Spatial and Spatio-temporal Epidemiology

Early learning and child care centres are critical settings to support childrens regular, repeated and quality time spent in outdoor play. Gibsons theory of affordances highlights the importance of the human-environment relationship, emphasizing how children use environmental information to inform their behaviour. This study aims to understand the association between childrens outdoor play behaviour and common environmental play features in early learning and child care outdoor play spaces, through the behaviour patterns of children. Childrens play behaviour was collected via observational behaviour mapping at eight early learning and child care centres in the Greater Vancouver region between September 2021 and November 2022, as part of the PROmoting Early Childhood Outside study. A multivariate logistic regression model examined the association between outdoor play behaviour and environmental play features, via odds ratio and 95% confidence intervals. The results indicate environmental play features, including gardening areas, playhouses, climbing structures and tricycle paths supported increased opportunities for childrens outdoor play. Gardening areas, playhouses, sandboxes, outdoor stages and fixed water features provided opportunities for exploratory play, while climbing structures and trike paths provided opportunities for physical play. Opportunities for diverse forms of play were less realized in dedicated open play areas, with the availability of loose parts and moveable equipment primarily influencing these spaces. The results of this study have important implications for future early learning and child care outdoor space design. Further research should consider childrens dynamic movement and transition between outdoor affordances, and the influence of loose parts on the use of environmental play features.

While numerous studies over the last decade have highlighted the important influence of environmental factors on mental health, globally applicable data on physical surroundings are still limited. Access to such data and the possibility to link them to epidemiological studies is critical to unlocking the relationship of environment, brain and behaviour and promoting positive future mental health outcomes. The Adolescent Brain Cognitive Development (ABCD) Study is the largest ongoing longitudinal and observational study exploring brain development and child health among children from 21 sites across the United States. Here we describe the linking of the ABCD study data with satellite-based "Urban-Satellite" (UrbanSat) variables consisting of 11 satellite-data derived environmental indicators associated with each subjects residential address at their baseline visit, including land cover and land use, nighttime lights, and population characteristics. We present these UrbanSat variables and provide a review of the current literature that links environmental indicators with mental health, as well as key aspects that must be considered when using satellite data for mental health research. We also highlight and discuss significant links of the satellite data variables to the default mode network clustering coefficient and cognition. This comprehensive dataset provides the foundation for large-scale environmental epidemiology research.

Infant mortality (IM), or death prior to the first birthday, is a key public health metric that increases with neighborhood structural inequities. However, neighborhood exposures shift as communities undergo gentrification, a pattern of neighborhood change defined by increasing affluence (in wealth, education, and housing costs). Gentrification has inconsistent associations with infant health outcomes like IM, which may be due to differing relationships between its composite measures and such outcomes. We designed a retrospective cohort analysis of all births and deaths from 2010-2019 across 4 metropolitan areas in Michigan to determine how gentrification and its neighborhood-change components are associated with risk of IM, using multilevel multivariable logistic regression models. Among 672,432 infants, 0.52% died before 1 year. IM was not associated with gentrification. Census tracts with greater increases in income and education had lower rates of IM, but tracts with greater increases in rent costs had higher rates of IM. In unadjusted models, odds of IM were 40% and 15% lower for infants living in tracts in the top quartile increase in household income and college completion, respectively, compared to infants from tracts with the least amount of change. Odds of IM were also increased 29% in infants from tracts with the most increases in rent, though these differences were attenuated when adjusting for individual social factors. Indicators of increasing community affluence have opposing relationships with IM. Policies and interventions that address rising housing costs may reduce IM.

ObjectiveThe aim of this study was to describe differences and trends in ATV-related hospitalizations for urban and rural-dwelling youth in Canada. MethodsWe conducted a cross-sectional study using administrative hospital abstract data all patients admitted for an ATV-related injury to hospitals in 9 provinces in Canada between 2002 and 2019. The primary exposure was rural residence, defined by postal code. Rural-urban comparisons were stratified by age group: children (<16 years), adolescents (16-20 years) and adults (>21 years). The primary outcome was the incidence of any hospitalization, secondary outcomes were head injury, fractures, crush injury and spinal cord injury.. ResultsAmong 34,390 patients with complete data, 17% were children younger than 16 yrs and 14% were adolescents 16-20 yrs; 78% of children and 85% of adolescents were male, and 47% lived rurally. The incident rate ratio (IRR) for being hospitalized for an ATV-related injury was 5-fold higher for rural children (5.59; 95% CI: 5.30-5.88) and adolescents (5.16; 95% CI: 4.88, 5.47) compared to urban children and adolescents, respectively. The 5-fold higher IRR was also evident for ATV-related fractures among rural children and adolescents. Adolescents had a particularly higher risk for ATV-related crush injuries (IRR: 10.43; 95% CI: 5.74-18.96) and spinal cord injuries (IRR: 5.21; 95% CI: 3.33-8.15) while children were at higher risk of ATV-related head injuries (IRR: 6.55; 95% CI: 5.76-7.46) compared to urban dwelling youth. ConclusionsIn Canada, rural children and adolescents were at a very elevated risk of ATV injuries compared to those living in urban centres.

Since the onset of the COVID-19 pandemic, researchers mainly examined how socio-economic, demographic, and environmental factors are related to disparities in SARS-CoV-2 infection rates. However, we dont know the extent to which racial disparities in environmental exposure are related to racial disparities in SARS-CoV-2 infection rates. To address this critical issue, we gathered black vs. white infection records from 1416 counties in the contiguous United States. For these counties, we used 30m-spatial resolution land cover data and racial mappings to quantify the racial disparity between black and white peoples two types of environmental exposure, including exposures to various types of landscape settings and urban development intensities. We found that racial disparities in SARS-CoV-2 infection rates and racial disparities in exposure to various types of landscapes and urban development intensities were significant and showed similar patterns. Specifically, less racial disparity in exposure to forests outside park, pasture/hay, and urban areas with low and medium development intensities were significantly associated with lower racial disparities in SARS-CoV-2 infection rates. Distance was also critical. The positive association between racial disparities in environmental exposures and racial disparity in SARS-CoV-2 infection rates was strongest within a comfortable walking distance (approximately 400m). HighlightsO_LIRacial dot map and landcover map were used for population-weighted analysis. C_LIO_LIRacial disparity in environmental exposures and SARS-CoV-2 infection were linked. C_LIO_LIForests outside park are the most beneficial landscape settings. C_LIO_LIUrban areas with low development intensity are the most beneficial urban areas. C_LIO_LILandscape and urban exposures within the 400m buffer distances are most beneficial. C_LI

BackgroundGreen and blue spaces can promote good physical and mental health and prevent the development of long-term conditions. Evidence suggests that not all green spaces affect health equally, and that certain types and properties of green spaces are stronger predictors of health than others. However, research into the causal mechanisms is limited in large cohorts due to lack of objective and comparable data on green space type, accessibility, and usage. MethodsWe used data from Urban Atlas to compute measures of urban park accessibility, street trees availability, and total green and blue space availability for 300,000 UK Biobank participants. Exposure metrics were computed using circular buffers with radii of 100 m to 3000 m. Pearson correlation coefficients and other descriptive statistical parameters were used to test agreement between variables and explore the utility of indictors in capturing different types of green spaces. ResultsStrong positive correlations were observed between variables of the same indicator with different buffer sizes. The presence of park and proportion of street tree canopy variables were negatively correlated with amount of total green space variables. This signifies distinct differences in type of green spaces captured by these variables. ConclusionsOverall, five distinct indicators of park accessibility, street trees availability, and total green and blue space availability have been integrated into a large sample of the UK Biobank. Our method is replicable to settings across Europe and facilitates evidence-based research on the roles of different green and blue spaces in health promotion and ill-health prevention. Key MessagesO_LIDifferent types of green spaces and their position in the neighbourhood can promote and protect health by mitigating pollution and increasing physical activity and socialisation. C_LIO_LIWe present the methods of constructing and linking data on urban green spaces, street trees and natural vegetation into a large health cohort, the UK Biobank. C_LIO_LIThe ability to distinguish between types of green spaces and their intended use can help inform public health interventions, influence urban policy, and aid urban planning in building sustainable and healthy cities. C_LIO_LIOur methods are transferable and will allow others to explore the links between environment and health in UK Biobank and other health cohorts. C_LI

IntroductionHigh-density congregate housing, including cruise ships, hotels, residence halls and correctional facilities are epidemiologically important, and key aspects of pathogen transmission have been elucidated in these environments. A range of methods have been developed to detect unusual clusters of infections in these settings; however use of explicitly 3-D (x,y,z) spatial data has received little attention. In this study, we use data collected during the COVID-19 pandemic to assess the fine-scale spatial epidemiology and the clustering of confirmed cases to better understand impacts of spatial resolution and aggregation on spatio-temporal cluster detection. MethodsData for this analysis combined the results from mandatory weekly viral testing during the 2020-2021 academic year with high-resolution spatial data from university students residing in high-rise residence halls at the University of Massachusetts, Amherst campus. These data were analyzed for statistically-significant clustering of SARS-CoV-2 cases in three-dimensional space as well as time, within and between the high-density buildings on campus. Two sets of analyses were conducted. The first used a Space-Time Permutation Model, which scans for areas with a greater than expected number of cases (SaTScan). To assess the impact of data aggregation, analysis was done at several levels of spatial resolution. Additionally, we performed sensitivity analyses using a purely temporal surveillance algorithm, CDCs Early Aberration Reporting System-EARS. Results and conclusionsAnalysis with SaTScan at the room- and floor-level identified multiple statistically significant clusters within one residence hall. Analyses with these same cases aggregated at the floor-level were found to be as sensitive, but far less computationally intensive, than room-level analysis. Analysis at both of these spatial scales was more sensitive than analysis aggregated at the street address-level. Two events exceeding alert thresholds were detected in the purely temporal analysis; one of which was also detected in spatio-temporal analyses. These results expand our understanding of spatio-temporal scan metrics to include 3-D analysis, and optimizing choice of spatial scales. These results have broad applicability in epidemiology in assessing the ability of spatio-temporal methods for public health surveillance, with potential expansion to ecological studies incorporating vertical movement.

Urban heat islands and climate change create increasingly hot environments that pose a threat to the health of the public in urban areas throughout the planet. In Maricopa County, Arizona, --- the hottest metropolitan area in the United States---we have previously shown that the effects of heat on mortality are greater in the social and built environments of low-income and communities of color (predominantly Hispanic/Latinx and Black neighborhoods). In this analysis of morbidity data from Maricopa County, we examined the relationship between heat-related hospitalization and summertime daily maximum air temperatures in groups defined at the census block group level as being at high, medium, or low vulnerability based on a Heat Vulnerability Index that was derived from socio-economic and built-environment data. For all three categories of census block group heat vulnerability, we identified 26{degrees}C as the daily maximum air temperature threshold beyond which heat-related hospitalization risk increased rapidly with each 1 {degrees}C increase in temperature. Compared to this baseline temperature, the relative risk of hospitalization was greatest in the high vulnerability census block groups and least in the low vulnerability census block groups with intermediate increases in the medium vulnerability census block groups. Specifically, with 26{degrees}C as the referent, the relative risks of heat-related hospitalization increased from 0.97 at 27{degrees}C to 15.71 at 46{degrees}C in the low vulnerability group, from 1.03 at 27{degrees}C to 53.97 at 46{degrees}C in the medium vulnerability group, and from 1.09 at 27{degrees}C to 162.46 at 46{degrees}C in the high vulnerability group. Our research helps identify areas with high heat population sensitivity and exposure that can be targeted for adaptation with policies and investments, which include, for example, improving public health safety nets and outcomes, access to affordable energy-efficient housing and health care, energy justice, and modifications to cool the urban built environment. Our hospitalization risk estimates can be incorporated into quantitative risk assessments of heat-related morbidity in Maricopa County.

ObjectiveAcetylcholinesterase (AChE) is an enzyme that metabolizes acetylcholine, an essential neurotransmitter, and is frequently used to monitor adult agricultural workers for exposure to cholinesterase inhibitor pesticides. Yet, there are no clear standards for AChE activity in children and adolescents, which prohibits evaluations of dangerous pesticide exposures in younger populations. MethodsWe measured AChE activity from a single finger stick blood sample data from 746 participants ages 4 to 26 years across 3,100 observations who resided in an agricultural county in Ecuador. We used generalized estimating equations to predict AChE activity levels in one year age increments from 5 to 25 years, accounting for nonlinear aging patterns and survey wave specific effects. We also decomposed variation in observed AChE activity levels into aging effects, differences in our recruited participants, and participant-specific aging patterns. ResultsAverage AChE activity levels across all observations were 3.88 U/mL (standard deviation [sd] = 0.67). AChE activity levels increased nonlinearly as participants aged. We found that males had higher AChE activity levels than females and that those levels established themselves later in age than females. AChE activity levels increased essentially linearly from ages 5 to 12 or 18 years, depending on sex, at which point levels did not meaningfully change with age. Most variation observed in AChE activity levels were due to aging effects. ConclusionOur findings provide reference levels for AChE activity across childhood, adolescence, and into early adulthood that can be used by clinicians and researchers in the context of assessing neurodevelopment and potential exposure to neurotoxicants.

ObjectiveIdentifying geographic-level prevalence of occupations associated with mobility during local stay-at-home pandemic mandate. MethodsA spatio-temporal ecological framework was applied to determine census-tracts that had significantly higher rates of occupations likely to be deemed essential: food-service, business and finance, healthcare support, and maintenance. Real-time mobility data was used to determine the average daily percent of residents not leaving their place of residence. Spatial regression models were constructed for each occupation proportion among census-tracts within a large urban area. ResultsAfter adjusting for demographics, results indicate census-tracts with higher proportion of food-service workers, healthcare support employees, and office administration staff are likely to have increased mobility. ConclusionsIncreased mobility among communities is likely to exacerbate COVID-19 mitigation efforts. This increase in mobility was also found associated with specific demographics suggesting it may be occurring among underserved and vulnerable populations. We find that prevalence of essential employment presents itself as a candidate for driving inequity in morbidity and mortality of COVID-19. Three-question SummaryO_LIEmployees and workers deemed essential during the COVID-19 pandemic are likely to endure additional risk of infection due to community exposure. While preliminary reports are still quantifying this risk, we set out to examine if prevalence of specific occupations could be used to evaluate overall community-level risk based on stay-at-home mandate adherence. C_LIO_LIStudy results suggest that that not only are certain occupation geo-spatially associated with movement outside the home but are also associated with demographic characteristics likely to contribute to inequity of COVID-19 morbidity. C_LIO_LIOften, nuanced inequities are lost in the larger data samples, being able to identify possible inequities from other sources such as prevalence of occupation among communities, remains an important and applicable alternative. C_LI

The extent to which urban vegetation improves environmental quality and affects the health of nearby residents is dependent on typological attributes of "greenness", such as canopy area to alleviate urban heat, grass to facilitate exercise and social interaction, leaf area to disperse and capture air pollution, and biomass to absorb noise pollution. The spatial proximity of these typologies to individuals further modifies the extent to which they impart benefits and influence health. However, most evaluations of associations between greenness and health utilize a single metric of greenness and few measures of proximity, which may disproportionately represent the effect of a subset of mediators on health outcomes. To develop an approach to address this potentially substantial limitation of future studies evaluating associations between greenness and health, we measured and evaluated distinct attributes, correlations, and spatial dependency of 13 different metrics of greenness in a residential study area of Louisville, Kentucky, representative of many urban residential areas across the Eastern United States. We calculated NDVI, other satellite spectral indices, LIDAR derived leaf area index and canopy volume, streetview imagery derived semantic view indices, distance to parks, and graph-theory based ecosystem connectivity metrics. We utilized correlation analysis and principal component analysis across spatial scales to identify distinct groupings and typologies of greenness metrics. Our analysis of correlation matrices and principal component analysis identified distinct groupings of metrics representing both physical correlates of greenness (trees, grass, their combinations and derivatives) and also perspectives on those features (streetview, aerial, and connectivity / distance). Our assessment of typological greenness categories contributes perspective important to understanding strengths and limitations of metrics evaluated by past work correlating greenness to health. Given our finding of inconsistent correlations between many metrics and scales, it is likely that many past investigations are missing important context and may underrepresent the extent to which greenness may influence health. Future epidemiological investigations may benefit from these findings to inform selection of appropriate greenness metrics and spatial scales that best represent the cumulative influence of the hypothesized effects of mediators and moderators. However, future work is needed to evaluate the effect of each of these metrics on health outcomes and mediators therein to better inform the understanding of metrics and differential influences on environments and health.

1Non-communicable diseases (NCDs) account for [~]71% of all deaths globally, including 15 million premature deaths each year (deaths between 30-69 years of age). Instead of waiting until the disease manifestation, focusing on the origins of NCDs during childhood offers a critical window of disease prevention and control for effective interventions. The CHILDREN_FIRST study aims to investigate how the spatio-temporal evolution of the childrens exposome profiles in the Mediterranean region influences the early-life programming of chronic disease risk during the unique critical window of susceptibility in the primary school years (6-11 years of age). The study protocol adopts the human exposome framework integrated with a personalized prevention approach using multi-omics platforms and advanced machine learning algorithms implemented across five Mediterranean countries, namely Cyprus, Greece, Spain, Israel, and Albania. The cohort will consist of children enrolled in the first grade of primary school, who will undergo annual follow-up assessments until completion of primary education. During the annual assessments, childrens exposome parameters from the three main exposome domains will be evaluated using different assessment types i.e., biospecimen, sensors, questionnaires. Standardized human sample and data collection methods will be employed following harmonized standardized operating procedures. The reference model of Observational Medical Outcomes Partnership - Common Data Model part of the Observational Health Data Sciences and Informatics will be used to conduct federated data analysis. This CHILDREN_FIRST study protocol is a human exposome-based initiative to establish a long-term prospective cohort infrastructure for biomedical research on childrens health within the Mediterranean region. The cohorts exposome-based findings will systematically feed into the evaluation and design of chronic disease prevention programs. Expected results would inform evidence-based policy making and the development of health interventions for reducing the risk of NCDs.

Leisure-time physical activity is beneficial for health and is associated with various urban characteristics. Using the exposome framework, the totality of the environment, this study investigated how urban physical environments were associated with leisure-time physical activity during early midlife. A total of 394 participants (mean age: 37, range 34-40) were included from the FinnTwin12 cohort residing in five major Finnish cities in 2020. We comprehensively curated 145 urban physical exposures at residential addresses of participants and measured three leisure-time physical activity measures: (1) total leisure-time physical activity (total LTPA) and its sub-domains (2) leisure-time physical activity without commuting activity (LTPA) and (3) commuting activity. Using K-prototypes cluster analysis, we identified three urban clusters: "original city center," "new city center," and "suburban". Results from adjusted linear regression models showed that participants in the "suburban" cluster had lower levels of total LTPA (beta: -0.13, 95% CI: -0.23, -0.03) and LTPA (beta: -0.17, 95% CI: -0.28, -0.05), compared to those in the "original city center" cluster. The eXtreme Gradient Boosting models ranked exposures related to greenspaces, pocket parks, and road junctions as the top important factors influencing outcomes, and their relationships with outcomes were largely non-linear. More road junctions and more pocket parks correlated with higher total LTPA and LTPA. When the all-year normalized difference vegetation index within a 500 m buffer fell below 0.4, it correlated with higher levels of total LTPA, whereas above 0.4, it correlated with lower levels. To conclude, our findings revealed a positive correlation between urbanicity and physical activity in Finnish cities and decomposed this complexity into crucial determinants. Importance rankings and nonlinear patterns offer valuable insights for future policies and projects targeting physical inactivity.

ObjectiveTo examine data on COVID-19 disease associated with a 10 percent increase in U.S. road deaths from 2020 to 2021 that raises the question of the potential effect of pandemic stress and neurological damage from COVID-19 disease. MethodsPoisson regression was used to estimate the association of recent COVID-19 cases, accumulated cases, maximum temperatures, truck registrations, and gasoline prices with road deaths monthly among U.S. states in 2021. Using the regression coefficients, changes in each risk factor from 2020 to 2021 were used to calculate expected deaths in 2021 if each factor had remained the same as in 2020. ResultsCorrected for the other risk factors, road deaths were associated with accumulated COVID-19 cases but not cases in the previous month. More than 20,700 road deaths were associated with the changes in accumulated COVID-19 cases but were substantially offset by about 19,100 less-than-expected deaths associated with increased gasoline prices. ConclusionsWhile more research is needed, the data are sufficient to warn people with "long COVID" to minimize road use. What is already known about this topicPrevious short-term fluctuations in road deaths are related to changes in temperature, fuel prices, and truck registrations. What this study addsCorrected for other risk factors, the monthly changes in road deaths from 2020 to 2021 in U.S. states were associated with cumulative COVID-19 cases. How this study might affect research, practice, or policyStudies are needed to distinguish the potential relative effects of neurological damage as well as the stress of coping with the pandemic on driving, walking, and bicyclist behavior. Warning people with "long covid" about road risk is warranted.

Housing infrastructure and quality is a major determinant of infectious disease risk and other health outcomes in regions of the world where vector borne, waterborne and neglected tropical diseases are endemic. It is important to quantify the geographical distribution of improvements to the major dwelling components to identify and target resources towards populations at risk. The aim of this study was to model the sub-national spatial variation in housing materials using covariates with quasi-global coverage and use the resulting estimates to map the predicted coverage across the worlds low- and middle-income countries (LMICs). Data relating to the materials used in dwelling construction were sourced from nationally representative household surveys conducted since 2005. Materials used for construction of flooring, walls, and roof were reclassified as improved or unimproved. Households lacking location information were georeferenced using a novel methodology, and a suite of environmental and demographic spatial covariates were extracted at those locations for use as model predictors. Integrated nested Laplace approximation (INLA) models were fitted to obtain and map predicted probabilities for each dwelling component. The dataset compiled included information from households in 283,000 clusters from 350 surveys. Low coverage of improved housing was predicted across the Sahel and southern Sahara regions of Africa, much of inland Amazonia, and areas of the Tibetan plateau. Coverage of improved roofs and walls was high in the Central Asia, East Asia and Pacific and Latin America and the Caribbean regions, while improvements in all three components, but most notably floors, was low in Sub-Saharan Africa. Human development was by far the strongest determinant of dwelling component quality, though vegetation greenness and land use were also relevant markers These findings are made available to the reader as files that can be imported into a GIS for integration into relevant analysis to derive improved estimates of preventable health burdens attributed to housing.

IntroductionFollowing its emergence in January 2020, SARS-CoV-2 diffusion occurred for a year with only non-pharmaceutical interventions (NPIs) available as mitigation tools. We aimed to assess the predictive capability of census-based indicators on the infection risk by SARS-CoV-2 in the French Auvergne-Rhone-Alpes region to assist NPIs allocation at the neighbourhood level. MethodsWe aggregated all counts of biologically confirmed cases of SARS-CoV-2 infection at the neighbourhood level between May 2020 and February 2021. 10 census-based ecological covariates were evaluated as predictors of case incidence using a Poisson regression with conditional autoregressive (CAR) spatial effects. Benefits of CAR effects and covariates on model fit were evaluated using pseudo-R{superscript 2} and Morans I statistics. Results438,992 infection cases over 5,410 neighbourhoods among 7,917,997 inhabitants were analysed. The association between covariates and case incidence was inconstant across time and space. Spatial correlation was estimated at high levels. Spatial CAR effects were necessary to improve on the pseudo-R2 and the Morans I statistics compared to the null model (intercept only). ConclusionThe ecological covariates assessed were insufficient to adequately model the distribution of cases at the neighbourhood level. Excess incidence was found mainly in metropolitan areas before the epidemic wave peak.

BackgroundDepressive symptoms lead to a serious public health burden and are considerably affected by the environment. Land use, describing the urban living environment, has an impact on mental health, but complex relationship assessment is rare. ObjectivesWe aimed to examine the complicated association between urban land use and depressive symptoms among young adults with differential land use environments, by applying multiple models, as an exposome study. MethodsWe included 1804 individual twins from the FinnTwin12 cohort, living in urban areas in 2012. There were 8 types of land use exposures in 3 buffer radii. The depressive symptoms were assessed through General Behavior Inventory (GBI) in young adulthood (mean age: 24.1). First, K- means clustering was performed to distinguish participants with differential land use environments. Then, linear elastic net penalized regression and eXtreme Gradient Boosting (XGBoost) were used to reduce dimensions or prioritize for importance and examine the linear and nonlinear relationships. ResultsTwo clusters were identified with notable differences in the percentage of high-density residential, low-density residential, and natural land use. One is more typical of city centers, and another of suburban areas. A heterogeneous pattern in results was detected from the linear elastic net penalized regression model among the overall sample and the two separated clusters. Agricultural residential land use in a 100 m buffer contributed to GBI most (coefficient: 0.097) in the "suburban" cluster among 11 selected exposures. In the "city center" cluster, none of the land use exposures was associated with GBI. From the XGBoost models, we observed that ranks of the importance of land use exposures on GBI and their nonlinear relationships are also heterogeneous in the two clusters. DiscussionAs a hypothesis-generating study, we found heterogeneous linear and nonlinear relationships between urban land use environment and depressive symptoms under different contexts in pluralistic exposome analyses.

Birth defects are a leading cause of infant mortality in the United States, but little is known about causes of many types of birth defects. Spatiotemporal disease mapping to identify high-prevalence areas, is a potential strategy to narrow the search for potential environmental and other causes that aggregate over space and time. We described the spatial and temporal trends of the prevalence of birth defects in North Carolina during 2003-2015, using data on live births obtained from the North Carolina Birth Defects Monitoring Program. By employing a Bayesian space-time Poisson model, we estimated spatial and temporal trends of non-chromosomal and chromosomal birth defects. During 2003-2015, 52,524 (3.3%) of 1,598,807 live births had at least one recorded birth defect. The prevalence of non-chromosomal birth defects decreased from 3.8% in 2003 to 2.9% in 2015. Spatial modeling suggested a large geographic variation in non-chromosomal birth defects at census-tract level, with the highest prevalence in south-eastern North Carolina. The strong spatial heterogeneity revealed in this work allowed to identify geographic areas with higher prevalence of non-chromosomal birth defects in North Carolina. This variation will help inform future research focused on epidemiologic studies of birth defects to identify etiologic factors.

This study analyzes the link between temperatures and COVID-19 contagions in a sample of Italian regions during the period ranging from February 24 to April 15. To that end, Bayesian Model Averaging techniques are used to analyze the relevance of the temperatures together with a set of additional climate, environmental, demographic, social and policy factors. The robustness of individual covariates is measured through posterior inclusion probabilities. The empirical analysis provides conclusive evidence on the role played by the temperatures given that it appears as the most relevant determinant of contagions. This finding is robust to (i) the prior distribution elicitation, (ii) the procedure to assign weights to the regressors, (iii) the presence of measurement errors in official data due to under-reporting, (iv) the employment of different metrics of temperatures or (v) the inclusion of additional correlates. In a second step, relative importance metrics that perform an accurate partitioning of the R2 of the model are calculated. The results of this approach support the evidence of the model averaging analysis, given that temperature is the top driver explaining 45% of regional contagion disparities. The set of policy-related factors appear in a second level of importance, whereas factors related to the degree of social connectedness or the demographic characteristics are less relevant.

Distributed lag non-linear models (DLNMs) offer a flexible approach towards modelling time-delayed exposures. They are popular to study the effect of environmental exposure on health outcomes, such as the effect of temperature on mortality. Conventional distributed lag non-linear models typically focus on a single exposure variable, potentially overlooking complex interactions between multiple predictors. In this paper, we propose a distributed lag non-linear model that captures the joint delayed impact of two exposure variables by incorporating their interaction through a tensor basis constructed from univariate P-splines. This model is compared to a model assuming an additive effect of delayed exposures. Our model is used to examine the joint impacts of maximum and minimum temperatures on all-cause mortality in Flanders during summer. The results show that our model provides a flexible strategy towards the analysis of two predictors with interacting time-delayed effects on an outcome of interest. The importance of both maximum and minimum temperatures in explaining variability in mortality is illustrated, and we show that the interaction effect varies across age and gender groups. A spatial risk analysis at the municipality level reveals that mortality is attributed differently to temperature exposure across different areas, due to temperature variations as well as spatial trends in age and gender.