- Split View
-
Views
-
Cite
Cite
Grace Koh, Michikazu Sekine, Masaaki Yamada, Yuko Fujimura, Takashi Tatsuse, Neighbourhood walkability and obesity among adults in rural Japan: results from a Japanese health database, Journal of Public Health, Volume 44, Issue 4, December 2022, Pages e467–e474, https://doi.org/10.1093/pubmed/fdab319
- Share Icon Share
Abstract
Walkable neighbourhoods promote physical activity and prevent obesity, but there is limited evidence to inform urban planning strategies for public health within the context of rural Japan. This study describes associations between neighbourhood walkability and obesity in Toyama, a regional municipality in Japan.
A cross-sectional analysis of the Toyama Prefecture National Health Insurance data (n = 3454) in 2016 using Analysis of Covariance (ANCOVA) and binary logistic regression. Walk Score® was used to estimate neighbourhood walkability.
Residents from highly walkable neighbourhoods generally had lower mean body mass index (BMI), but significant associations between neighbourhood walkability and BMI and prevalence of obesity were only observed in women (adjusted OR: 0.46, 95% CI: 0.26–0.80). Men below 65 years old had higher obesity prevalence (adjusted OR: 1.76, 95% CI: 1.34–2.30). Daily alcohol consumption was associated with lower odds of being obese among men (adjusted OR: 0.72, 95% CI: 0.55–0.95). Hypertension, diabetes mellitus and dyslipidaemia were associated with higher obesity prevalence among residents, regardless of gender.
Walkable environment may improve health outcomes for rural communities in Japan. Further studies are required to create equitable and inclusive living spaces for men and women to access healthier lifestyle choices.
Introduction
Obesity affects the health and well-being of millions of people worldwide. Although it is largely preventable, obesity still remains a challenging public health issue for governments across the globe.1 The last two decades saw an expansion of knowledge in the role of built environment, especially walkability, in preventing obesity and chronic diseases. Built environment is broadly defined as:2 ‘land use patterns, the transportation system, and design features that together provide opportunities for travel and physical activity.’ A physically active lifestyle is essential for good health. Residential environments which favour automobile use may diminish the need for walking or active transportation, giving rise to physical inactivity that could lead to weight gain.3–5 Implementing policy intervention, which creates walkable neighbourhoods (i.e. pedestrian-friendly residential environment), could promote physical activity and prevent obesity.6,7
Current evidence has established that neighbourhoods with high walkability scores are linked to lower levels of obesity and cardiometabolic diseases.8–12 However, a growing body of literature suggests that physical activity inequality and health disparity exist between sexes, especially those residing in rural communities. Women living in rural areas often have problems accessing walkable, safe and affordable spaces to be physically active.13 Additionally, Althoff et al.14 reported a worldwide ‘gender step gap’ phenomenon which demonstrated that: (i) a high level of physical activity inequality between men and women results in a rapid increase of obesity prevalence among women, and (ii) a country’s gender gap in physical activity is a better predictor of obesity. One way of closing this gap is through urban planning policies that improve cities’ walkability for male and female residents alike.13,14
The relationship between neighbourhood walkability (a measure of how well a neighbourhood facilitates walking) and population health has been studied mostly within Western contexts. Evidence relating to the role of built environment in health promotion within Asian context is developing at an early stage. Recent studies from Japan reported that adults and the elderly living in walkable neighbourhoods have lower body mass index (BMI).15,16 But information concerning neighbourhood walkability and health status of men and women living in rural Japan remains scarce.
We therefore aimed to contribute information to the body of evidence through our observation of Toyama, a regional municipality in Japan. Toyama has a land area of 1242 km2 and a population of ~420 000. Almost 70% of the municipality area are forest and agricultural lands, rendering car-dependence for transportation.17 We sought to understand (i) whether residents from more walkable neighbourhoods have lower BMI, and (ii) whether there is gender difference in the obesity prevalence associated with walkability of residential areas, age, presence of comorbidities and lifestyle factors.
Methods
Study population
This study was an observational cross-sectional analysis of health data from the Toyama Prefecture National Health Insurance (TPNHI) database. The TPNHI cohort consists of Toyama Prefecture residents who are insured under the Japanese National Health Insurance (NHI) policy. As eligibility criteria to quality for NHI benefits, insurees are under 75, unemployed or self-employed.18 This analysis pertains to 3454 Toyama residents who completed the Japanese Specific Health Check and Guidance System (SHC) in the year 2016. Designed for NHI insurees, the SHC is a strategy initiated by the Japanese Government in 2008 for early detection and treatment of metabolic syndrome.19 In this programme, participants undergo a health screening and answer a self-administered questionnaire, which include medical history and lifestyle habits. The 2016 SHC (response rate = 33.7%) datasets were anonymized and transferred to the Department of Epidemiology and Health Policy, University of Toyama. Participants without missing data were included in this study. Residents who had missing health and lifestyle information were excluded. The Ethics Committee of University of Toyama granted ethical approval for this study (approval number: R2019107).
Neighbourhood walkability
The walkability of the participants’ residential neighbourhood was estimated using Walk Score®, a publicly available algorithm which calculates neighbourhood walkability based on aspects of pedestrian friendliness and distance to amenities of an address within a 1-mile radius. Walk Score® first examines hundreds of walking routes to nine categories of amenities (parks, shopping, grocery stores, restaurant, bank services, coffee shops, schools, bookstores and entertainment) from an address. Maximum points are awarded to amenities within a 5-minute walk. For more distant amenities, Walk Score® uses a decay function for scoring walkability. Pedestrian friendliness is measured through road metrics (e.g. block length, intersection density) and population density analysis. A higher Walk Score indicates better walkability and it is divided into four categories: ‘Car-Dependent’ (Walk Score: 0–49), ‘Somewhat Walkable’ (50–69), ‘Very Walkable’ (70–89) and ‘Walker’s Paradise’ (90–100).20 The use of Walk Score® in Japan has been validated.21
Participants’ exact address was not available in order to maintain confidentiality. Therefore, the smallest administrative unit available—the chocho-aza—was used to determine the Walk Score. The 2010 population census of Japan defined chocho-aza as the objective unit of neighbourhood in Japan22 and this definition has been applied in epidemiological studies.23 Each participant was assigned a walkability score according to the latitude and longitude of the centroid of their chocho-aza.
Study variables
Participants’ age was divided into two categories: residents who were below the age of 65 (<65) and residents who were at least 65 years old (≥65). Sex was coded as a binary variable.
Lifestyle habits that may be related to unhealthy weight were assessed: current smoking (yes/no), alcohol consumption (daily, sometimes, rarely) and levels of physical activity (active/inactive). Physical activity was measured through these two questions: ‘Are you in the habit of exercising to sweat lightly for over 30 minutes, twice weekly, for over a year?’ and ‘In your daily life, do you walk and/or maintain physically active for more than one hour in a day?’ Participants who indicated ‘yes’ to at least one question were identified as physically active.
We referred to established clinical practice guidelines24–27 to determine the presence or absence of obesity and its associated comorbidities. A BMI of ≥25 kg/m2 was used to identify individuals with obesity. Hypertension was diagnosed through medical history, current intake of antihypertensive medications, measured systolic blood pressure was ≥140 mmHg or diastolic blood pressure was ≥90 mmHg. Diabetes mellitus was identified via medical history, current intake of antidiabetic medications, HbA1c of at least 6.5% or fasting plasma glucose of at least 126 mg/dl. Participants with dyslipidaemia were diagnosed through their medical history, current intake of lipid-lowering medications, triglyceride level was ≥150 mg/dl, low-density lipoprotein cholesterol was ≥140 mg/dl and/or high-density lipoprotein cholesterol was <40 mg/dl. These were coded as binary variables representing absence or presence of the condition.
Statistical analysis
Using the independent sample t-test, we compared the mean age and BMI of participants who were included and excluded from this study. This was carried out to examine whether there were differences between those who provided complete data for all variables and those who did not provide complete data for this study. We performed the Chi-square test to compare categorical data. Analysis of Covariance (ANCOVA) was used to compare mean BMI of residents between Walk Score categories while controlling for their age, comorbidities and lifestyle habits. Post-hoc Bonferroni test was conducted on observed pairwise differences from the ANCOVA analysis.
Binary logistic regression analysis was conducted to evaluate association of obesity prevalence with Walk Score categories between genders, adjusting for age, comorbidities and lifestyle habits. Unadjusted and adjusted odds ratio (OR) with 95% confidence interval (CI) from the logistic regressions are reported. All analyses were stratified by sex. Statistical analyses were conducted with SPSS version 25.0 (SPSS, Chicago, IL, USA) and statistical significance was set at a two-sided P-value of <0.05.
Results
There were significant differences in mean age and BMI between respondents who were included and excluded in this study (P < 0.05). The respondents included in this study were slightly younger (mean age = 65.7 ± 7.8 years) and had slightly lower BMI (mean BMI = 22.8 ± 3.2 kg/m2) compared with respondents excluded from this study (mean age = 67.3 ± 7.1 years, mean BMI = 22.9 ± 3.4 kg/m2).
A high proportion of residents was at least 65 years old (71.6%), and there were more men than women who were obese (28.1 versus 18.0%) in Toyama (Table 1). More men than women had comorbidities such as hypertension and diabetes mellitus but a higher proportion of women had dyslipidaemia. The percentage of male smokers was ~3.6 times higher than female smokers (15.4 versus 4.3%). Almost half (49.6%) of the male sample consumed alcohol daily, while 65.7% of females reported that they rarely drink. A lower proportion of women was physically active.
. | Total (N = 3454) . | Men (n = 1612) . | Women (n = 1842) . | P-value χ2 . | |||
---|---|---|---|---|---|---|---|
. | n . | % . | n . | % . | n . | % . | |
Age (years) | *** | ||||||
<65 | 980 | 28.4 | 374 | 23.2 | 606 | 32.9 | |
≥65 | 2474 | 71.6 | 1238 | 76.8 | 1236 | 67.1 | |
Obese | *** | ||||||
Yes | 785 | 22.7 | 453 | 28.1 | 332 | 18.0 | |
No | 2269 | 77.3 | 1159 | 71.9 | 1510 | 82.0 | |
Hypertension | *** | ||||||
Yes | 1317 | 38.1 | 725 | 45.0 | 592 | 32.1 | |
No | 2137 | 61.9 | 887 | 55.0 | 1250 | 67.9 | |
Diabetes | *** | ||||||
Yes | 377 | 10.9 | 236 | 14.6 | 141 | 7.7 | |
No | 3077 | 89.1 | 1376 | 85.4 | 1701 | 92.3 | |
Dyslipidaemia | ** | ||||||
Yes | 1955 | 56.6 | 869 | 53.9 | 1086 | 59.0 | |
No | 1499 | 43.4 | 743 | 46.1 | 756 | 41.0 | |
Smoking (current smoker) | *** | ||||||
Yes | 328 | 9.5 | 248 | 15.4 | 80 | 4.3 | |
No | 3126 | 90.5 | 1364 | 84.6 | 1762 | 95.7 | |
Alcohol consumption | *** | ||||||
Daily | 1036 | 30.0 | 800 | 49.6 | 236 | 12.8 | |
Sometimes | 777 | 22.5 | 382 | 23.7 | 395 | 21.4 | |
Rarely | 1641 | 47.5 | 430 | 26.7 | 1211 | 65.7 | |
Physical activity | ** | ||||||
Active | 1996 | 57.8 | 978 | 60.7 | 1018 | 55.3 | |
Inactive | 1458 | 42.2 | 634 | 39.3 | 824 | 44.7 |
. | Total (N = 3454) . | Men (n = 1612) . | Women (n = 1842) . | P-value χ2 . | |||
---|---|---|---|---|---|---|---|
. | n . | % . | n . | % . | n . | % . | |
Age (years) | *** | ||||||
<65 | 980 | 28.4 | 374 | 23.2 | 606 | 32.9 | |
≥65 | 2474 | 71.6 | 1238 | 76.8 | 1236 | 67.1 | |
Obese | *** | ||||||
Yes | 785 | 22.7 | 453 | 28.1 | 332 | 18.0 | |
No | 2269 | 77.3 | 1159 | 71.9 | 1510 | 82.0 | |
Hypertension | *** | ||||||
Yes | 1317 | 38.1 | 725 | 45.0 | 592 | 32.1 | |
No | 2137 | 61.9 | 887 | 55.0 | 1250 | 67.9 | |
Diabetes | *** | ||||||
Yes | 377 | 10.9 | 236 | 14.6 | 141 | 7.7 | |
No | 3077 | 89.1 | 1376 | 85.4 | 1701 | 92.3 | |
Dyslipidaemia | ** | ||||||
Yes | 1955 | 56.6 | 869 | 53.9 | 1086 | 59.0 | |
No | 1499 | 43.4 | 743 | 46.1 | 756 | 41.0 | |
Smoking (current smoker) | *** | ||||||
Yes | 328 | 9.5 | 248 | 15.4 | 80 | 4.3 | |
No | 3126 | 90.5 | 1364 | 84.6 | 1762 | 95.7 | |
Alcohol consumption | *** | ||||||
Daily | 1036 | 30.0 | 800 | 49.6 | 236 | 12.8 | |
Sometimes | 777 | 22.5 | 382 | 23.7 | 395 | 21.4 | |
Rarely | 1641 | 47.5 | 430 | 26.7 | 1211 | 65.7 | |
Physical activity | ** | ||||||
Active | 1996 | 57.8 | 978 | 60.7 | 1018 | 55.3 | |
Inactive | 1458 | 42.2 | 634 | 39.3 | 824 | 44.7 |
Abbreviations: χ2 Chi square test for difference in distribution by gender.
*P < 0.05, **P < 0.01, ***P < 0.001.
. | Total (N = 3454) . | Men (n = 1612) . | Women (n = 1842) . | P-value χ2 . | |||
---|---|---|---|---|---|---|---|
. | n . | % . | n . | % . | n . | % . | |
Age (years) | *** | ||||||
<65 | 980 | 28.4 | 374 | 23.2 | 606 | 32.9 | |
≥65 | 2474 | 71.6 | 1238 | 76.8 | 1236 | 67.1 | |
Obese | *** | ||||||
Yes | 785 | 22.7 | 453 | 28.1 | 332 | 18.0 | |
No | 2269 | 77.3 | 1159 | 71.9 | 1510 | 82.0 | |
Hypertension | *** | ||||||
Yes | 1317 | 38.1 | 725 | 45.0 | 592 | 32.1 | |
No | 2137 | 61.9 | 887 | 55.0 | 1250 | 67.9 | |
Diabetes | *** | ||||||
Yes | 377 | 10.9 | 236 | 14.6 | 141 | 7.7 | |
No | 3077 | 89.1 | 1376 | 85.4 | 1701 | 92.3 | |
Dyslipidaemia | ** | ||||||
Yes | 1955 | 56.6 | 869 | 53.9 | 1086 | 59.0 | |
No | 1499 | 43.4 | 743 | 46.1 | 756 | 41.0 | |
Smoking (current smoker) | *** | ||||||
Yes | 328 | 9.5 | 248 | 15.4 | 80 | 4.3 | |
No | 3126 | 90.5 | 1364 | 84.6 | 1762 | 95.7 | |
Alcohol consumption | *** | ||||||
Daily | 1036 | 30.0 | 800 | 49.6 | 236 | 12.8 | |
Sometimes | 777 | 22.5 | 382 | 23.7 | 395 | 21.4 | |
Rarely | 1641 | 47.5 | 430 | 26.7 | 1211 | 65.7 | |
Physical activity | ** | ||||||
Active | 1996 | 57.8 | 978 | 60.7 | 1018 | 55.3 | |
Inactive | 1458 | 42.2 | 634 | 39.3 | 824 | 44.7 |
. | Total (N = 3454) . | Men (n = 1612) . | Women (n = 1842) . | P-value χ2 . | |||
---|---|---|---|---|---|---|---|
. | n . | % . | n . | % . | n . | % . | |
Age (years) | *** | ||||||
<65 | 980 | 28.4 | 374 | 23.2 | 606 | 32.9 | |
≥65 | 2474 | 71.6 | 1238 | 76.8 | 1236 | 67.1 | |
Obese | *** | ||||||
Yes | 785 | 22.7 | 453 | 28.1 | 332 | 18.0 | |
No | 2269 | 77.3 | 1159 | 71.9 | 1510 | 82.0 | |
Hypertension | *** | ||||||
Yes | 1317 | 38.1 | 725 | 45.0 | 592 | 32.1 | |
No | 2137 | 61.9 | 887 | 55.0 | 1250 | 67.9 | |
Diabetes | *** | ||||||
Yes | 377 | 10.9 | 236 | 14.6 | 141 | 7.7 | |
No | 3077 | 89.1 | 1376 | 85.4 | 1701 | 92.3 | |
Dyslipidaemia | ** | ||||||
Yes | 1955 | 56.6 | 869 | 53.9 | 1086 | 59.0 | |
No | 1499 | 43.4 | 743 | 46.1 | 756 | 41.0 | |
Smoking (current smoker) | *** | ||||||
Yes | 328 | 9.5 | 248 | 15.4 | 80 | 4.3 | |
No | 3126 | 90.5 | 1364 | 84.6 | 1762 | 95.7 | |
Alcohol consumption | *** | ||||||
Daily | 1036 | 30.0 | 800 | 49.6 | 236 | 12.8 | |
Sometimes | 777 | 22.5 | 382 | 23.7 | 395 | 21.4 | |
Rarely | 1641 | 47.5 | 430 | 26.7 | 1211 | 65.7 | |
Physical activity | ** | ||||||
Active | 1996 | 57.8 | 978 | 60.7 | 1018 | 55.3 | |
Inactive | 1458 | 42.2 | 634 | 39.3 | 824 | 44.7 |
Abbreviations: χ2 Chi square test for difference in distribution by gender.
*P < 0.05, **P < 0.01, ***P < 0.001.
A significant difference was found for female residents in the fully adjusted model (P = 0.003) when comparing the adjusted mean BMI and Walk Score using ANCOVA (Table 2). Women from ‘Walker’s Paradise’ (adjusted mean = 21.7 ± 0.2 kg/m2; P < 0.01) and ‘Very Walkable’ neighbourhoods (adjusted mean = 22.0 ± 0.1 kg/m2; P < 0.05) had lower mean BMI as compared with those from ‘Car-Dependent’ areas (22.6 ± 0.2 kg/m2). In men, although there was no statistically significant association between Walk Score and the adjusted mean BMI, male residents in ‘Walker’s Paradise’ neighbourhoods had the lowest mean BMI (23.3 ± 0.2 kg/m2). These associations were independent of age, presence of hypertension, diabetes mellitus, dyslipidaemia, current smoking habit, alcohol consumption and physical activity.
. | Car-dependent (I) . | Somewhat walkable (II) . | Very walkable (III) . | Walker’s paradise (IV) . | ANCOVA P-value . | Pairwise comparisonsd . |
---|---|---|---|---|---|---|
. | Mean (SE) (kg/m2) . | Mean (SE) (kg/m2) . | Mean (SE) (kg/m2) . | Mean (SE) (kg/m2) . | ||
Men | n = 281 | n = 565 | n = 603 | n = 163 | ||
Model 1a | 23.6 (0.2) | 23.6 (0.1) | 23.7 (0.1) | 23.4 (0.2) | 0.865 | |
Model 2b | 23.6 (0.2) | 23.6 (0.1) | 23.6 (0.1) | 23.4 (0.2) | 0.835 | |
Model 3c | 23.7 (0.2) | 23.6 (0.1) | 23.6 (0.1) | 23.3 (0.2) | 0.621 | |
Women | n = 334 | n = 618 | n = 726 | n = 164 | ||
Model 1a | 22.6 (0.2) | 22.3 (0.1) | 22.0 (0.1) | 21.5 (0.3) | 0.001 | IV < I**, IV < II*, III < I* |
Model 2b | 22.6 (0.2) | 22.3 (0.1) | 22.0 (0.1) | 21.5 (0.3) | 0.002 | IV < I**, IV < II*, III < I* |
Model 3c | 22.6 (0.2) | 22.3 (0.1) | 22.0 (0.1) | 21.7 (0.2) | 0.003 | IV < I**, III < I* |
. | Car-dependent (I) . | Somewhat walkable (II) . | Very walkable (III) . | Walker’s paradise (IV) . | ANCOVA P-value . | Pairwise comparisonsd . |
---|---|---|---|---|---|---|
. | Mean (SE) (kg/m2) . | Mean (SE) (kg/m2) . | Mean (SE) (kg/m2) . | Mean (SE) (kg/m2) . | ||
Men | n = 281 | n = 565 | n = 603 | n = 163 | ||
Model 1a | 23.6 (0.2) | 23.6 (0.1) | 23.7 (0.1) | 23.4 (0.2) | 0.865 | |
Model 2b | 23.6 (0.2) | 23.6 (0.1) | 23.6 (0.1) | 23.4 (0.2) | 0.835 | |
Model 3c | 23.7 (0.2) | 23.6 (0.1) | 23.6 (0.1) | 23.3 (0.2) | 0.621 | |
Women | n = 334 | n = 618 | n = 726 | n = 164 | ||
Model 1a | 22.6 (0.2) | 22.3 (0.1) | 22.0 (0.1) | 21.5 (0.3) | 0.001 | IV < I**, IV < II*, III < I* |
Model 2b | 22.6 (0.2) | 22.3 (0.1) | 22.0 (0.1) | 21.5 (0.3) | 0.002 | IV < I**, IV < II*, III < I* |
Model 3c | 22.6 (0.2) | 22.3 (0.1) | 22.0 (0.1) | 21.7 (0.2) | 0.003 | IV < I**, III < I* |
Abbreviations: BMI, Body Mass Index; ANCOVA, Analysis of Covariance; SE, standard error.
*P < 0.05, **P < 0.01, ***P < 0.001.
aData adjusted for age.
bData adjusted for age, smoking, alcohol consumption and physical activity.
cData adjusted for age, smoking, alcohol consumption, physical activity, hypertension, diabetes and dyslipidaemia.
dPairwise comparisons calculated from the post-hoc Bonferroni test.
. | Car-dependent (I) . | Somewhat walkable (II) . | Very walkable (III) . | Walker’s paradise (IV) . | ANCOVA P-value . | Pairwise comparisonsd . |
---|---|---|---|---|---|---|
. | Mean (SE) (kg/m2) . | Mean (SE) (kg/m2) . | Mean (SE) (kg/m2) . | Mean (SE) (kg/m2) . | ||
Men | n = 281 | n = 565 | n = 603 | n = 163 | ||
Model 1a | 23.6 (0.2) | 23.6 (0.1) | 23.7 (0.1) | 23.4 (0.2) | 0.865 | |
Model 2b | 23.6 (0.2) | 23.6 (0.1) | 23.6 (0.1) | 23.4 (0.2) | 0.835 | |
Model 3c | 23.7 (0.2) | 23.6 (0.1) | 23.6 (0.1) | 23.3 (0.2) | 0.621 | |
Women | n = 334 | n = 618 | n = 726 | n = 164 | ||
Model 1a | 22.6 (0.2) | 22.3 (0.1) | 22.0 (0.1) | 21.5 (0.3) | 0.001 | IV < I**, IV < II*, III < I* |
Model 2b | 22.6 (0.2) | 22.3 (0.1) | 22.0 (0.1) | 21.5 (0.3) | 0.002 | IV < I**, IV < II*, III < I* |
Model 3c | 22.6 (0.2) | 22.3 (0.1) | 22.0 (0.1) | 21.7 (0.2) | 0.003 | IV < I**, III < I* |
. | Car-dependent (I) . | Somewhat walkable (II) . | Very walkable (III) . | Walker’s paradise (IV) . | ANCOVA P-value . | Pairwise comparisonsd . |
---|---|---|---|---|---|---|
. | Mean (SE) (kg/m2) . | Mean (SE) (kg/m2) . | Mean (SE) (kg/m2) . | Mean (SE) (kg/m2) . | ||
Men | n = 281 | n = 565 | n = 603 | n = 163 | ||
Model 1a | 23.6 (0.2) | 23.6 (0.1) | 23.7 (0.1) | 23.4 (0.2) | 0.865 | |
Model 2b | 23.6 (0.2) | 23.6 (0.1) | 23.6 (0.1) | 23.4 (0.2) | 0.835 | |
Model 3c | 23.7 (0.2) | 23.6 (0.1) | 23.6 (0.1) | 23.3 (0.2) | 0.621 | |
Women | n = 334 | n = 618 | n = 726 | n = 164 | ||
Model 1a | 22.6 (0.2) | 22.3 (0.1) | 22.0 (0.1) | 21.5 (0.3) | 0.001 | IV < I**, IV < II*, III < I* |
Model 2b | 22.6 (0.2) | 22.3 (0.1) | 22.0 (0.1) | 21.5 (0.3) | 0.002 | IV < I**, IV < II*, III < I* |
Model 3c | 22.6 (0.2) | 22.3 (0.1) | 22.0 (0.1) | 21.7 (0.2) | 0.003 | IV < I**, III < I* |
Abbreviations: BMI, Body Mass Index; ANCOVA, Analysis of Covariance; SE, standard error.
*P < 0.05, **P < 0.01, ***P < 0.001.
aData adjusted for age.
bData adjusted for age, smoking, alcohol consumption and physical activity.
cData adjusted for age, smoking, alcohol consumption, physical activity, hypertension, diabetes and dyslipidaemia.
dPairwise comparisons calculated from the post-hoc Bonferroni test.
In both men and women, the prevalence of obesity was the lowest in ‘Walker’s Paradise’ neighbourhoods (Table 3). Women living in ‘Walker’s Paradise’ (adjusted OR = 0.46, 95% CI = 0.26–0.80) and ‘Very Walkable’ (adjusted OR = 0.65, 95% CI = 0.47–0.91) neighbourhoods had lower obesity prevalence. In men, higher odds for being obese were associated with the younger age group (adjusted OR = 1.76, 95% CI = 1.34–2.30) and daily alcohol consumption was associated with lower obesity prevalence (adjusted OR = 0.72, 95% CI = 0.55–0.95). In men and women, the presence of hypertension (adjusted OR for men = 2.29, 95% CI: 1.80–2.89; adjusted OR for women = 2.14, 95% CI = 1.66–2.75), diabetes mellitus (adjusted OR for men = 1.48, 95% CI: 1.10–2.01; adjusted OR for women = 2.06, 95% CI = 1.40–3.03) and dyslipidaemia (adjusted OR for men = 1.70, 95% CI = 1.35–2.15; adjusted OR for women = 1.83, 95% CI = 1.39–2.41) were associated with higher prevalence of obesity. Although there were no statistically significant associations in the adjusted models, men and women who were physically active had lower prevalence of obesity.
. | . | Men . | Women . | ||||
---|---|---|---|---|---|---|---|
. | . | Obesity . | Unadjusted model . | Adjusted modela . | Obesity . | Unadjusted model . | Adjusted modela . |
. | . | % . | OR (95% CI) . | OR (95% CI) . | % . | OR (95% CI) . | OR (95% CI) . |
Walk score | Car-Dependent | 27.0 | 1 (referent) | 1 (referent) | 23.1 | 1 (referent) | 1 (referent) |
categories | Somewhat Walkable | 29.2 | 1.11 (0.81–1.53) | 1.10 (0.75–1.46) | 18.9 | 0.78 (0.56–1.08) | 0.77 (0.55–1.08) |
Very Walkable | 28.2 | 1.06 (0.77–1.46) | 0.96 (0.69–1.34) | 16.5 | 0.66 (0.48–0.91) | 0.65 (0.47–0.91) | |
Walker’s Paradise | 25.8 | 0.94 (0.60–1.45) | 0.82 (0.52–1.30) | 11.0 | 0.41 (0.24–0.72) | 0.46 (0.26–0.80) | |
Age (years) | <65 | 34.0 | 1.44 (1.12–1.84) | 1.76 (1.34–2.30) | 15.3 | 0.76 (0.58–0.98) | 1.01 (0.76–1.34) |
≥65 | 26.3 | 1 (referent) | 1 (referent) | 19.3 | 1 (referent) | 1 (referent) | |
Comorbidities | Hypertension | ||||||
Yes | 36.4 | 2.12 (1.70–2.64) | 2.29 (1.80–2.89) | 28.0 | 2.55 (2.00–3.24) | 2.14 (1.66–2.75) | |
No | 21.3 | 1 (referent) | 1 (referent) | 13.3 | 1 (referent) | 1 (referent) | |
Diabetes | |||||||
Yes | 37.3 | 1.65 (1.23–2.20) | 1.48 (1.10–2.01) | 34.8 | 2.67 (1.85–3.86) | 2.06 (1.40–3.03) | |
No | 26.5 | 1 (referent) | 1 (referent) | 16.6 | 1 (referent) | 1 (referent) | |
Dyslipidaemia | |||||||
Yes | 33.8 | 1.88 (1.50–2.35) | 1.70 (1.35–2.15) | 22.4 | 2.16 (1.66–2.81) | 1.83 (1.39–2.41) | |
No | 21.4 | 1 (referent) | 1 (referent) | 11.8 | 1 (referent) | 1 (referent) | |
Lifestyle habits | Current smoking | ||||||
Yes | 27.4 | 0.96 (0.71–1.30) | 0.94 (0.68–1.29) | 10.0 | 0.49 (0.24–1.03) | 0.55 (0.25–1.18) | |
No | 28.2 | 1 (referent) | 1 (referent) | 18.4 | 1 (referent) | 1 (referent) | |
Alcohol consumption | |||||||
Daily | 25.4 | 0.74 (0.57–0.95) | 0.72 (0.55–0.95) | 16.1 | 0.84 (0.57–1.22) | 1.04 (0.70–1.54) | |
Sometimes | 29.8 | 0.92 (0.68–1.24) | 0.95 (0.70–1.30) | 17.2 | 0.91 (0.67–1.22) | 0.97 (0.71–1.33) | |
Rarely | 31.6 | 1 (referent) | 1 (referent) | 18.7 | 1 (referent) | 1 (referent) | |
Physical activity Active | 26.1 | 0.78 (0.62–0.97) | 0.81 (0.64–1.02) | 16.8 | 0.83 (0.66–1.06) | 0.81 (0.63–1.04) | |
Inactive | 31.2 | 1 (referent) | 1 (referent) | 19.5 | 1 (referent) | 1 (referent) |
. | . | Men . | Women . | ||||
---|---|---|---|---|---|---|---|
. | . | Obesity . | Unadjusted model . | Adjusted modela . | Obesity . | Unadjusted model . | Adjusted modela . |
. | . | % . | OR (95% CI) . | OR (95% CI) . | % . | OR (95% CI) . | OR (95% CI) . |
Walk score | Car-Dependent | 27.0 | 1 (referent) | 1 (referent) | 23.1 | 1 (referent) | 1 (referent) |
categories | Somewhat Walkable | 29.2 | 1.11 (0.81–1.53) | 1.10 (0.75–1.46) | 18.9 | 0.78 (0.56–1.08) | 0.77 (0.55–1.08) |
Very Walkable | 28.2 | 1.06 (0.77–1.46) | 0.96 (0.69–1.34) | 16.5 | 0.66 (0.48–0.91) | 0.65 (0.47–0.91) | |
Walker’s Paradise | 25.8 | 0.94 (0.60–1.45) | 0.82 (0.52–1.30) | 11.0 | 0.41 (0.24–0.72) | 0.46 (0.26–0.80) | |
Age (years) | <65 | 34.0 | 1.44 (1.12–1.84) | 1.76 (1.34–2.30) | 15.3 | 0.76 (0.58–0.98) | 1.01 (0.76–1.34) |
≥65 | 26.3 | 1 (referent) | 1 (referent) | 19.3 | 1 (referent) | 1 (referent) | |
Comorbidities | Hypertension | ||||||
Yes | 36.4 | 2.12 (1.70–2.64) | 2.29 (1.80–2.89) | 28.0 | 2.55 (2.00–3.24) | 2.14 (1.66–2.75) | |
No | 21.3 | 1 (referent) | 1 (referent) | 13.3 | 1 (referent) | 1 (referent) | |
Diabetes | |||||||
Yes | 37.3 | 1.65 (1.23–2.20) | 1.48 (1.10–2.01) | 34.8 | 2.67 (1.85–3.86) | 2.06 (1.40–3.03) | |
No | 26.5 | 1 (referent) | 1 (referent) | 16.6 | 1 (referent) | 1 (referent) | |
Dyslipidaemia | |||||||
Yes | 33.8 | 1.88 (1.50–2.35) | 1.70 (1.35–2.15) | 22.4 | 2.16 (1.66–2.81) | 1.83 (1.39–2.41) | |
No | 21.4 | 1 (referent) | 1 (referent) | 11.8 | 1 (referent) | 1 (referent) | |
Lifestyle habits | Current smoking | ||||||
Yes | 27.4 | 0.96 (0.71–1.30) | 0.94 (0.68–1.29) | 10.0 | 0.49 (0.24–1.03) | 0.55 (0.25–1.18) | |
No | 28.2 | 1 (referent) | 1 (referent) | 18.4 | 1 (referent) | 1 (referent) | |
Alcohol consumption | |||||||
Daily | 25.4 | 0.74 (0.57–0.95) | 0.72 (0.55–0.95) | 16.1 | 0.84 (0.57–1.22) | 1.04 (0.70–1.54) | |
Sometimes | 29.8 | 0.92 (0.68–1.24) | 0.95 (0.70–1.30) | 17.2 | 0.91 (0.67–1.22) | 0.97 (0.71–1.33) | |
Rarely | 31.6 | 1 (referent) | 1 (referent) | 18.7 | 1 (referent) | 1 (referent) | |
Physical activity Active | 26.1 | 0.78 (0.62–0.97) | 0.81 (0.64–1.02) | 16.8 | 0.83 (0.66–1.06) | 0.81 (0.63–1.04) | |
Inactive | 31.2 | 1 (referent) | 1 (referent) | 19.5 | 1 (referent) | 1 (referent) |
Abbreviations: OR, odds ratio; CI, confidence interval.
Values in bold are significantly different.
aThe multivariate model is adjusted for walk score, age, comorbidities and lifestyle habits.
. | . | Men . | Women . | ||||
---|---|---|---|---|---|---|---|
. | . | Obesity . | Unadjusted model . | Adjusted modela . | Obesity . | Unadjusted model . | Adjusted modela . |
. | . | % . | OR (95% CI) . | OR (95% CI) . | % . | OR (95% CI) . | OR (95% CI) . |
Walk score | Car-Dependent | 27.0 | 1 (referent) | 1 (referent) | 23.1 | 1 (referent) | 1 (referent) |
categories | Somewhat Walkable | 29.2 | 1.11 (0.81–1.53) | 1.10 (0.75–1.46) | 18.9 | 0.78 (0.56–1.08) | 0.77 (0.55–1.08) |
Very Walkable | 28.2 | 1.06 (0.77–1.46) | 0.96 (0.69–1.34) | 16.5 | 0.66 (0.48–0.91) | 0.65 (0.47–0.91) | |
Walker’s Paradise | 25.8 | 0.94 (0.60–1.45) | 0.82 (0.52–1.30) | 11.0 | 0.41 (0.24–0.72) | 0.46 (0.26–0.80) | |
Age (years) | <65 | 34.0 | 1.44 (1.12–1.84) | 1.76 (1.34–2.30) | 15.3 | 0.76 (0.58–0.98) | 1.01 (0.76–1.34) |
≥65 | 26.3 | 1 (referent) | 1 (referent) | 19.3 | 1 (referent) | 1 (referent) | |
Comorbidities | Hypertension | ||||||
Yes | 36.4 | 2.12 (1.70–2.64) | 2.29 (1.80–2.89) | 28.0 | 2.55 (2.00–3.24) | 2.14 (1.66–2.75) | |
No | 21.3 | 1 (referent) | 1 (referent) | 13.3 | 1 (referent) | 1 (referent) | |
Diabetes | |||||||
Yes | 37.3 | 1.65 (1.23–2.20) | 1.48 (1.10–2.01) | 34.8 | 2.67 (1.85–3.86) | 2.06 (1.40–3.03) | |
No | 26.5 | 1 (referent) | 1 (referent) | 16.6 | 1 (referent) | 1 (referent) | |
Dyslipidaemia | |||||||
Yes | 33.8 | 1.88 (1.50–2.35) | 1.70 (1.35–2.15) | 22.4 | 2.16 (1.66–2.81) | 1.83 (1.39–2.41) | |
No | 21.4 | 1 (referent) | 1 (referent) | 11.8 | 1 (referent) | 1 (referent) | |
Lifestyle habits | Current smoking | ||||||
Yes | 27.4 | 0.96 (0.71–1.30) | 0.94 (0.68–1.29) | 10.0 | 0.49 (0.24–1.03) | 0.55 (0.25–1.18) | |
No | 28.2 | 1 (referent) | 1 (referent) | 18.4 | 1 (referent) | 1 (referent) | |
Alcohol consumption | |||||||
Daily | 25.4 | 0.74 (0.57–0.95) | 0.72 (0.55–0.95) | 16.1 | 0.84 (0.57–1.22) | 1.04 (0.70–1.54) | |
Sometimes | 29.8 | 0.92 (0.68–1.24) | 0.95 (0.70–1.30) | 17.2 | 0.91 (0.67–1.22) | 0.97 (0.71–1.33) | |
Rarely | 31.6 | 1 (referent) | 1 (referent) | 18.7 | 1 (referent) | 1 (referent) | |
Physical activity Active | 26.1 | 0.78 (0.62–0.97) | 0.81 (0.64–1.02) | 16.8 | 0.83 (0.66–1.06) | 0.81 (0.63–1.04) | |
Inactive | 31.2 | 1 (referent) | 1 (referent) | 19.5 | 1 (referent) | 1 (referent) |
. | . | Men . | Women . | ||||
---|---|---|---|---|---|---|---|
. | . | Obesity . | Unadjusted model . | Adjusted modela . | Obesity . | Unadjusted model . | Adjusted modela . |
. | . | % . | OR (95% CI) . | OR (95% CI) . | % . | OR (95% CI) . | OR (95% CI) . |
Walk score | Car-Dependent | 27.0 | 1 (referent) | 1 (referent) | 23.1 | 1 (referent) | 1 (referent) |
categories | Somewhat Walkable | 29.2 | 1.11 (0.81–1.53) | 1.10 (0.75–1.46) | 18.9 | 0.78 (0.56–1.08) | 0.77 (0.55–1.08) |
Very Walkable | 28.2 | 1.06 (0.77–1.46) | 0.96 (0.69–1.34) | 16.5 | 0.66 (0.48–0.91) | 0.65 (0.47–0.91) | |
Walker’s Paradise | 25.8 | 0.94 (0.60–1.45) | 0.82 (0.52–1.30) | 11.0 | 0.41 (0.24–0.72) | 0.46 (0.26–0.80) | |
Age (years) | <65 | 34.0 | 1.44 (1.12–1.84) | 1.76 (1.34–2.30) | 15.3 | 0.76 (0.58–0.98) | 1.01 (0.76–1.34) |
≥65 | 26.3 | 1 (referent) | 1 (referent) | 19.3 | 1 (referent) | 1 (referent) | |
Comorbidities | Hypertension | ||||||
Yes | 36.4 | 2.12 (1.70–2.64) | 2.29 (1.80–2.89) | 28.0 | 2.55 (2.00–3.24) | 2.14 (1.66–2.75) | |
No | 21.3 | 1 (referent) | 1 (referent) | 13.3 | 1 (referent) | 1 (referent) | |
Diabetes | |||||||
Yes | 37.3 | 1.65 (1.23–2.20) | 1.48 (1.10–2.01) | 34.8 | 2.67 (1.85–3.86) | 2.06 (1.40–3.03) | |
No | 26.5 | 1 (referent) | 1 (referent) | 16.6 | 1 (referent) | 1 (referent) | |
Dyslipidaemia | |||||||
Yes | 33.8 | 1.88 (1.50–2.35) | 1.70 (1.35–2.15) | 22.4 | 2.16 (1.66–2.81) | 1.83 (1.39–2.41) | |
No | 21.4 | 1 (referent) | 1 (referent) | 11.8 | 1 (referent) | 1 (referent) | |
Lifestyle habits | Current smoking | ||||||
Yes | 27.4 | 0.96 (0.71–1.30) | 0.94 (0.68–1.29) | 10.0 | 0.49 (0.24–1.03) | 0.55 (0.25–1.18) | |
No | 28.2 | 1 (referent) | 1 (referent) | 18.4 | 1 (referent) | 1 (referent) | |
Alcohol consumption | |||||||
Daily | 25.4 | 0.74 (0.57–0.95) | 0.72 (0.55–0.95) | 16.1 | 0.84 (0.57–1.22) | 1.04 (0.70–1.54) | |
Sometimes | 29.8 | 0.92 (0.68–1.24) | 0.95 (0.70–1.30) | 17.2 | 0.91 (0.67–1.22) | 0.97 (0.71–1.33) | |
Rarely | 31.6 | 1 (referent) | 1 (referent) | 18.7 | 1 (referent) | 1 (referent) | |
Physical activity Active | 26.1 | 0.78 (0.62–0.97) | 0.81 (0.64–1.02) | 16.8 | 0.83 (0.66–1.06) | 0.81 (0.63–1.04) | |
Inactive | 31.2 | 1 (referent) | 1 (referent) | 19.5 | 1 (referent) | 1 (referent) |
Abbreviations: OR, odds ratio; CI, confidence interval.
Values in bold are significantly different.
aThe multivariate model is adjusted for walk score, age, comorbidities and lifestyle habits.
Discussion
Main findings of this study
This study described the associations of neighbourhood walkability and obesity in Toyama based on datasets from the TPNHI database. We examined: (i) whether residents from more walkable neighbourhoods have lower BMI, and (ii) whether there is gender difference in the obesity prevalence associated with walkability of residential areas, age, presence of comorbidities and lifestyle factors. The prevalence of obesity in our sample is 23%, which is slightly lower than the prevalence of obesity in the general Japanese population (25%).28 This may be due to the voluntary nature of the SHC programme. Individuals who voluntarily participated in the SHC programme may be more health conscious than the general Japanese population.
We found that residents from highly walkable neighbourhoods generally had lower mean BMI but significant associations between neighbourhood walkability and BMI and prevalence of obesity were only observed in women. This finding aligns with a previous study15 that reported places with higher Walk Score ratings were associated with lower BMI among Japanese adults in both urban and rural environments. The positive effect of Walk Score on lowering obesity risk for women, but not for men, was reported in certain geographical settings with traditional domestic division of labour.29 In these settings, women are more likely than men to walk for daily errands such as shopping for grocery. Hence, improving neighbourhood walkability could influence women’s walking behaviour more than men. Given the cultural and traditional gender roles in our local community—women as caregivers and men as primary wage earners—the gender difference observed in this study may indicate a need for examining ways to create inclusive living spaces in Toyama, where men and women could have equitable access to healthier lifestyle choices.
Our findings show a higher prevalence of obesity among men below 65 and this correspond with a similar obesity trend in Japan observed two decades ago when there was an increase in the prevalence of obesity among younger men, especially among those in the middle age group.30 Major factors which may explain the obesity trend among Japanese men, particularly in the rural areas, include long working hours and a shift in traditional rural industry—from jobs that required more physical activity to more sedentary jobs in the service industry.31
Our results show that hypertension, diabetes mellitus and dyslipidaemia are associated with obesity among older Toyama residents, regardless of gender. These three conditions are known to be health consequences of obesity.32 Together with the obesity trend among younger Japanese men and the ageing population, obesity and multiple comorbidities may place a tremendous strain on the social and health insurance system in Toyama Prefecture. Loo et al. reported that neighbourhood walkability is associated with metabolic risk factors. Specifically, people living in residential areas with high Walk Score ratings have lower mean BMI, healthier blood pressure range and higher mean HDL.33 Outcomes from our study and the Loo et al. study indicate that there is a need for changing public health strategies to include urban planning actions that could improve walkability of residential areas. Creating walkable neighbourhoods in rural Japan could potentially encourage healthier lifestyle, prevent chronic disease and subsequently help to reduce escalating healthcare expenditures.
Our data show that daily alcohol consumption is associated with lower odds of being obese among men. This paradoxical finding may be linked to evidence which suggest light-to-moderate alcohol intake is unlikely a risk factor for obesity.34 Also, people who consume alcohol in moderate amount tend to enjoy a long-term moderate lifestyle which accommodates for their drinking habits.35 Results in this study show that smoking and physical activity are not associated with obesity. This finding is consistent with a previous study in Taiwan36 and we agree that further investigations are required to examine possible consistent or inconsistent associations between Walk Score and lifestyle behaviours in different countries, especially within Asian contexts.
What is already known on this topic?
The benefits of a walkable neighbourhood are well-established.37–39 Efforts to integrate and implement urban planning strategies in public health policies are actively encouraged at global, national and local levels. These efforts are reflected in: (i) the World Health Organization (WHO) Global Action Plan on Physical Activity 2018–2030,7 (ii) spatial planning for health evidence resource by Public Health England,40 (iii) healthy places resources by Centers for Disease Control and Prevention41 and (iv) healthy Toronto by design by the City of Toronto, Canada.42
In our local context, there is a clear dependency on cars for transportation. In fact, Toyama Prefecture has the second highest car ownership per household in Japan.17 However, the area of urban planning for creating healthy built environment has received little attention. This could be due to a lack of research evidence from national and local contexts that could drive interests and initiate dialogues among stakeholders.
What this study adds?
To our knowledge, this is the first large-sample study which focuses on the relationship between built environment and residents’ health in Toyama Prefecture, Japan. Findings from this study raise a new question which warrants further investigations within our local context. Specifically, the enablers and disablers of walking among residents require further examination. Understanding the daily opportunities and barriers for walking among men and women may: (i) clarify the gender difference in our observation and (ii) inform public policy and urban planning strategies towards creating healthy living spaces for people from all walks of life. More importantly, this study provides preliminary evidence towards enabling a conversation around the role of urban planning in improving health outcomes for rural communities in Japan.
Limitations of this study
We analysed information from the TPNHI database that was made available to us. We were unable to control for potential individual-level and neighbourhood-level confounding variables which we could not access. Also, self-reported physical activity may overestimate vigorous exercise. However, men and women do not appear to differ from one another when reporting their sedentary behaviour.43 Future studies would benefit from the use of device measures to evaluate physical activity/inactivity. As this was a cross-sectional analysis, limited inferences could be drawn from this study and the results could not be generalized across different settings and populations.
Conclusion
This study presents information relating to neighbourhood walkability and obesity within the context of rural Japan. Women living in highly walkable neighbourhoods had lower BMI and lower prevalence of obesity. No significant associations were found between neighbourhood walkability, mean BMI and prevalence of obesity among men. Further studies are required to explain these associations in Toyama as it may indicate further implications for public health strategies.
Acknowledgements
We thank the officers of the Toyama Prefecture Government for assistance in accessing the TPNHI database.
Data availability
The data for this article were provided by Toyama Prefecture Government and will be shared with permission from Toyama Prefecture Government upon request to the corresponding author.
Conflict of interest
None declared.
Funding
This study was supported by Toyama Prefectural Government and the discretionary budget of the President of University of Toyama. The funding organizations were not involved in the design, conduct, analysis and interpretation of this study nor review or approval of this manuscript.
Grace Koh, PhD student
Michikazu Sekine, Professor of Epidemiology and Health Policy
Masaaki Yamada, Assistant Professor in Epidemiology and Health Policy
Yuko Fujimura, Programme Coordinator in Community Medicine and Health Support
Takashi Tatsuse, Assistant Professor in Epidemiology and Health Policy