June 18 | 2019

Accelerometer and GPS Data to Analyze Built Environments and Physical Activity. 

Tamura K, Wilson JS, Goldfeld K, Puett RC, Klenosky DB, Harper WA, Troped PJ.

Res Q Exerc Sport. 2019 Jun 14:1-8. DOI: 10.1080/02701367.2019.1609649. [Epub ahead of print]


Purpose: Most built environment studies have quantified characteristics of the areas around participants’ homes. However, the environmental exposures for physical activity (PA) are spatially dynamic rather than static. Thus, merged accelerometer and global positioning system (GPS) data were utilized to estimate associations between the built environment and PA among adults. Methods: Participants (N = 142) were recruited on trails in Massachusetts and wore an accelerometer and GPS unit for 1-4 days. Two binary outcomes were created: moderate-to-vigorous PA (MVPA vs. light PA-to-sedentary); and light-to-vigorous PA (LVPA vs. sedentary). Five built environment variables were created within 50-meter buffers around GPS points: population density, street density, land use mix (LUM), greenness, and walkability index. Generalized linear mixed models were fit to examine associations between environmental variables and both outcomes, adjusting for demographic covariates. Results: Overall, in the fully adjusted models, greenness was positively associated with MVPA and LVPA (odds ratios [ORs] = 1.15, 95% confidence interval [CI] = 1.03, 1.30 and 1.25, 95% CI = 1.12, 1.41, respectively). In contrast, street density and LUM were negatively associated with MVPA (ORs = 0.69, 95% CI = 0.67, 0.71 and 0.87, 95% CI = 0.78, 0.97, respectively) and LVPA (ORs = 0.79, 95% CI = 0.77, 0.81 and 0.81, 95% CI = 0.74, 0.90, respectively). Negative associations of population density and walkability with both outcomes reached statistical significance, yet the effect sizes were small. Conclusions: Concurrent monitoring of activity with accelerometers and GPS units allowed us to investigate relationships between objectively measured built environment around GPS points and minute-by-minute PA. Negative relationships between street density and LUM and PA contrast evidence from most built environment studies in adults. However, direct comparisons should be made with caution since most previous studies have focused on spatially fixed buffers around home locations, rather than the precise locations where PA occurs.

June 10 | 2019

Associations of combined exposures to surrounding green, air pollution and traffic noise on mental health.

Klompmaker JO, Hoek G, Bloemsma LD, Wijga AH, van den Brink C, Brunekreef B, Lebret E, Gehring U, Janssen NAH.

Environ Int. 2019 May 31;129:525-537. DOI: 10.1016/j.envint.2019.05.040 [Epub ahead of print]



Evidence is emerging that poor mental health is associated with the environmental exposures of surrounding green, air pollution and traffic noise. Most studies have evaluated only associations of single exposures with poor mental health.


To evaluate associations of combined exposure to surrounding green, air pollution and traffic noise with poor mental health.


In this cross-sectional study, we linked data from a Dutch national health survey among 387,195 adults including questions about psychological distress, based on the Kessler 10 scale, to an external database on registered prescriptions of anxiolytics, hypnotics & sedatives and antidepressants. We added data on residential surrounding green in a 300 m and a 1000 m buffer based on the Normalized Difference Vegetation Index (NDVI) and a land-use database (TOP10NL), modeled annual average air pollutant concentrations (including particulate matter (PM10, PM2.5), and nitrogen dioxide (NO2)) and modeled road- and rail-traffic noise (Lden and Lnight) to the survey. We used logistic regression to analyze associations of surrounding green, air pollution and traffic noise exposure with poor mental health.


In single exposure models, surrounding green was inversely associated with poor mental health. Air pollution was positively associated with poor mental health. Road-traffic noise was only positively associated with prescription of anxiolytics, while rail-traffic noise was only positively associated with psychological distress. For prescription of anxiolytics, we found an odds ratio [OR] of 0.88 (95% CI: 0.85, 0.92) per interquartile range [IQR] increase in NDVI within 300 m, an OR of 1.14 (95% CI: 1.10, 1.19) per IQR increase in NO2 and an OR of 1.07 (95% CI: 1.03, 1.11) per IQR increase in road-traffic noise. In multi exposure analyses, associations with surrounding green and air pollution generally remained but attenuated. Joint odds ratios [JOR], based on the Cumulative Risk Index (CRI) method, of combined exposure to air pollution, traffic noise and decreased surrounding green were higher than the ORs of single exposure models. Associations of environmental exposures with poor mental health differed somewhat by age.


Studies including only one of these three correlated exposures may overestimate the influence of poor mental health attributed to the studied exposure, while underestimating the influence of combined environmental exposures.

May 27 | 2019

Association between exposure to the natural environment, rurality, and attention-deficit hyperactivity disorder in children in New Zealand: a linkage study.

Geoffrey H Donovan, PhD, Yvonne L Michael, ScD, Demetrios Gatziolis, PhD, Andrea ‘t Mannetje, PhD, Prof Jeroen Douwes, PhD

The Lancet Planetary Health Volume 3, Issue 5, May 2019, Pages e226-e234 DOI:https://doi.org/10.1016/S2542-5196(19)30070-1



Several small experimental studies and cross-sectional observational studies have shown that exposure to the natural environment might protect against attention-deficit hyperactivity disorder (ADHD) or moderate the symptoms of ADHD in children. We aimed to assess whether exposure to the natural environment protects against ADHD and whether this hypothesised protective effect varies across a child’s life course.


We did a longitudinal study with data collected from all children born in New Zealand in 1998, excluding those without an address history, those who were not singleton births, and those who died or emigrated before 18 years of age. We used Statistics New Zealand’s Integrated Data Infrastructure to identify children with ADHD and to define covariates. ADHD was defined according to hospital diagnosis or pharmacy records (two or more prescriptions for ADHD drugs). Exposure to green space for each year of a child’s life (from gestation to 18 years of age) was estimated at the meshblock level (the smallest geographical unit for which the New Zealand Census reports data) using normalised difference vegetation index (NDVI), and land-use data from Landcare Research New Zealand. We used logit models to assess the associations between ADHD prevalence and minimum, maximum, and mean lifetime NDVI, as well as rural living, controlling for sex, ethnicity, mother’s educational level, mother’s smoking status, mother’s age at parturition, birth order, antibiotic use, and low birthweight.


Of the 57 450 children born in New Zealand in 1998, 49 923 were eligible and had available data, and were included in the analysis. Children who had always lived in a rural area after 2 years of age were less likely to develop ADHD (odds ratio [OR] 0·670 [95% CI 0·461–0·974), as were those with increased minimum NDVI exposure after age 2 years (standardised OR for exposure vs first quartile: second quartile 0·841 [0·707–0·999]; third quartile 0·809 [0·680–0·963]; fourth quartile 0·664 [0·548–0·805]). In early life (prenatal to age 2 years), neither rural living nor NDVI were protective against ADHD. Neither mean nor maximum greenness was significantly protective against ADHD.


Rurality and increased minimum greenness were strongly and independently associated with a reduced risk of ADHD. Increasing a child’s minimum lifetime greenness exposure, as opposed to maximum or mean exposure, might provide the greatest increment of protection against the disorder.

May 21 | 2019

Complex relationships between greenness, air pollution, and mortality in a population-based Canadian cohort.

Crouse DL, Pinault L, Balram A, Brauer M, Burnett RT, Martin RV, van Donkelaar A, Villeneuve PJ, Weichenthal S.

Environ Int. 2019 Jul;128:292-300. Epub 2019 May 7. DOI:10.1016/j.envint.2019.04.047



Epidemiological studies have consistently demonstrated that exposure to fine particulate matter (PM2.5) is associated with increased risks of mortality. To a lesser extent, a series of studies suggest that living in greener areas is associated with reduced risks of mortality. Only a handful of studies have examined the interplay between PM2.5, greenness, and mortality.


We investigated the role of residential greenness in modifying associations between long-term exposures to PM2.5 and non-accidental and cardiovascular mortality in a national cohort of non-immigrant Canadian adults (i.e., the 2001 Canadian Census Health and Environment Cohort). Specifically, we examined associations between satellite-derived estimates of PM2.5 exposure and mortality across quintiles of greenness measured within 500 m of individual’s place of residence during 11 years of follow-up. We adjusted our survival models for many personal and contextual measures of socioeconomic position, and residential mobility data allowed us to characterize annual changes in exposures.


Our cohort included approximately 2.4 million individuals at baseline, 194,270 of whom died from non-accidental causes during follow-up. Adjustment for greenness attenuated the association between PM2.5 and mortality (e.g., hazard ratios (HRs) and 95% confidence intervals (CIs) per interquartile range increase in PM2.5 in models for non-accidental mortality decreased from 1.065 (95% CI: 1.056-1.075) to 1.041 (95% CI: 1.031-1.050)). The strength of observed associations between PM2.5 and mortality decreased as greenness increased. This pattern persisted in models restricted to urban residents, in models that considered the combined oxidant capacity of ozone and nitrogen dioxide, and within neighbourhoods characterised by high or low deprivation. We found no increased risk of mortality associated with PM2.5among those living in the greenest areas. For example, the HR for cardiovascular mortality among individuals in the least green areas was 1.17 (95% CI: 1.12-1.23) compared to 1.01 (95% CI: 0.97-1.06) among those in the greenest areas.


Studies that do not account for greenness may overstate the air pollution impacts on mortality. Residents in deprived neighbourhoods with high greenness benefitted by having more attenuated associations between PM2.5 and mortality than those living in deprived areas with less greenness. The findings from this study extend our understanding of how living in greener areas may lead to improved health outcomes.

May 13 | 2019

Liveable for whom? Prospects of urban liveability to address health inequities.

Badland H, Pearce J.

Soc Sci Med. 2019 May 2;232:94-105. DOI: 10.1016/j.socscimed.2019.05.001


The aspiration of liveable cities, underpinned by the New Urban Agenda, is gaining popularity as a mechanism to enhance population health and wellbeing. However, less attention has been given to understanding how urban liveability may provide an opportunity to redress health inequities. Using an environmental justice lens, this paper investigates whether urban liveability poses an opportunity or threat to reducing health inequities and outlines a future research agenda. Selected urban liveability attributes, being: education; employment; food, alcohol, and tobacco; green space; housing; transport; and walkability, were investigated to understand how they can serve to widen or narrow inequities. Some domains showed consistent evidence, others suggested context-specific associations that made it difficult to draw general conclusions, and some showed a reverse patterning with the social gradient, but with poorer outcomes. This suggests urban liveability attributes have equigenic potential, but operate within a complex system. We conclude more disadvantaged neighbourhoods and their residents likely have additional policy and design considerations for optimising outcomes, especially as changes to the contextual environment may impact neighbourhood composition through displacement and/or pulling up effects. Future research needs to continue to explore downstream associations using longitudinal and natural experiments, and also seek to gain a deeper understanding of the urban liveability system, including interactions, feedback loops, and non-linear and linear responses. There is a need to monitor neighbourhood population changes over time to understand how liveability impacts the most vulnerable. Other areas worthy of further investigation include applying a life course approach and understanding liveability within the context of other adversities and contextual settings.

May 6 | 2019

Early-life exposome and lung function in children in Europe: an analysis of data from the longitudinal, population-based HELIX cohort.

Agier L, Basagaña X, Maitre L, Granum B, Bird PK, Casas M, Oftedal B, Wright J, Andrusaityte S, de Castro M, Cequier E, Chatzi L, Donaire-Gonzalez D, Grazuleviciene R, Haug LS, Sakhi AK, Leventakou V, McEachan R, Nieuwenhuijsen M, Petraviciene I, Robinson O, Roumeliotaki T, Sunyer J, Tamayo-Uria I, Thomsen C, Urquiza J, Valentin A, Slama R, Vrijheid M, Siroux V.

Lancet Planet Health. 2019 Feb;3 (2):e81-e92. https://doi.org/10.1016/S2542-5196(19)30010-5



Several single-exposure studies have documented possible effects of environmental factors on lung function, but none has relied on an exposome approach. We aimed to evaluate the association between a broad range of prenatal and postnatal lifestyle and environmental exposures and lung function in children.


In this analysis, we used data from 1033 mother-child pairs from the European Human Early-Life Exposome (HELIX) cohort (consisting of six existing longitudinal birth cohorts in France, Greece, Lithuania, Norway, Spain, and the UK of children born between 2003 and 2009) for whom a valid spirometry test was recorded for the child. 85 prenatal and 125 postnatal exposures relating to outdoor, indoor, chemical, and lifestyle factors were assessed, and lung function was measured by spirometry in children at age 6-12 years. Two agnostic linear regression methods, a deletion-substitution-addition (DSA) algorithm considering all exposures simultaneously, and an exposome-wide association study (ExWAS) considering exposures independently, were applied to test the association with forced expiratory volume in 1 s percent predicted values (FEV1%). We tested for two-way interaction between exposures and corrected for confounding by co-exposures.


In the 1033 children (median age 8·1 years, IQR 6·5-9·0), mean FEV1% was 98·8% (SD 13·2). In the ExWAS, prenatal perfluorononanoate (p=0·034) and perfluorooctanoate (p=0·030) exposures were associated with lower FEV1%, and inverse distance to nearest road during pregnancy (p=0·030) was associated with higher FEV1%. Nine postnatal exposures were associated with lower FEV1%: copper (p=0·041), ethyl-paraben (p=0·029), five phthalate metabolites (mono-2-ethyl 5-carboxypentyl phthalate [p=0·016], mono-2-ethyl-5-hydroxyhexyl phthalate [p=0·023], mono-2-ethyl-5-oxohexyl phthalate [p=0·0085], mono-4-methyl-7-oxooctyl phthalate [p=0·040], and the sum of di-ethylhexyl phthalate metabolites [p=0·014]), house crowding (p=0·015), and facility density around schools (p=0·027). However, no exposure passed the significance threshold when corrected for multiple testing in ExWAS, and none was selected with the DSA algorithm, including when testing for exposure interactions.


Our systematic exposome approach identified several environmental exposures, mainly chemicals, that might be associated with lung function. Reducing exposure to these ubiquitous chemicals could help to prevent the development of chronic respiratory disease.


April 29 | 2019

Global, national, and urban burdens of paediatric asthma incidence attributable to ambient NO2pollution: estimates from global datasets.

Pattanun Achakulwisut PhD, Prof Michael Brauer ScD, Perry Hystad PhD, Susan C Anenberg PhD.

The Lancet Planetary Health Volume 3, Issue 4, April 2019, Pages e166-e178  https://doi.org/10.1016/S2542-5196(19)30046-4



Paediatric asthma incidence is associated with exposure to traffic-related air pollution (TRAP), but the TRAP-attributable burden remains poorly quantified. Nitrogen dioxide (NO2) is a major component and common proxy of TRAP. In this study, we estimated the annual global number of new paediatric asthma cases attributable to NO2 exposure at a resolution sufficient to resolve intra-urban exposure gradients.


We obtained 2015 country-specific and age-group-specific asthma incidence rates from the Institute for Health Metrics and Evaluation for 194 countries and 2015 population counts at a spatial resolution of 250 × 250 m from the Global Human Settlement population grid. We used 2010–12 annual average surface NO2concentrations derived from land-use regression at a resolution of 100 × 100 m, and we derived concentration-response functions from relative risk estimates reported in a multinational meta-analysis. We then estimated the NO2-attributable burden of asthma incidence in children aged 1–18 years in 194 countries and 125 major cities at a resolution of 250 × 250 m.


Globally, we estimated that 4·0 million (95% uncertainty interval [UI] 1·8–5·2) new paediatric asthma cases could be attributable to NO2 pollution annually; 64% of these occur in urban centres. This burden accounts for 13% (6–16) of global incidence. Regionally, the greatest burdens of new asthma cases associated with NO2exposure per 100 000 children were estimated for Andean Latin America (340 cases per year, 95% UI 150–440), high-income North America (310, 140–400), and high-income Asia Pacific (300, 140–370). Within cities, the greatest burdens of new asthma cases associated with NO2 exposure per 100 000 children were estimated for Lima, Peru (690 cases per year, 95% UI 330–870); Shanghai, China (650, 340–770); and Bogota, Colombia (580, 270–730). Among 125 major cities, the percentage of new asthma cases attributable to NO2 pollution ranged from 5·6% (95% UI 2·4–7·4) in Orlu, Nigeria, to 48% (25–57) in Shanghai, China. This contribution exceeded 20% of new asthma cases in 92 cities. We estimated that about 92% of paediatric asthma incidence attributable to NO2 exposure occurred in areas with annual average NO2 concentrations lower than the WHO guideline of 21 parts per billion.


Efforts to reduce NO2 exposure could help prevent a substantial portion of new paediatric asthma cases in both developed and developing countries, and especially in urban areas. Traffic emissions should be a target for exposure-mitigation strategies. The adequacy of the WHO guideline for ambient NO2 concentrations might need to be revisited.

April 22 | 2019

Estimated Long-term (1981-2016) Concentrations of Ambient Fine Particulate Matter across North America from Chemical Transport Modeling, Satellite Remote Sensing and Ground-based Measurements.

Meng J, Li C, Martin RV, van Donkelaar A, Hystad P, Brauer M.

Environ Sci Technol. 2019 Apr 17. doi: 10.1021/acs.est.8b06875 . [Epub ahead of print]



Accurate data concerning historical fine particulate matter (PM2.5) concentrations are needed to assess long-term changes in exposure and associated health risks. We estimated historical PM2.5 concentrations over North America from 1981-2016 for the first time by combining chemical transport modeling, satellite remote sensing and ground-based measurements. We constrained and evaluated our estimates with direct ground-based PM2.5 measurements when available and otherwise with historical estimates of PM2.5 from PM10 measurements or total suspended particles (TSP) measurements. The estimated PM2.5 concentrations were generally consistent with direct ground-based PM2.5 measurements over their duration from 1988 onward (R2 = 0.6-0.85) and to a lesser extent with PM2.5 inferred from PM10 measurements from 1985 to 1998 (R2 =0.5-0.6). The collocated comparison of the trends of population-weighted annual average PM2.5 from our estimates and ground-based measurements were highly consistent (RMSD = 0.66 μg m-3). The population-weighted annual average PM2.5 over North America decreased from 22 6.4 μg m-3 in 1981, to 12 3.2 μg m-3 in 1998, and to 7.9 2.1 μg m-3 in 2016, with an overall trend of -0.33 μg m-3 yr-1 (95% CI: -0.35 -0.30).

April 15 | 2019

Environmental Exposures and Depression: Biological Mechanisms and Epidemiological Evidence.

van den Bosch M, Meyer-Lindenberg A.

Annu Rev Public Health. 2019 Apr 1;40:239-259. DOI: 10.1146/annurev-publhealth-040218-044106


Mental health and well-being are consistently influenced-directly or indirectly-by multiple environmental exposures. In this review, we have attempted to address some of the most common exposures of the biophysical environment, with a goal of demonstrating how those factors interact with central structures and functions of the brain and thus influence the neurobiology of depression. We emphasize biochemical mechanisms, observational evidence, and areas for future research. Finally, we include aspects of contextual environments-city living, nature, natural disasters, and climate change-and call for improved integration of environmental issues in public health science, policies, and activities. This integration is necessary for reducing the global pandemic of depression.



New to data science or looking to pick up a few new skills? Don’t miss these free webinars, guided practical tutorials and online resources featuring CANUE data.

Developed in partnership with Population Data BC


Module 1: Introduction to Machine Learning

  • What is machine learning?
  • Supervised vs unsupervised learning
  • Model- and kernel-based methods
  • Measures of Accuracy (Test/train and cross-validation)
  • Causality and Accuracy
  • Unsupervised learning as feature reduction
Overview Webinar
Lab Session
Module 2: Regression and Regularization Algorithms

  • Regression with many correlated variables
  • Automatic variable selection, early approaches and problems
  • Gradient descent
  • Regularization  (L1 vs L2 vs ElasticNet)
Overview Webinar
Lab Session
Module 3: Advanced Supervised Learning 

  • Decision trees
  • Problems in overfit
  • Random Forest
  • Out-of-bag error vs cross-validation
Overview Webinar
Lab Session
Module 4: Advanced Unsupervised Learning 

  • Who uses unsupervised learning?
  • K-means
  • Expectation-maximization
  • Susceptibility to outliers
  • Dangers of labeling clusters
Overview Webinar
Lab Session

Dr. Aman Verma  is a Data Engineer with a PhD in Epidemiology from McGill University, and an undergraduate degree in Computer Science. He has experience in developing machine learning systems with large databases, particularly for scientific data in healthcare. While he’s comfortable learning any programming language, he’s recently become particularly interested in R. Aman is currently involved in a number of projects, including measuring how following opioid prescription guidelines can decrease the risk of opioid overdose, modelling trajectories of chronic obstructive pulmonary disease, and assessing how to best prioritize ambulance calls using secondary healthcare data.



This self paced free online course will provide you with an introduction to Data Management and Cleaning for Analysis using R Software. Each of the four modules includes a Power Point slide deck, CANUE training data, R code and associated exercises for practice.

To access this resource please create a Population Data BC account here: https://my.popdata.bc.ca/accounts/register/

Once your account has been approved you will be able to access the Education and Training site and self enroll in this and other free online courses.

Topics covered include:

  • Introduction and theory of data cleaning and management
  • Getting started with R software
  • Subsetting variables and data cleaning
  • Creating variables, subset observations and data cleaning
  • Merging, joining and reshaping data


Megan Striha currently works as a Data Analyst. She has a Masters of Public Health degree and three years of experience in health data analysis, including working with survey, administrative and census data.