Association between Land Use Mix and Respiratory Symptoms and Asthma in Children from the Generation XXI Birth Cohort

Abstract

Land use mix (LUM) in the neighbourhoods has been associated with healthier lifestyles. However, less is known about the association between LUM and health outcomes, namely during childhood. The objective of this study was to evaluate the association between different LUM indexes for Porto Metropolitan Area and asthma and respiratory symptoms in children. A cross-sectional analysis was performed involving 6260 children enrolled in Generation XXI. Land use around the child’s residence was assessed with the Portuguese official map of land cover using a GIS. Generalized linear mixed-effects models were fitted to estimate the association between LUM and respiratory symptoms and asthma at 7 years of age. Adjusted associations were quantified using odds ratio (OR) and 95% confidence interval (95% CI). After adjustment, LUM was associated with a lower odds of wheezing in the last 12 months [OR (95% CI) = 0.37 (0.15; 0.93) using Shannon’s Evenness Index within 500 m; and OR = 0.93 (0.89; 0.98) using the number of different land use types within 250 m]. Living in neighbourhoods with high LUM has a protective effect on current wheezing symptoms. Our results highlight the association between LUM and respiratory symptoms among children, suggesting that public health considerations should be incorporated in land use decision-making.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11524-021-00604-4.

Introduction

Urbanization is increasing rapidly worldwide, driving changes in the spatial composition, structure, and density of urban land uses largely due to social and economic drivers [1]. This trend of urban growth has led to the conversion of land of agriculture, forest, vegetation, and water bodies into artificial and impervious grey surfaces, including built-up land [2]. Recent studies have assessed health impacts of specific land use types, including grey, green, and blue spaces [3–5], but the effects of land use mix (LUM) on respiratory health have not been sufficiently explored.

LUM is an indicator for built environment, usually represented by the level of different land use types within a given neighbourhood, with higher values indicating greater land use heterogeneity [6, 7]. During most of the twentieth century, and supported by spatial policies, land uses became increasingly separated, namely as residential and work areas [8]. However, in the late 1980s and early 1990s, the concepts of diverse and multifunctional land use have become relevant aspects of spatial quality in urban areas [8] and a strategy to promote healthy community environments [9]. In the last decades, several measures have been proposed to quantify the diversity of land use [10–13], mainly derived from community ecology, used to quantify plant and animal species diversity [13, 14]. Nowadays, diversity of land use is most often assessed using entropy and dissimilarity scores, which measure the distribution of different land use types throughout the urban area [10, 12]. Song et al. [11] reviewed the conceptual and mathematical formulations of different measures of LUM, such as Atkinson index, balance, entropy, and Herfindahl–Hirschman indices, and suggested that each measure provides a distinct perspective on LUM based on the number of land use types, land use balance, or evenness.

Environmental factors, such as traffic-related air pollution, residential proximity to roads and heavy traffic, lower exposure to green spaces, and biodiversity, are associated with an increased risk of asthma-related symptoms [15, 16]. To the best of our knowledge, the association between LUM and asthma or respiratory symptoms in children was never reported. However, some studies suggest that the diversity of land uses is related to higher level of physical activity [17, 18] and a reduction in air pollution [19], well-established risk factors for asthma [20–22]. Mixed land use has been associated with increased levels of physical activity and active transportation, by providing a range of recreational and leisure facilities within a walking distance [17, 18, 23]. A higher level of regular physical activity is a protective factor for the development of asthma [20], by improving lung function and reducing airway inflammation [24]. Furthermore, neighbourhoods with a higher LUM have better access to services and resources, supporting sustainable forms of transport, such as walking and cycling, thereby reducing vehicle emissions [25, 26]. Mixing of land use was found to be associated with lower emissions of nitric oxides and particulate matter (PM_2.5) [19, 27]. A phytochemical simulation showed that cities with mixed land use may provide a better urban air quality than cities with disperse and segregated land use [28]. Accordingly, the US Centers for Disease Control and Prevention identified LUM as a strategy to promote healthy community environments, promoting active travel, increasing the viability of transportation alternatives, reducing private vehicle use and its associated impacts, and reporting that the potential benefits of LUM may also help to build a sense of place for local communities [9].

Epidemiological studies on the effect of LUM on health are few and inconclusive [18, 29–33]. Some reported a positive association between higher LUM and level of physical activity, with a lower risk of obesity among children; other found no such associations [18]. These results may reflect the differences in the approach to measuring LUM, the lack of specificity in the land use types included [11], the unavailability of detailed land use maps, and the variations in geographical scales used, such as census units, metropolitan regions, and buffers around individual neighbourhoods [12]. In addition, the choice of LUM measures is often based upon data availability and statistical criteria rather than theoretical models or empirical findings [34]. Thus, it is important to conduct comparison studies on how different measures of LUM affect respiratory health in order to provide evidence-based guidance for future studies on the relationship between the urban environment, diversity of land uses, and respiratory health.

Therefore, this cross-sectional analysis aimed to evaluate the association between different LUM indexes for Porto Metropolitan Area neighbourhoods and asthma and respiratory symptoms at 7 years of age.

Material and Methods

Study Participants

The study was developed based on data collected from the Generation XXI (G21) birth cohort, which includes 8647 newborns delivered in 2005/2006 in Porto Metropolitan Area, Northern Portugal [35]. Initial recruitment occurred in the five public maternity units, where 95% of the region’s births occurred. All participants were invited to be re-evaluated at 4 (2009/2011), 7 (2012/2014), and 10 (2015/2017) years of age. Ethical approval for the cohort recruitment was obtained from the University Hospital Center of São João Ethics Committee, and signed informed consent was obtained from all participants. All phases of the study followed the Ethical Principles for Medical Research Involving Human Subjects expressed in the Declaration of Helsinki.

Each child residential address has been collected during telephone calls with the caregiver at the time of each evaluation. At the 7-year evaluations, caregivers were asked to confirm the child’s residential address once again. Afterwards, residential addresses were processed and georeferenced using ArcGIS Online World Geocoding Service and Google Earth, due to their superior positional accuracy [36]. For the present study, the subsample of children living in Porto Metropolitan Area since birth to data collection and with information on at least one respiratory outcome at 7 years of age (clinical diagnosis of asthma, rhinitis, allergy or eczema, or respiratory symptoms) was included, totalling 6260 children.

Health Outcomes

Data on health outcomes were collected through questionnaires based on the International Study of Asthma and Allergies in Childhood (ISAAC) standardized questionnaire, which have been completed by the participants’ caregivers at the 7-year follow-up. Asthma, rhinitis, allergy, and eczema were defined as a positive answer to “Has your child ever been diagnosed with asthma by a physician?”, “Has your child ever been diagnosed with rhinitis by a physician?”, “Has your child ever been diagnosed with allergy by a physician?”, and “Has your child ever been diagnosed with eczema by a physician?”, respectively. The presence of respiratory symptoms up to age 7 years, as wheezing and rhinitis, was defined as a positive answer to “Has your child ever had wheezing or whistling in the chest at any time in the past?” and “Has your child ever had a problem with sneezing or a runny or blocked nose, when he/she did not have a cold or the flu?” questions. The presence of symptoms of wheezing, dry cough, and rhinitis in the past 12 months was defined as a positive answer to “Has your child experienced episodes of wheezing in the past 12 months, not associated with a cold or flu?”, “Has your child suffered from recurrent dry cough episodes in the past 12 months, not associated with a cold or flu”, and “Has your child had a problem with sneezing, or a runny, or blocked nose in the past 12 months, not associated with a cold or flu?”, respectively. The presence of respiratory symptoms in the past 12 months was defined as a positive answer to any of the questions related to wheezing, cough or rhinitis.

Exposure Assessment

Different LUM indexes which capture the two concepts related to urban land use—distance and quantity—were considered in this study [11]. We have selected four indexes: (a) Shannon’s Evenness Index, (b) number of different land use types, (c) proportion of largest function, and (d) Simpson’s diversity index S [8, 11]. In addition, the four chosen LUM measures are among the most commonly used in urban health scientific literature [25, 30, 34], which, as mentioned in the introduction, is predominantly focused on the associations between LUM, physical activity, and obesity.

These indexes have been defined as integral measures, since they are calculated based upon an area’s land use distribution as a whole, reflecting the land use balance or whether different land uses are present in equal proportion within a specific area [11]. Unlike divisional measures that consider land use measurements at the finer level of the district, integral measures are applied more often to small areas, such as neighbourhoods [11].

LUM was calculated using the 2015 Portuguese official map of land cover (called Carta de Uso e Ocupação do Solo (COS)) [37] at the scale of 1:25,000 and with a minimal cartographic unit of 1 hectare. These maps were produced by the Directorate-General for the Territory based on ortophotomaps and constitute the most detailed land use maps available for Continental Portugal [38]. Land use was calculated within 250 m and 500 m from the child’s residence at 7 years of age. In 2015, the land use was divided into five main classes (artificialized, agricultural and agroforestry, forests and natural areas, wetlands, and water bodies), subdividing these into 40 more detailed classes (online Supp. Table A.1) [37]. We used ArcMap 10.7.1 to extract the land use data within 250 m and 500 m from the child’s residence. These buffer sizes were selected to cover immediate and slightly more distant areas of exposure. The maximum threshold distance of 500 m was considered because children do not tend to transverse most of their local area. In fact, it was found that children only travel up to 25% of their traditional neighbourhood, commonly defined as an area within 800 m and 1600 m from residence [39]. Moreover, most of studies investigating the impacts of urban environmental exposures have used the same buffer sizes, allowing the comparison of our findings for the associations between respiratory health and LUM with the results for other urban exposures, such as green and blue spaces [5].

Shannon’s Evenness Index

The calculation of the LUM was based on Shannon’s Evenness Index, using the following equation (Eq. 1):

$$LUM=-\mathrm{\Sigma }\frac{pi\times \ln pi}{\ln n}$$ 1

LUM is calculated by multiplying each proportion of land use type by its logarithm and dividing the sum of all land use type products by the logarithm of the total possible land use types, n. LUM calculated based on entropy index provides the proportional abundance of each type of land use within an area, yielding a score between 0 and 1, with higher values corresponding to greater land use mixture [11].

Number of Different Land Use Types

This measure quantifies the number of different land use types within each polygon. The value ranges from 1 to n (which corresponds to the total number of land use types that are distinguished within a specific buffer size) [8]. Despite the similarities between the Shannon’s Evenness Index and the number of different land use types, the Shannon’s Evenness Index considers the relative proportion of two or more land use types within an area [11], being sensitive to the size distribution of all land uses within these areas [8], while the latter simply expresses the total number of land use types that are present [8].

Proportion of the Largest Function

This index corresponds to the proportion of the largest function within each polygon, ranging from $\frac{1}{n}$ (for an equal share of all land uses) to 1, if only a single land use is present [8], with a higher value indicating a greater land use homogeneity.

Simpson’s Diversity Index S

The LUM was calculated based on the Simpson’s diversity index S, using the following equation (Eq. 2):

$$S=1-\mathrm{\Sigma }{\mathit{pi}}^2$$ 2

The Simpson’s diversity index considers the proportional abundance of each type of land use within an area, providing an effective function richness, ranging from 0 to 1-$\frac{1}{n}$ [8], with a higher value indicating a greater land use heterogeneity.

Figure 1 shows the LUM according to each index for each neighbourhood in Porto Metropolitan Area.

Fig. 1.

Estimated land use mix indexes

Covariates

A comprehensive set of covariates was selected based on previous studies on the association between environmental factors and asthma and respiratory symptoms [17, 20, 40–45]. The following variables were considered: child’s sex, physical activity, maternal education level, household crowding, Normalized Difference Vegetation Index (NDVI), outdoor levels of nitrogen dioxide (NO₂), and neighbourhood socioeconomic deprivation.

The practice of physical activity was defined when participants were engaged in some sports and active leisure activities performed on a repeated and regular basis. Maternal education level was measured in years of schooling and categorized into two classes: primary (≤ 9 years of education, International Standard Classification of Education (ISCED) 2011 classes 0–2, corresponding to the compulsory education in Portugal in the age cohort of the G21 parents), and secondary/tertiary (> 9 years, ISCED class 3–6) [46]. Household crowding corresponded to the ratio between the number of household occupants and the number of rooms and was characterized as < 1.5 or ≥ 1.5 occupants per room [47]. NDVI was calculated based on land surface reflectance of visible red (VISR) and near-infrared (NIR) wavelengths using the ArcMap 10.5 to process satellite images and QGIS 3.8 to extract the average NDVI within 250 m and 500 m from the child’s residence [5, 16, 48]. NDVI identifies the presence of healthy vegetation, with higher values of NDVI corresponding to healthy and dense vegetation, and lower values showing burnt or sparse vegetation. On the other hand, land cover cartography distinguishes different features of land cover, including agricultural areas, forests and semi-natural areas, artificial surfaces, urban fabric, industrial, commercial, and transport units, bodies of water, wetlands, and other features [49], and provides no information of the coverage of healthy green vegetation. Only images with 5% or less cloud coverage from Landsat 5 (spatial resolution: 30 m) during the spring/summer period (peak of vegetation) of the baseline assessment (2005/2006) was used. Outdoor levels of NO₂ were obtained from AirBase—the European air quality database from the European Environment Agency (EEA). The EEA provides 1 km grids for NO₂, which were created based on monitoring station data. More details about this dataset can be found elsewhere [50]. Neighbourhood socioeconomic deprivation was determined based on the weighted sum of the following standardized variables obtained from census data: % of non-owned households, % of households without indoor flushing, % of households with ≤ 5 rooms, % of blue-collar (i.e. manual) occupations, % of residents with low education level (≤ 6^th grade), % of non-employers, % of unemployed individuals looking for a job, and % of foreign residents [51]. The index was then categorized into quintiles (the first quintile corresponds to least deprived to the fifth quintile to most deprived).

Statistical Analysis

Distribution of continuous variables was checked for normality. Since non-Gaussian distributions were observed, non-parametric tests were performed for inferential statistics: Mann–Whitney tests to analyse differences between continuous variables and chi-square test to compare proportions.

Generalized linear mixed-effects models were fitted to estimate the adjusted association between the different indexes of LUM and respiratory symptoms, allergic sensitization, allergic diseases, and asthma at 7 years of age, considering the neighbourhood as a random effect (R Package lme4) [52]. Adjusted models were fitted using child’s sex, socioeconomic conditions, measured by the maternal education level and household crowding, and neighbourhood characteristics, such as the NDVI, and neighbourhood socioeconomic deprivation. Results were expressed as odds ratios (OR) and 95% confidence intervals (95% CI).

We hypothesized that physical activity and outdoor levels of NO₂ could be potential mechanisms underlying the association between LUM and asthma and respiratory symptoms [17, 20, 40, 43]. The mediation effects were calculated based on the structural equation modelling (SEM). SEM methods are based on path analysis and simultaneously test both the direct and indirect effects of predictive variables on the outcome [53]. A SEM was fitted using logistic regression analysis to estimate the indirect effects of physical activity and outdoor levels of NO₂ (R Package lavaan) [54]. Statistical analysis was performed using R software version 3.6.2.

Results

Participants

Participants’ and exposure characteristics are presented in Table 1. Among the 6260 included participants, 3214 (51.3%) were female. The prevalence of asthma was 6.2%, and the period prevalence of individual respiratory symptoms at the age of 7 ranged in the last 12 months between 32.5 and 85.9%, being higher for symptoms related to rhinitis.

Table 1.

Characteristics of participants (Generation XXI, n = 6260)

Characteristic	n (%)
At birth
Maternal education
≤ 9 years	2872 (45.9)
> 9 years	3388 (54.1)
Household crowing (occupants per room) (n = 6065)
< 1.5	5104 (84.2)
≥ 1.5	961 (15.8)
Neighbourhood socioeconomic deprivation
Q1 (least deprived)	836 (13.4)
Q2	1282 (20.5)
Q3	1591 (25.4)
Q4	1391 (22.2)
Q5 (most deprived)	1160 (18.5)
NDVI*
250 m	0.201 (0.143; 0.251)
500 m	0.223 (0.166; 0.273)
At 7 years
Outdoor levels of NO2*	15.8 (15.3; 18.5)
Practice of physical activity (n = 5322)	4546 (85.4)
Clinical diagnosis (ever)
Asthma (n = 6826)	426 (6.2)
Rhinitis (n = 6805)	523 (7.7)
Allergy (n = 6792)	1143 (16.8)
Eczema (n = 6575)	882 (13.4)
Ever symptoms
Wheezing (n = 5805)	2872 (49.5)
Rhinitis (n = 4863)	1451 (25.0)
Symptoms in past 12 months
Wheezing (n = 2872)	995 (34.6)
Rhinitis (n = 1457)	1252 (85.9)
Dry cough (n = 5796)	1883 (32.5)

*Median (25th; 75th percentile)

Table A.2 (online Supp.) provides a summary for each LUM index by buffer size. The mean LUM values increased by buffer size, except for the proportion of largest function.

Associations with LUM Indexes

The mean values of the proportion of largest function within 250 m were higher among children with eczema (0.57 ± 0.18 vs. 0.55 ± 0.18, p = 0.020). Lower mean values based on Simpson’s diversity index S within 250 m were observed among children with no eczema (0.57 ± 0.19 vs. 0.56 ± 0.19, p = 0.050). No differences were observed between mean LUM values and clinical diagnosis of asthma, rhinitis, and allergy at 7 years of age (online Supp. Table A.3). The mean LUM values based on the number of different land use types within 250 m were significantly lower among children with wheezing (4.83 ± 1.72 vs. 4.97 ± 1.70, p = 0.033) in the last 12 months at 7 years of age (online Supp. Table A.4). The mean values of Shannon’s Evenness Index and Simpson’s diversity index S within 250 m were also lower among children with cough in the last 12 months (0.29 ± 0.11 vs. 0.30 ± 0.11, p = 0.023; and 0.56 ± 0.19 vs. 0.57 ± 0.19, p = 0.048, respectively). Additionally, the mean number of different land use types within 250 m and 500 m were also significantly lower among children with cough (4.82 ± 1.70 vs. 4.93 ± 1.72, p = 0.017; and 8.09 ± 2.08 vs. 8.21 ± 2.15, p = 0.047, respectively) in the last 12 months.

Tables 2 and 3 show the crude and adjusted associations between residential exposure to different indexes of LUM at specific time points and clinical diagnosis of asthma and allergic diseases and respiratory symptoms at 7 years of age. A higher land use diversity using Shannon’s Evenness Index within 500 m and was associated with a lower odds of wheezing in the last 12 months [OR (95% CI) = 0.37 (0.15; 0.93)] (Table 3). Additionally, higher exposure to land use diversity using the number of different land use types, within 250 m from the residence at 7 years of age, was also associated with a lower odds of wheezing in the last 12 months [OR (95% CI) = 0.93 (0.89; 0.98)]. No significant associations were observed between LUM indexes and clinical diagnosis of asthma and allergic diseases up to age 7 years (Table 2).

Table 2.

Associations (OR, 95% confidence intervals (95% CI)) between land use mix indexes by buffer size and asthma, allergy, and eczema at 7 years of age

Land use mix indexes	Asthma		Rhinitis		Allergy		Eczema
	Crude OR	Adjusted ORa	Crude OR	Adjusted ORa	Crude OR	Adjusted ORa	Crude OR	Adjusted ORa
Shannon’s Evenness Index
250 m	1.10 (0.44; 2.78)	0.89 (0.33; 2.40)	0.86 (0.37; 1.96)	1.02 (0.42; 2.50)	0.99 (0.55; 1.80)	1.23 (0.64; 2.33)	0.62 (0.32; 1.18)	0.69 (0.34; 1.40)
500 m	1.31 (0.45; 3.83)	1.04 (0.33; 3.26)	0.75 (0.29; 1.94)	0.74 (0.27; 2.05)	0.78 (0.40; 1.54)	0.87 (0.42; 1.82)	0.67 (0.32; 1.41)	0.68 (0.30; 1.53)
Number of different land use types
250 m	1.02 (0.96; 1.09)	1.01 (0.95; 1.08)	0.98 (0.92; 1.03)	0.99 (0.93; 1.04)	0.99 (0.95; 1.03)	1.00 (0.96; 1.04)	0.97 (0.93; 1.01)	0.98 (0.93; 1.02)
500 m	1.02 (0.97; 1.07)	1.02 (0.97; 1.07)	1.00 (0.96; 1.05)	0.99 (0.95; 1.04)	1.01 (0.98; 1.04)	1.01 (0.98; 1.05)	0.99 (0.96; 1.02)	0.99 (0.95; 1.02)
Proportion of largest function
250 m	0.97 (0.55; 1.71)	1.11 (0.61; 2.04)	1.03 (0.62; 1.71)	0.90 (0.52; 1.56)	0.93 (0.65; 1.34)	0.81 (0.54; 1.20)	1.46 (0.98; 2.18)	1.37 (0.88; 2.11)
500 m	1.09 (0.55; 2.14)	1.30 (0.63; 2.66)	1.30 (0.71; 2.38)	1.32 (0.69; 2.51)	1.29 (0.84; 1.98)	1.23 (0.77; 1.96)	1.41 (0.88; 2.27)	1.39 (0.83; 2.33)
Simpson’s diversity index S
250 m	1.06 (0.61; 1.84)	0.93 (0.52; 1.68)	0.95 (0.58; 1.56)	1.07 (0.63; 1.82)	1.04 (0.73; 1.48)	1.19 (0.81; 1.75)	0.76 (0.52; 1.11)	0.81 (0.53; 1.24)
500 m	1.12 (0.53; 2.39)	0.93 (0.42; 2.09)	0.74 (0.39; 1.42)	0.73 (0.36; 1.48)	0.78 (0.49; 1.25)	0.82 (0.49; 1.37)	0.74 (0.44; 1.24)	0.74 (0.42; 1.32)

^aAdjusted sex, maternal education level, household crowding, NDVI, and neighbourhood socioeconomic deprivation

Table 3.

Associations (OR, 95% confidence intervals (95% CI)) between land use mix indexes by buffer size and respiratory symptoms at 7 years of age

Land use mix indexes	Ever symptoms				Symptoms in the past 12 months
	Wheezing		Rhinitis		Wheezing		Rhinitis		Cough
	Crude OR	Adjusted ORa	Crude OR	Adjusted ORa	Crude OR	Adjusted ORa	Crude OR	Adjusted ORa	Crude OR	Adjusted ORa
Shannon’s Evenness Index
250 m	1.45 (0.89; 2.35)	1.69 (0.99; 2.86)	0.71 (0.41; 1.23)	0.87 (0.48; 1.57)	0.68 (0.32; 1.45)	0.44 (0.20; 1.00)	1.06 (0.25; 4.42)	1.08 (0.24; 4.83)	0.61 (0.37; 1.01)	0.65 (0.37; 1.12)
500 m	1.18 (0.68; 2.06)	1.62 (0.89; 2.96)	0.58 (0.31; 1.09)	0.69 (0.35; 1.36)	0.65 (0.28; 1.53)	0.37 (0.15; 0.93)	0.63 (0.12; 3.30)	0.52 (0.09; 2.90)	0.61 (0.34; 1.09)	0.78 (0.42; 1.46)
Number of different land use types
250 m	1.01 (0.98; 1.05)	1.02 (0.99; 1.06)	0.97 (0.93; 1.00)	0.98 (0.94; 1.01)	0.95 (0.91; 1.00)	0.93 (0.89; 0.98)	1.02 (1.01; 1.03)	1.01 (0.92; 1.11)	0.96 (0.93; 0.99)	0.97 (0.94; 1.01)
500 m	1.02 (0.99; 1.05)	1.04 (0.99; 1.06)	0.98 (0.95; 1.01)	0.98 (0.95; 1.01)	0.98 (0.94; 1.02)	0.96 (0.92; 1.00)	1.02 (0.94; 1.09)	1.01 (0.94; 1.09)	0.97 (0.95; 1.00)	0.99 (0.96; 1.01)
Proportion of largest function
250 m	0.78 (0.58; 1.05)	0.72 (1.15; 0.99)	1.12 (0.80; 1.57)	0.99 (0.69; 1.42)	1.17 (0.74; 1.86)	1.52 (0.93; 2.51)	1.06 (0.44; 2.53)	1.04 (0.42; 2.58)	1.24 (0.91; 1.69)	1.21 (0.87; 1.70)
500 m	0.97 (0.68; 1.39)	0.80 (0.55; 1.18)	1.41 (0.95; 2.10)	1.28 (0.83; 1.96)	1.04 (0.60; 1.78)	1.42 (0.79; 2.54)	1.62 (0.56; 4.71)	1.78 (0.59; 5.38)	1.30 (0.90; 1.88)	1.14 (0.76; 1.69)
Simpson’s diversity index S
250 m	1.32 (0.99; 1.75)	1.45 (1.06; 1.99)	0.88 (0.64; 1.22)	1.00 (0.71; 1.43)	0.86 (0.55; 1.35)	0.68 (0.42; 1.11)	0.90 (0.39; 2.12)	0.92 (0.38; 2.23)	0.81 (0.60; 1.09)	0.83 (0.60; 1.15)
500 m	1.03 (0.69; 1.52)	1.28 (0.84; 1.95)	0.73 (0.47; 1.12)	0.82 (0.51; 1.32)	0.86 (0.48; 1.55)	0.60 (0.31; 1.13)	0.56 (0.17; 1.84)	0.50 (0.14; 1.72)	0.78 (0.52; 1.17)	0.92 (0.59; 1.43)

^aAdjusted sex, maternal education level, household crowding, NDVI, and neighbourhood socioeconomic deprivation

The mediation analysis (online Supp. Table A.5) revealed that associations between LUM indexes and wheezing in the last 12 months do not seem to be mediated by physical activity and outdoor levels of NO₂. Significant direct and total effects were found for LUM using Shannon’s Evenness Index within 250 m and 500 m and for LUM using the number of different land use types within 250 m (online Supp. Table A.5).

Discussion

In this study, we assessed the association between different land use indexes and asthma and respiratory symptoms in children, using data from a birth cohort. Our results suggest that children with a greater exposure to LUM calculated based on Shannon’s Evenness Index and number of different land use types had a lower odds of wheezing in the last 12 months at 7 years of age. The results suggest that both Shannon’s Evenness Index and number of different land use types have a bigger influence on current wheezing symptoms at 7 years of age.

Although our findings did not show an association between LUM and asthma, the presence of respiratory symptoms, including wheezing, are hallmarks of asthma [55] also among children [56]. A previous study in the adult population reported an inverse association between a higher land use diversity based on Shannon index and the prevalence of respiratory diseases, including asthma and chronic obstructive pulmonary disease (COPD) [(OR (95% CI) = 0.77 (0.61–0.97)], in particular among occupationally active individuals [(OR (95% CI) = 0.72 (0.53–0.98)] [57]. Authors suggested that LUM may reflect the diversity of all types of land use, including both natural and build-up areas. In addition, most of the land use types were related to green spaces, and some of the classes included in built areas may also be perceived as green, such as grass and forest within built-up areas [57]. However, no previous studies have assessed the association between LUM and asthma and respiratory symptoms among children. Our results showed that the diversity of land use in the children’s residence was protective against wheezing in the last 12 months at 7 years of age. It is important to refer that our results are not consistent for all land use indexes, being only statistically significant for the Shannon’s Evenness Index and the number of different land use types. This variability in the results may reflect the differences in the approach to measuring LUM [11]. In particular, Shannon’s Index and the number of different land use types are measures of diversity, while the proportion of the largest function is simply a measure of homogeneity. Given these empirical results, it is advisable that future studies on the relationship between LUM and respiratory health use, at least, one of the first two measures—Shannon’s Index and/or the number of different land use types—of land use diversity.

Several mechanisms may be implicated in the association observed between LUM and respiratory symptoms. The mediation analysis revealed that associations between LUM indexes and wheezing in the last 12 months do not seem to be mediated by physical activity or outdoor levels of NO₂. However, the higher LUM may affect the number and dispersal of animals and plants through the landscape by increasing the number of land use types they come into contact [14], which has been suggested to influence the development of wheezing [58] through the exposure to more diverse macro- and microbiota that modulate the immune system [59, 60]. According to the biodiversity hypothesis, the contact with natural environmental features and biodiversity, including environmental microbial exposures, may influence the composition of human microbiota and, consequently, their ability to stimulate the immunoregulatory circuits and other antigen-recognizing receptors to prevent inappropriate inflammatory responses [60]. Moreover, a high LUM may be also characterized by a wide variety and higher number of green spaces, recreational and leisure facilities, and consequently a high amount of time spent outdoors [18], which may also be associated with a lower risk of respiratory symptoms. A previous study among adults reported a correlation between LUM and green space higher than 0.5, suggesting that land use variability is explained by variability in the amount of green spaces [57]. Recently, the results from the INMA birth cohort showed an inverse association between the higher proximity to green spaces and wheezing among children [61]. These results provide support for our findings, which highlight the effects of exposure to higher diversity of land use in each neighbourhood on respiratory symptoms in the last 12 months among children.

Our results suggested that the association between LUM and wheezing was not mediated by physical activity. However, our study considered only the practice of sports and active leisure activities performed on a repeated and regular basis, not considering children’s active travel or walkability index. Previous studies have found that LUM may play an important role on children’s active travel [7, 62], being positively associated with walking and cycling [62], and thus contributing to residents’ active lifestyles that, in turn, may reduce the prevalence of asthma symptoms, such as wheezing [63, 64], by improving lung function and reducing airway inflammation [24].

Our results may also contribute to the implementation of urban planning policies and practices to actively pursue land use organizing and urban design interventions, such as increasing land use diversity around residences, to improve children’s health, promote a healthy lifestyle, and create healthier urban environments. The planners should consider distributing the land use in a way that helps to create greater proximity to destinations and an environment that promotes and facilitates increased levels of physical activity and a lower emission of air pollutants. In addition, growing evidence suggests that land use is a key component in effective urban planning [13], which can have benefits beyond health for environmental sustainability, social interactions, and production of economic value [65].

The current study has several limitations that should be discussed. Integral measures used in our study only account for the diversity of land uses within each buffer size. As previously reported, LUM is not sensitive to the arrangement, shape, proximity, relative importance, or interaction between land uses [12, 66]. Integral measures are also sensitive to the size of the areas under analysis [11, 67], where larger areas may appear more mixed than smaller areas because they cover a broader scale [11]. Moreover, Shannon’s Evenness Index lacks a probabilistic basis, being difficult to assess differences in diversity values between areas [67]. However, LUM has been previously associated with an important path for promoting healthier lifestyles [18, 68] being a valuable tool for assessing the association between the diversity of built environment features and respiratory outcomes. Compared with those excluded from the analysis, the children included lived in less deprived neighbourhoods (13.4% among included participants vs. 11.4% among those excluded, p < 0.05) and in households with fewer occupants (household crowding < 1.5 was 84.2% among included participants vs. 76.7% among those excluded, p < 0.05). The included children also belonged to families with higher educational levels (maternal educational level > 9 years was 54.1% among included participants vs. 42.0% in those excluded, p < 0.05). Additionally, the percentage of georeferencing losses correspond to 0.3%, and despite being small, compared with those excluded from the analysis, the included children belonged to families with higher educational levels (maternal educational level > 9 years was 54.1% among included participants vs. 32.1% in those excluded, p = 0.022). So, although the differences between the included and excluded children were relatively small, our sample may not fully represent the socioeconomic conditions of the initial cohort participants. Finally, similar to other studies on neighbourhood health effects, our study may be affected by the uncertain geographic context problem (UGCoP). The UGCoP occurs whenever the method to delimitate a person’s neighbourhood (e.g. the use of a certain buffer size/shape) does not capture an individual’s true geographic context, since individuals are not only influenced by their local residential environment, but also by other neighbourhoods where their activities take place [69].

This study also presents important strengths. Our study included multiple measures of LUM at 7 years of age, allowing us to assess the relationship between different LUM indexes and respiratory symptoms among children. Comparing with other studies that included land uses positively or negatively related to specific outcomes, such as walkability or obesity, [7, 30] we have considered all land use types present within each residential neighbourhood. Additionally, in Shannon’s Evenness Index, the number of land use types in each buffer size is constant and equivalent to the total number of land use types, being a more accurate reflection of LUM and also allow comparisons of LUM across neighbourhoods within our study. A previous study demonstrated that using a variable rather than a constant definition of the total number of land use types systematically underestimated by almost 26% the association between LUM and daily step counts; and this underestimation may be greater for neighbourhoods with fewer land use types [66]. Finally, sensitivity analyses were conducted to assess the impact of the selected distance buffers (250 m and 500 m).

Conclusions

This is the first study suggesting that high land use diversity may have a protective effect on wheezing in the last 12 months among children. Our findings could have important public health and urban planning implications, suggesting that land use mix may be more conducive to active lifestyles. Moreover, the results may assist informed decisions about the planning and design of healthy neighbourhoods, which may be helpful in preventing allergic and respiratory diseases.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contribution

Inês Paciência: Methodology, formal analysis, investigation, writing-original draft, and writing-review and editing. André Moreira: Writing-review and editing and supervision. João Cavaleiro Rufo: Formal analysis and writing-review and editing. Ana Cristina Santos: Writing-review and editing and supervision. Henrique Barros: Supervision and writing-review and editing. Ana Isabel Ribeiro: Methodology, formal analysis, investigation, supervision, and writing-review and editing.