|
Activity Number:
|
472
- Statistical Methods for Causal Inference
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, July 31, 2019 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
|
Abstract #305247
|
|
|
Title:
|
Contamination in Stepped-Wedge Randomized Trials and Its Impact on Public Health Interventions
|
|
Author(s):
|
Lior Rennert* and Moonseong Heo and Victor De Gruttola
|
|
Companies:
|
Clemson University and Clemson University and Harvard T.H. Chan School of Public Health
|
|
Keywords:
|
Stepped-wedge design;
contamination;
opioid;
public health
|
|
Abstract:
|
A stepped-wedge (SW) design is proposed to implement a comprehensive external facilitation intervention in 15 South Carolina communities with the aim of reducing opioid overdose mortality. The SW design randomly selects each community to receive the intervention at a prespecified time point. This allows for all communities to eventually receive the intervention and may alleviate ethical concerns of withholding lifesaving interventions from control communities. However, they may lead to contamination (also called interference) across randomized communities during the intervention roll-out periods. This can occur in the current setting if success of intervention in treated communities influence the control communities to preemptively adopt the intervention. Based on simulations using data from the 15 communities, we find that even relatively small incremental contamination can cause severe attenuation of the intervention effect and reduction in power. In this talk, we explore the bias in the intervention effects under various contamination schemes, and propose strategies for minimizing this bias in these settings.
|
Authors who are presenting talks have a * after their name.
