Mobile Phone-Based Neighborhood Audits

We recently published a paper describing our efforts to adapt street audit strategies for use in a large informal community, Rio das Pedras (RdP) located in Rio de Janeiro, Brazil.  We developed a smartphone-based systematic observation protocol to gather street-level information for a high-density convenience sample of street segments in RdP (N = 630, estimated as 86% of all street segments in the community).  The street audit protocol was built on the Fulcrum app deployed on smart phones and allowed street auditors to record observational data into forms on the phone’s screen and to capture geotagged images from the phone’s camera. The app also had a mapping interface to help guide the field team around RdP to the selected street segments that were being audited.

Posted in Uncategorized | Leave a comment

Free-floating Bike Share in Seattle

BEH member Steve Mooney recently led two studies regarding the use of the free-floating bike share system in Seattle.  (Free-floating bike share systems are systems that allow users to pick up and leave bikes anywhere within a service area rather than at dedicated docking stations). These studies showed two things: a) that few people riding free-floating bike share rentals in Seattle are wearing helmets and b) that bikes were usually available in all Seattle neighborhoods across economic, racial and ethnic lines.

The helmet study, recently published in the Journal of Community Health, suggested that bike share systems, as compared with private bikes, may facilitate a more casual approach to cycling that makes helmet usage more challenging.  The team, mostly members of the INSIGHT program at the Harborview Injury Prevention & Research Center in Seattle, counted the number of cyclists – noting helmet use –at four strategic locations around Seattle: the Fremont bridge, the Burke-Gilman Trail, Broadway Bike Lane and NW 58th Street at 22nd Avenue NW.  They found that only one in five people riding bike shares wore helmets, as compared with nine in ten people riding private bikes.  Intriguingly, they also found that fewer private bike cyclists wore helmets in locations where there were more bike share users.

Other studies have found similarly low statistics of bike helmet use in other bike share systems, with about 15 percent helmet use in New York and about 39 percent in Boston compared with Seattle’s estimated 20 percent use of helmets. However, Vancouver’s system, Mobi, provides helmets and boosts usage there as much as three times Seattle, at 64 percent.

In a separate study, published recently in the Journal of Transport Geography, Mooney (and frequent BEH collaborator Jana Hirsch) again looked at bike share programs. This time they wanted to see if the benefits of bike share were available to all neighborhoods regardless of economic, racial and ethnic composition.  Prior studies have shown that docked bike share systems, which are geographically constrained by station locations, tend to favor advantaged neighborhoods.

What they found was nuanced: a baseline level of free-floating bikes were available in all neighborhoods across the city.  When they looked a little more closely at where bikes tended to be ridden to and ‘rebalanced’ to (i.e. where the operating companies moved the bikes around the city) and merged that with neighborhood based census data, they did find a higher concentration of bikes in more affluent areas of the city.  This inequity appeared to be driven by unequal demand across the city.  That is, bike share operators did a good job of rebalancing bikes to places where they did not remain idle long – but those places tended to have wealthier and more educated residents than the city as a whole.

Their future work will dig more deeply into barriers to bike share usage and how much the low helmet use affects injury risk.

Posted in Active Transport, Bike Share, Bikeshare, Injury, Physical Activity, Safety, Socioeconomic status, Transportation | Leave a comment

Launching the Interactive-Pedestrian Injury Mapper (I-PIM)

In 2015 in the U.S. 5,376 pedestrians were killed and 70,000 were injured. The Built Environment and Health Research Group has just launched the Interactive-Pedestrian Injury Mapper (I-PIM) website (HERE), to crowd source the collection of data on locations where pedestrians have been hit by automobiles.  Our goal is to collect location data on intersections where pedestrians have been injured so that built environment, side-walk and road-way risk factors for pedestrian injuries can be identified.

I-PIM has a Google maps based tool that allows a user to place a pin on the intersection or street where they were hit and to then map the route they walked prior to getting hit. I-PIM then asks some questions about the user, such as age and gender and then asks some questions about the collision, such as the type of vehicle that hit the person and whether they received medical care at the scene or had to go to the hospital.

Once injury location and route data from a lot of people have been entered into I-PIM, we plan to use Google Street View to virtually walk the routes and collision site and collect data on built environment, side-walk and road-way risk factors such as the presence of cross-walks, stop signs, visual distractions and blocked sight lines.  Our goal is to understand the ways in which pedestrian and roadway infrastructure at intersections that people crossed without being hit differs from intersections where they were hit.

If you have been hit by an automobile and would like to enter your information into I-PIM click HERE.

Posted in Pedestrian Injury | Leave a comment

How and where patterns of activity among older adults change over time

At BEH, we’re interested in how your residential neighborhood affects how physically active you are. But we’ve come to understand that being active as not just one thing and not merely a matter of expending calories. That is, walking is different from yoga is different from basketball is different from a seven-minute workout every morning.  They’re all activity, and for the most part, they’re all good for you, but two people could both meet the recommended number of minutes of activity per week, but arrive there through very different patterns of activity. We focus a lot on walking, because there’s extensive evidence that walking is one of the easiest ways to integrate activity into your daily life. As a result, we expect that making our cities more walkable could be a great way to get our population more active.

But for any given person, meeting activity recommendations may be easier for some activities than for others.  Few older adults play basketball, say, but a lot of older adults say they’d like low-cost, low-impact activities, like walking and gardening.  A few years ago, we started characterizing the patterns of types of activity older adults engage in, focusing on the NYCNAMES-II cohort, a cohort of older adults in New York we surveyed three times over five years. We found that adhering to specific patterns of activity predicted lower BMI better than simply analyzing total activity alone, and also explained prevalence of depressive symptoms. We also know this cohort’s neighborhood conditions (including neighborhood disorder, a conditionwe’vestudiedextensively) are correlated with their total activity and with their depressive symptoms.

So we thought we might learn something by looking at change through time in activity patterns in the NYCNAMES-II cohort, and whether those changes we correlated with neighborhood conditions.  In a paper that’s just hitting the web now at the American Journal of Epidemiology, we used a latent transition analysis to explore how the activity patterns of the study participants changed through time, and then explored individual and neighborhood predictors of those changes.

Activity classes are shown in boxes, and common transitions (among those who changed classes) are indicated by arrows between boxes

We found that there were 7 common activity patterns and that the most common changes between those patterns came from adding or removing one activity type (e.g. walking, sports & exercises, etc.). Neighborhood unemployment rate was the only neighborhood level factor we found to be consistently associated with transitioning between activity patterns.

Our results were encouraging. The latent transition analysis gave us a better picture of the kinds of changes older adults in New York City make in their activity, and we learned that, for example, engaging in gardening at any given time is not just a matter of having access to gardening space and an inclination to garden.  We anticipate that future latent transition analyses, with more people followed over more years could give us further insight into which activities we might suggest to which older adults to best increase population activity levels.

Posted in Adults, Methods, Physical Activity, Physical Disorder | Leave a comment

Neighborhood Conditions Influence the Ability of Diabetics to Control Their Blood Sugar

In collaboration with researchers from the New York City Department of Health and Mental Hygiene we recently published an article in the American Journal of Epidemiology showing that diabetics living in neighborhoods with more advantaged economic environments, greater walkability and healthier retail food environments have an improved ability to achieve glycemic control.  Hemoglobin A1C data from the New York City A1C Registry for 182,756 adults who had 1,273,801 A1C tests from 2007 to 2013 were analyzed along with data describing the neighborhood contexts they lived in.  The odds of individuals achieving glycemic control in the most advantaged residential neighborhoods (better economic conditions, greater walkability, healthier retail food profiles) was two and a half times greater than in the least advantaged.  Furthermore, individuals who lived in the most advantaged residential neighborhoods achieved glycemic control in a shorter period of time than individuals who lived in the least advantaged neighborhoods.  For those who moved during the 2007 to 2014 period, moving from less advantaged neighborhoods to more advantaged neighborhoods was associated with improved diabetes control, while moving from more advantaged areas to less advantaged areas was related to worsening diabetes control.  This is the first longitudinal study to examine the relationship between residential neighborhood environments and individual’s ability to control their diabetes.

Posted in Diabetes, Food Environment, Parks, Social Determinants, Socioeconomic status, Urban Design, Walkability | Leave a comment

Teaching Epidemiology to Undergraduate Students

Undergraduate programs in public health are proliferating (and see here), and increasing numbers of undergraduate students are receiving training in epidemiology.  James Stark, a BEH alum and now a Director of Epidemiology at Pfizer and Adjunct Professor at NYU’s College of Global Public Health, just published a paper in the American Journal of Epidemiology on approaches to teaching epidemiology at the undergraduate level.

Epidemiology courses introduce undergraduate students to a population health perspective and provide opportunities for these students to build essential skills and competencies such as ethical reasoning, teamwork, comprehension of scientific methods, critical thinking, quantitative and information literacy, ability to analyze public health information, and effective writing and oral communication.

Posted in Uncategorized | Leave a comment

The Built Environment and Health Research Group is looking for Post-Docs.

We are looking for candidates to fill a post-doctoral researcher position at the Department of Epidemiology at the Columbia University, Mailman School of Public Health.

The post-doc position will be at Columbia University, but we are a multi-disciplinary team of faculty at the Mailman School, the Columbia University School of Social Work, Drexel University, American University and the University of Washington.  Strong candidates will have a Doctorate and experience and interest in social epidemiology, spatial epidemiology, neighborhood health research, GIS and/or urban health.  Post-Docs will be able to collaborate on on-going and developing BEH projects and to develop their own research projects.  On-going and developing BEH projects focus; 1) on the links between neighborhood built environments and obesity, physical activity, pedestrian safety, asthma, cancer, and cardio vascular disease, and 2) on developing new methods for characterizing neighborhood environments.

Please contact Andrew Rundle at

Posted in Uncategorized | Leave a comment

When geographic proximity and access to medical services is not enough

Although many health determinants are outside of the health care sector, quality health care is crucial to population health.  Recently, we included a look at perceptions of local health care as part of a community needs assessment ( in Rio de Janeiro’s third largest favela.  At the time of collecting data in 2015, only a geographic area with approximately 40% of the population was covered by the local family health clinic, and we thus expected that those outside of the coverage area might perceive greater barriers to care.

Family health clinic service area

However, when talking about public health care with residents during 14 semi-structured interviews, our colleague Débora Castiglione noted other salient concerns.  She led the qualitative analyses and coauthored with Dr. Lovasi a paper just published in Qualitative Health Research.  This paper highlights that residents felt disrespected or dehumanized in the process of seeking health care from the public system.  Substantial delays and appointments missed due to the doctor’s absence were perceived as exacerbating vulnerability faced by pregnant women or during injury recovery, extending periods of uncertainty, elevated risk, or disability.  Even in the face of scarce financial resources, residents would pay for private care if they could, in order to get timely care and feel well-received.


Posted in Community Needs Assessment, Health Care Access | Leave a comment

Man on the Street or Google Street View to Measure Neighborhood Physical Disorder

We’ve done a lot with Street View at the BEH, and we think the CANVAS application we developed to help teams do reliable and efficient virtual audits works pretty well.  But we never really knew what we might be missing by not being on the street in person.

Fortunately, we stumbled across an opportunity to investigate what we might be missing.  It happens that our friends at the Detroit Neighborhood Health Study (DNHS) had conducted an in-person systematic audit of Detroit streets only one year prior to when Google captured the imagery that we’d audited on Street View, and the DNHS team had constructed a physical disorder measure from their work.

So we took a look at how their measure aligned with ours, and together, we and the DNHS team recently wrote a paper on what we found. Spolier alert: both methods showed pretty similar spatial patterns of disorder — the final measures were significantly positively correlated at census block centroids (r=0.52), identified the same general regions as highly disordered (see attached image) and displayed comparable leave-one-out cross-validation accuracy.

But the two methods didn’t take the same amount of auditor time – the virtual audit required about 3% of the time of the in-person audit, largely because the virtual audit was able to take a more diffuse sample of the streets because travel time between segments was not a factor in developing an audit sample.

There were a number of other differences between the audit designs, including that the CANVAS audit included more disorder indicators and the DNHS audit aggregated street-level measures to create neighborhood area measures before interpolating.  So it wasn’t a completely apples-to-apples comparison and the 97% of auditor time saved might not apply for other audit contexts.  Nonetheless, virtual audits do appear to permit comparable validity with more diffuse samples.

Ultimately, we concluded that the virtual audit-based physical disorder measure could substitute for the in-person one with little to no loss of precision.  Jackelyn Hwang wrote a thoughtful commentary on our paper and on technological innovation in neighborhood research more generally, and we responded to her thoughts.



Posted in CANVAS, Mehtods, Physical Disorder, Street View | Leave a comment

Webinar Online – Urban Informatics: Studying How Urban Design Influences Health in New York City

Dr. Rundle’s March 2nd webinar for the ISBNPA webinar has been posted online at ISBNPA’s web site (Here and embedded below).

His talk covered different approaches to assessing neighborhood walkability and the link between urban design and resident’s physical activity using New York City as a case study.  He highlighted the challenges to measuring neighborhood form across multiple municipal jurisdictions and retrospectively over the past three decades.

Posted in Physical Activity, Urban Design, Walkability | Leave a comment