Relationships among interventions and health literacy outcomes for sub-populations: A data-driven approach

Abstract

ObjectivesThe goals of this study were to examine relationships among health literacy and outcomes for sub-populations identified within a large, multi-dimensional Omaha System dataset. Specific aims were to extract sub-populations from the data using Latent Class Analysis (LCA); and quantify the change in knowledge score from pre- to post-intervention for common sub-populations. DesignData-driven retrospective study using statistical modeling methods. SampleA set of admission and discharge cases, captured in the Omaha System, representing 65,468 cases from various health care providers. MeasuresDemographic information and the Omaha System terms including problems, signs/symptoms, and interventions were used as the features describing cases used for this study. Development of a mapping of demographics across health care systems enabled the integration of data from these different systems. ResultsKnowledge scores increased for all five sub-populations identified by latent class analysis. Effect sizes of interventions related to health literacy outcomes varied from low to high, with the greatest effect size in populations of young at-risk adults. The most significant knowledge gains were seen for problems including Pregnancy, Postpartum, Family planning, Mental health, and Substance use. ConclusionsThis is the first study to demonstrate positive relationships between interventions and health literacy outcomes for a very large sample. A deeper analysis of the results, focusing on specific problems and relevant interventions and their impact on health literacy is required to guide resource allocation in community-based care. As such, future work will focus on determining correlations between interventions for specific problems and knowledge change post-intervention.