, gait maneuverability score, vision, number of comorbidities, Geriatric Depression Scale (GDS) score, Short Portable Mental Status Questionnaire (SPMSQ) score, and fall frequency. Full MLR models were constructed by replacing the combinations with their multilevel dummy variables. The MLR models for physical component summary (PCS) and mental component summary (MCS) were further stratified by the combinations.

Results

The participants had a mean age of 76.6 years, 49.2% were male, and 20.2% had both FOF and low PA. ANOVA results showed significant cross-combination differences across all QoL subscales (p < 0.001). The partial regression coefficients of the combinations indicated consistent cross-combination trends for each QoL subscale.

Conclusion

The combined presence of FOF and low PA was associated with broad negative effects on all QoL subscales among community-dwelling older adults. Interventions targeting fall risk reduction and QoL improvement should be implemented concurrently to maximize outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12877-026-07632-2.

Introduction

With the world’s rapidly aging population, the prevalence of geriatric syndromes—including falls, fear of falling (FOF), and low physical activity (PA)—has become a global public health challenge. A 1990-2021 analysis revealed that age-standardized fall-related mortality initially declined but later increased, with projections suggesting a continued rise by 2040, particularly among older adults [1]. In the United States, the proportion of older adults reporting a fall in the past year increased from 27.9% in 2012 to 29.6% in 2016, before declining slightly to 27.4% in 2018, indicating a likely upward trend in fall-related injuries as the population ages [2]. Given the global objective of promoting healthy longevity and improved quality of life (QoL) in older adults, both FOF and low PA warrant attention due to their adverse impacts on multiple health domains.

Regardless of whether they lead to injury, falls pose an emerging threat to the QoL of older adults, as they evoke feelings of helplessness, trigger FOF, increase healthcare utilization [3], and reduce social participation while heightening the need for social support [4]. A recent systematic review revealed that the global pooled prevalence of FOF was 49.60%, with higher rates observed in Africa and Asia, in developing countries, and among patients, especially for those with hip fracture, knee osteoarthritis, diabetes, etc., compared with community residents [5]. Potential risk factors for FOF include demographic variables, chronic diseases, impaired physical function, and mental health issues [5].

Low PA is not only associated with numerous chronic conditions, such as respiratory disease, obesity, cancer, cardiovascular disease, diabetes, and stroke [6], but also predisposes older adults to falls and related injuries, likely due to muscle weakness [7].

Incorporating regular PA and exercise into daily routines reduces the risk of chronic diseases and mortality, serving as a primary disease prevention strategy [6]. Moreover, recent systematic reviews, meta-analyses, and pooled analyses from 2006 to 2018 demonstrated that PA enhances QoL and overall well-being among older adults compared with minimal or no-treatment controls [8].

Although FOF and low PA may synergistically increase fall risk among community-dwelling older adults [9], their combined association with QoL remains underexplored. This study aims to assess the joint effects of FOF and low PA on QoL and to examine potential cross-combination trends across specific QoL domains. The findings may inform policy development and guide integrated strategies for simultaneous fall risk reduction and QoL improvement.

Methods and measures

Study design and participants

The data used were obtained from a two-wave cohort study conducted in Hunei, Kaohsiung, in southern Taiwan, as detailed in the preceding study [9]. In each wave, interviewees were required to recall their fall experience in the past year. In brief, this study included community-dwelling individuals aged ≥ 70 years, residing in the locality for > 6 months. Instead of being selected through random sampling, eligible participants were recruited following the household registration system. A total of 1,253 subjects received an information letter and were further contacted by the health station nurses and community volunteers. Informed consent to participate was waived. The study excluded individuals who were hospitalized, had moved, or lived in institutions. The first-wave survey that collected data on age, sex, gait maneuverability, vision, fall experience among others was completed by 1,092 older adults from September 1996 to April 1997. The second-wave questionnaire that included the SF-36 [10] and WHOQOL-BREF, Taiwan [11], as well as measures of comorbidities, depressive symptoms, cognitive impairment, FOF, physical activity, and fall experience was completed by 600 (54.9%) participants from November 1999 to May 2000, with 492 individuals lost to follow-up due to either death (n = 304) or unknown reasons (n = 188).

Measurement instruments of quality of life

The QoL of participants was measured using SF-36 [10] and WHOQOL-BREF, Taiwan [11]. The SF-36 measures eight health dimensions: physical functioning (PF), role physical (RP), body pain (BP), general health (GH), vitality (VT), social functioning (SF), role emotional (RE), and mental health (MH). The raw scores were converted to a 0–100 scale, with higher scores indicating better health status. The PF, RP, BP, and GH scores were combined to form the physical component summary (PCS) score, whereas the VT, SF, RE, and MH scores comprised the mental component summary (MCS) score [12]. The SF-36, which was translated and developed following the standard methodology of the International Quality of Life Assessment project [10], has been validated for both reliability and feasibility [10, 12–15]. Relatively minor modifications were made to enhance the cultural relevance of the SF-36. For example, Tai Chi Chuan was substituted for playing golf, and “mile” was changed to “kilometer.”

In contrast, the Taiwanese version of the WHOQOL-BREF used here was originally developed by Yao et al. [11]. from the original WHOQOL-BREF [16–18]. It comprises 28 items, including the 26 standard WHOQOL-BREF items for assessing the QoL in four domains: physical (Phy), psychological (Psy), social relations (Soc), and environmental (Env). Despite two additional items added to reflect the unique cultural characteristics in Taiwan (being respected or accepted by others and eating foods one enjoys), its reliability and validity are analogous to the global version [19], and suitable for assessing the health-related QoL for the community-dwelling older persons [20]. Each scale item used a 5-point Likert-type response measurement. Item scores are equivalent to facet scores; however, the domain scores were calculated from the average of the item/facet scores in the same domain multiplied by 4, such that the final range of the transformed domain score lies between 4 and 20. A higher score indicates a better QoL [11].

Combinations of FOF and low PA

The FOF was evaluated with a Yes/No question [21], instead of the Falls Efficacy Scale [22]. A respondent’s PA level was measured with the Physical Activity Scale for the Elderly [PASE] [23, 24]. The PASE scores were categorized into tertiles as low (0–8.6), moderate (8.7–64.9), and high (≥ 65.0), and were further dichotomized as low versus moderate-to-high (M-H) PA. The combinations of FOF and low PA [9] were coded in the following numeric order: 1 for no FOF/M-H PA (base group), 2 for no FOF/Low PA (i.e., Low PA only), 3 for FOF/M-H PA (i.e., FOF only), and 4 for FOF/Low PA.

Covariates

Covariates that were chosen from the previous study [9] for the correlation analyses with QoL included sociodemographic factors (age, sex), gait maneuverability [25], vision, number of comorbidities, psychological factors (depressive symptoms measured by GDS [26], and cognitive impairment by SPMSQ [27, 28]), and fall frequency (as a sum of the number of falls reported in the two surveys).

Two modifications were introduced: gait maneuverability was scored as a continuous variable for multiple linear regression analysis, and the SPMSQ was administered via face-to-face interviews to evaluate cognitive function. For example, gait maneuverability comprised two groups, with a total of nine items. One group included gait initiation, step height, step length, step symmetry, step continuity, and walk stance, while the other included path deviation, trunk stability, and turning while walking. Lichtenstein et al.’s methods (1990) [25] were modified to score the items in each group separately. The first-group items were scored just once without separate counting for the left and right legs as 1 if normal, and 0 if abnormal. The second-group items were scored as 2 if normal, and 0 if abnormal. The item scores were summed to provide a gait maneuverability score with a maximum of 12, with a higher score indicating a better gait function. Although cognitive function was measured with ten questions adopted from the SPMSQ [27], each question score was reversed. Each question was given 1 score if answered correctly, and 0 if erroneous. The number of correct answers was counted, with unanswered items treated as missing. Question scores of right answers were summed to provide the cognitive function score, which was divided and modulated according to educational level [28]. For people of primary school educational level, their cognition was categorized by the cognitive function score as normal if scored 7–10, mildly impaired if 5–6, moderately impaired if 2–4, and severely impaired if 0–1. The corresponding categories are respectively 6–10, 4–5, 1–3, and 0 for the illiterate, and 8–10, 6–7, 3–5 and 0–2 for those of secondary school educational level or higher.

Statistical analyses

Among the covariates, sex and vision were treated as categorical variables to examine their cross-combination differences in distribution with the chi-square test [9]. The other covariates, including age, gait maneuverability score, number of comorbidities, GDS score, SPMSQ score, and frequency of falls were treated as continuous variables to examine their cross-combination differences in mean score with one-way ANOVA respectively. The mean scores for each subscale QoL across the combinations were also examined using one-way ANOVA. The Scheffé multiple-comparison test was used to make pairwise comparisons for mean QoL score between any two combinations. The MLR models were, in addition to the combinations, adjusted for the sociodemographic variables (age and sex), gait maneuverability score, vision, number of comorbidities, GDS and SPMSQ scores, and frequency of falls. The full MLR models were established by replacing the combinations with their multilevel dummy variable to differentiate the association of each combination with the QoL, and a test for trend was performed. The MLR models for PCS and MCS were stratified by the combinations. All statistical analyses were conducted using STATA/SE 17 for Windows. The statistical significance level (α) was set at 0.05.

Results

A total of 600 participants completed both surveys, and their background characteristics were detailed in the previous study [9]. In brief, among them, the average baseline age was 76.6 years (standard deviation: 4.6), and 49.2% were male. They also reported various health issues, including unclear vision (20.5%), impaired gait maneuverability (20.0%), three or more chronic conditions (15.0%), depressive symptoms (35.2%), cognitive impairment (41.7%), FOF (40.7%), and low PA (35%).

Table 1 reveals the observed number and percentage was 267 (44.5%), 89 (14.8%), 123 (20.5%), and 121 (20.2%) respectively by the combinations: Base, low PA only, FOF only, and FOF/low PA. Participant characteristics that significantly varied across the combinations (p < 0.001) included age, sex, gait maneuverability score, vision, GDS score, SPMSQ score, and frequency of falls. Note that, across the combinations, the FOF/low PA group tended to have a higher mean age (77.9 ± 5.4 years), mean GDS score (7.5 ± 3.9), and mean frequency of falls (0.7 ± 1.4), but a lower mean gait maneuverability score (8.3 ± 4.9) and mean SPMSQ score (7.1 ± 3.2). However, there was short of significant cross-combination difference in the mean number of comorbidities (p = 0.096).

Table 1.

Distribution and mean score of participant characteristics by the combinations of FOF and low PA

	Total (N = 600)		Base group (N = 267)		Low PA only (N = 89)		FOF only (N = 123)		FOF/Low PA (N = 121)		p value
	n*	%	n	%	n	%	n	%	n	%
Age (years)#	76.2 ± 4.6		75.1 ± 4.1		77.2 ± 5.0		76.0 ± 3.9		77.9 ± 5.4		< 0.001
Sex#											< 0.001
Male	295	49.2	146	54.7	52	58.4	42	34.2	55	45.5
Female	305	50.8	121	45.3	37	41.6	81	65.9	66	54.6
Gait maneuverability score#§	10.6 ± 3.2§		11.4 ± 1.9§		10.8 ± 2.9§		10.8 ± 3.0§		8.3 ± 4.9§		< 0.001
Vision#											< 0.001
Clear	362	60.3	174	65.2	58	65.2	75	61.0	55	45.5
Average	115	19.2	48	18.0	9	10.1	30	24.4	28	23.1
Unclear	123	20.5	45	16.9	22	24.7	18	14.6	38	31.4
No. of comorbidities	1.1 ± 1.1		1.0 ± 1.0		1.1 ± 1.1		1.2 ± 1.1		1.3 ± 1.2		0.096
GDS score	4.3 ± 3.8		2.5 ± 2.3		3.7 ± 2.9		5.5 ± 4.2		7.5 ± 3.9		< 0.001
SPMSQ score	8.2 ± 2.5		8.7 ± 1.9		7.8 ± 2.8		8.6 ± 2.2		7.1 ± 3.2		< 0.001
Frequency of falls	0.4 ± 1.2		0.3 ± 1.1		0.2 ± 0.6		0.5 ± 1.3		0.7 ± 1.4		< 0.001

The cross-combination differences in the distribution or mean score of each explanatory variable were examined with the chi-square test or one-way ANOVA respectively, as appropriately. n* (the observed number) and corresponding percentage of participants was provided for categorical variables. Mean±standard deviation was provided for continuous variables. Base group: no FOF/M-H PA.§Gait maneuverability scores were, with 34 missing values, obtained in 566 persons only, to whom the corresponding number observed by the combinations were 261, 83, 120, and 102 participants respectively. The second-wave survey measured the explanatory variables other than those specified with the # sign

FOF Fear of falling, PA Physical activity, M–H Moderate to high

Table 2 shows that every mean subscale QoL score significantly varied across the combinations (one-way ANOVA; p < 0.001). The highest and lowest mean subscale QoL scores were 74.0 ± 25.4 and 50.8 ± 17.9, respectively, for BP and GH in SF-36, whereas the corresponding mean subscale scores were 13.8 ± 2.1 for the social domain and 12.9 ± 2.7 for the psychologic domain in the WHOQOL-BREF, Taiwan. These also present that the PCS and the MCS had a lower mean score than the eight subscales of SF-36. Although the older adults in the Low PA only, FOF only, and FOF/Low PA groups tended to have a lower mean subscale QoL score than the base group, no significant difference was observed between the low PA only and FOF only groups (Scheffé multiple-comparison test; p > 0.05, shown in Table S1), except for the following subscales: VT, MH, and psychological, social, and environmental domains.

Table 2.

Mean subscale scores of SF-36 and WHOQOL-BREF, Taiwan by the combinations of FOF and low PA

Total (N = 600)a, b, c	Total (N = 600)	Combinations of FOF and low PA				p value
		Base group (N = 267)	Low PA only (N = 89)a, b, c	FOF only (N = 123)	FOF/low PA (N = 121)
SF-36
PFa	61.7 ± 32.0	78.8 ± 21.4	60.4 ± 29.6 a	62.2 ± 26.5	24.2 ± 25.7	< 0.001
RP	53.8 ± 46.5	74.1 ± 39.7	48.0 ± 48.3 a	49.8 ± 43.9	17.1 ± 36.4	< 0.001
BP	74.0 ± 25.4	83.7 ± 20.7	77.4 ± 21.6	70.4 ± 23.7	53.8 ± 26.6	< 0.001
GHb	50.8 ± 17.9	59.1 ± 15.1	50.2 ± 16.5 b	45.7 ± 16.3	38.0 ± 16.5	< 0.001
VTa	65.0 ± 18.9	74.5 ± 15.2	65.3 ± 17.4 a	58.2 ± 17.6	50.8 ± 16.7	< 0.001
SF	72.5 ± 27.4	85.8 ± 17.6	73.9 ± 23.1	71.5 ± 25.1	43.1 ± 27.4	< 0.001
RE	64.9 ± 45.5	87.3 ± 31.1	58.1 ± 48.3	63.4 ± 43.6	22.3 ± 39.5	< 0.001
MHa	72.9 ± 16.3	79.2 ± 13.3	77.3 ± 11.7 a	65.6 ± 15.7	63.3 ± 18.1	< 0.001
PCSc	42.5 ± 11.7	48.0 ± 9.8	41.7 ± 10.4 c	42.0 ± 10.7	31.5 ± 9.5	< 0.001
MCSc	47.8 ± 11.4	53.6 ± 8.1	47.9 ± 11.1c	44.9 ± 11.1	37.7 ± 10.6	< 0.001
WHOQOL-BREF, Taiwan
Phy	13.7 ± 2.9	15.2 ± 2.0	13.5 ± 2.5	13.2 ± 2.6	10.9 ± 2.6	< 0.001
Psy	12.9 ± 2.7	14.3 ± 2.0	13.4 ± 2.0	11.7 ± 2.6	10.9 ± 2.6	< 0.001
Soca	13.8 ± 2.1	14.5 ± 1.8	14.1 ± 1.8a	12.9 ± 2.3	12.8 ± 2.0	< 0.001
Env	13.4 ± 2.5	14.4 ± 1.9	13.9 ± 1.8	12.0 ± 2.8	12.1 ± 2.4	< 0.001

Base group: no FOF/M-H PA. Mean ± standard deviation is presented for each subscale according to the combinations

PF Physical functioning, RP Role limitations due to physical health problems, BP Bodily pain, GH General health perceptions, VT Vitality, SF Social functioning, RE Role limitations due to emotional problems, MH Mental health, PCS Physical component summary, MCS Mental component summary, Phy Physical domain, Psy Psychological domain, Soc Social domain, Env environmental domain

N = 599 for PF, VT, MH, and Soc; N = 598 for GH; N = 597 for PCS and MCS; a = one missing value; b = two missing values; c = three missing values

p values indicate the mean subscale score difference across the combinations of FOF and low PA using one-way analysis of variance

Table 3 presents the results of the simple and multiple linear regression models for each subscale QoL. With a sequential entry of the combinations, age, sex, and other selected covariates, the partial regression coefficients of the combinations for each subscale QoL progressively shrank from − 15.84 in simple regression to − 8.00 in MLR for PF, whereas the corresponding values were − 12.56 and − 5.77 for SF, − 4.90 and − 2.23 for PCS, − 5.06 and − 2.35 for MCS, − 1.33 and − 0.51 for the physical domain, and − 1.18 and − 0.51 for the psychological domain.

Table 3.

Partial regression coefficients for multiple linear regression models for each subscale of SF-36 and WHOQOL-BREF, Taiwan

	SF-36										WHOQOL-BREF, Taiwan
	PF	RP	BP	GH	VT	SF	RE	MH	PCS	MCS	Phy	Psy	Soc	Env
Full-Full-factor combinations	−15.84***	−17.21***	−9.25***	−6.91***	−7.96***	−12.56***	−19.19***	−5.70***	−4.90***	−5.06***	−1.33***	−1.18***	−0.64***	−0.87***
Full-factor combinations	−8.00***	−7.26***	−4.20***	−3.14***	−3.60***	−5.77***	−8.99***	−2.84***	−2.23***	−2.35***	−0.51***	−0.51***	−0.25**	−0.36***
Low PA only#	−10.02***	−17.33**	−3.45	−6.01**	−5.84**	−5.70*	−21.59***	0.44	−3.82**	−3.72**	−0.96***	−0.40	0.08	0.16
FOF only#	−5.40*	−7.78	−3.99	−6.56***	−7.52***	−3.09	−8.23*	−6.71***	−1.87	−3.60***	−0.74**	−1.53***	−0.99***	−1.42***
FOF and Low PA#	−31.59***	−26.99***	−15.66***	−9.34***	−10.68***	−23.29***	−34.40***	−7.64***	−8.60***	−7.90***	−1.76***	−1.17***	−0.38	−0.57*
Age	−0.98***	−0.82*	0.01	0.10	−0.22	−0.47*	−0.20	0.02	−0.27**	0.00	−0.02	0.03	0.02	−0.03
Sex	−5.53**	−8.94*	−6.88***	−0.14	−3.32	−3.73*	−6.87*	−3.01*	−2.17*	−1.82*	−0.15	0.22	0.43**	0.16
Gait maneuverability score	0.98**	0.90	−0.16	0.20	−0.38	0.49	0.76	−0.69***	0.30*	−0.16	0.08**	0.05	0.03	0.01
Vision	−2.00	−6.45**	−0.77	−0.92	0.94	−1.42	−4.43*	1.02	−1.11*	−0.09	−0.17	−0.09	0.04	0.18
No. of comorbidities	−0.73	−2.50	−1.40	−2.82***	−1.79**	−1.70*	−1.49	−1.88**	−0.67	−0.94**	−0.19*	−0.09	−0.01	−0.03
GDS score	−2.85***	−3.90***	−2.61***	−1.82***	−2.64***	−3.01***	−4.41***	−2.18***	−0.97***	−1.58***	−0.37***	−0.39***	−0.25***	−0.33***
SPMSQ score	0.93	−0.14	−0.21	0.44**	−0.29	0.09	0.06	−0.39	0.21	−0.18	0.04***	0.02	0.16**	0.09
Fall frequency	−1.05	−0.96	−1.02	−1.83***	0.10	−0.56	−0.70	0.28	−0.67*	0.08	−0.17*	−0.03	0.06	0.12
R 2	0.55	0.33	0.33	0.39	0.48	0.51	0.39	0.42	0.42	0.49	0.54	0.50	0.30	0.38

#base: no FOF\M-H PA. The second and third rows present partial regression coefficients of the combinations in simple and multiple linear regression models, respectively. Each multiple linear regression model was adjusted for the full combinations of FOF and low PA, age, sex, gait maneuverability score, vision, number of comorbidities, GDS score, SPMSQ score, and fall frequency. Sex was coded as 1 for male and 2 for female

*p < 0.05; **p < 0.01; ***p < 0.001

When the multilevel dummy variable was introduced to substitute for the combinations, in the full MLR models, the proportions of the total QoL variation collectively explained (R²) were ≥ 0.5 for PF, SF, and the physical and psychologic domains but were lying in between 0.3 and 0.5 for the other subscales. Relative to the base group, one level change across the combinations was significantly associated with a loss of the QoL. For example, the FOF/Low PA group suffered a heavier QoL loss than the other two groups in each subscale, except for the psychological, social, and environmental domains. However, the same pattern was not applicable to the Low PA only and FOF only groups.

Concerning the other covariates, depressive symptoms and cognitive impairment showed more pervasive impacts on the QoL. Accordingly, the GDS scores were negatively correlated with each subscale QoL. In contrast, the SPMSQ scores were positively correlated with the QoL in terms of GH, and physical and social domains. As for the sociodemographic variables, age was negatively correlated with PF, RP, SF, and PCS. Compared to old men, old women tended to be negatively correlated with PF, RP, BP, SF, RE, MH, PCS, and MCS but positively correlated with the social domain.

Moreover, the gait maneuverability scores were positively correlated with PF, PCS, and the physical domain but negatively correlated with MH. Vision was negatively correlated with RP, RE, and PCS. The number of comorbidities depicted a negative correlation with GH, VT, SF, MH, MCS, and the physical domain.

Table 4 reveals the results of the MLR models adjusted for age, sex, gait maneuverability score, vision, number of comorbidities, GDS and SPMSQ scores, and fall frequency for PCS and MCS stratified by the combinations. Among the base group, the proportions of the total QoL variation collectively explained (R²) for PCS and MCS were 0.25. The corresponding figures were 0.38 and 0.26 for the Low PA only group, 0.34 and 0.54 for the FOF only group, and 0.15 and 0.36 for the FOF/Low PA group, respectively. Furthermore, compared with the base group, in terms of PCS, both the Low PA only and FOF only groups depicted more modifiable risk factors of the QoL identified, including the number of comorbidities, GDS, and fall frequency, while the FOF/Low PA group had only the number of comorbidities and GDS score identified. In terms of MCS, the risk factors identified for the base and FOF only groups were the number of comorbidities and GDS score, and the GDS score for the Low PA only and FOF/Low PA groups.

Table 4.

Multiple linear regression models for PCS and MCS stratified by the combinations of FOF and low PA

	Base group (N = 261)		Low PA only (N = 80)		FOF only (N = 120)		FOF/Low PA (N = 102)
	PCS	MCS	PCS	MCS	PCS	MCS	PCS	MCS
Age	−0.45**	0.07	−0.48*	−0.58*	−0.21	0.08	−0.02	0.05
Sex	−4.32***	−2.29*	−1.61	4.52	−0.87	−4.74**	1.69	−0.56
Gait maneuverability score	0.85**	−0.04	−0.04	−0.07	0.53	−0.16	0.48	−0.19
Vision	−1.52*	−0.76	−1.97	−0.61	−0.15	0.47	−0.05	0.60
No. of comorbidities	−0.65	−1.07*	−2.11*	−0.28	−1.80*	−1.68*	2.23*	−0.85
GDS score	−1.12***	−1.44***	−0.91**	−1.40***	−0.92***	−1.63***	−0.55*	−1.56***
SPMSQ score	−0.28	0.20	−0.05	0.16	0.98	0.31	0.39	−0.84
Fall frequency	0.17	−0.01	−4.09*	−1.17	−1.36*	−0.17	−0.80	0.54
R 2	0.25	0.25	038	0.26	0.34	0.54	0.15	0.36

Each multiple linear regression model was adjusted for age, sex, gait maneuverability score, vision, number of comorbidities, GDS and SPMSQ scores, and fall frequency. Sex was coded as 1 for male and 2 for female

*p < 0.05; **p < 0.01; ***p < 0.001

Discussion

Our study is the first to demonstrate that the combinations of the FOF and the low PA were associated with a pervasive impact on the QoL by SF-36 and WHOQOL-BREF, Taiwan. Their partial regression coefficients indicated consistent cross-combination trends for each QoL subscale. Fall risk reduction and QoL improvement must be addressed in parallel for effective fall prevention.

Three possible explanations were posed to the cross-subscale variation in terms of the cross-combination trends of the partial regression coefficients in the full MLR models. Primarily, the synergy between FOF and low PA likely amplified fall risk [9], resulting in a more significant decline in QoL due to the compounding physical, mental, and social consequences of falls [4]. However, compared with the Low PA only group with Scheffé multiple-comparison tests, the FOF only group might likely suffer from more losses of the QoL in terms of VT, MH, and the psychological, social, and environmental domains. These findings mentioned above are consistent with the clinical implications, derived from a systematic review [29], that the relationship between FOF and QoL appears to be partially mediated by physical and cognitive functioning and by higher levels of PA, therefore FOF becomes debilitating, initiating a vicious cycle of activity restriction and physical and mental function decline. [29] Second, psychological impacts of the combinations on QoL tended to lag behind the physical impacts [9], especially when the follow-up period was too short to observe the aftereffect. Third, the cross-subscale discrepancy in response to the combinations might lie in the fact that SF-36 and WHOQOL-BREF, Taiwan were developed from different conceptual frameworks. SF-36 and WHOQOL-BREF appear to measure different constructs: SF-36 measures the health-related QoL, while the WHOQOL-BREF measures the global QoL [30].

With regard to sociodemographic variables, despite age-related declines in physical performance, older adults may acquire emotional and behavioral adaptive strategies that help offset functional limitations. Instead of using primary control strategies that change a situation itself, older adults are more likely to use secondary control strategies, such as emotion regulation, which aims to change the self to adjust to a given situation [31]. Women are in themselves keener to seek psychological support and social contact. Even if women have more physical disadvantages than men, they might obtain important and beneficial support from their social ties (for example, high self-disclosure, intimacy, etc.) [32].

Moreover, the finding that a higher gait maneuverability score tended to increase the QoL in terms of PF, PCS, and the physical domain is plausible. Pain in the lower extremities may lead to deterioration in balance, gait and QoL [33]. The negative impact of vision on RP, RE, and PCS might reflect the following: a declining visual acuity in older participants restricts the role they could play [34], in contrast to the systematic impact of the comorbidities. However, the unusual finding of the fall frequency demonstrating a significant negative correlation with GH, PCS, and the physical domain, instead of the psychological domain, might be because of a lag of falls’ psychological impacts and underestimation of fall frequency.

Concerning the psychological variables, the depressive symptoms had comprehensive impacts on the QoL, which went far beyond the psychological subscales. The possible explanations are two-fold. First, a depressive mood triggers an intrinsic biochemical change to slow down the neuromuscular system and accordingly downgrade the health-related status and activity performance of older adults [35]. Second, depressive symptoms are apt to restrict or isolate an older adult from a social circle [36]. In contrast, the positive correlation of the SPMSQ scores with most subscale QoL displayed higher SPMSQ scores, a better cognitive function for conducting daily activities.

The stratified MLR models showed that more modifiable risk factors of the PCS can be identified for the Low PA only and FOF only groups, while those of the MCS were identified for the base and FOF only groups. Notably, for the FOF/Low PA group, the number of comorbidities and GDS score were the only significant risk factors for PCS, while only the GDS score was a significant risk factor for MCS. The under-presentation of the risk factors associated with the QoL might be due to the survival selection and downsizing sample size for each stratified model, especially for the FOF/Low PA group [9]. In a future study, it might take a larger sample size for a longer follow-up to identify more risk factors for the QoL subscales.

This study reveals an issue worthy of attention related to weighing the balance between fall risk reduction and QoL improvement. On one hand, in relation to the risk factors of PCS, more attention must be paid to the Low PA only and FOF only groups for the QoL improvement. On the other hand, judging from the fall risk stratification, the FOF/Low PA group must take precedence over the other groups because of being at a higher fall risk [9]. However, following the trend of patient-centered care, we cannot separate strategies coping with PCS-related risk factors from those with MCS-related risk factors. Accordingly, we consider the following typology of the four paradigms of preventive strategies designated as I, II, III, and IV: (1) Paradigm I is supposed to be effective in both fall prevention and QoL improvement, such as taking a multiple-component intervention program [37] or Multi-System Physiological Exercise Intervention [38] tailored to each individual. Moreover, combining food in high nutritional values with adaptations to the physiological and sensory needs of seniors, it could not only reduce the healthcare and societal burdens associated with aging [39], but also improve their QoL. (2) Paradigm II is supposed to be effective in fall prevention but inconclusive for QoL improvement. For example, a gradual withdrawal of psychotropic medications can significantly reduce the falling risk for older persons taking these drugs. However, participants might go back to their original medication strength during stressful times [40]. (3) Paradigm III is supposed to be effective in QoL improvement but inconclusive in fall prevention. For instance, direct-to-patient interventions enabling transitions from a long-term benzodiazepine receptor agonist (BZRA) use to cognitive behavioral therapy for insomnia by older adults can potentially reduce the BZRA use and related harms [41]. A study found that social network size did not meaningfully modify the association between FOF and fall prevention measures such as bathroom modification [42]. (4) Paradigm IV is assumed to be ineffective for both, for instance, body restraint or confined to bed [43].

Instead of a sharp line, the distinction between the paradigms is often blurry and depends on the evidence-based context and practical concerns. Future policies must integrate fall risk reduction and QoL improvement by creating comprehensive, evidence-based programs addressing multiple risk factors to maximize program perks, considering that both are parallel goals in supporting healthy aging.

Our study has three strengths. First, it included the FOF and the PASE scale, enabling a comparison across the combinations. Second, both SF-36 and WHOQOL-BREF, Taiwan were considerably modified and validated to reflect the unique local culture. Third, full MLR models made it possible to differentiate the cross-combination effects on each subscale QoL. Nevertheless, several study limitations are worth considering.

First, causal relationships cannot be established due to a lack of temporal relevance. Second, approximately half of the participants were lost to follow-up. The high attrition rate might be likely attributed to survival selection. If patients with a better QoL survived longer than those with a poorer QoL, this study might overestimate the overall QoL. Third, the data collection based on self-perception is susceptible to recall bias, especially when several assessment item batteries make interviewees committed to different recalling periods, such as 7 days for the GDS and PASE scales, 2 weeks for the WHOQOL-BREF, Taiwan, and 1 month for SF-36. Fourth, the lack of repeated QoL measurements and the selected covariates makes the detection of their time-dependent changes difficult. Fifth, no fall ascertainment was performed between the first- and second-wave surveys when some non-fallers or single fallers could have experienced one or more falls and resulted in an underestimation of fall frequency.

Sixth, the lump sum of comorbidities does not reflect the real impact of each disease because not all chronic conditions affect QoL with similar magnitudes. For example, stroke might lower down QoL more severely than simple hypertension or metabolic syndrome, whereas diabetes without any complication might not affect QoL at all. Therefore, a careful generalization of the study results to other community-dwelling older adults must be considered.

Abbreviations

QoL

Quality of life

GDS

Geriatric Depression Scale

SPMSQ

Short Portable Mental Status Questionnaire

PASE

Physical Activity Scale in the Elderly

SF-36

Short Form 36 Health Survey

PF

Physical functioning

RP

Role limitations due to physical health problems

BP

Bodily pain

GH

General health perceptions

VT

Vitality

SF

Social functioning

RE

Role limitations due to emotional problems

MH

Mental health

PCS

Physical component summary

MCS

Mental component summary

one-way ANOVA

One-way analysis of variance

MLR

multiple linear regression

Authors' contributions

YJT had full access to the Hunei cohort study data and took responsibility for the data integrity and analysis. YJT, YCY, and WJS were responsible for the study design. WJS was in charge of data collection and curation. DCY and MCL helped prepare the study proposal for grant support. YJT drafted the manuscript and was responsible for the data interpretation. DCY, YCY, WJS, and MCL helped with the study methodology and participated in the manuscript revision, review, and editing. All authors contributed to the critical revision of the manuscript for important intellectual content. All approved the article submitted for publication.

Funding

The data collection in this study was supported by the Taiwan Provincial Government (Technical Contract No. 007843) and the National Science Council, Taiwan (NSC 89-2320-B-065-001-M56). However, the in-depth analyses of the QoL were supported in part by the Research and Development Working Circle within the Cross-University Research Project on Humanities and Social Sciences by the Taiwan Comprehensive University System (TCUS), Taiwan under Project No. GM1104-126.

Data availability

YJT keeps the Hunei cohort study data.

Declarations

Ethical approval and consent to participate

The study was submitted to and approved by the Institutional Review Board of National Cheng Kung University Hospital (Protocol/IRB No.: B-ER-115-018; Document No.: 8800-4-07-001). The IRB approved a waiver of informed consent, in accordance with the Ministry of Health and Welfare and related regulations.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.