Dr. Geoffrey “Jeff” Whitfield, PhD, MEd, was our PAPREN Grand Rounds speaker in November. Dr. Whitfield leads the Epidemiology and Surveillance Team in the CDC’s Physical Activity and Health Branch. His presentation The Future of National Physical Activity Surveillance: Emerging Technologies and Methods reviewed current efforts on data modernization. You can view the presentation recording here and access the slides here. Dr. Whitfield graciously agreed to answer the questions we couldn’t get to in the session. We edited questions lightly for brevity and clarity.
It was very exciting to hear about the interest in computational text analysis and machine learning for policy surveillance. The built environment and policies you talked about all impact the food environment, sustainability outcomes, and broader health equity/environmental justice concerns as well. Are there opportunities to partner with other branches or organizations who may be interested in related components or outcomes?
Whitfield: Yes! Many of the policy approaches that we are reviewing/analyzing can be applied to a range of topics. As an agency, CDC has a data modernization coordination effort already underway so that these surveillance methods do not remain siloed.
Acknowledging sidewalks as part of the transportation network at a policy level could help to incorporate their mapping, planning, and construction into the systems already in place for vehicle transportation. Why not move away from assignment of responsibility for sidewalk construction and maintenance to private property owners, like we did with cars?
Whitfield: This is a good example of the type of question (and resultant action) that might be possible with advanced policy surveillance/research tools. Can we identify the policies, such as Complete Streets, that clearly place sidewalk construction and maintenance in the public sphere? Can we then go another step and actually measure sidewalk construction and use in places with this type of policy language versus without? That could be a huge quantitative step forward.
Has there been any work to combine location-based services data with accelerometry or gyroscopic smart phone data to get more precise information about physical activity modality?
Whitfield: This type of work is being done in research studies, but to my knowledge, not for national-level surveillance purposes.
We may be interested in equity of these resources across neighborhoods, but ACS/Census data is for a year or multiple years and Census tracts vs. more granular data. How do you think our field will navigate issues of differences in granularity that can result in defaulting to largest spatial or temporal scale?
Whitfield: This will continue to be an area of investigation. What is the appropriate trade-off between spatial and temporal granularity? How does this differ by population density or urban/rural areas? There is still much to learn, but early indications from our Technical Expert Panels indicate that Census tract-level estimates are probably possible for most of the United States, and perhaps even tract-level estimates more than once per year. Below the tract, the issues with sparse data in low population density areas, potential identifiability, and data size (storage and analysis) become more and more salient for a national surveillance system.
Isn’t it important to capture the context of vehicle travel, such as with StreetLight Data, to understand more deeply where active transportation is or is not happening?
Whitfield: There are definitely active transportation uses for vehicular data, in particular identifying areas or roadways that present particular risks for pedestrians and bicyclists. This is an area of interest for us.