Assessing the distribution and drivers of vaccine hesitancy using medical claims data
AbstractObjectiveThe purpose of this study was to investigate the use of large-scalemedical claims data for local surveillance of under-immunizationfor childhood infections in the United States, to develop a statisticalframework for integrating disparate data sources on surveillance ofvaccination behavior, and to identify the determinants of vaccinehesitancy behavior.IntroductionIn the United States, surveillance of vaccine uptake for childhoodinfections is limited in scope and spatial resolution. The NationalImmunization Survey (NIS) - the gold standard tool for monitoringvaccine uptake among children aged 19-35 months - is typicallyconstrained to producing coarse state-level estimates.1In recent years,vaccine hesitancy (i.e., a desire to delay or refuse vaccination, despiteavailability of vaccination services)2has resurged in the United States,challenging the maintenance of herd immunity. In December 2014,foreign importation of the measles virus to Disney theme parks inOrange County, California resulted in an outbreak of 111 measlescases, 45% of which were among unvaccinated individuals.3Digitalhealth data offer new opportunities to study the social determinantsof vaccine hesitancy in the United States and identify finer spatialresolution clusters of under-immunization using data with greaterclinical accuracy and rationale for hesitancy.4MethodsOur U.S. medical claims data comprised monthly reports ofdiagnosis codes for under-immunization and vaccine refusal(Figure 1). These claims were aggregated to five-digit zip-codes bypatient age-group from 2012 to 2015. Spatial generalized linear mixedmodels were used to generate county-level maps for surveillanceof under-immunization and to identify the determinants of vaccinehesitancy, such as income, education, household size, religious grouprepresentation, and healthcare access. We developed a Bayesianmodeling framework that separates the observation of vaccinehesitancy in our data from true underlying rates of vaccine hesitancyin the community. Our model structure also enabled us to borrowinformation from neighboring counties, which improves predictionof vaccine hesitancy in areas with missing or minimal data. Estimatesof the posterior distributions of model parameters were generated viaMarkov chain Monte Carlo (MCMC) methods.ResultsOur modeling framework enabled the production of county-levelmaps of under-immunization and vaccine refusal in the UnitedStates between 2012-2015, the identification of geographic clustersof under-immunization, and the quantification of the associationbetween various epidemiological factors and vaccination status.In addition, we found that our model structure enabled us to accountfor spatial variation in reporting vaccine hesitancy, which improvedour estimation.ConclusionsOur work demonstrate the utility of using large-scale medicalclaims data to improve surveillance systems for vaccine uptake andto assess the social and ecological determinants of vaccine hesitancy.We describe a flexible, hierarchical modeling framework forintegrating disparate data sources, particularly for data collectedthrough different measurement processes or at different spatial scales.Our findings will enhance our understanding of the causes of under-immunization, inform the design of vaccination policy, and aid inthe development of targeted public health strategies for optimizingvaccine uptake.Figure 1. Instances of vaccine refusal (per 100,000 population) for UnitedStates counties in 2014 as observed in medical claims data.
How to Cite
Goldlust, S., Lee, E., & Bansal, S. (2017). Assessing the distribution and drivers of vaccine hesitancy using medical claims data. Online Journal of Public Health Informatics, 9(1). https://doi.org/10.5210/ojphi.v9i1.7590
Novel algorithms, statistical or mathematical methods