This study is a secondary analysis of data from the 2024 National Survey on Sports Participation for Persons with Disabilities (Approval No. 113020), a nationally approved statistical survey conducted in South Korea. This cross-sectional study used individual respondents as the unit of analysis. Sampling weights were applied to produce population-representative estimates. Because the source survey operationalized participation through a survey-defined concept of exercise, the term “physical activity” in this study refers to exercise participation captured by the survey rather than to all bodily movement encompassed by the broader public-health definition of physical activity. The dataset did not include any personally identifiable information, and no direct recruitment of participants or collection of biological samples was involved. Accordingly, this study did not require Institutional Review Board (IRB) review.

The study included adults aged ≥60 years with disabilities identified in the survey (SQ2). In the publicly available source dataset, respondents aged 60 years or older, including those aged ≥70 years, were combined into a single age category. Therefore, further age stratification within the analytic sample was not possible. Among them, individuals who reported no exercise participation during the previous year based on the survey participation screening item (C1) were selected, resulting in a final sample of 1,865 participants. The screening question asked, “During the past year (September 2023 to August 2024), have you exercised?” Respondents were instructed to regard movements undertaken for physical conditioning or health as exercise. Walking solely for commuting or transportation was excluded, whereas intentional activities such as using stairs or walking one or two transit stops for exercise purposes were included. Accordingly, non-participation in this study refers specifically to non-participation under this survey-based operational definition and does not indicate the absence of all forms of bodily movement or incidental daily activity. Disability type (HSQ4) was categorized into six groups: physical, visual, hearing and speech, developmental, brain lesion, and internal/health disabilities. Missing values were minimal (<5%), and no cases were excluded after listwise deletion. The analytic sample for regression remained at 1,865 participants. Analysis of preferred participation modes was restricted to 890 respondents with participation intention, and cells with zero observations were treated as not estimable.

Sociodemographic and disability-related characteristics included sex (SQ1), region of residence (SQ3), disability severity (SQ5), disability onset (SQ6), ambulation (SQ7), employment status (DQ4), and household income (DQ5). Region of residence was recoded into metropolitan areas (Seoul, Incheon, and Gyeonggi Province) and non-metropolitan area. Household income was originally measured in six categories and was recoded into four categories by combining the upper three categories into a single group (≥3,000,000 Korean won [KRW]/month). Ambulation was recoded into two categories: independent ambulation and assisted ambulation, with the assisted ambulation category including respondents requiring partial or full assistance for walking.

Barriers to participation were assessed using item D1, which asked respondents to select a single primary reason for non-participation. Because only one response could be selected, this measure represents the respondent’s primary reported barrier and does not capture co-occurring barriers or interactions among multiple barriers within an individual. Preferred participation mode was assessed using item D2, which asked, “If you were to start exercising, how would you like to exercise?” Response options were facility-based individual exercise, individual home training, lessons or classes, sports club participation, outdoor walking or strolling in nearby areas such as parks or hiking trails, and other modes. Participation intention was assessed using item D3, which asked whether respondents intended to start exercising. The four response options were intention to exercise within 1 month, intention to exercise within 6 months, intention to exercise someday but not within 6 months, and no current intention to exercise. For the regression analysis, the first three categories were combined as “intention” (1), whereas “no current intention to exercise” was coded as “no intention” (0). Barriers, participation intention, and preferred participation modes were therefore treated as separately measured but complementary aspects of non-participation. No integrated causal structure or interaction among these domains was assumed or statistically tested.

All statistical analyses were conducted using IBM SPSS Statistics version 28.0 (IBM Corp., Armonk, NY, USA). Because the dataset did not provide stratification or cluster variables, analyses were conducted using sampling weights only. Weighted estimates were used to reflect population-representative distributions.

First, weighted cross-tabulations were used to describe sociodemographic and disability-related characteristics and barriers, and differences across disability types were tested using the Rao–Scott adjusted chi-square test (reported as an F statistic).

Second, to identify factors associated with the intention to participate in exercise, a weighted logistic regression analysis was conducted among the 1,865 respondents who reported not participating in physical activity during the past year. Logistic regression analysis was used because the dependent variable was binary, allowing for the estimation of the odds of reporting an intention to participate in exercise while controlling for multiple sociodemographic and disability-related factors. Independent variables included sex, region of residence, disability severity, disability onset, walking ability, employment status, household income, and disability type. Age could not be included as a covariate because the source dataset did not provide further age information within the single category of respondents aged ≥60 years. Physical disability was used as the reference group because it represented the largest subgroup in the sample. Multicollinearity among independent variables was assessed separately using variance inflation factors (VIFs) from an equivalent non-complex diagnostic procedure.

Finally, preferred participation modes were analyzed using weighted cross-tabulations and Rao–Scott tests. Results are presented as estimated population percentages (EP%) with 95% confidence intervals (CI), and statistical significance was set at p<0.05.

Table 1 presents the sociodemographic and disability-related characteristics of 1,865 older adults with disabilities who reported no exercise participation under the survey-based definition during the previous year, stratified by disability type. Regarding sex, males accounted for a larger proportion of the sample than females, representing 60.2% of non-participating older adults with disabilities. The distribution differed significantly across disability types (Rao–Scott F=3.59, p=0.021). Region of residence also differed significantly across disability groups (Rao–Scott F=5.08, p=0.004), with a larger proportion of respondents residing in non-metropolitan areas. Disability severity showed significant differences according to disability type (Rao–Scott F=77.12, p<0.001). Individuals with brain lesion disabilities showed a relatively higher proportion of severe disability. Disability onset also differed significantly across disability types (Rao–Scott F=34.66, p<0.001), with most respondents reporting acquired disabilities. Ambulation status differed significantly across disability groups (Rao–Scott F=31.67, p<0.001), with a higher proportion of assisted walking observed among individuals with brain lesion disabilities. Employment status also differed significantly according to disability type (Rao–Scott F=20.10, p<0.001), with most respondents reporting no employment. Household income showed significant differences across disability groups (Rao–Scott F=2.59, p=0.011).

Table 2 presents the barriers to exercise participation among older adults with disabilities who did not participate in exercise during the previous year, according to disability type. Because the survey captured a single primary reason for non-participation, the percentages represent primary reported barriers rather than the prevalence of all barriers experienced by respondents. The most frequently reported barriers were lack of motivation (27.1%), poor health condition (23.9%), and lack of time to exercise (11.5%). In addition, having no one to assist with exercise (8.3%) and severity of disability (8.3%) were reported at similar levels, followed by not knowing how to exercise (7.2%).

Several barriers differed significantly across disability types, including lack of motivation (Rao–Scott F=3.93, p=0.014), lack of time to exercise (Rao–Scott F=10.34, p<0.001), poor health condition (Rao–Scott F=7.21, p<0.001), disability severity (Rao–Scott F=8.49, p<0.001), and financial constraints (Rao–Scott F=2.62, p=0.033). The hearing and speech disability group showed the highest estimated proportion of lack of motivation as the primary reported barrier (40.0%).

In contrast, other barriers including lack of exercise programs, lack of sports instructors, lack of sports facilities, lack of exercise information, concerns about infectious diseases, not knowing how to exercise, and having no one to assist with exercise did not show statistically significant differences across disability types.

Before conducting the regression analysis, multicollinearity among the independent variables was examined using VIF. The VIF values ranged from 1.026 to 1.512, indicating no evidence of multicollinearity.

To identify factors associated with the intention to participate in exercise, a weighted logistic regression analysis was conducted among 1,865 older adults with disabilities who reported no exercise participation during the previous year. The overall model was statistically significant (Wald F=7.82, p<0.001). The dependent variable was the intention to participate in exercise (yes=890, no=975; total analytic sample=1,865).

Disability severity was significantly associated with participation intention (p=0.003). Individuals with severe disabilities had lower odds of reporting participation intention than those with mild disabilities (odds ratio [OR]=0.62, 95% CI 0.46–0.85). Employment status was also significantly associated with participation intention (p=0.011).

Employed respondents had higher odds of reporting participation intention than unemployed respondents (OR=1.44, 95% CI 1.09–1.91). Household income was not significantly associated with participation intention across all income categories (all p>0.05).

Disability type showed a statistically significant overall effect (p=0.019). Compared with individuals with physical disabilities (reference group), those with hearing and speech disabilities had higher odds of reporting participation intention (OR=1.43, 95% CI 1.08–1.89), whereas other disability types did not show significant differences. Sex, region of residence, disability onset, walking ability were not significantly associated with participation intention. Table 3 presents the results of the logistic regression analysis.

Table 4 presents the preferred modes of exercise participation among older adults with disabilities who reported an intention to participate in exercise. Outdoor walking was the most frequently preferred participation mode (57.5%), followed by facility-based individual exercise (15.4%), home-based exercise (13.6%), and structured exercise programs (11.8%). Relatively small proportions of respondents reported a preference for sports club membership (1.4%) or other forms of participation (0.2%). When preferred participation modes were compared across disability types, no statistically significant differences were observed (Rao–Scott F=1.705, p=0.060).

Among older adults with disabilities reporting no exercise participation under the survey-based definition, lack of motivation was the most frequent primary barrier (27.1%), followed by poor health condition (23.9%) and lack of time to exercise (11.5%). Having no one to assist with exercise and severe disability as a barrier to participation were each reported by 8.3%, followed by not knowing how to exercise (7.2%). Because respondents selected only one primary reason, these percentages represent the distribution of primary reported barriers rather than the prevalence of all barriers experienced by respondents.

The prominence of both motivational and health-related barriers indicates that non-participation cannot be characterized by a single dominant type of constraint. This pattern is broadly consistent with previous research identifying physical limitations, lack of time, accessibility, social support, and motivation as relevant barriers to exercise or physical activity participation among persons with disabilities.^13,14,16) However, the present data do not establish how these different constraints combine within individuals.

Importantly, the distribution of barriers differed across disability types, indicating that reported patterns of non-participation were not uniform across disability groups. For example, lack of motivation as the primary reported barrier was particularly prominent among individuals with hearing and speech disabilities (40.0%), whereas poor health condition was more pronounced among those with internal/health disabilities (36.0%) and brain lesion disabilities (26.8%). In addition, severity of disability as the primary reported barrier was more frequent among individuals with brain lesion disabilities (18.9%). These differences highlight disability-type heterogeneity in the primary reasons reported for non-participation, although they do not by themselves identify the mechanisms producing these patterns.

Primary reported barriers should also be interpreted in light of differences in functional context. Previous studies have highlighted health and functional factors as important determinants of mobility in older populations.^24,25) Nevertheless, because only one barrier could be selected, the present study cannot establish whether multiple barriers co-occurred or interacted within individuals. Similarly, although contextual conditions such as accessibility, mobility, and social support have been identified as relevant participation constraints,^13,14,26) the present analyses did not directly test whether such conditions contributed to motivational or other reported barriers.

Barriers such as lack of time (11.5%) and financial constraints (2.5%) further indicate that some respondents identified circumstances beyond health or disability severity as their primary reason for non-participation. Previous studies have identified environmental and transportation-related barriers as relevant constraints on physical activity and social participation among persons with mobility disabilities.^27,28) However, these contextual factors were not comprehensively assessed in the present study and should not be interpreted as direct causes of non-participation. Overall, the findings indicate heterogeneity in primary reported barriers rather than a multidimensional or interactive structure among barriers.

Disability severity and employment status were significantly associated with participation intention. Individuals with severe disabilities had lower odds of reporting intention to participate in exercise (OR=0.62, 95% CI 0.46–0.85), whereas employed respondents had higher odds of reporting such intention (OR=1.44, 95% CI 1.09–1.91). Individuals with hearing and speech disabilities also had higher odds of participation intention than those with physical disabilities (OR=1.43, 95% CI 1.08–1.89). These associations should be interpreted within the cross-sectional design and do not establish causal or temporal relationships.

The negative association between disability severity and participation intention may reflect greater anticipated difficulty in initiating exercise, although perceived feasibility was not directly measured. This interpretation is broadly consistent with previous research linking disability severity, ambulation, and assistive-device-related functional conditions with participation characteristics among persons with disabilities.^22,24,25)

In contrast, the positive association between employment and participation intention may be related to broader differences in daily routines, social participation, or functional circumstances. Previous research has identified employment status, economic hardship, and social relationships as factors associated with leisure-time physical activity among middle-aged and older adults with disabilities.^16,26) However, these mechanisms were not directly assessed, and employment itself should not be interpreted as causally increasing intention.

The hearing and speech disability group showed a particularly notable pattern: this group had the highest proportion reporting lack of motivation as the primary barrier (40.0%), while also showing higher odds of participation intention than the physical disability group. These findings should not be interpreted as an individual-level contradiction because the analyses do not establish that the same respondents who selected lack of motivation also reported future participation intention. Rather, the coexistence of these group-level patterns may indicate heterogeneity within the hearing and speech disability group and suggest that a single primary barrier response may not fully capture differences in readiness or future-oriented intention.

In addition, caution is needed when interpreting intention as a direct proxy for future behavior. Previous evidence has emphasized that exercise intention is an important predictor of exercise behavior, but that an “intention–behavior gap” remains a persistent issue.²³⁾ In this respect, the present findings help identify characteristics associated with retaining an intention to exercise despite current non-participation, but they do not, by themselves, demonstrate that such intention will necessarily translate into actual participation.

In the present study, outdoor walking was the most preferred mode of exercise among older adults with disabilities who expressed an intention to participate (57.5%), followed by facility-based individual exercise (15.4%), home-based exercise (13.6%), and structured exercise programs (11.8%). The predominance of walking may reflect a preference for a form of exercise that can be undertaken outside formal exercise facilities and structured programs. However, preference alone does not establish actual accessibility or feasibility. This finding is consistent with previous research indicating that walking is a feasible and sustainable form of physical activity for older adults, particularly in environments that support accessibility and walkability.²⁹⁾

No statistically significant differences in preferred participation modes were observed across disability types (p=0.060). Thus, the present data do not support disability-type differences in preferred participation modes. In contrast to the significant variation observed for several primary barriers, outdoor walking was the most frequently preferred option overall among respondents reporting participation intention.

When considered together, the separately measured findings also suggest that preference should not be equated with feasibility. Although outdoor walking was the most frequently preferred mode among respondents with participation intention, poor health condition, disability severity, and lack of someone to assist with exercise were also reported as primary barriers in the broader non-participating sample, and severe disability was associated with lower participation intention. These findings do not demonstrate direct relationships between barriers, intention, and preferred modes because such relationships were not statistically tested. However, they indicate that the strong overall preference for walking should be interpreted alongside heterogeneity in health, functional, and support-related circumstances.

From a policy perspective, efforts to promote exercise participation among older adults with disabilities may benefit from considering both shared preferences and heterogeneity in reported barriers. The strong preference for outdoor walking suggests that community-based options may be considered alongside facility-based opportunities, while disability-type differences in several primary barriers indicate the potential value of identifying salient constraints when planning outreach or participation support. The lower odds of participation intention among individuals with severe disabilities further suggest that general promotional approaches may not address all participation-related conditions. However, because the study did not test specific interventions, environmental modifications, assistance services, or disability-tailored programs, these implications should be regarded as considerations for planning and future evaluation rather than evidence for particular policy interventions.

Several limitations should be considered. First, the cross-sectional design precludes causal inference and temporal ordering. Second, self-reported data may be subject to recall or reporting bias. Third, heterogeneity within disability categories may not have been fully captured. Individuals within the same category may differ in the timing and duration of disability, including ageing with long-standing disability and disability acquired later in life, but these life-course trajectories could not be examined.^6,7) The publicly available dataset did not provide further age differentiation among respondents aged ≥60 years, including those aged ≥70 years. Consequently, within-group age differences could not be described or adjusted for, and residual confounding by age cannot be excluded. Fourth, barriers were measured using a single primary response. The study therefore could not determine whether multiple barriers co-occurred, how they were combined, or whether interactions were present. Fifth, participation intention does not necessarily translate into actual behavior, and preferred participation modes were assessed only among respondents reporting intention. These preferences should therefore not be generalized to all non-participating older adults with disabilities.

Future research should use longitudinal designs to clarify temporal relationships and examine changes in participation over time. Multiple-response assessments may better characterize co-occurring barriers, while more detailed measures of age, timing and duration of disability, functional characteristics, support needs, and environmental context could further clarify heterogeneity. Future studies should also directly test individual-level relationships among barriers, participation intention, and preferred modes.