The Effect of Exercise Intervention on Reducing the Fall Risk in Older Adults: A Meta-Analysis of Randomized Controlled Trials

Exercise intervention has a positive effect on reducing the fall risk in older adults. To investigate the effect of different factors of exercise intervention (type, duration, and frequency) on reducing the fall risk in older adults, a meta-analysis was performed in this study. According to the PRISMA®, two researchers independently searched PubMed, Web of Science, and the China National Knowledge Infrastructure databases to assess the quality of the studies using the PEDro scale. A total of 648 subjects in 10 randomized controlled trials were included in this study, and the exercise interventions included integrated training (resistance training, core training, and balance training), balance training, core training, Pilates, Ba Duan Jin, and Tai Chi. These studies show that exercise intervention has a huge and significant impact on reducing the risk of falls of the elderly. In conclusion, an integrated intervention with a frequency of more than five times a week and a duration of more than 32 weeks are more effective in reducing the fall risk.


Introduction
Although falls can occur at all ages, older adults are more prone to injury due to physiological changes and delayed functional recovery caused by aging [1]. Currently, falls are the leading cause of injury and death in people over 65 years old [2]. It has been shown that among older adults aged 65-74 years, 25% experience a fall each year, rising to 29% among those aged 75-84 years and up to 39% in the older adults over 85 years old [3]. Falls in older adults can cause irreversible physical injuries [4], which may even lead to disability or death [5]. The disadvantages of falls affect their quality of life [6] and impose a heavy economic burden on the health care system [4]. In addition, older adults who experience falls also develop a significant fear of falling. Their social isolation and depression can increase the risk of falling, thus creating a vicious cycle [7]. Therefore, the prevention of falls in the elderly is of great significance, and it is urgent to study how to reduce the risk of falls in the elderly. However, the current design of the city is not conducive to the prevention of falls for the elderly to a certain extent, and has a negative impact on the well-being of the elderly, including the lack of access to daily services, and the buildings cannot provide enough space for the elderly and people with mobility impairments [8]. One way to solve the issues above and support the elderly to live a satisfied life is to build elderly-friendly cities and communities [9]. Studies have shown that communities with better street connectivity and daily life services tend to improve physical activities of the elderly [10], and communities roads with low-lying edges and barrier-free sidewalks performance. However, the study did not indicate which sports the elderly participate in can reduce the risk of falls in the elderly [26].
Although researchers have done a large number of studies investigating the effects of exercise interventions on reducing the fall risk in older adults, few studies have examined the specific effects of different types of exercise interventions. Some studies have pointed out that different forms of exercise have different effects on reducing the risk of falls, but this study only reports the effects of balance, functional exercises and Tai Chi exercises in reducing the risk of falls. It does not reduce the risk of falls for exercises such as resistance training, dancing or walking [16]. Therefore, the aim of this meta-analysis is to investigate the effect of different exercise interventions on reducing the fall risk in older adults.

Materials and Methods
The systematic evaluation and meta-analysis followed the Preferred Reporting Items for Systematic Evaluation and Meta-Analysis (PRISMA) guidelines and the Cochrane Collaboration protocol. The scheme for this meta-analysis was registered with PROSPERO under the registration number CRD42021279113.

Search Strategy
Two researchers (M.S. and L.M.) independently searched the China National Knowledge Infrastructure (CNKI) database using the Chinese keywords sports, sports activities, physical exercises, fall risk, older adults, middle-aged and older adults. The time frame for the literature search was 1 January 1990 to 31 July 2021. In addition, the researchers searched PubMed and Web of Sciences databases using the keywords "sport", "physical activity", "physical fitness", "fall risk", "the aged", "elder population", "the old", "middle-aged and senior people". The time frame was also 1 January 1990 to 31 July 2021.

Inclusion Criteria and Exclusion Criteria
The specific criteria for inclusion and exclusion are shown in Table 1, and the screening process is shown in Figure 1. Table 1. Inclusion criteria and exclusion criteria.

Criteria
Type Definition

Inclusion Criteria
Study type Randomized controlled trial (RCT) studies in Chinese and English.

Study subject
Middle-aged and older adults (over 50 years old) without disabilities and other diseases that make them unsuitable for exercise interventions.

Intervention requirement
Physical activity at baseline levels is the same in both groups, and the duration of the intervention is greater than or equal to 4 weeks.

Outcome indicators
Fall risk index, fall risk score.

Exclusion Criteria
Indicators without clear comparability.
Baseline level data not available.
Missing full text or incomplete data on outcome indicators.

Quality Assessment
Two researchers (M.S. and L.M.) independently used the PEDro scale [27] to assess the quality of the included studies. The scale contains 11 items, each item is scored 1 point for meeting the criteria, and the maximum score is 10 (the first item is not scored). Studies were classified as high quality (≥7 points), medium quality (5-6 points), and low quality (≤4 points) based on the scores. When the two researchers disagree, they joined a third party (N.X.), and finally discussed and agreed.

Data Analysis
The data on outcome indicators extracted from the studies were processed using Revmansoftware (version5; https://training.cochrane.org/online-learning/core-softwarecochrane-reviews/revman, accessed on 9 January 2021). The outcome variables in this study were continuous. Due to the non-uniformity of unit magnitudes, the standardized mean difference (SMD) was chosen for processing. The specific data were processed according to the following equations: where Mchange is the mean of change; Mbaseline is the mean before the experiment; Mfinal is the mean after the experiment; SDchange is the standard deviation of change; SDbaseline is the standard deviation before the experiment; SDfinal is the standard deviation after the experiment; r is the correlation coefficient before and after the experiment, which was set to a constant of 0.5 for simplicity of calculation.

Quality Assessment
Two researchers (M.S. and L.M.) independently used the PEDro scale [27] to assess the quality of the included studies. The scale contains 11 items, each item is scored 1 point for meeting the criteria, and the maximum score is 10 (the first item is not scored). Studies were classified as high quality (≥7 points), medium quality (5-6 points), and low quality (≤4 points) based on the scores. When the two researchers disagree, they joined a third party (N.X.), and finally discussed and agreed.

Data Analysis
The data on outcome indicators extracted from the studies were processed using Revmansoftware (version5; https://training.cochrane.org/online-learning/core-softwarecochrane-reviews/revman, accessed on 9 January 2021). The outcome variables in this study were continuous. Due to the non-uniformity of unit magnitudes, the standardized mean difference (SMD) was chosen for processing. The specific data were processed according to the following equations: where M change is the mean of change; M baseline is the mean before the experiment; M final is the mean after the experiment; SD change is the standard deviation of change; SD baseline is the standard deviation before the experiment; SD final is the standard deviation after the experiment; r is the correlation coefficient before and after the experiment, which was set to a constant of 0.5 for simplicity of calculation.
The statistic I 2 was chosen to indicate the heterogeneity among the included studies. I 2 reflects the degree of overlap of the credible interval. It does not depend on the true effect size and distribution, but reflect the size of the variance on a relative scale [28]. When I 2 = 0, there was no heterogeneity among studies; when 0 < I 2 ≤ 20%, there was low heterogeneity among studies; when 20% < I 2 ≤ 50%, there was medium heterogeneity among studies; when I 2 > 50%, there was high heterogeneity among studies. When I 2 > 50%, a randomeffects model was chosen, and subgroup analysis and sensitivity analysis were performed to explore the sources of heterogeneity. Funnel plots were constructed for the risk of bias assessment.

Characteristics of the Included Studies
The search results were screened and reviewed according to the inclusion and exclusion criteria in this study. A total of 10 studies were finally selected, including four in Chinese and six in English. Participants were 60-95 years old, and different exercise interventions were involved ( Table 2).  [32] 2016 Portugal The aerobic exercise group trained twice per week in a land environment (Mondays and Wednesdays) and once per week in an aquatic environment (Fridays). All aerobic training sessions consisted of: (i) a 10-min warm-up period, which included walking and flexibility exercises; (ii)a 30-min cardiorespiratory period, including walking and/or brisk walking; (iii) a 10-min muscular endurance, which included three exercises (three sets of 15-20 repetitions) using only bodyweight and gravity for strengthening the lower and upper limbs in a land environment, and water resistance in an aquatic environment; and (iv) a 5-min cool-down period, which included breathing and stretching exercises.  [32] 2016 Portugal the aerobic training session on Mondays were replaced by a resistance exercise session. The intensity of the resistance training sessions was defined by the pyramidal method set to 65% of 1-RM in the first eight weeks; 75% of 1-RM for Weeks 8-24; 70% of 1-RM for Weeks 24-28; and 65% of 1-RM for Weeks 28-32 (three sets of 10-12 repetitions in all sessions). Each session always began with a 10-min warm-up and ended with a cool-down period. The main part of the sessions consisted of a circuit of seven exercises: bench press, leg press, lateral pull-down, leg extension, military press, leg curl and arm curl, in this order, and carried out with conventional variable resistance devices (PANATTA, Fitline 2000 series, Italy

Characteristics of the Interventions in Included Studies and Outcome Indicators
The duration of the interventions in the included studies ranged from 8 weeks to 1 year. The frequencies of the intervention were mostly three times a week. Outcome indicators included PPA test score, Sensory Organization Test (SOT) test score, 30s chair stand (CS-30) test score, Tetra fall index, Morse Fall Scale (MFS), and fall risk score (Table 3). Comparison of disease efficacy, disease awareness, fall prevention awareness, psychological anxiety level, body immunity status, bone density before and after care, and incidence of falls between the two groups of older adults with osteoporosis. SOT score of the NeuroCom ® system, Timed Up and Go (TUG), and Activity Specific Balance Confidence (ABC) Scale.
SOT score 10 weeks Once a week Pilates modified exercises once a week for 10 weeks can reduce the fall risk in adults aged 65 and older who are at risk of falls. This exercise significantly improved static and dynamic balance, functional mobility, balance self-efficacy, and lower-extremity AROM. In contrast, the control group showed no significant improvement in any of the indicators. Pilates modified exercises were more effective than no exercise intervention in improving AROM in the hip and ankle joints.

CS-30 test 12 weeks 3 times a week
This study provided an effective, evidence-based fall prevention program that can be implemented to improve the physical health of older adults in the community and reduce the fall risk. In the assessment of fall risk in older adults, the SEBT might be a sensitive measure of physical performance.

The Effect of Exercise Intervention on Reducing the Fall Risk in Older Adults
A combined effect size test was performed on 10 included studies using Revman 5.3 software. The results showed high heterogeneity among studies (I 2 = 92%, p < 0.00001). Therefore, a random-effects model was selected for effect size combining. The results showed that SMD = 1.40, 95% CI interval is (0.75, 2.05), and total effect test Z = 4.22, p < 0.0001. Therefore, it can be concluded that exercise intervention can reduce the fall risk in older adults. However, due to the high overall heterogeneity, subgroup analysis was needed to explore the source of heterogeneity (Table 4 and Figure 2).

Publication Bias Assessment
Funnel plots were constructed with the outcome indicators in the included studies, as shown in Figure 3. If the points on the funnel chart are symmetrically scattered around the estimated true value of each independent research effect point, there is an inverted symmetrical funnel shape showing that the included study is unbiased and vice versa. For a symmetrical funnel chart, the greater the degree of bias. The data are largely well symmetrical, indicating that there was some publication bias in the study, but the degree of bias was minimal.

The Effect of Different Types of Exercise Interventions on Reducing the Fall Risk in Older Adults
Exercise interventions can effectively reduce the fall risk in older adults [38], but the value of different types of exercise activities differs for older adults [39]. Therefore, it was assumed in this study that differences in the type of exercise interventions could have different effects on reducing the fall risk in older adults. According to the different types of exercise interventions, we divided them into three subgroups: integrated training (resistance training, core training, and balance training), physical training and fitness training (Pilates, Ba Duan Jin, and Tai Chi). The subgroup analysis results showed that the within-group heterogeneity was high for integrated training (I 2 = 95%) and physical training (I 2 = 85%), and it was low for fitness training (I 2 = 31%), indicating that all kinds of exercise interventions were effective in reducing the fall risk in older adults. In terms of the SMD, integrated training (SMD = 3.16) > physical training (SMD = 0.88) > fitness train-

Publication Bias Assessment
Funnel plots were constructed with the outcome indicators in the included studies, as shown in Figure 3. If the points on the funnel chart are symmetrically scattered around the estimated true value of each independent research effect point, there is an inverted symmetrical funnel shape showing that the included study is unbiased and vice versa. For a symmetrical funnel chart, the greater the degree of bias. The data are largely well symmetrical, indicating that there was some publication bias in the study, but the degree of bias was minimal.

Publication Bias Assessment
Funnel plots were constructed with the outcome indicators in the included studies, as shown in Figure 3. If the points on the funnel chart are symmetrically scattered around the estimated true value of each independent research effect point, there is an inverted symmetrical funnel shape showing that the included study is unbiased and vice versa. For a symmetrical funnel chart, the greater the degree of bias. The data are largely well symmetrical, indicating that there was some publication bias in the study, but the degree of bias was minimal.

The Effect of Different Types of Exercise Interventions on Reducing the Fall Risk in Older Adults
Exercise interventions can effectively reduce the fall risk in older adults [38], but the value of different types of exercise activities differs for older adults [39]. Therefore, it was assumed in this study that differences in the type of exercise interventions could have different effects on reducing the fall risk in older adults. According to the different types of exercise interventions, we divided them into three subgroups: integrated training (resistance training, core training, and balance training), physical training and fitness training (Pilates, Ba Duan Jin, and Tai Chi). The subgroup analysis results showed that the within-group heterogeneity was high for integrated training (I 2 = 95%) and physical training (I 2 = 85%), and it was low for fitness training (I 2 = 31%), indicating that all kinds of

The Effect of Different Types of Exercise Interventions on Reducing the Fall Risk in Older Adults
Exercise interventions can effectively reduce the fall risk in older adults [38], but the value of different types of exercise activities differs for older adults [39]. Therefore, it was assumed in this study that differences in the type of exercise interventions could have different effects on reducing the fall risk in older adults. According to the different types of exercise interventions, we divided them into three subgroups: integrated training (resistance training, core training, and balance training), physical training and fitness training (Pilates, Ba Duan Jin, and Tai Chi). The subgroup analysis results showed that the within-group heterogeneity was high for integrated training (I 2 = 95%) and physical training (I 2 = 85%), and it was low for fitness training (I 2 = 31%), indicating that all kinds of exercise interventions were effective in reducing the fall risk in older adults. In terms of the SMD, integrated training (SMD = 3.16) > physical training (SMD = 0.88) > fitness training (SMD = 0.57), as shown in Table 5 and Figure 4.

Effect of Different Intervention Durations on Reducing the Fall Risk in Older Adults
From a long-term perspective, the effectiveness of exercise interventions to reduce the fall risk in older adults is related to the duration of the intervention [21]. The included studies were analyzed in subgroups according to different durations of interventions. The results showed that there was a high within-group heterogeneity for intervention durations of less than 12 weeks (I 2 = 77%), 12-32 weeks (I 2 = 88%), and more than 32 weeks (I 2 = 96%), indicating that different intervention durations have a significant effect on reducing the fall risk in older adults. In terms of the SMD, intervention durations more than 32 weeks (SMD = 2.92) > 12-32 weeks (SMD = 0.98) > less than 12 weeks (SMD = 0.68), as shown in Table 6 and Figure 5.

Effect of Different Intervention Durations on Reducing the Fall Risk in Older Adults
From a long-term perspective, the effectiveness of exercise interventions to reduce the fall risk in older adults is related to the duration of the intervention [21]. The included studies were analyzed in subgroups according to different durations of interventions. The results showed that there was a high within-group heterogeneity for intervention durations of less than 12 weeks (I 2 = 77%), 12-32 weeks (I 2 = 88%), and more than 32 weeks (I 2 = 96%), indicating that different intervention durations have a significant effect on reducing the fall risk in older adults. In terms of the SMD, intervention durations more than 32 weeks (SMD = 2.92) > 12-32 weeks (SMD = 0.98) > less than 12 weeks (SMD = 0.68), as shown in Table 6 and Figure 5.

Effect of Different Intervention Frequencies on Reducing the Fall Risk in Older Adults
Similarly, the effects of exercise interventions on reducing the fall risk in older adults are related to the frequency of the intervention [21]. By subgroup analysis of the included studies according to different exercise intervention frequencies, it was shown that a frequency of 3-5 times a week (I 2 = 85%) and a frequency of more than 5 times a week (I 2 = 96%) had high within-group heterogeneities. This result indicated that different intervention frequencies have significant effects on reducing the fall risk in older adults. In terms of the SMD, more than five times a week (SMD = 2.39) > 3-5 times a week (SMD = 1.17), as shown in Table 7 and Figure 6.

Effect of Different Intervention Frequencies on Reducing the Fall Risk in Older Adults
Similarly, the effects of exercise interventions on reducing the fall risk in older adults are related to the frequency of the intervention [21]. By subgroup analysis of the included studies according to different exercise intervention frequencies, it was shown that a frequency of 3-5 times a week (I 2 = 85%) and a frequency of more than 5 times a week (I 2 = 96%) had high within-group heterogeneities. This result indicated that different intervention frequencies have significant effects on reducing the fall risk in older adults. In terms of the SMD, more than five times a week (SMD = 2.39) > 3-5 times a week (SMD = 1.17), as shown in Table 7 and Figure 6.

Discussion
Normal human aging is associated with the decline of many physiological functions, including the decline of skeletal muscles, cardiovascular, vision, vestibular system, and proprioception. These declines result in decreased body coordination, slower postural responses, and cognitive decline, which are associated with increased fall risk [40]. Therefore, a comprehensive fall prevention program is essential for older adults [41]. Currently, assessing the fall risk of older adults is a valuable tool for caregivers to take targeted and effective measures [42]. However, in different situations, researchers tend to choose different fall risk assessment tools [2]. For example, the TUG test has been widely adopted by researchers due to its lower cost, simplicity of the procedure, and excellent sensitivity [43]. Besides, gait balance evaluation based on MoS [44] and Tinetti test [45] are more suitable for individual balance assessment. The fall risk assessment in some studies has also included both physical and perceived dimensions [46]. In the included studies, a wide variety of test metrics were involved, and the scales were not uniform. It can be speculated that the different fall risk assessments are a source of heterogeneity between studies, which needs to be verified in future studies.
This study shows that integrated training (resistance training, core training, and balance training) is more effective in reducing the fall risk in older adults, consistent with previous studies [13]. Previous studies have also shown that adherence rates of participants may be higher in exercises with shorter durations [47], while the present study shows that exercise interventions are more effective in reducing the fall risk in older adults when the duration of training is 32 weeks or more. Considering whether exercise interventions produce a conscious continuation of exercise behavior, future studies should focus on the relationship between the effectiveness and the cost-effectiveness of exercise interventions on reducing the fall risk in older adults.
Previous studies have demonstrated that the effect of exercise interventions in reducing the fall risk in older adults is closely related to the exercise load [16]. Few studies have

Discussion
Normal human aging is associated with the decline of many physiological functions, including the decline of skeletal muscles, cardiovascular, vision, vestibular system, and proprioception. These declines result in decreased body coordination, slower postural responses, and cognitive decline, which are associated with increased fall risk [40]. Therefore, a comprehensive fall prevention program is essential for older adults [41]. Currently, assessing the fall risk of older adults is a valuable tool for caregivers to take targeted and effective measures [42]. However, in different situations, researchers tend to choose different fall risk assessment tools [2]. For example, the TUG test has been widely adopted by researchers due to its lower cost, simplicity of the procedure, and excellent sensitivity [43]. Besides, gait balance evaluation based on MoS [44] and Tinetti test [45] are more suitable for individual balance assessment. The fall risk assessment in some studies has also included both physical and perceived dimensions [46]. In the included studies, a wide variety of test metrics were involved, and the scales were not uniform. It can be speculated that the different fall risk assessments are a source of heterogeneity between studies, which needs to be verified in future studies.
This study shows that integrated training (resistance training, core training, and balance training) is more effective in reducing the fall risk in older adults, consistent with previous studies [13]. Previous studies have also shown that adherence rates of participants may be higher in exercises with shorter durations [47], while the present study shows that exercise interventions are more effective in reducing the fall risk in older adults when the duration of training is 32 weeks or more. Considering whether exercise interventions produce a conscious continuation of exercise behavior, future studies should focus on the relationship between the effectiveness and the cost-effectiveness of exercise interventions on reducing the fall risk in older adults.
Previous studies have demonstrated that the effect of exercise interventions in reducing the fall risk in older adults is closely related to the exercise load [16]. Few studies have shown that exercises with high loads produce better results [48]. In this study, most of the selected studies addressed exercise duration and frequency, and only a few included studies were related to the intervention load. Therefore, the intervention load should be considered in future studies.
The research is more for relatively healthy elderly people. Our results are not applicable to elderly people who are dementia, stroke rehabilitation, disabled or other people who cannot exercise independently. Future studies can be done to explore more rehabilitation options for elderly people who cannot exercise independently. In addtion, the study lacks more detailed subgroup analysis such as the duration of each training and the chronic disease of the trainer due to the limited number of included studies.
In conclusion, reducing the fall risk in older adults should be aimed at reducing and eliminating fall risk factors [49]. Since the causes of falls in older adults are usually the result of multiple factors, exercise as a single intervention has reduced the incidence of falls by more than 36% [25], some studies have pointed out that multifactorial interventions are more effective than single-factor interventions [50]. Interventions such as dual-task training [51], multiple standardized prevention programs (including exercise components and individualized design) [13], and multifactorial fall prevention programs with exercise interventions [52] are beneficial for older adults. Factors can include sports, education, environmental transformation, drug treatment, walking aids, vision and psychological management, etc. [24]. Therefore, multiple factors should also be considered to explore the effect of exercise interventions on reducing the fall risk in older adults.

Conclusions
The exercise intervention is effective in reducing the fall risk in older adults. Integrated training (resistance training, core training, and balance training) with a duration of over 32 weeks and a frequency of more than five times a week is more effective in reducing the fall risk in older adults. It is a long-term and continuous method to reduce the risk of falls of the elderly through exercise. Researchers need to consider economic conditions and the willingness of the elderly when choosing exercise methods for intervention.