A1213
Title: A doubly hierarchical changepoint model for multi-unit interrupted time series
Authors: Maricela Cruz - Kaiser Permanente Washington Health Research Institute (United States) [presenting]
Andrew Holbrook - UCLA (United States)
Abstract: Interrupted time series (ITS) designs are aptly situated for studying the impacts of large-scale public health policies, as they borrow from case-crossover designs and can retrospectively assess the impact of an intervention. Recent advances in ITS methods include formal tests for the existence of a change point, procedures to estimate a change point when appropriate, models that capture post-intervention changes in higher-order moments, and models that estimate marginal effects. However, no ITS methods that estimate a change point also quantify the uncertainty around those estimates. DHITS is proposed, a Bayesian doubly hierarchical change point model that identifies unit-specific change points and quantifies their uncertainty, while sharing information across units. The model also estimates a global change point across all units (along with its variance) and accounts for changes in temporal dependence following the intervention. The methodology is demonstrated by analyzing multi-unit patient centered data from a hospital that implemented a new care delivery model.
