We conducted a retrospective observational study using data from a community-based physical activity program, “Tec-Tec Zushi”,³¹⁾ and the National Health Insurance (NHI) claims database (Kokuho Database [KDB]). The Tec-Tec Zushi program was implemented from August to December 2021, whereas the health insurance claims data were analyzed from April 2017 to March 2023.

As the program was designed to encourage the development of exercise habits among older adults, a control group was not established within the initial program framework. Therefore, to enable a comparative analysis, a comparison group comprising older adults residing in the same municipality during the same period as the intervention was retrospectively constructed using data from the KDB.

The intervention group consisted of older adults who voluntarily enrolled in the Tec-Tec Zushi physical activity program, were not advised by a physician to restrict their physical activity, and provided written informed consent for secondary use of their data. The pre-program questionnaire and pedometer³²⁾ were administered during a pre-implementation orientation session organized by the municipality. The post-program questionnaire was administered when participants visited the city office to report program completion and return their pedometers.

The control group comprised NHI enrollees aged ≥65 years who resided in Zushi City during the program implementation period in 2021, underwent health check-up before and after the intervention period, and had available check-up data. Individuals were excluded if they moved out of Zushi City, relocated to the city during the study period, died during the study period, or had missing data on key variables.

Two primary data sources were used: the program records, which included participant identification documents (IDs), step count logs, and responses to behavioral questionnaires administered before and after the program,^33,34) and the NHI claims database (KDB), which included demographic information and health checkup results, such as body mass index (BMI) and self-reported physical activity. For analysis, 2021 was used as a reference point; the most recent health checkup data obtained between April 2017 and March 2021 were used as pre-intervention data, and those obtained between April 2022 and March 2023 were used as post-intervention data.

These datasets were linked using unique identifiers by a certified third-party agency, and all data were anonymized prior to the analysis.

The primary outcome was the change in behavioral stage.^35-37) The TTM describes how individuals progress toward adopting healthier behaviors in five stages. The concept of “stage of change” provides a useful framework for evaluating an individual’s readiness to initiate exercise and maintain physical activity.³⁶⁾ Participants’ behavioral stages related to lifestyle improvement were assessed using a five-point scale: (1) no intention to act (pre-contemplation), (2) intention to act (contemplation), (3) preparation, (4) action, and (5) maintenance for more than 1 year. The behavioral stage was measured before and after the interventions. Stages 4 and 5 correspond to the “action” and “maintenance” stages, respectively, and are interpreted as indicating regular engagement in physical activity.

The covariates included age (year), sex, height (cm), weight (kg), BMI (kg/m²), the presence or absence of a habitual exercise routine lasting at least 30 minutes, and the pre-intervention TTM stage. For the intervention group, height and weight were measured using a bioelectrical impedance analyzer and recorded by municipal staff, whereas for the control group, these values were obtained from health checkup records.

Information on habitual exercise and the pre-intervention TTM stage was obtained from self-administered questionnaires. These variables encompassed all available data obtained by linking records from the physical activity program with health checkup information from the NHI database.

Continuous variables are expressed as medians and ranges, and categorical variables as counts and percentages. Ordinal logistic regression was applied to assess the association between program participation and the five-level TTM stage. Three models were constructed: an unadjusted model, a model adjusted for basic demographic characteristics, and a fully adjusted model that controlled for pre-intervention TTM stage and pre-intervention habitual exercise.

The TTM scale was then dichotomized into lower (Stages 1–3) and higher (Stages 4–5) stages, and binary logistic regression analyses were conducted using the same three models. Results are reported as odds ratios (ORs) and 95% confidence intervals (CIs). All statistical analyses were performed using R (version 4.3.3; R Foundation for Statistical Computing, Vienna, Austria), and statistical significance was defined as a two-sided p-value <0.05.

At the time of recruitment, municipal staff provided verbal and written explanations to prospective participants regarding the purpose of the study, the scope of data use, the voluntary nature of participation, and the right to decline participation. Only those who provided written informed consent were included in the analysis, and their program participation data were solely used for research purposes.

Data related to the physical activity program were provided to the research team through a third-party data processor, the Japan Medical Information Management Organization. All personally identifiable information, such as names and addresses, was removed prior to data transfer. The data were fully anonymized, and individual participants were not identifiable at any stage of the analysis.

In the NHI database, anonymized health checkup data were provided by a third-party organization. Although all personally identifiable information was removed, each dataset was linked using an anonymized ID, allowing researchers to distinguish between program participants and non-participants. Accordingly, the program participation status was identifiable, but the individual identities remained confidential. Data handling and storage complied fully with Japanese laws and regulations. This study was approved by the Ethics Committee of the University of Electro-Communications (Approval No. 220119).

Among the 18,101 residents aged ≥65 years living in Zushi City in 2021, 150 individuals (0.8%) voluntarily participated in the municipality-led walking promotion program (Fig. 1).

In the intervention group, 20 participants (13%) were excluded from the analysis due to missing questionnaire responses or program dropout, leaving 130 participants (87%). For the non-intervention group, we extracted data of 1,292 residents who had received health checkups during the pre- and post-intervention periods from the NHI database. In total, 1,422 individuals were included in the pre-intervention analytic dataset. Table 1 presents the baseline characteristics.

At the pre-intervention time point (January 2021), the mean age was 69.8 years in the non-intervention group and 78.0 years in the intervention group. The non-intervention group included 527 men (41%) and the intervention group included 40 men (31%). Regarding exercise habits, 673 individuals (52%) in the non-intervention group and 77 individuals (59%) in the intervention group reported engaging in at least 30 minutes of regular exercise (Table 1).

Regarding the ordinal logistic regression analyses, participants had higher odds of being at a higher stage of readiness than non-participants (OR=2.91, 95% CI 2.08–4.07) in the unadjusted model (Model 1). This association remained in Model 2 after adjusting for age, sex, and BMI (OR=2.99, 95% CI 1.96–4.58). In Model 3, event participants continued to show higher odds of being in a higher readiness stage (OR=2.98, 95% CI 1.94–4.60) (Table 2).

To further examine the association between program participation and progression toward higher behavioral readiness for lifestyle improvement, we conducted logistic regression analyses using a dichotomized TTM variable, classifying Stages 1–3 (pre-contemplation, contemplation, preparation) and Stages 4–5 (action, maintenance). In the unadjusted model, participants had significantly higher odds of being classified in Stages 4–5 compared with non-participants (OR=2.53, 95% CI 1.75–3.66). This association remained significant in the model adjusted for basic demographic characteristics and in the fully adjusted model, suggesting that program participation was consistently associated with higher odds of being in the action or maintenance stages. Taken together, these analytical results indicate that the direction of the estimates was consistent across all models, suggesting that participation in the program may be associated with higher behavioral readiness for lifestyle improvement.

Program participation was consistently associated with higher odds of being in a more advanced behavioral stage across all models. Although the magnitude of the association was modest, the direction and consistency of the findings suggest a potential beneficial influence of participation on behavioral readiness.

The participants in the intervention group were older on average than the comparison group. Since advanced age is generally associated with physical limitations and greater difficulty in adopting new behaviors,³⁸⁾ this age imbalance is likely to bias the estimated effect toward the null rather than inflate it. Thus, the fact that an association was observed even under these conditions suggests that program participation may have supported progression in the behavioral stage among community-dwelling older adults. It also indicates that the program may have been feasible and sustainable even for adults in the later stages of old age.

Additionally, the intervention group included a higher proportion of women, consistent with commonly observed trends in voluntary participation in community-based programs.³⁹⁾ The control group was composed of a greater proportion of men, resulting in a sex imbalance between the two groups. Although these demographic differences may affect comparability, the consistently higher odds of being in more advanced behavioral stages among program participants support the potential value of this community-based intervention.

Participants in the intervention group were more likely to be at higher stages of the TTM at baseline than participants in the control group, likely reflecting the characteristics of individuals who voluntarily enroll in community programs.⁴⁰⁾ However, after accounting for these initial differences, the ordinal logistic regression analyses consistently demonstrated that program participation was associated with higher odds of being in more advanced behavioral stages. Although the magnitude of these associations was modest, the consistent direction of the estimates suggests that participation may contribute to improvements in behavioral readiness for lifestyle change.

In contrast, many individuals in the control group remained in the contemplation stage (Stage 2) throughout the observation period, recognizing the need for behavioral change but not progressing toward concrete action. This stagnation aligns with previous findings that age-related declines in motivation and physical capacity can impede the transition from intention to action.³⁸⁾ Moreover, behavioral change is unlikely to occur without targeted support.⁴¹⁾ Prolonged inactivity contributes to cognitive decline and deterioration in activities of daily living (ADL) among older adults. In Japan, many older people experienced substantial reductions in daily activity due to prolonged stay-at-home recommendations during the coronavirus disease 2019 (COVID-19) pandemic.⁴²⁾ Previous studies reported elevated levels of anxiety among residents during the COVID-19 pandemic.⁴³⁾ This may have significantly increased older adults’ tendency to stay indoors, potentially leading to reduced physical activity. The timing of this program coincided with these social circumstances, which may further explain the observed stagnation in behavioral readiness in the control group. Although the distributional shift in TTM stages appeared modest, the ordinal logistic regression showed higher odds of being in advanced TTM stages among program participants. These findings suggest that even short-term, municipality-led interventions may support early behavioral shifts in older adults. This interpretation is consistent with evidence that even small increases in physical activity yield health benefits in sedentary older adults.⁵⁾ Guidelines also emphasize that reducing sedentary time and incorporating activity of any intensity contributes to better health in later life.⁴⁴⁾

Several key program components may have contributed to the observed behavioral shift: self-monitoring through step-count visualization, immediate feedback mechanisms, and social connectedness within the community. The ability to monitor one’s step count likely promotes self-awareness and self-regulation, which have been associated with enhanced self-efficacy in older adults. Previous intervention studies have reported that sending regular short message service reminders to older adults engaged in exercise programs improves attitudes toward physical activity and promotes participation.⁴⁴⁾ Similarly, feedback such as verbal encouragement from municipal staff and incentive-based rewards (e.g., coupons linked to step counts) may have served as positive reinforcement, thereby enhancing motivation for continued engagement.⁴⁵⁾ Additionally, shared participation among residents may have fostered a sense of social connectedness, mitigating feelings of isolation and promoting sustained adherence to the program. Given that limited access to exercise facilities and social isolation are well-documented barriers to physical activity in this population, the program’s accessible, community-based approach may have helped overcome these obstacles.^24,25) These elements likely worked synergistically to support participants’ progression from contemplation and preparation to action, underscoring the potential of short-term, low-cost interventions to stimulate early behavioral changes among older adults.

The findings of this study indicate that short-term, low-cost, municipality-led interventions can effectively promote early-stage behavioral changes among older adults, providing valuable insights for community health promotion and long-term care prevention strategies. Previous research on physical activity has predominantly focused on long-term outcomes such as mortality, fractures, and long-term care certification.⁴⁶⁾ However, this study focuses on behavioral change as an intermediate outcome that precedes these endpoints. This study elucidates the mechanisms through which physical activity yields health benefits and highlights the importance of promoting lifestyle modifications before clinical deterioration occurs.

These findings suggest the importance of municipality-led exercise programs on public health. Prior studies show that behavioral change functions as an important pathway linking community interventions with reduced health complications.^29,47) Sustained increases in physical activity can reduce the risk of chronic diseases, functional decline, and the need for long-term care.^6,48) Therefore, demonstrating that municipality-led programs can promote health-related behavioral change suggests that these initiatives may effectively improve population health at the community level.

Additionally, although participants in this study tended to be at a relatively advanced stage in their readiness to change their behavior, the results remained similar after adjusting for baseline behavioral stage. This suggests that the intervention could benefit individuals with lower readiness for behavioral change. If participation among individuals with lower levels of interest in physical activity could be increased, the potential population-level health impact might be even greater.³⁰⁾ Individuals with higher behavioral readiness may be able to maintain physical activity independently, whereas those with lower readiness may be less responsive to such programs; therefore, stage-tailored motivational approaches and intervention designs may be necessary.⁴⁹⁾ These results imply that encouraging participation among residents in earlier stages of behavioral readiness through alternative outreach or engagement strategies could enable a more inclusive, population-based approach and enhance the effectiveness of municipality-led health promotion programs.

This study contributes to the relatively limited body of literature that rigorously evaluates the effectiveness of municipality-led health initiatives. Although community-level interventions have proliferated in recent years, systematic reviews have noted a scarcity of peer-reviewed evaluations of such programs. This study attempted to evaluate a community-implemented, non-structured program in a comparable manner by linking administrative program data with health checkup records.⁵⁰⁾ The methodological approach employed in this study can serve as a practical tool for other municipalities seeking to evaluate the effectiveness of their health interventions.

The program centered on walking demonstrated potential health benefits. Accumulating evidence suggests that increased step counts are associated with a reduced risk of lifestyle-related diseases, fractures, and long-term care dependency, even in the absence of complex or high-intensity exercise regimens. This supports the feasibility and long-term sustainability of walking-based interventions for promoting healthy aging.

The program encouraged self-management among participants by minimizing their reliance on professional instructors and structured sessions. This characteristic enhances both the scalability and sustainability of the intervention, making it feasible within the resource constraints commonly faced by municipal governments. The combination of potential cost-effectiveness and promotion of autonomous health-behavior change underscores the practical utility of community-based walking programs and supports further evaluation within broader policy and research frameworks.

This study has some limitations. Although data were obtained from the NHI database and program records, only a small set of standardized variables—age, sex, BMI, and baseline exercise habits—could be included in the adjusted models. Important factors that should ideally be considered, such as ADL, general health status, and chronic conditions, were not available. Additionally, commonly examined socioeconomic variables—including household income, living arrangements, and educational background—could not be incorporated. These data limitations may have introduced residual confounding; therefore, the findings should be interpreted with caution. However, because they lived in the same area during the same period and regularly underwent health checkups at relatively frequent intervals, their living circumstances were likely broadly similar to those of the program participants.

The intervention group included a higher proportion of older-old adults and women, indicating potential baseline differences between groups that may have influenced the estimated effects. To improve comparability, future studies may consider examining more homogeneous populations, such as individuals with similar motivations for participating in exercise programs. However, because municipal programs are generally implemented for the broader older adult population, constructing a strictly comparable control group is inherently challenging. This issue has been highlighted in prior studies, and improving methods for identifying and constructing comparable populations remains an important direction for future research.

The primary outcome was based on self-administered questionnaires, which may have been influenced by respondents’ emotional states or the context in which the survey was completed. Social desirability bias may have led to overreporting of favorable behaviors. Additionally, the baseline survey was administered immediately after the program orientation, and the follow-up survey was completed when returning the pedometer in the intervention group—situations in which participants may have been more motivated. Contrarily, the control group completed surveys during routine health checkups, without exposure to a program context.

The intervention period was limited to 5 months, making it impossible to evaluate the long-term sustainability of behavioral changes. Although short-term improvements were observed, it remains unclear whether these changes persisted over time or led to downstream outcomes such as reduced fall incidence or lower rates of long-term care certification. Longitudinal studies are needed to determine the durability and broader health implications of these behavioral changes.

In this study, 13.3% of participants did not complete the questionnaire. Notably, 13 of these individuals did not respond at all, suggesting they may have declined to participate or have withdrawn during the intervention period. Previous studies of municipal programs have reported high dropout rates (22%–76%), yet factors influencing participation, adherence, and withdrawal remain insufficiently examined.⁵⁰⁾

In conclusion, our findings provide a valuable foundation for the development of future long-term care prevention strategies and community health initiatives. Short-term, low-cost community-based interventions, such as this walking program, may be practical for encouraging lifestyle modification among older adults. However, future studies should incorporate a broader set of influential factors and a longer follow-up period to determine whether the observed behavioral changes are sustained.