Effects of walking exercise on cognitive and physical functions: ­meta-analysis of older adults

Abstract

Aging-related diseases are a significant social concern due to rapid aging as life expectancy continues to increase and birth rates decline. This study conducted a meta-analysis to suggest guidelines maximizing effects of forest walking exercises on older adults’ cognitive and physical functions. We collected articles from national and international electronic search engines (RISS, NDSL, KISS, Pubmed, Embase, CINAHL, Riss, and Web of Science) published until February 28, 2021, wherein older adults participated in walking exercises. The overall effect sizes of walking exercise on the Mini-Mental State Examination (MMSE) and performance of activities of daily living (ADL), which are measures of cognitive function, were 0.772 and 0.527, respectively. The overall effect sizes of agility, cardiorespiratory endurance, and upper and lower body strengths, which are measures of physical function, were –0.670, 0.698, 0.685, and 0.433, respectively. The greatest moderation effect was observed for an experimental group size of 20–29 subjects, with complex exercises, exercising for 10 to 19 weeks, exercising three times per week, and for 60 minutes. The results of this study can help minimize errors when designing or assessing research on forest walking exercise programs or selecting appropriate assessment tools or measures.

Keywords:

• elderly people
• meta-analysis
• cognitive functions
• body functions
• walking

1. Introduction

Medical advances have increased human life expectancy; however, the birth rate continues to decrease, accelerating global population aging (KOSTAT 2021; Jeon et al. 2012). The Organization for Economic Cooperation and Development (OECD) projected South Korea to have the highest percentage of the older adult population in the world by 2045 and the highest old-age dependency ratio among the OECD countries by 2060 (OECD 2020; Jeon and Khang 2012). Health spending for older adults aged above 65 years reached 37.6 trillion won in 2021, accounting for over 43% of the total health spending (HIRA and NHIS 2021). This suggests that the social costs generated by the older adult population require particular attention. The incidence of long-term treatment and chronic diseases requiring high costs increases with age. Therefore, along with providing treatment, preventive measures must be taken to reduce older adult patients’ financial burdens. In an aging society, functional independence is crucial in the health management of older adults compared to increasing life expectancy. Regular exercise is vital to prevent physical impairment and functional loss resulting from natural aging (Daley and Spinks 2000; Schoenfelder and Rubenstein 2004). In a study on older adult patients with chronic stroke, participating in weight shift, balance, and walking exercises for at least one year after the disease onset improved the performance of activities of daily living (ADLs), while regular walking in the older adults prevented chronic diseases and physical impairment (ACSM 2006). Another study reported that although individuals’ physical function and health deteriorate with age, regular exercise can help improve muscle strength, muscular endurance, flexibility, and balance reactions in older adults (Mills 1994; Kang 2007). Son and Lee (2006) demonstrated that walking exercise significantly improves cardiorespiratory function in older adults. NIFoS (2018) reported a study that 2 hours of forest walking improves concentration. Reports consistently show that walking positively affects various indices, including mood state, heart rate, blood pressure, cognitive function, body fat mass, and muscle mass (Shimada et al. 2014; Zheng et al. 2021; Bishnoi and Hernandez 2021). Walking is a free and easily accessible exercise that requires minimal preparation and walking in the forest can produce greater effectiveness through combination with environmental factors. It causes a lower impact on the musculoskeletal system and joints than running; thus, it is recommended for obese individuals or older adults. Walking improves cognitive function by affecting brain activation. In addition, regular physical activity helps maintain and improve brain function in older adults (Bherer et al. 2013; Hilman et al. 2008). Hence, it is necessary to synthesize the results of previous papers and examine the effect size through meta-analysis to verify effect of walking exercise affects the improvement of physical function in older adults (Roh and Park 2013). A meta-analysis determines the relationship between variables and their level of impact based on the results of studies using various methodologies, thereby deriving an evidence-based conclusion regarding an intervention’s effectiveness (Park and Shin 2011). Borenstein et al. (2009) explained that due to their post-hoc nature, meta-analyses can help determine and provide future research directions rather than present a conclusion. To date, meta-analyses have conducted to measure the effect sizes of various kinds of exercise on older adults’ physical function (Gu and Conn 2008; Jung 2006). However, there is a lack of meta-analyses specifically investigating the effect of walking on older adults’ cognitive and physical functions. While studies have measured muscle strength and flexibility following an exercise intervention in older adults, further research examining upper and lower body strengths is needed since the aging-related loss of lower body strength is a significant risk factor for falls (Schoenfelder and Rubenstein 2004). Therefore, it is necessary to examine the effect size of walking on cognitive function and the function such as agility, cardiorespiratory endurance, and upper and lower body strengths in older adults based on multiple studies.

Hence, this study aimed to systematically analyze studies that examine the effect of walking exercise on older adults’ cognitive and physical functions and determine overall effect of walking exercises. This study’s results can help develop guidelines on walking exercises to improve older adults’ cognitive and physical functions.

2. Materials and methods

2.1. Research design

This study was a meta-analysis that examined the effect of national and international walking exercise intervention programs, including forest walking, on older adults’ cognitive and physical functions. The Preferred Reporting Items for Systematic and Meta-Analysis (PRISMA) guideline was followed when conducting this meta-analysis (Page et al. 2021).

2.2. Literature selection criteria

Before conducting a literature search, studies satisfying the inclusion and exclusion criteria were selected from national and international data search engines following the PICOS (Participants, Intervention, Comparisons, Outcomes, Study Design) scheme (Choi et al. 2020). The participants were older adults. The older adults were selected to be 60 or more years of age. The interventions were programs involving forest walking or indoor walking. The comparator was a group that did not participate in a walking exercise program. The outcome was the effect of walking exercises on cognitive and physical functions, and only the studies that provided numerical data were included. Finally, the study designs were randomized controlled trials (RCTs) and non-RCTs (NRCTs). Domestic and foreign studies published in Korean or English were included to investigate the effect of walking exercise on disease prevention in older adults considering the global population aging. Case studies, meta-analyses, studies without original text, and single-group studies were excluded. Once the inclusion criteria were created, studies were searched and selected following the PRISMA Flow Diagram (Page et al. 2021).

2.3. Data search and selection process

Studies wherein a walking program was conducted among older adult subjects were searched without restricting the publication year. National and international studies published until February 28, 2021, the data of search date, were searched using Korean search engines, including RISS, NDSL, and KISS, and global search engines, including Pubmed, Embase, CINAHL, Riss, and Web of Science. The search was performed using the “AND” and “OR” operators. The search words used in the national and international search engines were as follows: (“walking” OR “forest walking” OR “walking in urban parks” OR “pedometer” OR “accelerometor” OR “accelerometer” OR “green space walking” OR “walking exercise in forest” OR “forest physical activity” OR “physical activity in forest” OR “forest exercise” OR “forest activity” OR “shinrin-yoku” OR “forest basking” OR “forest bathing” OR “forest therapy”) AND Mini-Mental State Examination (“MMSE”). Additionally, studies were added by manually searching the references of the searched studies. Duplicate articles were removed, and RefWorks was used for article management. The initial search was independently performed by two researchers. After reviewing the titles and abstracts of the studies, they selected studies following the inclusion and exclusion criteria. In case disagreements, they consulted a third researcher to conclude. The results of the final selected studies were extracted and analyzed.

2.4. Assessment of bias

An assessment of bias was conducted on the final selected studies. For, RCTs the Cochrane’s Risk-of-bias tool (RoB2) was used (sterne et sl. 2019). For NRCTs, the Risk of Bias Assessment tool for Non-randomized Study (RoBANS) was used (Kim et al. 2011). Two researchers collaboratively reviewed the assessment results for each domain.

2.5. Data management and analysis

A data format was mutually determined by the researchers. The format included study information (authors, publication year, and study country), subjects (age and sex), intervention method (number of intervention groups, exercise type, duration, frequency, time, and intensity), and outcomes (cognitive and physical functions).

A meta-analysis package for the R program was used to conduct a meta-analysis. Since studies vary in subjects, intervention methods, and research environments, a random effect model was used to calculate the mean effect size. The effect size was calculated for cognitive function (MMSE and ADL) and physical function (agility, cardiorespiratory endurance, and upper and lower body strengths). MMSE scores are a scale for assessing cognitive function. This consist of a total of 30 questions in 23 items. total score is 30 points. 9 out of 30 points or less are suspected of severe dementia, 10 ∼ 14 points are suspected of moderate dementia, 15 ∼ 19 points are suspected of mild dementia, 20 ∼ 24 points are borderline area, and 25 points or more mean normal state. ADL performance is an assessment of daily activities, which is performed to determine the ability to take care of oneself and maintain independence in basic daily life and is an important basis for assessing early-stage illness. Agility was measured by timed up and go (TUG); this is an indicator that measures of how quickly it takes person to stand up from a 43 cm chair, go and return to the target 2.4 m, and sit down on the chair. It is a fall risk assessment scale that determines that the longer it takes to perform, the higher the risk of decline in physical function and dementia. Cardiorespiratory endurance was measured by 6-minute walking test (6MWT); this is an indicator that measures of how far person walked a 45.7m rectangular course in 6 minutes. Upper body strength is a 30 lb dumbbell repetitive lift for 30 seconds. Lower body strength was measured by 30s sit to stand test; this is an indicator that measures the number of times person sitting down and standing up from a 43 cm chair for 30 seconds. Normal standards are different for each gender and age, and healthy middle-aged people can perform it more than 20 times.

The overall effect size was determined based on the obtained effect sizes. The outcome variable was the difference between the post-test and pre-test means. The effect size was converted to Hedge’s g, which reflects a corrected standardized mean difference to account for the varying sample size across the studies and confidence interval was 95% (Borenstein et al. 2009). A forest plot was used to graphically present each study’s effect size, statistical significance, and weightage, overall (mean) effect size, and statistical significance of the combined studies (Hwang 2020). Cohen (1988) interpreted that an effect size is small if it is less than 0.10 ∼ 0.30, medium if it is less than 0.40 to 0.70, and large if it is 0.80 or higher. Heterogeneity was examined by creating a forest plot of average effect sizes and analyzed using the Q value representing the total variance and the I² value representing the true variance. An I² value ≤25% indicates low heterogeneity, 25-50% indicates moderate heterogeneity, and ≥75% indicates high heterogeneity (Higgins and Green 2016).

A meta-ANOVA was performed using frequency variables from walking programs to determine heterogeneity. To assess publication bias, the symmetry of a funnel plot was examined and visualized, followed by a test to assess the results’ consistency (Borenstein et al. 2009).

3. Results

3.1. Article selection

Including one from a reference review, a total of 2,123 studies were reviewed, selected, and organized using an article management program (ProQuest RefWorks). Among these, 138 duplicate studies, 102 studies with unmatched targets, 103 studies without original text, 1,463 studies with unmatched subjects, 170 studies without an experiment, 18 studies with unmatched study periods, 1 study written in a non-target language, and 101 studies without a comparator group were excluded. Two researchers independently collected and selected the studies. During disagreements, the researchers read the entire paper and made the final decision after reviewing the inclusion and exclusion criteria. If a study was excluded, they recorded the reason for the exclusion. A total of 27 papers eligible for a meta-analysis were eventually selected (Figure 1).

Figure 1. PRISMA flow chart

3.2. Characteristics of selected articles

Table 1 summarizes the characteristics of the 27 studies included in the meta-analysis, including authors, publication year, study country, sample size, study design, participants, intervention type, duration, frequency, time, intensity of exercise, and measurement tool. Publication years were evenly distributed from 2008 to 2021. One study (3.7%) was conducted before 2010, and 26 (96.3%) from 2010 to 2021. Twenty-one studies (77.8%) were published in South Korea, two (7.4%) in the United States, one (3.7%) in Finland, one (3.7%) in Germany, one (3.7%) in Brazil, and one (3.7%) in France. Fourteen studies (51.9%) were RCTs, and 13 (48.1%) were NRCTs. Fourteen studies (51.9%) had female subjects only, and 13 (48.1%) had mixed or unspecified sex. The subjects’ ages were evenly distributed from 60s to 70s years. Nineteen studies (70.4%) included healthy older adult subjects, and eight (29.6%) included older adult patients with cognitive decline, dementia, or Alzheimer’s disease. The intervention type was a complex exercise in 20 studies (74.1%) and walking exercise in 7 (25.9%). Complex exercise refers to exercises that combine aerobic and strength training. The intervention period ranged from 8 to 92 weeks. An intervention period of 12 weeks was the most common, used in 14 studies (51.9%). The weekly exercise frequency was twice in six studies (22.2%), three times in sixteen studies (59.3%), four times in two studies (7.4%), and five times or more in three studies (11.1%). The exercise time was 60 minutes in most studies (n = 16, 59.3%). Exercise intensity was presented in the heart rate reserve (HRR) and rating of perceived exertion (RPE). Nine studies (33.3%) did not describe how exercise intensity was determined (Table 1).

Table 1. Characteristics of selected articles

No	Authors (year)	Study country	Sample size(n)		Study design	M:W	Participants	Intervention					Outcome variable
			EG	CG				Type	Duration (weeks)	Frequency	Time (min)	Intensity	Cognitive function	Physical function
1	Song and Kim (2018)	South Korea	10	10	RCT	All women	Age > 70years women	Complex	12	3	60	THR 50 ∼ 70%	MMSE-K	–
2	Hong (2012)	South Korea	16	10	NRCT	EG 11:5 CG 5:5	Age > 60years	Complex	8	2-3	60	–	MMSE-K	Agility, upper body strength, lower body strength
3	Jeong and Oh (2013)	South Korea	16	19	NRCT	All women	Age > 65years with dementia(mmse11-23) or moderate dementia(10m walk)	Complex	12	3	60	RPE 9-11	MMSE-K	Agility, cardiorespiratory endurance, upper body strength, lower body strength
4	Huh et al. (2014)	South Korea	14	14	RCT	All women	Age > 65years women	Walking	13	3	60	HRR 70%	–	Agility, upper body strength, lower body strength
5	Chu and Shon (2012)	South Korea	14	11	NRCT	All women	Age > 65years women	Complex	16	3	60	–	–	cardiorespiratory endurance, upper body strength, lower body strength
6	Kim et al. (2014)	South Korea	10	10	NRCT	All women	Age > 65years women	Walking	12	5	60	HRR 50-80%	MMSE-K	–
7	Zhang (2021)	South Korea	14	12	NRCT	All women	Age > 70years women	Complex	12	3	60	HRR 40-60%	MMSE-K	Agility, cardiorespiratory endurance, upper body strength, lower body strength
8	An et al. (2019)	South Korea	21	17	NRCT	All women	Age > 70years women	Complex	16	3	60	HRR 60-70%	MMSE-DS	Agility, upper body strength(2), lower body strength
9	An and Choi (2019)	South Korea	10	10	NRCT	–	Age > 70years	Complex	8	3	60	RPE≦13	–	Agility, cardiorespiratory endurance, upper body strength, lower body strength
10	Kim (2017)	South Korea	15	15	NRCT	All women	Age > 65years women	Walking	12	3	50	RPE 10-12	MMSE-K	–
11	Lee and Joo (2021)	South Korea	20	23	NRCT	–	Age > 65years	Walking	12	2	20	HRR 60-70%	MMSE-DS	–
12	Park et al. (2014)	South Korea	20	20	NRCT	All women	Age > 65years women	Complex	12	3	80	HRR 60-70%	–	cardiorespiratory endurance, upper body strength(2)
13	Kim (2008)	South Korea	10	10	NRCT	All women	Age > 60-65years women	Walking	12	3	60	HRR 50-60%	–	Agility, lower body strength
14	Cho and Baek (2017a)	South Korea	9	9	RCT	All women	Age > 65 years women	Walking	12	4	60	RPE 9-11	–	Agility, cardiorespiratory endurance, upper body strength, lower body strength
15	Cho and Baek (2017b)	South Korea	9	9	RCT	All women	Age > 65 years women	Complex	12	4	60	RPE 9-11	–	Agility, cardiorespiratory endurance, upper body strength, lower body strength
16	Lee and Lee (2012)	South Korea	14	13	RCT	All women	Age > 65 years women	Complex	12	3	60	HRR 40-70%	MMSE-K	–
17	Hong (2013)	South Korea	15	7	NRCT	–	Dementia	Complex	24	2	60	–	MMSE-K	Agility, cardiorespiratory endurance, upper body strength, lower body strength
18	Yang and Park (2018)	South Korea	9	10	NRCT	All women	Age > 60 years women	Complex	12	3	70	RPE 6-17	–	Agility, cardiorespiratory endurance, lower body strength
19	Park et al. (2019)	South Korea	25	24	RCT	–	Age > 60 years	Complex	24	–	110	HRR 55-80%	MMSE-K	Agility, lower body strength
20	Fonte et al. (2019)	United States	27	30	RCT	–	Age > 60 years	Complex	24	3	90	HRR 70%	IADL	Cardiorespiratory endurance
21	Sanders et al. (2020)	United States	39	30	RCT	–	Mild and moderate dementia	Complex	24	3	90	RPE 9-11	MMSE	Agility(2), cardiorespiratory endurance, lower body strength
22	Holthoff et al. (2014)	Germany	15	15	RCT	–	Mild and moderate dementia	Complex	12	3	30	–	MMSE, ADCS-ADL	Agility
23	Barreto et al. (2017)	France	44	47	RCT	–	Alzheimer or dementia	Complex	24	2	60	–	MMSE, IADL	Cardiorespiratory endurance
24	Pitkänen et al. (2019)	Finland	86	89	RCT	–	Dementia	Complex	92	6	45	–	ADCS-ADL	BMI
25	Jeoug et al. (2021)	South Korea	13	13	RCT	–	Mild dementia	Complex	12	2	90	–	MMSE-K, ADCS-ADL	–
26	Aguiar et al. (2014)	Brazil	17	17	RCT	–	Alzheimer	Complex	24	2	60	–	MMSE, ADCS-ADL	Agility
27	Lee (2016)	South Korea	15	17	RCT	–	Age > 65years	Walking	5	3	50	–	–	Agility, cardiorespiratory endurance, upper body strength(2), lower body strength

EG: Experimental group; CG: Control group; RCT: Randomized Controlled Trial; NRCT: Non-randomized Controlled Trial; Complex: Interventions that combine walking with other exercise or activities; THR: Target Heart Rate; RPE: Rating of Perceived Exertion; HRR: Heart Rate Reserve; MMSE: Mini Mental State Examination; ADL: Activities of Daily Living; IADL: Instrumenta-ADL; ADCS: Alzheimer’s Disease Cooperative Study; BMI: Body Mass Index; Complex: walking and strength

3.3. Quality assessment

A risk of bias assessment was performed to assess the methodological quality of the 27 selected studies to ensure the risk of biases and misinterpretations that stem from a study design or occur during a research process is minimized. The 14 RCTs assessed with RoB2 showed a low risk of bias (Table 2). The 13 NRCTs assessed with RoBANS also showed a low risk of bias (Table 3).

Table 2. Characteristics of selected articles

Domain	Risk of bias
	Unclear	Low	High
Random sequence generation (selection bias)	–	13	1
Allocation concealment (selection bias)	3	11	–
Blinding of participants and personnel (performance bias)	–	14	–
Blinding of outcome assessment (detection bias)	–	14	–
Incomplete outcome data (attrition bias)	–	14	–
Selective reporting (reporting bias)	–	14	–
Other bias	–	14	–

Table 3. Risk of bias for non-randomized controlled trial studies (RoBANS)

Domain	Risk of bias
	Unclear	Low	High
Participants selection	–	13	–
Confounding variables	4	9	–
Measurement of exposure	–	13	–
Blinding of the outcome assessments	–	13	–
Incomplete outcome data	–	11	2
Selective outcome reporting (reporting bias)	–	13	–

3.4. Effect size of walking exercise

Hedge’s g, or a corrected standardized mean difference, was calculated using the difference between the pre-test and post-test means, the standard deviations of the difference, and the sample size of the experimental and control groups from the 27 selected studies, which was visualized using a Forest plot to see the overall trend (see Figures 2–7).

Figure 2. Effect size with MMSE scores

Figure 3. Effect size scores on activities of daily living

Figure 4. Effect size on agility

Figure 5. Effect size on cardiorespiratory endurance

Figure 6. Effect size on upper body strength

Figure 7. Effect size on lower body strength

1) Effect size of walking exercise on cognitive function

The effect of walking exercise on the MMSE scores and ADL performance, examined as measures of cognitive function, was examined.

Sixteen studies were included in the MMSE calculation of the effect sizes of walking. The overall effect size on MMSE scores was medium at 0.772 (95% Confidence Interval (CI): 0.419 ∼ 1.125) and was statistically significant as it did not include 0. This indicates that walking exercise substantially affects MMSE scores. The I² value, which represents the heterogeneity in the overall effect size of MMSE scores, was 76%, indicating high heterogeneity (Q = 62.50, p <. 0001) (see Figure 2). Seven studies were included in calculating the effect size of walking on the ADL performance. The overall effect size was 0.527 (95% CI: –0.820 ∼ 1.873) and was not statistically significant. The I² value was 87%, indicating high heterogeneity (Q = 45.73, p <. 0001) (see Figure 3).

2) Effect size of walking exercise on physical function

In 17 studies that examined agility, the overall effect size of walking exercise was –0.671 (95% CI: –1.284∼0.058), indicating that walking significantly affects agility. The I² value was 82%, indicating high heterogeneity (Q = 88.87, p <. 0001; see Figure 4). The overall effect size of walking exercise on cardiorespiratory endurance examined in 13 studies was 0.698 (95% CI: 0.383 ∼ 1.041), indicating that walking significantly affects cardiorespiratory endurance. The I² value of these studies was 56% (Q = 27.03, p = 0.0076; see Figure 5). The overall effect size of walking exercise on upper body strength examined in 15 studies was 0.685 (95% CI: 0.295 ∼ 1.074), indicating that walking significantly affects upper body strength. The I² value of these studies was 70% (Q = 46.24, p <. 0001; see Figure 6). The overall effect size of walking exercise on lower body strength examined in 15 studies was 0.433 (95% CI: 0.080 ∼ 0.787), indicating that walking significantly affects lower body strength. The I² value of these studies was 67% (Q = 41.91, p <. 0001; see Figure 7).

3.5. Heterogeneity test of effect size: moderation effect analysis

The MMSE scores (I²=76%, Q = 62.50, p <. 0001) and agility scores (I²=82%, Q = 88.87, p <. 0001) showed statistically significant effect sizes and high heterogeneity in the previous analysis. When effect sizes are heterogeneous across studies, a moderation effect analysis can identify the sources of heterogeneity (Page et al. 2021). Thus, a meta-ANOVA was conducted with sample size (experimental group), the type, duration, frequency, and time of exercise as the moderators (Table 4). During the analysis of the moderation effect on MMSE scores, walking had the highest effect size on MMSE scores when the sample size was 20-29, and a significant difference was found between the experimental group sizes (Q_b=20.28, p   =.   0001). For exercise types, walking had a higher effect size than complex exercise, with a significant difference between the experimental group sizes (Q_b = 13.50, p =. 0002). An exercise duration of 10 ∼ 19 days had the largest effect size, with a significant difference between the experimental group sizes (Q_b = 25.56, p =. 0001). An exercise time of 60 minutes had a largest effect size, with a significant difference between the experimental group sizes (Q_b = 32.61, p =. 0001). Exercise frequency did not affect the mean effect size (Q_b = 8.39, p =. 0387).

Table 4. Moderation effects on MMSE scores

Variables	Categories	k	Hedges's g	95% CI	I^2 (%)	Qb (p)
Size of experimental group	10-19	11	0.83	0.5240 ∼ 1.2262	50.1	20.28 (.0001)
	20-29	3	0.93	0.0915 ∼ 2.0873	86.4
	30-39	1	−0.06	0.5240 ∼ 1.2262	–
	40<	1	−0.19	−0.6062 ∼ 0.2181	–
Exercise type	Complex	13	0.56	0.2308 ∼ 0.8891	67.3	13.50 (.0002)
	Walking	3	1.68	1.1817 ∼ 2.1795	11.8
Exercise duration (weeks)	<10	1	0.92	0.0854 ∼ 1.7555	–	25.56 (.0001)
	10-19	10	1.14	0.7658 ∼ 1.5219	54.0
	20-29	5	0.00	−0.2420 ∼ 0.2450	0
Exercise frequency (weekly)	2	5	0.41	−0.1961 ∼ 1.0183	77.9	8.39 (.0387)
	3	9	1.01	0.5501 ∼ 1.4707	73.4
	5	1	1.38	0.3831 ∼ 2.3762	–
	<5	1	0.12	−0.4368 ∼ 0.6846	–
Exercise time (minute)	20	1	1.49	0.8100 ∼ 2.1787	–	32.61 (.0001)
	30	1	1.10	0.3241 ∼ 1.8740	–
	50	1	2.3	1.3507 ∼ 3.2467	–
	60	10	0.76	0.3575 ∼ 1.1588	70.4
	90	2	−0.01	−0.4169 ∼ 0.3929	0
	110	1	0.12	−0.4368 ∼ 0.6846	–

Note. k = number of studies; CI = Confidence Interval; Q_b = Q-value between subgroups.

During a moderation effect analysis on agility, the size of the experimental group (Q_b = 3.54, p =. 3514), exercise type (Q_b = 2.91, p =. 0879), the exercise duration (Q_b = 1.46, p =. 4811), and exercise frequency (Q_b = 9.50, p =. 0234) did not affect the mean effect size (Table 5). The largest effect size was observed for an exercise time of 60 minutes, with a significant difference between the experimental group sizes (Q_b = 37.18, p =. 0001).

Table 5. Moderation effect on agility

Variables	Categories	k	Hedges's g	95% CI	I^2 (%)	Qb (p)
Sample size (Experimental group)	10<	3	0.23	−0.7771 ∼ 1.2450	69.9	3.54(.3514)
	10-19	9	−1.20	−2.3355 ∼ −0.0698	88.3
	20-29	3	−0.45	−1.0256 ∼ 0.1344	52.0
	30-39	2	−0.34	−0.6791 ∼ −0.0001	0.0
Exercise type	Complex	13	−0.88	−1.6839 ∼ −0.0736	84.8	2.91(.0879)
	Walking	4	−0.88	−0.5220 ∼ 0.3657	25.2
Exercise duration (weeks)	<10	2	−1.90	−1.2916 ∼ −0.0425	96.2	1.46(.4811)
	10 ∼ 19	10	−0.69	−0.8169 ∼ −0.2836	82.9
	20 ∼ 29	5	−0.26	−0.5189 ∼ −0.0055	13.4
Exercise frequency (weekly)	2	2	−0.17	−1.1930 ∼ 0.8623	69.0	9.50(.0234)
	3	12	−1.03	−1.8308 ∼ −0.2264	83.7
	4	2	0.69	−0.0650 ∼ 1.4466	17.8
	–	1	−0.27	−0.8329 ∼ 0.2928	0
Exercise time (minutes)	30	1	−4.87	−6.3624 ∼ −3.3669	–	37.18(.0001)
	50	1	0.14	−0.5597 ∼ 0.8307	–
	60	11	−0.53	−1.2028 ∼ 0.1432	79.7
	70	1	−0.69	−1.6230 ∼ 0.2434	–
	90	2	−0.34	−0.6791 ∼ −0.0001	0.0
	110	1	−0.27	−0.8329 ∼ 0.2928	–

Note. k = number of studies; CI = Confidence Interval; Q_b = Q-value between subgroups

3.6. Publication Bias

A funnel plot analysis, generally recommended for visually analyzing plot symmetry, was used to assess publication bias, and the analysis results were validated using Egger’s regression test (Borenstein et al. 2009). A symmetrical funnel plot with studies evenly distributed throughout the funnel plot was observed for MMSE, agility, cardiorespiratory endurance, and upper and lower body strengths. Egger’s regression test was performed to confirm the result’s statistical significance, which showed a bias of 5.87 (t = 4.67, df = 14, p =. 0514) and –3.7001 (t=–2.05, df = 15, p =. 0584) for MMSE and agility, respectively; no significant asymmetry was observed, indicating no publication bias. Regarding cardiorespiratory endurance, upper body strength, and lower body strength, a bias of 3.0175 (t = 2.83, df = 11, p =. 0164), 6.3397 (t = 3.25, df = 13, p =. 0063), and 3.9287 (t = 2.22, df = 13, p =. 0446) was observed, respectively; publication bias existed for these variables. Publication bias was not assessed for ADL using the funnel plot and Egger’s regression analysis since there were fewer than 10 studies.

Discussion

Cognitive effects of walking fognitive effects of walkingalking with other exercise or

The overall effect size of walking exercise on MMSE scores as a measure of cognitive function in older adults was moderate at 0.772 (95% CI: 0.419 ∼ 1.125), indicating that these can improve cognitive function in older adults. Regular physical activity increases neuroplasticity in the brain, thereby positively affecting cognitive function (Voelcker‐Rehage et al. 2010, Hillman et al. 2008). This supports the results of this study wherein walking improved cognitive function (Colombe and Kramer 2003). A moderation effect analysis performed due to high heterogeneity showed significant differences in the effect of walking on MMSE scores according to the experimental group size and the type, duration, and time of exercise. Therefore, it is considered that a group of 20 ∼ 29 subjects and a combined exercise program including forest walking exercise are suitable for improving the cognitive function of the elderly (Kim and Yoo 2013, Spina et al. 1993). Walking exercise had the highest effect on MMSE scores for an exercise duration of 10 ∼ 19 weeks, an exercise frequency of three times per week, and an exercise time of 60 minutes. Walking can be recommended to anyone since it is an easily accessible aerobic exercise that can be performed anywhere at any time and does not require special skills. Han et al. (2007) reported that an exercise program for older adults is the most effective if conducted for 8 to 12 weeks, consistent with our findings. The effect size of walking exercise on ADL scores as a measure of cognitive function was moderate at 0.527 (95% CI: -0.820 ∼ 1.873) and not statistically significant. Chu et al. (2007) conducting a cognitive behavioral program among older adults with cognitive decline twice a week for six weeks (a total of 12 50-minute sessions) reported significant improvements in cognitive function by the program end but no significant difference in ADL performance, consistent with our results.

Second, the overall effect size of walking exercise on older adults’ agility was -0.671 (95% CI: -1.284∼-0.058) and was statistically significant. The I2 value for studies on agility was 82%, indicating high heterogeneity (Q = 88.87, p <. 0001). A moderation effects analysis showed that the experimental group size and the type, duration, and frequency of exercise did not affect the mean effect size; however, the exercise time was the only variable that had an effect. Since agility is highly associated with walking speed, balance, and ADL measures, our result demonstrates that regular exercise can promote normal walking and improve body stability (Rikli and Jones 2013). The effect size of walking exercise on cardiorespiratory endurance as a measure of physical function was 0.698 (95% CI: 0.383 ∼ 1.041) and was statistically significant. Cardiorespiratory endurance is crucial to older adults’ ADL performance, which can be improved by participating in an exercise program (Kim et al. 2015); our results thus confirm this finding by showing that participating in a walking program improves one’s ADL performance by improving cardiorespiratory function. The overall effect size of walking exercise on upper body strength was 0.685 (95% CI: 0.295 ∼ 1.074) and was statistically significant, and lower body strength was 0.433 (95% CI: 0.080 ∼ 0.787) and was statistically significant. Aging-related muscle mass and strength loss lead to muscle functional deterioration, reducing physical activity. Regular exercise delays functional muscle deterioration or improves muscle function and reduces the risk of falls, leading to a safer life (Kim 2013, Seo et al. 2014, Haskell and Phillips 1995). Our results show that improving muscle strength and muscular endurance positively affects daily life and that exercise promotes healthy living in older adults.

Our study is meaningful as we determined the effect sizes of walking exercise on older adults’ cognitive functions, agility, cardiorespiratory endurance, upper body strength, and lower body strength and the overall effect of all studies combined. Additionally, we analyzed the effect size of walking exercise for varying sample sizes, study designs, sexes, ages, exercise types, durations, frequencies, and time, and obtained essential data for developing a forest walking program for older adults. The effects of walking may be maximized by using the appropriate exercise duration and intensity for each individual to improve their cardiorespiratory function and adding flexibility exercises, such as stretching. We combined and quantitatively analyzed previous findings to produce more generalizable results, thereby reducing the need for repeated research and providing data for better decisions when developing a forest walking program. We also explored various study designs used to assess the effectiveness of walking exercise programs through a literature review. Our findings can help minimize errors when designing or assessing research on walking exercise programs and forest walking program or selecting appropriate assessment tools or measures.

Conclusion

This study performed a meta-analysis to assess the effects of walking exercise programs for older adults based on the RCTs and NRCTs results assessing cognitive and physical functions. Our analysis demonstrates that walking exercise programs can improve older adults’ cognitive and physical functions, which is a significant finding considering the ongoing global population aging. This result indicates that walking exercise improves overall cognitive function and physical fitness and creates positive changes of body composition in older adults, which can further improve their ability to perform daily activities and, consequently, improve their quality of life and reduce the medical costs of aging societies.

Declaration of interest

The authors report there are no competing interests to declare.

Additional information

Funding

This work was supported by the Forest Science Technology R&D Project under Grant C1016811-01-02 by the Korea Forestry Promotion Institute (KOFPI).