Primary Submission Category: Interference and Consistency Violations
Statistical Methods for Causal Effects of Multi-component Interventions in Longitudinal Observational Studies with Interference
Authors: Ke Zhang, Ashley Buchanan, Laura Forastiere, Natallia Katenka, Donna Spiegelman, Collins lwuji,
Presenting Author: Ke Zhang*
Spillover effects arise when an intervention received by one unit affects the outcomes of units within a predefined group, referred to as an interference set. Such effects commonly occur in sociometric clusters/networks, such as HIV prevention programs, due to interactions with HIV transmission risk among individuals. HIV prevention programs are often delivered as intervention packages to reduce HIV incidence in the community, which is a combination of single components to prevent or treat HIV through multiple pathways simultaneously. Disentangling component-specific effects of intervention packages at the individual and community level is essential for fully understanding the effectiveness of HIV interventions and improving package interventions. However, existing causal methods for estimating spillover effects are typically limited to settings with a single intervention (whether static or time-varying) or to multiple interventions that are analyzed as a whole, thereby unable to provide insights into which components were driving (or hindering) the effectiveness. In this study, we develop novel causal methods for cluster randomized trials with time-varying exposure to intervention package components in the presence of interference and non-compliance. We expand partial interference assumption to temporal partial interference to account for interference in time-varying settings, use marginal structure models to estimate the marginal potential outcome for the intervention compo