B1315
Title: Latent variable modeling to understand characteristics associated with disparities in adult asthma
Authors: Knashawn Morales - University of Pennsylvania (United States) [presenting]
Abstract: Asthma, a chronic treatable disease, disproportionately affects low-income and minority adults. Improving access to care and patient-provider communication is believed to be essential for better outcomes. Understanding the trajectory of improvement and subgroups that benefit from service interventions is a necessary step following the conduct of randomized trials. Latent variable modeling can capture both heterogeneous patterns in outcomes over time as well as subgroups that demonstrate more or less improvement in outcomes. We discuss how the flexible class of models applied to longitudinal data can be used to quantify these patterns while taking into account patient characteristics and social determinants of health. We then illustrate the method using pooled data from two randomized trials of service interventions for adults with uncontrolled asthma. Identifying characteristics of individuals who are less likely to respond to treatment may signal a need for more intensive interventions and be a stepping stone to reducing disparities.
