Abstract
Background
Reducing socioeconomic inequalities in obesity is a public health priority. Limited research exists on the role of neighbourhood environmental attributes in mitigating these inequalities. However, it has been shown that neighbourhoods with more greenery tend to have lower levels of socioeconomic inequalities in non-obesity health outcomes. We examined whether neighbourhood greenery moderates the association of area-level socioeconomic status (SES) with waist circumference.
Methods
Data from 3,261 middle-aged and older adults who participated in a national cohort study conducted in Australia (2011–12) were used. The outcome was objectively measured waist circumference. For area-level SES, a composite index of disadvantage based on census data was used. We used two measures of neighbourhood greenery: mean greenness and geographic size of greenspace. They were assessed using the Normalized Difference Vegetation Index (NDVI) within 0.5, 1, and 2 km radius buffers around participants’ homes. The mean NDVI value within each buffer area was used for the former, and the geographic size of the area with NDVI ≥ 0.6 (dense greenery) was used for the latter.
Results
There was a significant negative association between area-level SES and waist circumference: one standard deviation higher score in the area-level SES indicator (less disadvantage) was associated with 1.76 cm (95% CI: -2.68, -0.83) lower waist circumference. Analyses stratified by greenery levels found similar significant associations in the areas with low and high levels of greenery but not in the areas with medium levels of greenery for both greenery measures within 1 km and 2 km buffers.
Conclusions
Consistent with previous studies, our study found that participants living in disadvantaged suburbs were likely to have a larger waist circumference than those living in advantaged suburbs. However, we also found that such socioeconomic inequalities in obesity were mitigated in the areas with medium levels of greenery for this sample of Australian adults. Our findings suggest that there may be an optimum level of greenery where inequalities in obesity are alleviated. Further research is needed to understand the mechanisms underlying these findings.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12889-024-20711-6.
Keywords: Greenspace, Environment, Health disparities, Socioeconomic status, Moderation
Background
There has been a worldwide rise in the burden of overweight and obesity. The prevalence of obesity has doubled since 1980 and is expected to increase in the US and other high-income countries [1]. Overweight and obesity (henceforth referred collectively as obesity) have a substantial impact on people and society, in terms of life expectancy, healthcare cost and productivity [2]. It is well documented that obesity is not distributed evenly, with significant disparities existing between disadvantaged and advantaged neighbourhoods [3–5]. Such inequalities in obesity by socioeconomic status (SES) also contribute to the subsequent inequalities in morbidity and mortality [6]. It is a public health priority to enhance population health through addressing such inequalities in health risk [7].
Neighbourhood built environment attributes are known to be related to residents’ obesity risk [8–10] and unevenly distributed within society [11, 12]. Since neighbourhoods vary both in socioeconomic and environmental characteristics, it can be argued that the disparities in obesity between disadvantaged and advantaged areas may depend to some extent on their environmental characteristics. This suggests a possibility that the presence of specific built environment attributes may mitigate or magnify area-level socioeconomic inequalities in obesity. A recent review on this topic found that this research is still in its early stages, as it identified only nine studies examining environmental moderation of the relationship between area-level SES and obesity [13]. A few such studies have shown that associations of area-level SES with obesity can be moderated by neighbourhood environmental characteristics, specifically urbanicity [14] and perceptions of aesthetics and safety [15], with mixed findings for neighbourhood walkability [16, 17]. No evidence of moderation was found for access to fast food outlets [18, 19] and street connectivity [20]. The review did not find studies examining potential moderation by other environmental attributes such as green spaces, availability of public transport and micro-scale features. Given the dearth of studies, research is needed to explore whether the association of area-level SES with obesity varies across different levels of exposure to environmental features to produce evidence that can inform environmental policies and initiatives for health equity.
A potential environmental attribute that may reduce obesity inequalities is neighbourhood greenery. This refers to vegetation (e.g., trees, grass, shrubs) in local parks, gardens, nature reserves, and along streets. Exposure to greenery is known to be beneficial to health through promoting physical activity and reducing stress [21]. Since these factors are also related to obesity [22, 23], neighbourhood greenery could reduce socioeconomic inequalities if greenery is more closely associated with them in lower SES areas. In addition, increasing the amount of greenery in a neighbourhood can be feasible, as it can be implemented by renovating existing parks or increasing the tree canopy along the streets. In contrast, modifying macro-scale environmental attributes, such as street connectivity and public transport accessibility, can be more challenging [24]. However, there is limited research examining the role of neighbourhood greenery in moderating socioeconomic inequalities in obesity. One study in Brisbane, Australia identified that the size of park area within a 1.6 km buffer did not moderate socioeconomic inequalities in obesity [25]. However, it found that socioeconomic inequalities in obesity (between least disadvantaged and moderate levels of disadvantage) were lower among those who had parks with more park facilities and amenities, such as walking/biking paths, benches, picnic tables, car parks, drinking fountains, toilets, and lighting [25]. Another study, conducted in the UK, found that area-level SES moderated the association of park attributes (access, quality) with body mass index (BMI) [26]. These studies suggest the possibility that park-based attributes may act as moderators of the association between SES and BMI. However, it remains unclear to what extent neighbourhood greenery, which is distinct from park features (e.g., facilities, access), would help to reduce socioeconomic inequalities in obesity.
Previous studies have shown that greenery can moderate the socioeconomic inequalities of non-obesity health outcomes. For instance, an English study found that inequalities in mortality from circulatory disease and all-cause mortality were lower in populations living in the greenest areas compared to those living in the least green areas [27]. Similarly, a Chinese study showed that SES-related inequalities in heart disease were less pronounced in neighbourhoods with larger greenspaces [28]. These significant findings suggest a possibility that a contribution of neighbourhood greenery to health outcomes may be more prominent in lower SES areas. Since socioeconomic disparities in obesity are a significant public health problem for which any potential approach for reduction is worth testing, it is of interest to investigate the role of neighbourhood greenery in the disparities.
It is important to note that previous studies examining the moderation of the association between area-level SES and a measure of obesity by park attributes employed BMI calculated from self-reported height and weight [25, 26]. Since self-reported measures are often susceptible to reporting bias [29], objective measures of obesity may yield more accurate results. Additionally, BMI does not fully capture cardiometabolic risk as it does not consider visceral fat nor abdominal adiposity [30]. On the other hand, waist circumference is an indicator of abdominal obesity, which plays an important role in the adverse metabolic effects of obesity and has been strongly associated with all-cause and cardiovascular mortality [31]. Thus, objectively measured waist circumference is more suitable to accurately understand whether the relationships between SES and obesity vary by environmental attributes.
We examined whether measures of neighbourhood greenery moderate the association between area-level SES and objectively measured waist circumference in Australian adults.
Methods
Data source and study participants
The study used data from the third wave of the Australian Diabetes, Obesity, Lifestyle (AusDiab) study, which was conducted in 2011–12. This is a national study (with representation from six Australian states and the Northern Territory) originally designed to investigate the prevalence and incidence of diabetes and associated risk factors in a population-based sample of adults [32]. This dataset, which was collected more than 10 years ago, was used as it provided an opportunity to use a large national sample recruited from diverse localities in socioeconomic status and geographic region. There was a total of 11,247 participants in the baseline (1999–2000) data collection, of which 4,614 participated in the third wave. Participants residing in outer-regional or remote areas (n = 1,018) were excluded, since greenery in such areas could include farmland and forests. Further exclusions were made for those who did not have geocoded addresses (n = 134); had missing values for the variables used (n = 28); and were pregnant (n = 7). Participants reporting difficulties walking more than 100 m (n = 166) were also excluded, as greenery in outdoor environments may not be relevant to them. The final sample size was 3,261. Ethics approval for the AusDiab study was granted by the Alfred Hospital Ethics Committee (no. 39/11). Written informed consent was obtained from all participants.
Measures
Outcome: waist circumference
We used participants’ measured waist circumference as the outcome for this study. It was measured by trained personnel according to the field survey guidelines recommended by the WHO [32]. Waist circumference was measured twice with a measuring tape, halfway between the lower border of the ribs and the iliac crest on a horizontal plane, and their mean value was used. Participants’ height and weight were also measured in the AusDiab study. However, since waist circumference and BMI showed similar patterns in their associations with area-level SES and interactions with greenery measures, we presented the results only for waist circumference.
Exposure: area-level SES
Area-level SES was identified using the 2011 Index of Relative Socio-economic Disadvantage (IRSD), which is a census-based index consisting of 16 area-level disadvantage measures such as the percentage of low-income households, unemployed people, people with low-level education, households without cars and households with a single parent [33]. This is a commonly used indicator of area-level socioeconomic disadvantage in Australia, which is also used by a government agency to assess socioeconomic inequalities in health [34]. It is standardised to a distribution where the average is 1,000 and the standard deviation is 100, with lower values representing greater disadvantage. We extracted the IRSD score for each suburb [35] where participants resided.
Potential moderators: neighbourhood greenery
Greenery was assessed in two ways: the mean level of greenness; and the geographic size of greenspace. The Normalized Difference Vegetation Index (NDVI) was used to assess neighbourhood greenery. This is a widely used indicator of greenness in forestry and agriculture, typically assessed using satellite images [36, 37]. NDVI has also been commonly used in public health research as a measure of environmental exposure [37]. For instance, an Australian study found that NDVI-based measures were associated with obesity [38]. NDVI is a suitable measure of greenery for this study, as retrospective satellite images are available at the national scale. NDVI values range from − 1 to 1, with higher values representing greener areas. The median annual NDVI data for the year 2012 was obtained through Google Earth Engine using Landsat 7 tier 1 surface reflectance data as the source with a 10-m resolution [39]. The annual NDVI score for each pixel was determined by calculating the median NDVI value from all Landsat images taken in 2012. This method was selected to ensure a consistent measure of greenery with reduced seasonal influence [40].
The mean level of greenness was calculated as the average of NDVI values of all pixels within the target area. The negative NDVI values (water) were set to zero before calculating the mean NDVI as this study was interested in the level of greenness. This parsimonious measure is often used in research but does not consider the distribution of greenery. For instance, a neighbourhood with dense vegetation only in a limited area (e.g., parks) can have similar mean NDVI to a neighbourhood covered evenly with sparse greenery. Given that the presence of large parks can encourage residents to walk [41], which can be protective against obesity, the geographic size of greenspace may represent an aspect of greenery distinct from the mean level of greenness. For the geographic size of greenspace, we calculated the size of the area (in hectare) covered with denser vegetation, which was defined as NDVI ≥ 0.6, within the buffers. This threshold was chosen as higher NDVI values over 0.6 are indicative of dense vegetation [42] and generally capture tree canopies, shrubs/forbs, and grassland, provided they are densely vegetated [37].
We calculated the neighbourhood greenery measures for each participant using a 0.5 km, 1 km, and 2 km circular buffer around participant’s home address. A 0.5 km buffer represents the immediate neighbourhood of residence [43], while a 1 km buffer has been typically used as a neighbourhood area in previous research on health and place [44]. We also used a 2 km buffer, since a previous Australian study reported that the distance most adults walked to reach natural features was around 2 km [45]. It was also reported that the likelihood of greenery predicting physical health increased as the size of the buffer increased but plateaued at approximately 2 km [46]. We used circular buffers around homes instead of network buffers because the size of the latter depends on the street layout pattern (i.e., the size of network buffers would be smaller in neighbourhoods with lower street connectivity). If a network buffer is used, a smaller size greenspace could be either due to a smaller area of vegetation or a smaller buffer size (or both). Since areas with lower street connectivity may be more conducive to obesity [47], we considered that it would be better not to confound the greenery measure with street connectivity. QGIS v.3.16 (Open-Source Geospatial Foundation Project, 2021) was used to calculate these greenery measures.
Data analysis
To present descriptive statistics, the sample was divided into two groups: lower and higher SES areas, using the median split of the IRSD scores of their suburbs. Participants were assigned to these two categories according to their suburb, ensuring an equal number of suburbs (but not participants) in each of these SES categories. Two-level linear regression models were then used to examine the association between the IRSD score and waist circumference, accounting for clustering of participants (level 1) within suburbs (level 2). These models adjusted for individual-level demographic and socioeconomic factors (age, gender, education status, employment status, and household income) and suburb-level population density, which was obtained from the 2012 census data. Population density was considered as a confounder since it is known to be related to obesity, IRSD and greenery. To examine the moderating role, each greenery measure was first categorised into tertiles of low, medium and high levels. Tertiles may capture potential threshold effects more effectively than a continuous measure, which assumes a linear moderation across all levels. We chose tertiles as a greater number of levels (i.e., quartiles, quintiles) could reduce sample size within each stratum, lowering statistical power. To examine the interaction between IRSD and each greenery measure, we included an interaction term in the relevant model and performed the Wald test, which is to assess the overall significance of the interaction between these two variables [48]. A significant Wald test result suggests that the effect of IRSD on waist circumference differs across levels of greenery. Stratified analyses examined the association between IRSD and waist circumference for each tertile of greenery measures. The statistical analyses were conducted using RStudio v2022.12.0.
Results
Table 1 shows the characteristics of the suburbs, study participants and the greenery measures in lower and higher SES areas. Waist circumference was higher among those living in lower SES areas, compared to those living in higher SES areas. The overall mean NDVI was 0.45, indicating that the study areas, on average, had a moderate level of greenness. There were only small differences in mean NDVI between the SES categories. However, for the size of greenspace, higher SES areas had a larger area with dense vegetation compared to lower SES areas in all three buffers.
Table 1.
Characteristics of study areas and participants in lower and higher SES areas
| N, Mean (SD), or % | |||
|---|---|---|---|
| Lower SES | Higher SES | Total | |
| N (suburbs) | 347 | 346 | 693 |
| IRSD score | 974 (57) | 1080 (25) | 1040 (67) |
| IRSD median [range] | 993 [714, 1034] | 1081 [1035, 1140] | 1035 [714, 1140] |
| Suburb geographic size, km2 | 21.4 (49.9) | 9.8 (22.2) | 14.6 (36.7) |
| Population density, persons/ha | 14.1 (10.6) | 15.2 (11.7) | 14.8 (11.3) |
| N (participants) | 1,341 | 1,920 | 3,261 |
| Age | 61.4 (11.7) | 60.3 (10.8) | 60.8 (11.2) |
| Gender, % women | 56.5 | 54.3 | 55.2 |
| Education status | |||
| % high school or less | 38.6 | 25.9 | 31.1 |
| % technical or vocational | 43.8 | 43.9 | 43.8 |
| % Bachelor’s degree or higher | 17.7 | 30.2 | 25.1 |
| Employment status, % working | 49.0 | 56.0 | 53.1 |
| Household income, % | |||
| % <$600 pw | 22.2 | 12.2 | 16.3 |
| % $600–1500 pw | 36.5 | 32.4 | 34.1 |
| % >$1500 pw | 28.6 | 44.7 | 38.1 |
| % Not reported | 12.7 | 10.7 | 11.5 |
| Waist circumference, cm | 94.8 (13.8) | 93.9 (13.8) | 94.3 (13.8) |
| Greenness a | |||
| 0.5 km buffer | 0.43 (0.10) | 0.46 (0.14) | 0.45 (0.13) |
| 1 km buffer | 0.44 (0.11) | 0.46 (0.14) | 0.45 (0.13) |
| 2 km buffer | 0.45 (0.11) | 0.45 (0.14) | 0.45 (0.13) |
| Geographic size of greenspace b, ha | |||
| 0.5 km buffer | 12.6 (14.5) | 19.8 (20.4) | 16.8 (18.5) |
| 1 km buffer | 60.1 (58.6) | 86.0 (80.2) | 75.4 (73.2) |
| 2 km buffer | 282.0 (247.0) | 354.0 (305.0) | 324.0 (285.0) |
a Mean NDVI; b Area with NDVI ≥ 0.6
Supplementary Table 1 shows correlation coefficients between key variables. Significant but low positive correlations were observed between IRSD and all greenery measures (range: 0.09 to 0.19, p < 0.001), which means that areas with higher SES tended to have more greenery. There were weak negative correlations between waist circumference and the greenery measures (range: -0.06 to -0.05, p < 0.01). Supplementary Table 2 is a descriptive summary of each greenery tertile. The median of mean NDVI values in each tertile ranged from around 0.3 in lower greenery, 0.45 in medium greenery, to 0.6 in higher greenery areas in all buffer sizes. The median size of greenspace ranged from 8 ha in lower greenery, 46 ha in medium greenery to 159 ha in higher greenery areas in the 1 km buffer.
There was a significant negative association between area-level SES and waist circumference. One SD higher score in IRSD (= 100) was associated with 1.76 cm lower waist circumference (95% CI: -2.68, -0.83, p < 0.001). Table 2 shows the results of the Wald tests, indicating the level of significance for the interaction terms between area-level SES and greenery. A significant interaction between IRSD and greenery measures (mean greenness and geographic size of greenspace) was observed for the 2 km buffer but not for the 500 m or 1 km buffer.
Table 2.
Significance of the interaction between area-level SES and greenery measures: results of Wald tests
| Greenery measure | Buffer size | Wald chi-square | P value |
|---|---|---|---|
| Mean greenness (mean NDVI) | 0.5 km | 0.04 | 0.98 |
| 1 km | 1.19 | 0.55 | |
| 2 km | 14.28 | < 0.001 | |
| Geographic size of greenspace (area with NDVI ≥ 0.6) | 0.5 km | 0.42 | 0.81 |
| 1 km | 0.43 | 0.81 | |
| 2 km | 6.02 | 0.049 |
Bold figures denote significant interaction
Models adjusted for age, gender, education status, employment status, household income, population density and corrected for suburb-level clustering
Table 3 show the results of regression analyses stratified by the levels of greenery. Differing patterns emerged for the results from the 0.5 km buffer, and the 1 km and 2 km buffers. Within the 0.5 km buffer, the association between area-level SES and waist circumference remained significant across all tertiles of greenery, except for the lower greenery tertile of mean NDVI, which had a p-value approaching significance (p = 0.064). Although the interaction was significant only for the greenery measures within the 2 km buffer, the same pattern of findings emerged for the 1 km and 2 km buffer greenery measures: there were significant negative associations between area-level SES and waist circumference for both the lower and higher greenery tertiles, but not for the medium greenery tertile.
Table 3.
Associations of area-level SES with waist circumference stratified by tertiles of the greenery measures
| Greenery measure | Buffer size | B (95% CI) | ||
|---|---|---|---|---|
| Lower greenery | Medium greenery | Higher greenery | ||
| Mean greenness (mean NDVI) | 0.5 km | -1.50 (-3.09, 0.09)† | -1.61 (-3.21, -0.01)* | -1.63 (-3.19, -0.08)* |
| 1 km | -2.05 (-3.60, -0.50)** | -0.70 (-2.42, 1.03) | -2.37 (-3.84, -0.89)** | |
| 2 km | -2.98 (-4.64, -1.31)*** | 0.38 (-1.26, 2.03) | -2.59 (-4.12, -1.07)** | |
| Geographic size of greenspace (area with NDVI ≥ 0.6) | 0.5 km | -1.55 (-3.08, -0.02)* | -1.99 (-3.67, -0.32)* | -1.80 (-3.35, -0.26)* |
| 1 km | -2.13 (-3.87, -0.38)* | -1.09 (-2.67, 0.49) | -1.95 (-3.36, -0.54)** | |
| 2 km | -2.61 (-4.06, -1.16)*** | -0.07 (-1.93, 1.80) | -2.84 (-4.33, -1.34)*** | |
† p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001, Bold figures denote significant coefficients
Models adjusted for age, gender, education status, employment status, household income, population density and corrected for suburb-level clustering. Regression coefficients correspond to one SD increment in IRSD score (= 100)
Supplementary Table 3 was produced to better make sense of the findings of moderation. It shows how the lower, medium, and higher greenery areas at lower and higher levels of SES differ in IRSD scores and population density (only for 2 km buffer where the interaction was significant). It was found that lower greenery areas tend to have higher population density (over 20 persons/ha) in both lower and higher SES suburbs. On the other hand, higher greenery areas were lower in population density (below 10 persons/ha), and lower SES areas seemed to be slightly lower in density than higher SES areas. It should be noted that stratification by greenery levels did not markedly reduce the variability in IRSD scores. For instance, the SD of IRSD for the whole sample was 67 in Table 1. This was 57, 62 and 77 in lower, medium, and higher greenery areas (in mean greenness), respectively. Thus, it can be argued that variability in IRSD is unlikely to be a factor contributing to the lack of significant associations for the medium level of greenery.
Discussion
Our study found that participants living in disadvantaged suburbs were likely to have a larger waist circumference than those living in advantaged suburbs. Such uneven distribution of waist circumference is consistent with socioeconomic inequalities of obesity reported in previous studies [3–5]. The study also found that the level of greenery, both greenness (mean NDVI) and geographic size of greenspace (area with NDVI ≥ 0.6) measured within the 2 km buffer zones, appeared to moderate the association between area-level SES and waist circumference. There was a significant socioeconomic gradient in waist circumference in the areas with lower and higher levels of greenery, but this was not the case in the areas with medium levels of greenery. These findings suggest that socioeconomic inequalities in obesity may be reduced by providing a certain level of greenery.
It is challenging to interpret this pattern of moderation. However, it is possible to speculate that areas defined by levels of SES and greenery may have distinct characteristics. While our models did adjust for population density, each greenery stratum has a different range of population density, and this variation was not considered in the stratified analysis. Hence population density may still be a factor in the observed moderating effect. As shown in Supplementary Table 3, lower greenery areas tended to be higher in population density than higher greenery areas. It may be argued that neighbourhoods with lower SES and lower levels of greenery may involve industrial areas, which may have limited resources that can help residents reduce obesity risk (e.g., recreational facilities, fruit/vegetable grocers). Residents may rely on cars for transport in such areas. On the contrary, neighbourhoods with higher SES and lower levels of greenery may be urban centres, where residents are likely to have access to various health-enhancing resources. People living in such areas may be more active for their daily transport and may benefit from better access to and availability of healthy food options or physical activity facilities. It is thus important to understand in what context greenspace exists, e.g., within mixed-use (residential and commercial) neighbourhoods or in predominantly industrial areas. Research incorporating in what way land outside greenspace is used may help to advance the understanding of this topic. The inequalities in obesity may be more pronounced in lower greenery areas, since disadvantaged and advantaged areas may differ in the availability of non-green resources that can help residents to reduce obesity risk. In the areas with higher levels of greenery, lower SES suburbs may be located in urban fringes where greenery may not be fully accessible (e.g., natural reserves, woodland). Even though we removed participants living in outer-regional and remote areas from the study, some inner-regional and peri-urban areas may have such greenery. A study in Ireland found that participants inhabiting peri-urban areas are more likely to have greenery that is inaccessible or not conducive to physical activity [49]. Areas with higher levels of greenery and higher SES may be affluent residential areas, possibly with large parks and well-developed street trees. Thus, in the case of higher levels of greenery, the differences in the types of greenery between lower and higher SES suburbs may explain area-level socioeconomic inequalities in obesity.
Moderation by the medium level of greenery was observed in the 2 km buffers but not in the 0.5 km buffer. (The greenery measures within the 1 km buffer did not show a significant interaction with area-level SES but exhibited the same pattern of associations in stratified analyses with the 2-km buffer greenery measures.) This may indicate that the presence of medium level of greenery in the immediate vicinity of home and that in larger neighbourhood areas are likely to play a different role in affecting the relationships between SES and obesity. Having this level of greenery in a larger area (up to 2 km) around the home may be equally protective against obesity both in disadvantaged and advantaged areas. This suggests that an optimal level of greenery may exist—where insufficient greenery fails to mitigate the impact of SES, and excessive greenery does not provide additional benefits, possibly due to variations in the type, access, or quality of green spaces. Medium levels of greenery might offer a balance that encourages physical activity, reduces stress, and provides environmental benefits in a way that is particularly effective in disadvantaged areas. Further research needs to explore what types of greenery in local neighbourhoods can be conducive to lower levels of obesity, in particular among those living in disadvantaged areas. Research is also needed to understand mechanisms through which neighbourhood greenery may reduce socioeconomic inequalities in obesity. Greenery is known to be associated with a higher level of physical activity and lower level of stress [21]. Research can examine whether such associations are stronger in low SES areas than in high SES areas.
Strengths of the study include participants recruited from diverse urban and inner regional areas across Australia and the use of objective measures for obesity. Even though the AusDiab data were gathered more than 10 years ago, it remains a key national dataset providing a unique opportunity to explore the relationships investigated in our study. Varying buffer sizes have been applied to examine the moderating role of greenery more comprehensively. Additionally, our study stratified the sample into different levels of greenery to explore the moderating role of greenery on the relationship between area-level SES and waist circumference, whereas most studies examining the role of environmental attributes on socioeconomic inequality in obesity have only focused on interaction effects [13]. Such stratification analysis can provide a better understanding of the moderation effect compared to the use of interaction terms [50]. Limitations include the use of NDVI, which captures an area’s greenness based on satellite images. Although this is a common measure of greenery used in diverse disciplines, it may not accurately represent the level of greenery from the viewpoint of human eyes nor reflect the volume of greenery that exists on the ground level. An indicator based on street-level imagery, such as the green view index [51], may be worth exploring. Our greenery measures are concerned only with the quantity. It has been shown that quality aspects of greenery, such as maintenance, aesthetics, and safety, can affect residents’ use of greenery, their physical activity levels, and BMI [52]. Future research needs to investigate whether area-level inequalities in obesity are smaller in the neighbourhoods with higher quality greenspace. Our study did not differentiate types of greenery nor whether it is in the public or private domains, but lower and higher SES areas may differ in the distribution of these greenspaces. It is possible that higher SES areas may have more private greenery such as gardens and backyards. The greenery measures used also did not assess how large each patch of greenspace is. This can be relevant, as the presence of large contiguous greenspace (e.g., large parks) is known to be associated with a higher level of physical activity [41]. Future research could explore whether better access to different types of greenspaces (e.g., gardens, parks, nature reserves) and the presence of large blocks of greenspace may help to reduce area-level socioeconomic inequalities in obesity. This study focused on area-level inequalities of obesity building on a review that reported neighbourhood-level socioeconomic inequalities in obesity [4]. However, such disparities in obesity also occur between individuals with different levels of SES [53]. Research is needed to understand whether environmental attributes may be involved in individual-level socioeconomic inequalities in obesity.
Conclusion
We confirmed what existing studies reported, i.e., socioeconomic inequalities of obesity, in this national sample of middle-aged and older Australian adults. However, we also found that this SES-related inequality was mitigated in areas with the medium level of greenery. Although it is not clear what mechanisms are responsible for such patterns of moderation, our findings suggest a possibility that there is an optimal level of greenery where socioeconomic inequlaities in obesity may be reduced. Given that this study focused on quantity aspects of greenery, which may not be sufficient to fully understand how greenery can be involved in socioeconomic inequalities in obesity, future research should explore how different types and quality aspects of neighbourhood greenery affect such disparities in obesity.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Acknowledgements
The AusDiab study was co-coordinated by the Baker IDI Heart and Diabetes Institute. We gratefully acknowledge the support and assistance given by: K. Anstey, B. Atkins, B. Balkau, E. Barr, A. Cameron, S. Chadban, M. de Courten, A. Kavanagh, D. Magliano, S. Murray, K. Polkinghorne, J. Shaw, T. Welborn, P. Zimmet and all the study participants. For funding orlogistical support, we are grateful to: National Health and Medical Research Council (NHMRC: #233200, #1007544), Australian Government Department of Health and Ageing, Abbott Australasia Pty Ltd, Alphapharm Pty Ltd, Amgen Australia, AstraZeneca, Bristol-Myers Squibb, City Health Centre-Diabetes Service-Canberra, Department of Health and Community Services - Northern Territory, Department of Health and Human Services – Tasmania, Department of Health – New South Wales, Department of Health – Western Australia, Department of Health – South Australia, Department of Human Services –Victoria, Diabetes Australia, Diabetes Australia Northern Territory, Eli Lilly Australia, Estate of the Late Edward Wilson, GlaxoSmithKline, Jack Brockhoff Foundation, Janssen-Cilag, Kidney Health Australia, Marian & FH Flack Trust, Menzies Research Institute, Merck Sharp & Dohme, Novartis Pharmaceuticals, Novo Nordisk Pharmaceuticals, Pfizer Pty Ltd, PrattFoundation, Queensland Health, Roche Diagnostics Australia, Royal Prince Alfred Hospital, Sydney, Sanofi Aventis, and sanofi-synthelabo.
Abbreviations
- AusDiab
Australian diabetes, obesity, lifestyle study
- BMI
Body mass index
- IRSD
Index of relative socio-economic disadvantage
- NDVI
Normalized difference vegetation index
- SES
Socio-economic status
Author contributions
SS and TS wrote the main manuscript, which was reviewed and edited by NH, MC, SM and NO. SS conducted data analysis with help from MC. NDVI data were acquired by SM. All authors contributed to conceptualisation of the study and interpretation of the results. All authors read and approved the final manuscript.
Funding
SS is supported by the Swinburne University Postgraduate Research Award (SUPRA) scholarship by Swinburne University of Technology. Research at the Murdoch Children’s Research Institute is supported by the Victorian Government’s Operational Infrastructure Program.
Data availability
The data that support the findings of this study are available from the authors upon reasonable request and with the permission of Baker Heart & Diabetes Institute.
Declarations
Ethics approval
Ethics approval for the AusDiab study was granted by the Alfred Hospital Ethics Committee (no. 39/11). Written informed consent was obtained from all participants.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Venegas O, Mehrzad R. Prevalence and trends in obesity in the United States and affluent countries. Obesity: Elsevier; 2020. pp. 19–41. [Google Scholar]
- 2.Organisation for Economic Co-operation and Development (OECD). The heavy burden of obesity: The economics of prevention. OECD Publishing, Paris. 2019. https://www.oecd.org/en/publications/2019/10/the-heavy-burden-of-besity_0beeff1b.html. Accessed 15 Jan 2024.
- 3.Keramat SA, Alam K, Gow J, Biddle SJ. Impact of disadvantaged neighborhoods and lifestyle factors on adult obesity: evidence from a 5-year cohort study in Australia. AM J Health Promot. 2021;35(1):28–37. [DOI] [PubMed] [Google Scholar]
- 4.Mohammed SH, Habtewold TD, Birhanu MM, Sissay TA, Tegegne BS, Abuzerr S, et al. Neighbourhood socioeconomic status and overweight/obesity: a systematic review and meta-analysis of epidemiological studies. BMJ Open. 2019;9(11):e028238. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Sepúlveda-Peñaloza A, Cumsille F, Garrido M, Matus P, Vera-Concha G, Urquidi C. Geographical disparities in obesity prevalence: small-area analysis of the Chilean National Health surveys. BMC Public Health. 2022;22(1):1443. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Hoffmann R, Eikemo TA, Kulhánová I, Kulik MC, Looman C, Menvielle G, et al. Obesity and the potential reduction of social inequalities in mortality: evidence from 21 European populations. Eur J Public Health. 2015;25(5):849–56. [DOI] [PubMed] [Google Scholar]
- 7.Marmot M, Bell R. Fair society, healthy lives. Public Health. 2012;126(Suppl 1):S4–10. [DOI] [PubMed] [Google Scholar]
- 8.Chandrabose M, Rachele JN, Gunn L, Kavanagh A, Owen N, Turrell G, et al. Built environment and cardio-metabolic health: systematic review and meta-analysis of longitudinal studies. Obes Rev. 2019;20(1):41–54. [DOI] [PubMed] [Google Scholar]
- 9.Hadgraft N, Chandrabose M, Bok B, Owen N, Woodcock I, Newton P, et al. Low-carbon built environments and cardiometabolic health: a systematic review of Australian studies. Cities Health. 2022;6(2):418–31. [Google Scholar]
- 10.McGowan V, Buckner S, Mead R, McGill E, Ronzi S, Beyer F, et al. Examining the effectiveness of place-based interventions to improve public health and reduce health inequalities: an umbrella review. BMC Public Health. 2021;21:1888. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Cowie CT, Ding D, Rolfe MI, Mayne DJ, Jalaludin B, Bauman A, et al. Neighbourhood walkability, road density and socio-economic status in Sydney, Australia. Environ Health. 2016;15:58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Schüle SA, Gabriel KM, Bolte G. Relationship between neighbourhood socioeconomic position and neighbourhood public green space availability: an environmental inequality analysis in a large German city applying generalized linear models. Int J Hyg Environ Health. 2017;220(4):711–8. [DOI] [PubMed] [Google Scholar]
- 13.Selvakumaran S, Lin C-Y, Hadgraft N, Chandrabose M, Owen N, Sugiyama T. Area-level socioeconomic inequalities in overweight and obesity: systematic review on moderation by built-environment attributes. Health Place. 2023;83:103101. [DOI] [PubMed] [Google Scholar]
- 14.Feuillet T, Valette J-F, Charreire H, Kesse-Guyot E, Julia C, Vernez-Moudon A, et al. Influence of the urban context on the relationship between neighbourhood deprivation and obesity. Soc Sci Med. 2020;265:113537. [DOI] [PubMed] [Google Scholar]
- 15.Oyeyemi AL, Adegoke BO, Oyeyemi AY, Deforche B, Bourdeaudhuij ID, Sallis JF. Environmental factors associated with overweight among adults in Nigeria. Int J Behav Nutr Phy. 2012;9:32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.McCormack GR, Shiell A. In search of causality: a systematic review of the relationship between the built environment and physical activity among adults. Int J Behav Nutr Phy. 2011;8:125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Sallis JF, Saelens BE, Frank LD, Conway TL, Slymen DJ, Cain KL, et al. Neighborhood built environment and income: examining multiple health outcomes. Soc Sci Med. 2009;68(7):1285–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.van Diepen RJ, van Erpecum CL, Tabak D, van Zon SKR, Bultmann U, Smidt N. Neighborhood socioeconomic differences in BMI: the role of fast-food outlets and physical activity facilities. Obesity. 2023;31(2):506–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Wang MC, Kim S, Gonzalez AA, MacLeod KE, Winkleby MA. Socioeconomic and food-related physical characteristics of the neighbourhood environment are associated with body mass index. J Epidemiol Commun Health. 2007;61(6):491–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Ball K, Lamb K, Travaglini N, Ellaway A. Street connectivity and obesity in Glasgow, Scotland: impact of age, sex and socioeconomic position. Health Place. 2012;18(6):1307–13. [DOI] [PubMed] [Google Scholar]
- 21.Yang B-Y, Zhao T, Hu L-X, Browning MH, Heinrich J, Dharmage SC, et al. Greenspace and human health: an umbrella review. Innovation. 2021;2(4):100164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Wanner M, Richard A, Martin B, Faeh D, Rohrmann S. Associations between self-reported and objectively measured physical activity, sedentary behavior and overweight/obesity in NHANES 2003–2006. Int J Obes. 2017;41(1):186–93. [DOI] [PubMed] [Google Scholar]
- 23.Tomiyama AJ. Stress and obesity. Ann Rev Psychol. 2019;70(1):703–18. [DOI] [PubMed] [Google Scholar]
- 24.Stankov I, Howard NJ, Daniel M, Cargo M. Policy, research and residents’ perspectives on built environments implicated in heart disease: a concept mapping approach. Int J Env Res Pub Health. 2017;14(2):170. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Zhang B, Loh V, Reid RA, Nyanhanda T, Jamalishahni T, Turrell G, et al. The moderating effects of greenspace on the association between neighbourhood disadvantage and obesity among mid-to-older aged Australian adults. Health Place. 2024;87:103263. [DOI] [PubMed] [Google Scholar]
- 26.Hobbs M, Green MA, Griffiths C, Jordan H, Saunders J, Grimmer H, et al. Access and quality of parks and associations with obesity: a cross-sectional study. SSM Popul Health. 2017;3:722–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Mitchell R, Popham F. Effect of exposure to natural environment on health inequalities: an observational population study. Lancet. 2008;372(9650):1655–60. [DOI] [PubMed] [Google Scholar]
- 28.Wang Q, Lan Z. Park green spaces, public health and social inequalities: understanding the interrelationships for policy implications. Land Use Pol. 2019;83:66–74. [Google Scholar]
- 29.Stommel M, Schoenborn CA. Accuracy and usefulness of BMI measures based on self-reported weight and height: findings from the NHANES & NHIS 2001–2006. BMC Public Health. 2009;9:421. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Janssen I, Katzmarzyk PT, Ross R. Waist circumference and not body mass index explains obesity-related health risk. Am J Clin Nutr. 2004;79(3):379–84. [DOI] [PubMed] [Google Scholar]
- 31.Ross R, Neeland IJ, Yamashita S, Shai I, Seidell J, Magni P, et al. Waist circumference as a vital sign in clinical practice: a consensus statement from the IAS and ICCR Working Group on visceral obesity. Nat Rev Endocrinol. 2020;16(3):177–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Dunstan DW, Zimmet PZ, Welborn TA, Cameron AJ, Shaw J, De Courten M, et al. The Australian diabetes, obesity and lifestyle study (AusDiab)—methods and response rates. Diabetes Res Clin Pr. 2002;57(2):119–29. [DOI] [PubMed] [Google Scholar]
- 33.Australian Bureau of Statistics. Census of Population and Housing: Socio-Economic Indexes for Areas (SEIFA), Australia, 2011. Canberra: Australian Bureau of Statistics. 2011. https://www.abs.gov.au/ausstats/abs@.nsf/DetailsPage/2033.0.55.0012011. Accessed 10 Nov 2023.
- 34.Australian Institute of Health and Welfare. Health across Socioeconomic Groups. Canberra: Australian Institute for Health and Welfare. 2022. https://www.aihw.gov.au/reports/australias-health/health-across-socioeconomic-groups. Accessed 5 Feb 2024.
- 35.Australian Bureau of Statistics. Australian Statistical Geography Standard (ASGS): Volume 3 - Non-ABS Structures. July 2016. Canberra: Australian Bureau of Statistics. 2016. https://www.abs.gov.au/ausstats/abs@.nsf/Lookup/by Subject/1270.0.55.003~July 2016~Main Features~State Suburbs (SSC)~9. Accessed 10 Nov 2023.
- 36.Aryal J, Sitaula C, Aryal S. NDVI threshold-based urban green space mapping from sentinel-2a at the local Governmental Area (LGA) level of Victoria, Australia. Land. 2022;11(3):351. [Google Scholar]
- 37.de la Iglesia Martinez A, Labib S. Demystifying normalized difference vegetation index (NDVI) for greenness exposure assessments and policy interventions in urban greening. Environ Res. 2023;220:115155. [DOI] [PubMed] [Google Scholar]
- 38.Pereira G, Christian H, Foster S, Boruff BJ, Bull F, Knuiman M, et al. The association between neighborhood greenness and weight status: an observational study in Perth Western Australia. Environ Health. 2013;12:49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Ramsey S, Mavoa S. Annual Normalized Difference Vegetation Index time-series data for Australian statistical areas. OSF Preprints; 2021. Available from: osf.io/ryqvs.
- 40.Chandrabose M, Hadgraft N, Owen N, Mavoa S, Sugiyama T. Joint associations of neighbourhood walkability and greenery with walking among middle-aged and older adults: findings from diverse urban settings in Australia. Health Place. 2024;89:103334. [DOI] [PubMed] [Google Scholar]
- 41.Sugiyama T, Francis J, Middleton NJ, Owen N, Giles-Corti B. Associations between recreational walking and attractiveness, size, and proximity of neighborhood open spaces. Am J Public Health. 2010;100(9):1752–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.US Geological Survey, Remote sensing phenology. 2018. https://www.usgs.gov/special-topics/remote-sensing-phenology. Accessed 10 Nov 2023.
- 43.Su JG, Dadvand P, Nieuwenhuijsen MJ, Bartoll X, Jerrett M. Associations of green space metrics with health and behavior outcomes at different buffer sizes and remote sensing sensor resolutions. Environ Int. 2019;126:162–70. [DOI] [PubMed] [Google Scholar]
- 44.Adams MA, Frank LD, Schipperijn J, Smith G, Chapman J, Christiansen LB, et al. International variation in neighborhood walkability, transit, and recreation environments using geographic information systems: the IPEN adult study. Int J Health Geogr. 2014;13:43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Sugiyama T, Kubota A, Sugiyama M, Cole R, Owen N. Distances walked to and from local destinations: age-related variations and implications for determining buffer sizes. J Transp Health. 2019;15:100621. [Google Scholar]
- 46.Browning M, Lee K. Within what distance does greenness best predict physical health? A systematic review of articles with GIS buffer analyses across the lifespan. Int J Env Res Pub Health. 2017;14(7):675. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Leonardi C, Simonsen NR, Yu Q, Park C, Scribner RA. Street connectivity and obesity risk: evidence from electronic health records. Am J Prev Med. 2017;52(1):S40–7. [DOI] [PubMed] [Google Scholar]
- 48.Agresti A. Foundations of linear and generalized linear models. Hoboken, New Jersey: Wiley; 2015. [Google Scholar]
- 49.Dempsey S, Lyons S, Nolan A. Urban green space and obesity in older adults: evidence from Ireland. SSM Popul Health. 2018;4:206–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Supplee LH, Kelly BC, MacKinnon DM, Barofsky MY. Introduction to the special issue: subgroup analysis in prevention and intervention research. Prev Sci. 2013;14:107–10. [DOI] [PubMed] [Google Scholar]
- 51.Zhu H, Nan X, Yang F, Bao Z. Utilizing the green view index to improve the urban street greenery index system: a statistical study using road patterns and vegetation structures as entry points. Landsc Urban Plan. 2023;237:104780. [Google Scholar]
- 52.Knobel P, Maneja R, Bartoll X, Alonso L, Bauwelinck M, Valentin A, et al. Quality of urban green spaces influences residents’ use of these spaces, physical activity, and overweight/obesity. Environ Pollut. 2021;271:116393. [DOI] [PubMed] [Google Scholar]
- 53.Ogden CL. Prevalence of obesity among adults, by household income and education—United States, 2011–2014. MMWR-Morbid Mortal W. 2017;66:1369–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available from the authors upon reasonable request and with the permission of Baker Heart & Diabetes Institute.