Evaluating the Impact of Missing Data When Assessing Disseminated Effects in Network-Based Studies

• Activity Number: 263 - Addressing Incomplete Data in Public Health Studies: New Frontiers for Network-Based Studies and Meta-Analyses
• Type: Topic Contributed
• Date/Time: Tuesday, August 4, 2020 : 1:00 PM to 2:50 PM
• Sponsor: ENAR
• Author(s): Ashley Buchanan and Natallia Katenka and Benjamin Skov and TingFang Lee* and M. Elizabeth Halloran and Samuel Friedman and Georgios Nikolopoulos
• Companies: University of Rhode Island and University of Rhode Island and University of Rhode Island and University of Rhode Island and Fred Hutchinson Cancer Research Center and New York University and University of Cyprus
• Keywords: Causal Inference; Dissemination; Network Studies; People who Use Drugs; HIV/AIDS; Missing data

Abstract

People who inject drugs (PWID) are embedded in social networks and exert influence on the members of their networks. The disseminated effect is the effect on the unexposed participants who shared a network with those who received the intervention. We assumed that smaller neighborhoods for each individual can be identified in the data. However, the outcomes may be missing due to administrative end of the study or participants being lost to follow-up, known as censoring. We used an inverse probability weighted approach with censoring weights to quantify the disseminated effects in a network-based study. We conducted a simulation study to evaluate the performance of the inverse probability weighted estimator with both exposure and censoring weights. The Transmission Reduction Intervention Project (TRIP), 2013 to 2015, Athens, Greece, was an observed network-based study with a nonrandomized exposure, where information is available on each participant and their connections with other participants. We analyzed a network of PWID from TRIP to evaluate the effects community alerts on HIV risk behavior, where links were defined by shared sexual/injection behaviors.