December 17 | 2018

The influence of social networks and the built environment on physical inactivity: A longitudinal study of urban-dwelling adults.

Josey MJ, Moore S.

Health Place. 2018 Nov;54:62-68. Epub 2018 Sep 21. DOI: 10.1016/j.healthplace.2018.08.016



Policies targeting the built environment to increase physical activity may be ineffective without considering personal social networks. Physical activity and social network data came from the Montreal Neighborhood Networks and Healthy Aging Panel; built environment measures were from geolocation data on Montreal parks and businesses. Using multilevel logistic regression with repeated physical inactivity measures, we showed that adults with more favorable social network characteristics had lower odds of physical inactivity. Having more physical activity facilities nearby also lowered physical inactivity, but not in socially-isolated adults. Community programs that address social isolation may also benefit efforts to increase physical activity.


December 10 | 2018

Residential green space and pathways to term birth weight in the Canadian Healthy Infant Longitudinal Development (CHILD) Study.

Cusack L, Sbihi H, Larkin A, Chow A, Brook JR, Moraes T, Mandhane PJ, Becker AB, Azad MB, Subbarao P, Kozyrskyj A, Takaro TK, Sears MR, Turvey SE, Hystad P; CHILD Study Investigators.

Int J Health Geogr. 2018 Dec 4;17(1):43. doi: 10.1186/s12942-018-0160-x



A growing number of studies observe associations between the amount of green space around a mother’s home and positive birth outcomes; however, the robustness of this association and potential pathways of action remain unclear.


To examine associations between mother’s residential green space and term birth weight within the Canadian Healthy Infant Longitudinal Development (CHILD) study and examine specific hypothesized pathways.


We examined 2510 births located in Vancouver, Edmonton, Winnipeg, and Toronto Canada. Green space was estimated around mother’s residences during pregnancy using Landsat 30 m normalized difference vegetation index (NDVI). We examined hypothesized pathways of: (1) reduction of environmental exposure; (2) built environment features promoting physical activity; (3) psychosocial conditions; and (4) psychological influences. Linear regression was used to assess associations between green space and term birth weight adjusting first for a comprehensive set of confounding factors and then incrementally for pathway variables.


Fully adjusted models showed non-statistically significant increases in term birth weight with increasing green space. For example, a 0.1 increase in NDVI within 500 m was associated with a 21.5 g (95% CI - 4.6, 47.7) increase in term birth weight. Associations varied by city and were most robust for high-density locations. For the two largest cities (Vancouver and Toronto), we observed an increase in birth weight of 41.2 g (95% CI 7.8, 74.6) for a 0.1 increase in NDVI within 500 m. We did not observe substantial reductions in the green space effect on birth weight when adjusting for pathway variables.


Our results highlight the need to further characterize the interactions between green space, urban density and climate related factors as well as the pathways linking residential green space to birth outcomes.

December 3 | 2018

A picture tells a thousand…exposures: Opportunities and challenges of deep learning image analyses in exposure science and environmental epidemiology.

Weichenthal S, Hatzopoulou M, Brauer M. 

Environ Int. 2018 Nov 22. pii: S0160-4120(18)32200-1. [Epub ahead of print] DOI:10.1016/j.envint.2018.11.042




Artificial intelligence (AI) is revolutionizing our world, with applications ranging from medicine to engineering.


Here we discuss the promise, challenges, and probable data sources needed to apply AI in the fields of exposure science and environmental health. In particular, we focus on the use of deep convolutional neural networks to estimate environmental exposures using images and other complementary data sources such as cell phone mobility and social media information.


Characterizing the health impacts of multiple spatially-correlated exposures remains a challenge in environmental epidemiology. A shift toward integrated measures that simultaneously capture multiple aspects of the urban built environment could improve efficiency and provide important insights into how our collective environments influence population health. The widespread adoption of AI in exposure science is on the frontier. This will likely result in new ways of understanding environmental impacts on health and may allow for analyses to be efficiently scaled for broad coverage. Image-based convolutional neural networks may also offer a cost-effective means of estimating local environmental exposures in low and middle-income countries where monitoring and surveillance infrastructure is limited. However, suitable databases must first be assembled to train and evaluate these models and these novel approaches should be complemented with traditional exposure metrics.


The promise of deep learning in environmental health is great and will complement existing measurements for data-rich settings and could enhance the resolution and accuracy of estimates in data poor scenarios. Interdisciplinary partnerships will be needed to fully realize this potential.

November 26 | 2018

Association of Prenatal Exposure to Air Pollution With Autism Spectrum Disorder. 

Pagalan L, Bickford C, Weikum W, Lanphear B, Brauer M, Lanphear N, Hanley GE, Oberlander TF, Winters M. 

JAMA Pediatr. 2018 Nov 19. doi: 10.1001/jamapediatrics.2018.3101. [Epub ahead of print] 10.1001/jamapediatrics.2018.3101



IMPORTANCE: The etiology of autism spectrum disorder (ASD) is poorly understood, but prior studies suggest associations with airborne pollutants.

OBJECTIVE: To evaluate the association between prenatal exposures to airborne pollutants and ASD in a large population-based cohort.

DESIGN, SETTING, AND PARTICIPANTS: This population-based cohort encompassed nearly all births in Metro Vancouver, British Columbia, Canada, from 2004 through 2009, with follow-up through 2014. Children were diagnosed with ASD using a standardized assessment with the Autism Diagnostic Interview-Revised and Autism Diagnostic Observation Schedule. Monthly mean exposures to particulate matter with a diameter less than 2.5 µm (PM2.5), nitric oxide (NO), and nitrogen dioxide (NO2) at the maternal residence during pregnancy were estimated with temporally adjusted, high-resolution land use regression models. The association between prenatal air pollution exposures and the odds of developing ASD was evaluated using logistic regression adjusted for child sex, birth month, birth year, maternal age, maternal birthplace, and neighborhood-level urbanicity and income band. Data analysis occurred from June 2016 to May 2018.

EXPOSURES: Mean monthly concentrations of ambient PM2.5, NO, and NO2 at the maternal residence during pregnancy, calculated retrospectively using temporally adjusted, high-resolution land use regression models.

MAIN OUTCOMES AND MEASURES: Autism spectrum disorder diagnoses based on standardized assessment of the Autism Diagnostic Interview-Revised and Autism Diagnostic Observation Schedule. The hypothesis being tested was formulated during data collection.

RESULTS: In a cohort of 132 256 births, 1307 children (1.0%) were diagnosed with ASD by the age of 5 years. The final sample size for the PM2.5-adjusted model was 129 439 children, and for NO and NO2, it was 129 436 children; of these, 1276 (1.0%) were diagnosed with ASD. Adjusted odds ratios for ASD per interquartile range (IQR) were not significant for exposure to PM2.5 during pregnancy (1.04 [95% CI, 0.98-1.10] per 1.5 μg/m3 increase [IQR] in PM2.5) or NO2 (1.06 [95% CI, 0.99-1.12] per 4.8 ppb [IQR] increase in NO2) but the odds ratio was significant for NO (1.07 [95% CI, 1.01-1.13] per 10.7 ppb [IQR] increase in NO). Odds ratios for male children were 1.04 (95% CI, 0.98-1.10) for PM2.5; 1.09 (95% CI, 1.02-1.15) for NO; and 1.07 (95% CI, 1.00-1.13) for NO2. For female children, they were for 1.03 (95% CI, 0.90-1.18) for PM2.5; 0.98 (95% CI, 0.83-1.13) for NO; and 1.00 (95% CI, 0.86-1.16) for NO2.

CONCLUSIONS AND RELEVANCE: In a population-based birth cohort, we detected an association between exposure to NO and ASD but no significant association with PM2.5 and NO2.

November 19 | 2018

The health risks and benefits of cycling in urban environments compared with car use: health impact assessment study.

David Rojas-Rueda, Audrey de Nazelle, Marko Tainio, Mark J Nieuwenhuijsen


BMJ 2011; 343 doi:



Objective To estimate the risks and benefits to health of travel by bicycle, using a bicycle sharing scheme, compared with travel by car in an urban environment.

Design Health impact assessment study.

Setting Public bicycle sharing initiative, Bicing, in Barcelona, Spain.

Participants 181 982 Bicing subscribers.

Main outcomes measures The primary outcome measure was all cause mortality for the three domains of physical activity, air pollution (exposure to particulate matter <2.5 μm), and road traffic incidents. The secondary outcome was change in levels of carbon dioxide emissions.

Results Compared with car users the estimated annual change in mortality of the Barcelona residents using Bicing (n=181 982) was 0.03 deaths from road traffic incidents and 0.13 deaths from air pollution. As a result of physical activity, 12.46 deaths were avoided (benefit:risk ratio 77). The annual number of deaths avoided was 12.28. As a result of journeys by Bicing, annual carbon dioxide emissions were reduced by an estimated 9 062 344 kg.

Conclusions Public bicycle sharing initiatives such as Bicing in Barcelona have greater benefits than risks to health and reduce carbon dioxide emissions.


November 12 | 2018

A Web-Based Survey of Residents’ Views on Advocating with Patients for a Healthy Built Environment in Canada. 

Matthew Cruickshank and Marcus Law

International Journal of Family Medicine Volume 2014, Article ID 458184, 7 pages



Purpose. To determine family medicine residents’ perceived knowledge and attitudes towards the built environment and their responsibility for health advocacy and to identify their perceived educational needs and barriers to patient education and advocacy.

Methods. A web-based survey was conducted in Canada with University of Toronto family medicine residents. Data were analyzed descriptively.

Results. 93% agreed or strongly agreed that built environment significantly impacts health. 64% thought educating patients on built environment is effective disease prevention; 52% considered this a role of family physicians. 78% reported that advocacy for built environment is effective disease prevention; 56% perceived this to be the family physician’s role. 59% reported being knowledgeable to discuss how a patient’s environment may affect his/her health; 35% reported being knowledgeable to participate in community discussions on built environment. 78% thought education would help with integration into practice. Inadequate time (92%), knowledge (73%), and remuneration (54%) were barriers.

Conclusions. While residents perceived value in education and advocacy as disease prevention strategies and acknowledged the importance of a healthy built environment, they did not consider advocacy towards this the family physician’s role. Barrier reduction and medical education may contribute to improved advocacy, ultimately improving physical activity levels and patient health outcomes.

November 5 | 2018

Mapping Air Pollution with Google Street View Cars: Efficient Approaches with Mobile Monitoring and Land Use Regression.

Messier KP, Chambliss SE, Gani S, Alvarez R, Brauer M, Choi JJ, Hamburg SP, Kerckhoffs J, LaFranchi B, Lunden MM, Marshall JD, Portier CJ, Roy A, Szpiro AA, Vermeulen RCH, Apte JS.

Environ Sci Technol. 2018 Oct 24. doi: DOI:10.1021/acs.est.8b03395 [Epub ahead of print]



Air pollution measurements collected through systematic mobile monitoring campaigns can provide outdoor concentration data at high spatial resolution. We explore approaches to minimize data requirements for mapping a city’s air quality using mobile monitors with “data-only” versus predictive modeling approaches. We equipped two Google Street View cars with 1-Hz instruments to collect nitric oxide (NO) and black carbon (BC) measurements in Oakland, CA. We explore two strategies for efficiently mapping spatial air quality patterns through Monte Carlo analyses. First, we explore a “data-only” approach where we attempt to minimize the number of repeated visits needed to reliably estimate concentrations for all roads. Second, we combine our data with a land use regression-kriging (LUR-K) model to predict at unobserved locations; here, measurements from only a subset of roads or repeat visits are considered. Although LUR-K models did not capture the full variability of on-road concentrations, models trained with minimal data consistently captured important covariates and general spatial air pollution trends, with cross-validation R2 for log-transformed NO and BC of 0.65 and 0.43. Data-only mapping performed poorly with few (1-2) repeated drives but obtained better cross-validation R2 than the LUR-K approach within 4 to 8 repeated drive days per road segment.



October 29 | 2018

A machine learning approach to estimate hourly exposure to fine particulate matter for urban, rural, and remote populations during wildfire seasons. 

Yao J, Brauer M, Raffuse SM, Henderson S.

Environ Sci Technol. 2018 Oct 24. Epub ahead of print] DOI:10.1021/acs.est.8b01921


Exposure to wildfire smoke averaged over 24-hour periods has been associated with a wide range of acute cardiopulmonary events, but little is known about the effects of sub-daily exposures immediately preceding these events. One challenge for studying sub-daily effects is the lack of spatially and temporally resolved estimates of smoke exposures. Inexpensive and globally applicable tools to reliably estimate exposure are needed. Here we describe a Random Forests machine learning approach to estimate 1-hour average population exposure to fine particulate matter during wildfire seasons from 2010 to 2015 in British Columbia, Canada, at a 5km by 5km resolution. The model uses remotely sensed fire activity, meteorology assimilated from multiple data sources, and geographic/ecological information. Compared with observations, model predictions had a correlation of 0.93, root mean squared error of 3.2 µg/m3, mean fractional bias of 15.1%, and mean fractional error of 44.7%. Spatial cross-validation indicated an overall correlation of 0.60, with an interquartile range from 0.48 to 0.70 across monitors. This model can be adapted for global use, even in locations without air quality monitoring. It is useful for epidemiologic studies on sub-daily exposure to wildfire smoke, and for informing public health actions if operationalized in near-real-time.

October 22 | 2018

Socioeconomic status and environmental noise exposure in Montreal, Canada.

Dale LM, Goudreau S, Perron S, Ragettli MS, Hatzopoulou M, Smargiassi A.

BMC Public Health. 2015 Feb 28;15:205. doi: 10.1186/s12889-015-1571-2




This study’s objective was to determine whether socioeconomically deprived populations are exposed to greater levels of environmental noise.


Indicators of socioeconomic status were correlated with LAeq24h noise levels estimated with a land-use regression model at a small geographic scale.


We found that noise exposure was associated with all socioeconomic indicators, with the strongest correlations found for median household income, proportion of people who spend over 30% of their income on housing, proportion of people below the low income boundary and with a social deprivation index combining several socio-economic variables.


Our results were inconsistent with a number of studies performed elsewhere, indicating that locally conducted studies are imperative to assessing whether this double burden of noise exposure and low socioeconomic status exists in other contexts. The primary implication of our study is that noise exposure represents an environmental injustice in Montreal, which is an issue that merits both investigation and concern.



October 15 | 2018

Association between residential self-selection and non-residential built environment exposures.

Howell NA, Farber S, Widener MJ, Allen J, Booth GL.

Health Place. 2018 Oct 1;54:149-154 DOI: 10.1016/j.healthplace.2018.08.009


Studies employing ‘activity space’ measures of the built environment do not always account for how individuals self-select into different residential and non-residential environments when testing associations with physical activity. To date, no study has examined whether preferences for walkable residential neighborhoods predict exposure to other walkable neighborhoods in non-residential activity spaces. Using a sample of 9783 university students from Toronto, Canada, we assessed how self-reported preferences for a walkable neighborhood predicted their exposure to other walkable, non-residential environments, and further whether these preferences confounded observed walkability-physical activity associations. We found that residential walkability preferences and non-residential walkability were significant associated (β = 0.42, 95% CI: (0.37, 0.47)), and further that these preferences confounded associations between non-residential walkability exposure and time spent walking (reduction in association = 10.5%). These results suggest that self-selection factors affect studies of non-residential built environment exposures.