Can electric-powered carts used in public transportation change long-term care risk among older adults?: evidence from a 2-year follow-up survey

Abstract

Background

Amid rapid population aging in Japan, declining public transport use has created substantial barriers to older adults’ mobility and independence. Such services address first-/last-mile gaps and enable social participation, which are key determinants of healthy aging in community settings. We examined whether Green Slow Mobility (GSM)—a low-speed electric cart used as community public transport—was followed by differences in future long-term care need.

Methods

We analyzed data from 599 community-dwelling older adults (59.3% women; mean age 74.6 years) in two municipalities with different service conditions: Oji town, where GSM operation was stably maintained, and Kawachinagano city, where GSM experienced a 4-month suspension followed by reduced frequency. At the 1-year survey, participants using GSM at least once per month were classified as users; others as nonusers. Long-term care risk was measured at baseline and at 2 years with a validated 0–48-point scale (higher scores indicate greater risk); the 2-year score served as the primary outcome. To address confounding, we conducted propensity-score matching on sex, age, and baseline risk (standardized difference < 0.25) and then estimated municipality-stratified linear regressions. Model 1 adjusted for baseline risk; Model 2 additionally adjusted for education, subjective economic status, household composition, and marital status. Robustness analyses applied stricter matching (standardized difference < 0.10).

Results

In Oji town, users had significantly lower risk scores at 2 years than nonusers (B = − 2.31, 95% CI − 4.35 to − 0.28, p = 0.027). In Kawachinagano city, no significant difference was observed (B = 0.28, 95% CI − 0.74 to 1.32, p = 0.581). Findings were directionally consistent in the robustness analysis with stricter matching: Oji town B = − 2.72 (95% CI − 5.25 to − 0.18, p = 0.037); Kawachinagano city B = 0.38 (95% CI − 0.69 to 1.44, p = 0.490).

Conclusions

Under conditions of stable operation, regular GSM use was followed by lower long-term care risk among older adults. The absence of a difference in the municipality with suspended and reduced service suggests that continuity and frequency of operation are important factors. These results underscore the importance of maintaining reliable, sustainable community transport when designing mobility systems for super-aged societies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13690-026-01931-1.

Text box 1. Contributions to the literature
• Declining public transportation poses mobility issues for healthy aging populations.
• We examined whether using electric-powered carts was followed by differences in subsequent long-term care risk among older adults in Japan.
• Using electric-powered carts at least once a month showed lower long-term care risk.
• Users compared with nonusers may save 15–20 million yen in long-term care benefits.
• Community implementation of electric-powered carts may be followed by lower long-term care risk among older adults.

Introduction

With the global aging of the population, the declining use of public transportation is posing increasingly significant mobility challenges for older adults around the world [1]. This issue is particularly acute in Japan, the world’s most aged society, and raising concerns about potential increases in long-term care expenditures due to functional decline [2, 3]. Within the World Health Organization’s “Age-Friendly Cities” framework, transportation has been identified as a core domain essential for healthy aging [4]. Building on this foundation, community-based transport solutions have emerged as promising interventions. In Japan, these services utilize small electric-powered carts termed “Green Slow Mobility” (GSM), which operate at speeds less than 20 km/h within residential areas and have been implemented across 130 regions throughout the country [5–8].

Previous research on community-based transport solutions has examined various effects on older adults following implementation. Studies in Florida and the southern regions of the United States have reported that electric carts enhance the safety of mobility for older adults, individuals with disabilities, and those unable to drive automobiles, thereby serving as an effective means to maintain independence in such communities [9]. Research on sustainable transportation systems has shown that community-based transport solutions may reduce environmental impacts while securing mobility for older adults and revitalizing local communities [10]. In Japan, these services utilize GSM, which are promoted as a sustainable mobility solution by the Ministry of Land, Infrastructure, Transport and Tourism. Recent studies have demonstrated that the regular use of community-operated transport services can help reduce loneliness and strengthen social networks, contributing to improved social relationships and well-being among older adults [11, 12]. A recent short-term before-and-after study in Japan reported that the utilization of GSM promotes outings and social behavior and enhances positive emotions [13]. That study primarily assessed subjective changes following GSM implementation rather than long-term care risk outcomes [13]. Furthermore, in a 1-year longitudinal study, individuals with GSM-induced positive emotional improvements demonstrated a reduced risk for functional decline [14]. In that study, the risk reduction was reported mainly among those with improved positive emotions rather than at the overall user level [14].

These previous findings suggest that beyond merely providing mobility support, GSM may promote physical activity and social participation, which are essential for successful aging. However, in previous studies investigating the impact of GSM use on subsequent long-term care risks for short periods of 2–3 months and 1 year [13, 14], the effects showing positive emotional changes have been limited to subgroups rather than the overall user population, and therefore lacked sufficient longitudinal verification. In addition, it remains unclear how implementation features, such as operational continuity and the operating body, shape longer-term outcomes.

Given this background, this study aimed to examine whether GSM use was followed by differences in subsequent long-term care risk by analyzing data from two regions in Japan where GSM has been implemented, which differed in implementation conditions. Because GSM users and nonusers have different baseline characteristics [13, 14], we applied propensity score matching (PSM) based on preexposure baseline variables to address potential selection bias [15–17].

Methods

Study participants

This observational study leveraged variation between municipalities to compare differences in functional disability risk—assessed as long-term care risk [18]—within each municipality at 2 years after operation, based on whether participants used GSM at the 1-year follow-up in two Japanese regions where GSM had been implemented. The two municipalities differed in terms of the continuity and management of GSM operations, which enabled a comparison between stable and interrupted service provision.

The target regions were the Myojin and Taishi communities of Oji town, Nara Prefecture (population 3449; proportion of residents aged ≥ 65 years: 50.9%) [19], and the Nitto and Daishi districts of Kawachinagano city, Osaka Prefecture (population 2640; proportion of residents aged ≥ 65 years: 47.2%) [20]. Both communities are hilly residential areas in metropolitan areas that were developed as new towns in the 1970s and have been experiencing population aging and mobility challenges.

GSM drivers were contracted as local transportation operators for the on-demand operation in Oji town and Kawachinagano city, while the fixed-route operation in Oji town was managed by community volunteers. Operations in Kawachinagano city, which were managed by contracted local transportation operators, were temporarily suspended on July 22, 2023, and when resumed in December 2023, were reduced to 2 days a week with eight services per day as a result of operational changes.

Figure 1 shows a flowchart of the study participants. A total of 1669 questionnaires were distributed at explanatory meetings (n = 217) and by mail to randomly selected residents (n = 1452) from July to August 2022. We received 1137 responses at baseline (response rate: 68.1%). At the 1-year follow-up (July to August 2023), 796 individuals responded (follow-up rate: 70.0%). For the 2-year follow-up (July to August 2024), we excluded participants who had moved, received long-term care certification, or died, and distributed questionnaires to the remaining participants. We received 599 valid responses (follow-up rate: 52.7%). These 599 individuals (59.3% women, mean age: 74.6 years) were included in the final analysis after excluding those with sex/age discrepancies, those with unlinkable data across the three time points, and those who had not provided consent.

Fig. 1.

Flowchart of the study participants

GSM operation

The GSM used in this study was the AR-07 manufactured by Yamaha Motor Co., Ltd. (Iwata, Japan). This electric cart has a seating capacity of seven and features an open design without doors. During rainy weather, a vinyl enclosure can be used to keep the passengers from getting wet. In both municipalities, GSM operations were designed to improve convenience and encourage uptake. Oji town operated one vehicle under resident-led management due to constraints in vehicle storage and driver availability. By contrast, Kawachinagano city initially operated two vehicles because it had secured storage space for two vehicles and drivers for both, but reduced operations to one vehicle when services resumed.

Table 1 shows the operational status of GSM in the two municipalities. In Oji town, a fixed-route operation was adopted 4 days a week from 09:30 to 16:00 on three routes (1.6–3.0 km, 9–16 min each), with 12–15 services per day. Additionally, Oji town provided on-demand operation 3 days a week from 10:00 to 17:00 until February 2023, and then continued with fixed-route services better suited to local needs. In Kawachinagano city, GSM operated 4 days a week from 09:30 to 12:00 and 13:00 to 16:00 on fixed routes. A 20–25 min, 4.2 km route ran twice per hour in both directions, providing approximately 20 services per day.

Table 1.

Operating schedules of Green Slow Mobility

Oji town experienced one interruption lasting 1 month. Kawachinagano city faced two interruptions, first for 1 month, and then for 4 months

On-demand operation in Oji town has been discontinued

In Kawachinagano city, service has resumed after the interruption, but with significantly reduced frequency

Both municipalities avoided busy main roads and operated free of charge along fixed routes within residential areas, with passengers able to board and alight at designated stops or by hailing vehicles along the route. Routes included supermarkets as key destinations.

Measurements

Outcome: risk score

The outcome variable was the risk of functional disability, measured using the Long-Term Care Risk Assessment Scale (Supplementary Table 1) [18]. Throughout this paper, the score on this scale is referred to as the “risk score”. The Long-Term Care Risk Assessment Scale is composed of 12 items: sex, age, and 10 items on physical function such as “Do you normally stand up from a chair without any aids?” and “Are you very worried about falls?”. The risk score ranges from 0 to 48, with higher scores indicating higher risk. Previous follow-up studies have confirmed the scale’s predictive validity for the incidence of long-term care certification among community-dwelling older adults and its ability to predict subsequent cumulative long-term care costs [21–23]. As a sensitivity analysis, we used a revised 2-year risk score that excluded the item “Do you go out alone by bus or train?” as the outcome.

Explanatory variable: GSM use

The explanatory variable was GSM use. At the 1-year follow-up, the participants were asked about their frequency of GSM use from August 2022 until the survey date using the following six categories: “4 or more times per week”, “2–3 times per week”, “once a week”, “1–3 times per month”, “several times per year”, and “no use” (Supplementary Table 2). Supplementary Table 3 shows that higher usage frequency was associated with higher baseline risk scores, and the number of users decreased as usage frequency increased. For the analysis, participants who reported using GSM at least once a month were classified as users, whereas those who used GSM less than once a month were classified as nonusers.

Covariates

Covariates were selected based on previous research on public transportation and disability (Supplementary Table 2) [13, 14, 24]. The covariates used for calculating propensity scores were sex, age, and baseline risk score. For the linear regression analysis after PSM, Model 1 was adjusted for baseline risk score only, while Model 2 was adjusted for education level, subjective economic status, living arrangement, marital status, and baseline risk score (Supplementary Table 4). The baseline risk score was included as an adjustment variable to reduce any remaining imbalance after matching. This approach follows recommendations that additional regression adjustment for prognostically important variables is appropriate even after PSM. Matching creates comparable groups, while regression analysis adjusts for remaining differences within the matched cohort [25–27].

Statistical analysis

To align the background characteristics of the user and nonuser groups, PSM was performed based on sex, age, and risk score at baseline. Propensity scores were estimated using a logistic regression model. For the matching algorithm, nearest-neighbor matching was used, selecting one person randomly from the user group and pairing them with the nonuser who had the closest propensity score. The matching ratio was 1:1. For the individual extraction of the nonuser group after matching, sampling without replacement was used. In this analysis, two matching patterns were created to ensure the robustness of the findings. To verify the balance between groups after matching, standardized differences (StD) were calculated, with StD < 0.25 considered acceptable and StD < 0.1 considered an ideal balance [28, 29]. Based on these criteria, the caliper method was used, setting the caliper width based on the standard deviation of the logit-transformed propensity scores. As the propensity score distribution and sample size differed by region, the caliper width was adjusted to ensure appropriate matching. The caliper widths were adjusted by region based on the propensity score distribution. In Oji town, a caliper width of 0.001 was used for the StD < 0.1 criterion and 0.2 for the StD < 0.25 criterion. In Kawachinagano city, a caliper width of 0.2 was used for the StD < 0.1 criterion and 0.3 for the StD < 0.25 criterion. Next, linear regression analysis was performed to estimate effect sizes (B) and confidence intervals (CIs). The risk score was used as the outcome variable and GSM use as the explanatory variable. Two models were fitted: Model 1 adjusted for baseline risk score only, and Model 2 adjusted for education level, subjective economic status, living arrangement, marital status, and baseline risk score. The data sets used for the analysis were constructed from the full sample before PSM, the primary analysis using the sample matched with StD < 0.25, and the robustness analysis using the sample matched with StD < 0.1. To account for regional differences in operation continuity and periods, data from each region were stratified for analysis. Two-sided p-values < 0.05 were considered statistically significant. All analyses were performed using STATA (Ver.16.1; Stata Corp LLC, College Station, TX, USA).

Ethics approval statement

This study was approved by the Ethics Committee of the Graduate School of Medicine, Chiba University (approval date: August 24, 2021, approval No.: M10065), and conducted in accordance with the Declaration of Helsinki. All participants were informed that participation was voluntary and that refusing to participate would not result in any disadvantage. Participants received written information about the study, and informed consent was obtained through questionnaire return or written consent forms.

Results

Baseline characteristics

Table 2 shows the GSM usage status at the 1-year follow-up, with those who used GSM at least once a month accounting for 13.2% in Oji town and 37.2% in Kawachinagano city. Table 3 shows the participants’ baseline characteristics. The user group included more women, individuals aged ≥ 75 years, and those with higher baseline risk scores.

Table 2.

Green Slow Mobility usage status at the time of the 1-year follow-up

		Oji town		Kawachinagano city
		(n = 289)		(n = 310)
		N	%	N	%
Usage frequency	4 or more times per week	0	0	7	2.3
	2–3 times per week	8	2.8	28	9
	Once a week	7	2.4	20	6.5
	1–3 times per month	23	8	60	19.4
	Several times per year	42	14.5	47	15.2
	No use	195	67.5	135	43.6
	Missing	14	4.8	13	4.2

Table 3.

Participant characteristics before and after propensity score matching (PSM)

Oji town		Before PSM				After PSM (StD ≤ 0.25)b				After PSM (StD ≤ 0.1)b
		User	Nonuser	p	StD	User	Nonuser	p	StD	User	Nonuser	p	StD
		groupa	groupa			groupa	groupa			groupa	groupa
		(n = 38)	(n = 237)			(n = 36)	(n = 36)			(n = 27)	(n = 27)
		%	%			%	%			%	%
Sex	Men	31.6	46.4	0.087	0.390	33.3	25	0.437	0.182	33.3	33.3	0.776	0.076
	Women	68.4	53.6			66.7	75			66.7	66.7
Age (years)	65–69	7.9	35	< 0.001	0.548	8.3	8.3	0.994	0.000	8.3	11.1	0.994	0.037
	70–74	39.5	42.2			38.9	38.9			38.9	51.9
	75–79	26.3	12.7			27.8	25			27.8	18.5
	80–84	13.2	8.4			13.9	19.4			13.9	14.8
	85+	13.2	1.7			11.1	8.3			11.1	3.7
	Mean ± SD	76.2 ± 6.0	71.8 ± 5.1	-	-	76.0 ± 5.8	75.9 ± 6.3	-	-	74.2 ± 4.9	74.1 ± 5.7	-	-
Risk score	Mean ± SD	18.2 ± 9.1	11.4 ± 7.6	< 0.001	0.681	18.2 ± 9.1	17.9 ± 9.3	0.899	0.030	15.1 ± 8.0	14.6 ± 7.5	0.807	0.067

Kawachinagano city		Before PSM				After PSM (StD ≤ 0.25)b				After PSM (StD ≤ 0.1)b
		User	Nonuser	p	StD	User	Nonuser	p	StD	User	Nonuser	p	StD
		groupa	groupa			groupa	groupa			groupa	groupa
		(n = 115)	(n = 182)			(n = 107)	(n = 107)			(n = 99)	(n = 99)
		%	%			%	%			%	%
Sex	Men	32.1	39.6	0.198	0.089	30.8	38.3	0.250	0.157	33.3	35.4	0.765	0.042
	Women	67.8	60.4			69.2	61.7			66.7	64.7
Age (years)	65–69	10.4	16.5	0.158	0.280	11.2	6.5	0.312	0.009	12.1	6.1	0.335	0.019
	70–74	16.5	24.7			16.8	26.2			17.1	26.3
	75–79	30.4	25.8			30.8	29.9			30.3	30.3
	80–84	35.7	26.4			34.6	28.0			34.3	29.3
	85+	7	6.6			6.5	9.4			6.1	8.1
	Mean ± SD	77.6 ± 5.5	75.8 ± 5.8	-	-	77.5 ± 5.5	77.2 ± 5.2	-	-	77.2 ± 5.5	77.2 ± 5.2	-	-
Risk score	Mean ± SD	19.5 ± 8.1	15.9 ± 8.3	< 0.001	0.435	19.0 ± 7.5	19.2 ± 7.9	0.790	0.036	18.6 ± 7.8	18.9 ± 7.6	0.775	0.041

GSM Green Slow Mobility, PSM Propensity score matching, StD Standardized difference, SD Standard deviation

^aThe user group included those who used GSM at least once per month; the nonuser group included those who used GSM less than once a month or never, excluding participants with missing GSM usage data (n = 14 and n = 13, respectively)

^bPSM was conducted to achieve standardized differences (StD) ≤ 0.25 and ≤ 0.1 between groups

Results of PSM

Table 3 shows the results of PSM. In PSM with StD < 0.25, 36 pairs in Oji town and 107 pairs in Kawachinagano city were matched. After matching, the groups were well-balanced between the user and nonuser groups in both Oji town (18.2 ± 9.1 vs. 17.9 ± 9.3, respectively) and Kawachinagano city (19.0 ± 7.5 vs. 19.2 ± 7.9, respectively). With StD < 0.1, 27 pairs in Oji town and 99 pairs in Kawachinagano city were matched, despite the reduced sample sizes. Risk scores were well-balanced between the user and nonuser groups in both Oji town (15.1 ± 8.0 vs. 14.6 ± 7.5, respectively) and Kawachinagano city (18.6 ± 7.8 vs. 18.9 ± 7.6, respectively).

GSM use and risk score differences

Table 4 shows the risk score differences between the GSM user and nonuser groups. The analysis revealed different patterns between the two regions at the 2-year follow-up. After PSM with StD < 0.25, Oji town showed a significant risk difference in risk scores (B = − 2.31; 95% confidence interval [CI], − 4.35 to − 0.28; p = 0.027), with consistent results across adjustment models. Similar significant differences were observed in the robustness analysis with StD < 0.1 (B = − 2.72; 95% CI, − 5.25 to − 0.18; p = 0.037). By contrast, Kawachinagano city showed no significant differences in any matching criterion (StD < 0.25; B = 0.28; 95% CI, − 0.74 to 1.32; p = 0.581; StD < 0.1; B = 0.38; 95% CI, − 0.69 to 1.44; p = 0.490). The sensitivity analysis using the revised risk score yielded results consistent with the main analysis (Table 4).

Table 4.

Differences in long-term care risk assessment scale scores between Green Slow Mobility (GSM) users and nonusers

		Model 1				Model 2
		After PSM		After PSM		After PSM		After PSM
		[StD ≤ 0.25]b		[StD ≤ 0.1]b		[StD ≤ 0.25]b		[StD ≤ 0.1]b
		B	p	B	p	B	p	B	p
		(95% CI)		(95% CI)		(95% CI)		(95% CI)
Oji town	Nonuser groupa	ref.		ref.		ref.		ref.
	User groupa	–2.34	0.021	–2.76	0.025	–2.31	0.027	–2.72	0.037
		(–4.30 to –0.37)		(–5.16 to 0.37)		(–4.35 to –0.28)		(–5.25 to –0.18)
Sensitivity analysisc	Nonuser groupa	ref.		ref.		ref.		ref.
	User groupa	–2.00	0.032	–2.39	0.032	–2.05	0.032	–2.42	0.038
		(–3.82 to –0.18)		(–4.57 to 0.21)		(–3.91 to –0.19)		(–4.69 to –0.15)
Kawachinagano city	Nonuser groupa	ref.		ref.		ref.		ref.
	User groupa	0.31	0.548	0.46	0.391	0.28	0.581	0.38	0.490
		(–0.71 to 1.34)		(–0.60 to 1.52)		(–0.74 to 1.32)		(–0.69 to 1.44)
Sensitivity analysisc	Nonuser groupa	ref.		ref.		ref.		ref.
	User groupa	0.41	0.428	0.52	0.328	0.39	0.449	0.46	0.397
		(–0.61 to 1.43)		(–0.53 to 1.57)		(–0.63 to 1.41)		(–0.60 to 1.52)

PSM Propensity score matching, StD Standardized difference, CI Confidence interval

Model 1 adjusted for baseline risk score

Model 2 adjusted for education level, subjective economic status, living arrangement, marital status, and baseline risk score (see Supplementary Table 4)

Sample sizes after propensity score matching: Oji town: 36 pairs (StD ≤ 0.25), 27 pairs (StD ≤ 0.1); Kawachinagano city: 107 pairs (StD ≤ 0.25), 99 pairs (StD ≤ 0.1)

^aThe user group included those who used GSM at least once per month; the nonuser group included those who used GSM less than once a month or never

^bPSM was conducted to achieve standardized differences (StD) ≤ 0.25 and ≤ 0.1 between groups

^cSensitivity analysis used a revised risk score excluding the item “Do you go out alone by bus or train?” as the outcome variable

Discussion

In this study, we examined whether GSM use was followed by differences in subsequent long-term care risk in two communities in hilly residential areas of Japan. Using PSM to compare user and nonuser groups across three-point survey data, we found lower risk scores at the 2-year follow-up among users compared with nonusers in Oji town. By contrast, no significant differences were observed in Kawachinagano city.

Regarding the lower long-term care risk observed among GSM users in Oji town, while previous research has demonstrated lower risk only among subgroups with increased positive emotions during a 1-year follow-up, the present findings extend this difference to the overall user population through longer-term follow-up [13]. The findings suggest that sustained GSM use may be accompanied by positive attitudes and lower long-term care risk across all users. Furthermore, Rambaldini-Gooding et al. [30] suggested that free or subsidized public transportation services may help improve the health of older adults and support preventive care. Building on this foundation, the present study revealed a risk score difference of approximately 2.3 points over the 2 years following GSM introduction. Based on this risk score differential, calculations based on previous research suggest that the GSM user group may incur approximately 15–20 million yen less in long-term care benefits over the next 6 years compared with the nonuser group [23]. This indicates that transportation investment can be economically rational in terms of long-term care costs. These results suggest that regional transportation sustainability can help reduce older adults’ risk of functional disability while empirically supporting the importance of transportation’s role in the “Age-Friendly Cities” framework [4].

Moreover, while Oji town continued operations without interruption, Kawachinagano city experienced service interruptions in July 2023, with reduced frequency following resumption in December 2023. Kobayashi et al. [31] reported deteriorated risk scores among high-frequency GSM users during service interruptions. Previous studies have shown that continuous access to public transportation improves mental health and maintains physical function among older adults, whereas restricted access may increase health-care and long-term care costs through functional decline [32, 33]. These findings suggest that ensuring regular and stable operational systems is important for preventive care. In Kawachinagano, although heavier GSM use was observed, risk scores did not show a clear difference, whereas in Oji town, risk scores were lower among GSM users despite less overall use. This suggests that service continuity may be more important than frequency of use alone.

Additionally, Oji town’s operational approach might have been advantageous. While Oji town’s fixed-route operations were volunteer-managed, Kawachinagano city used contracted local transportation operators. Previous studies have shown that social interaction and participation among older adults contribute to reducing long-term care risks [34, 35]. A systematic review by Zhang and Yang reported that age-friendly transportation systems play important roles in older adults’ health and social participation [1]. These findings suggest that when community residents participate as volunteer operators, interactions may occur more readily not only among users, but also between users and drivers, potentially strengthening social relationships.

Regarding the frequency of appropriate GSM usage, we established monthly or more frequent use as our criterion. Given that our lowest frequency option was “1–3 times per month”, we infer that most users in our sample likely used GSM several times per month rather than just once. GSM facilitates social interaction during rides, which may help prevent social withdrawal and isolation, which are typically measured by weekly thresholds [36, 37]. Additionally, considering research showing that proximity to a positive neighborhood food environment may reduce the risks of dementia and functional disability [38, 39], GSM-facilitated shopping trips may reduce the risk of functional disability by improving access to food resources. Taken together, while not as active as weekly usage, using services several times per month may represent a moderately active lifestyle indicator.

From a gerontological perspective, our findings provide important implications for the implementation of “Age-Friendly Cities”. The concept of successful aging emphasizes three elements: avoiding disease and disability, maintaining high physical and cognitive function, and active engagement with life. GSM use may contribute to successful aging particularly by promoting active engagement with life [40]. Furthermore, in Baltes’ Selection, Optimization, and Compensation model [41], the importance of environments supporting older adults’ adaptation to changes is emphasized, and regular mobility service provision may support this process.

This study has several limitations. First, while we attempted to adjust for confounding factors using PSM, unmeasured confounding factors may have influenced the relationship between GSM use and the subsequent risk of functional disability. Additionally, although our 2-year follow-up is longer than previous studies, the effects may be underestimated; longer follow-up periods are needed to capture fully the benefits of GSM use. Second, as this study targeted specific regions, caution is needed in regard to generalizability, and verification across multiple municipalities with different characteristics is expected. Third, data from self-reported surveys are based on subjective assessments, which raises concerns about various biases. Additionally, we analyzed only three-wave completers; because healthier individuals are more likely to respond to surveys, attrition bias may have led to underestimation of the differences following GSM use and limited the generalizability of the findings. Therefore, additional research utilizing objective usage records for more precise analysis is needed. Despite these limitations, a major strength of this study is our use of a 2-year longitudinal follow-up and PSM with real-world community data, which enabled us to compare GSM users and nonusers across two municipalities that differed in terms of the continuity and management of GSM operations. These findings were consistent in sensitivity analyses using a revised risk score that excluded the item on independent outings by public transportation.

Conclusions

In this study, we examined whether GSM was followed by differences in long-term care risk through a 2-year longitudinal analysis in suburban regions of Japan. Using PSM, we found lower risk scores among GSM users compared with nonusers in Oji town, where services operated continuously. By contrast, in Kawachinagano city, where operations were temporarily suspended and later resumed at a reduced frequency, a similar difference was less evident. Taken together, these findings suggest that the difference with GSM use may depend on implementation conditions such as operational continuity rather than service availability alone. When scaling up GSM as a community mobility approach, maintaining continuous operations and securing an appropriate operating structure may be important to achieve longer-term benefits.

Abbreviations

GSM

Green Slow Mobility

PSM

Propensity Score Matching

CI

Confidence Interval

StD

Standardized Difference

Authors’ contributions

Conceptualization: MF, MS, RW, SK, KI, and KK; Data curation: MF, SK, and KI; Formal analysis: MF; Funding acquisition: MS, KI, and KK; Investigation: MS, RW, SK, and KI; Project administration: MS and KK; Supervision: MS, RW, and KK; Validation: MF; Visualization: MF; Writing original draft: MF; Writing review & editing: MS, RW, SK, KI, and KK. All authors read and approved the final manuscript.

Funding

This study was funded by Yamaha Motor Co., Ltd., JST (JPMJOP1831) and JSPS KAKENHI (JP23H00060) through a collaboration between Chiba University and Yamaha Motor Co., Ltd. The funders of this study had no role in the study design, data collection, data analysis, data interpretation, or writing of the paper.

Data availability

The datasets are not publicly available due to ethical or legal restrictions arising from sensitive human participant information, but are available from the corresponding author on reasonable request, subject to institutional and ethics approvals and a data use agreement.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of the Graduate School of Medicine, Chiba University (approval date: August 24, 2021, approval No.: M10065), and conducted in accordance with the Declaration of Helsinki. All participants were informed that participation was voluntary and that refusing to participate would not result in any disadvantage.

Consent for publication

Not applicable.

Competing interests

This study was conducted through the JST-OPERA program, during which, Nihon Fukushi University served as a partner institution. Following the program’s completion, a collaborative research agreement was established between Chiba University and Yamaha Motor Co., Ltd., with Chiba University entering into a subcontract agreement with Nihon Fukushi University. Saito M received funding from Chiba University, while Kondo K and Ide K received funding from Yamaha Motor Co., Ltd.