Approaches to enhance interpretability and meaningful use of big data in population health practice and research
MetadataShow full item record
While many public health and medical studies use big data, the potential for big data to further population health has yet to be fully realized. Because of the complexities associated with the storage, processing, analysis, and interpretation of these data, few research findings from big data have been translated into practice. Using small area estimation synthetic data and electronic health record (EHR) data, the overall goal of this dissertation research was to characterize health-related exposures with an explicit focus on meaningful data interpretability. In our first aim, we used regression models linked to population microdata to respond to high-priority needs articulated by our community partners in New Bedford, MA. We identified census tracts with an elevated percentage of high-risk subpopulations (e.g., lower rates of exercise, higher rates of diabetes), information our community partners used to prioritize funding opportunities and intervention programs. In our second and third aims, we scrutinized EHR data on children seen at Boston Medical Center (Boston, MA), New England’s largest safety-net hospital, from 2013 through 2017 and uncovered racial/ethnic disparities in asthma severity and residential mobility using logistic regression. We built upon a validated asthma computable phenotype to create a computable phenotype for asthma severity that is based in clinical asthma guidelines. We found that children for whom severity could be ascertained from these EHR data were less likely to be Hispanic and that Black children were less likely to have lung function testing data present. Lastly, we constructed contextualized residential mobility and immobility metrics using EHR address data and the Child Opportunity Index 2.0, identified opportunities and challenges EHR address data present to study this topic, and found significant racial/ethnic disparities in access to neighborhood opportunity. Our findings highlighted the perpetuation of residence in low opportunity areas among non-White children. The main challenge of this dissertation, to work within the limitations inherent to big data to extract meaningful knowledge from these data and by linking to external datasets, turned out to be an opportunity to engage in solutions-oriented research and do work that, to quote Aristotle, “…is greater than the sum of its parts”. Through strategies ranging from engaging with community partners to examining who and what data are captured (and not captured) in EHR health and address data, this dissertation demonstrated potential ways to leverage big data sources to further public health and health equity.