Purpose:  This presentation will focus on methodological lessons learned from the Houston TRAIN Project and present recommendations for future work.

Methods:  Natural experiments have the potential to add valuable answers to scientific questions that may not be able to be addressed with experimental or traditional observational studies. In a natural experiment, participants are exposed to a “treatment” that is not under the control of the investigator. The Houston Travel-Related Activity In Neighborhoods (TRAIN) Project was a natural experiment designed to take advantage of a Light Rail Train (LRT) expansion project throughout the city of Houston Texas (USA) to determine if increased access to mass transit would result in an increase in transportation related physical activity. Four waves of recruitment resulted in a cohort of 865 participants who were followed for varying amounts of time as four expansions of the LRT were completed. Physical activity was assessed via questionnaire and accelerometry.

Results:  Although there was no association between the LRT expansion proximity and changes in travel-related physical activity, sub-analyses suggested that usual travel mode was a potential confounding factor. Several unexpected and uncontrollable outside influences occurred during the project that could explain the findings. Participant recruitment and retention, construction delays, a natural disaster, and potential delays or lags between environmental change and behavior change were unique influences that could explain the findings from this study.

Conclusions:  The utility of natural experiments to improve physical activity must be balanced with unexpected and uncontrollable external influences.

Purpose: To describe the practical and methodological challenges faced and lessons learned from designing and implementing a natural experiment assessing the effects of a large-scale expansion of cycling infrastructure in Mexico City, Mexico, on active travel.

Methods: Funds for conducting a two-year natural experiment study were secured in mid-2019. The original study was slated to examine the effects of a planned expansion of Mexico City’s public bicycle-sharing program (EcoBici) on active travel. Our research team designed a rigorous study, including intervention and control neighborhood samples, based on the expansion timelines and maps provided by the city. Data collection was initiated in October 2019. However, in March 2020, the COVID-19 emergency reached Mexico, and with this came a major shift in urban infrastructure resources and priorities. EcoBici expansion plans were delayed and modified multiple times while other cycling infrastructure improvements were funded and rolled out. These new plans included major and rapid expansions to the City’s protected cycling lane infrastructure, emphasizing adding high-quality cycle lanes to high-capacity roads traversing the city North-to-South and East-to-West. Our research team worked closely with the funding agency to adapt the original study design to capture the effect of these COVID-related infrastructure improvements on active travel patterns.

Results: The original aims and hypotheses focused on the effects of bicycle-sharing programs remain unrealized. However, pandemic-induced expansions of protected cycle lane infrastructure provided an excellent platform for assessing the impacts of a different type of cycling infrastructure on active travel. Due to the disruption caused by the pandemic, the study duration was expanded to 3 years. Preliminary results suggest significant increases in cycling for transportation associated with new infrastructure. Despite the intention for the new protected cycle lanes rolled out during COVID-19 in high-capacity roads to be a temporary pandemic-response measure, many have been converted into permanent infrastructure given high user demand.

Conclusions: Rigid designs, protocols, timelines, and funding schemes may be incompatible with successful and informative natural experiments. Finding a balance between scientific rigor and flexibility is key for successfully conducting natural experiments that can inform better urban design for health.

Purpose: This presentation will describe the study design and methods of the STREETS study, with examples of challenges faced and lessons learned. Preliminary data for the project will also be shown.

Methods: Evaluations to determine the efficacy of urban infrastructure interventions to increase active commuting to schools (ACS) have been small in scope and often lack objective measurements of physical activity.  In 2016, a mobility bond in Austin, Texas, USA, allocated $27.5 million (USD) for changes in the built environment around schools to facilitate ACS. The Safe TRavel Environment Evaluations in Texas Schools (STREETS) study was designed to determine the health effects of these infrastructure changes. The STREETS Study comprises two designs: 1) a serial cross-sectional design to assess changes in the prevalence of ACS in participating schools over time, and 2) a quasi-experimental, prospective cohort to examine changes in child physical activity and ACS.  Measures include ACS tally measures across grades 3, 4, and 5; school-level policy and programs surveys; parent surveys; child surveys; child physical activity as measured using accelerometers; and school-level audits.

Results:  Study schools were recruited in 2018 (n = 88 for the cross-sectional study, 45 for the quasi-experimental study).  Both studies included comparison schools that did not receive funds for infrastructure changes.  Initial recruitment and measurements were robust, but the COVID pandemic in 2020 necessitated changes in data collection procedures. During this time, school-level measures were paused, and individual-level child and parent measures were transitioned to home-based measures.  COVID also affected the schedule for the infrastructure improvements. Strategies to address these changes included: (1) working with school districts to develop alternate measurement protocols; (2) alterations in study design to account for school attrition and missing data; (3) regular meetings and communication with city urban planners to track changes in construction schedules; and (4) flexibility with the study timeline.

Conclusions:  Flexibility in study design, data collection protocols, and funding timeline is crucial for natural experiments that evaluate infrastructure changes.  Since the ‘intervention’ e.g., infrastructure change, is not within the control of study investigators, a close working relationship with city planners is necessary.