Group therapy is a central treatment modality for alcohol or drug (AOD) disorders. Clients are often admitted to therapy groups under a rolling admissions basis, so that new members enter the group while others leave. Failure to properly adjust for this correlation could lead to biased statistical tests of treatment effects, thus impeding efforts to make AOD group therapy more effective. AOD treatment innovations for rolling groups should be guided by improved knowledge of the complex dynamics of client interactions. Despite the ubiquity of rolling groups in standard clinical practice, there is very little guidance available to researchers regarding proper analysis of these data. The key innovation of this project is that we develop an explicit model for the contribution of session participation on outcomes. Our model exploits similarities between sessions, such as their proximity to each other in time and the degree of overlap in participating clients. We do so by using statistical techniques developed to model data that are related due to geographic locations of sampled units. We take advantage of the conceptual similarity between measuring distance between geographic locations and measuring the closeness of sessions. These spatial statistical techniques provide a rich &powerful set of tools to model correlations among client outcomes in rolling groups. Our hierarchical (multilevel) models capture session-level effects and allow them to be correlated. This project provides a unique opportunity to integrate AOD treatment research with state-of-the-art statistical modeling to develop appropriate analytic techniques for rolling therapy group data.
Specific Aims are to: 1) develop a hierarchical modeling framework to estimate the impact of rolling group admissions on treatment outcomes that incorporates methods initially developed for spatial data analysis;2) extend this modeling framework to test whether client outcomes vary with rolling therapy session features and to identify which session-level features lead to improved client outcomes;3) develop and disseminate analytic tools for study design, sample size determination, and analysis for rolling group studies.
This proposed research project is relevant to public health because its ultimate goal is to improve group-based treatments of alcohol and other drug (AOD) disorders. Motivated by the ubiquity of group therapy for treating AOD disorders, this project develops a statistical approach that can be used to test the effectiveness of AOD group therapies in a variety of realistic settings.
|Paddock, Susan M; Leininger, Thomas J; Hunter, Sarah B (2016) Bayesian restricted spatial regression for examining session features and patient outcomes in open-enrollment group therapy studies. Stat Med 35:97-114|
|Osilla, Karen Chan; Paddock, Susan M; Leininger, Thomas J et al. (2015) A pilot study comparing in-person and web-based motivational interviewing among adults with a first-time DUI offense. Addict Sci Clin Pract 10:18|
|Savitsky, Terrance D; Paddock, Susan M (2014) Bayesian Semi- and Non-parametric Models for Longitudinal Data with Multiple Membership Effects in R. J Stat Softw 57:1-35|
|Paddock, Susan M; Hunter, Sarah B; Leininger, Thomas J (2014) Does group cognitive-behavioral therapy module type moderate depression symptom changes in substance abuse treatment clients? J Subst Abuse Treat 47:78-85|
|Paddock, Susan M; Savitsky, Terrance D (2013) Bayesian Hierarchical Semiparametric Modelling of Longitudinal Post-treatment Outcomes from Open Enrolment Therapy Groups. J R Stat Soc Ser A Stat Soc 176:|
|Savitsky, Terrance D; Paddock, Susan M (2013) Bayesian Non-Parametric Hierarchical Modeling for Multiple Membership Data in Grouped Attendance Interventions. Ann Appl Stat 7:|
|Paddock, Susan M; Savitsky, Terrance D (2013) Discussion of 'Bayesian Nonparametric Inference - Why and How', by Peter MÃ¼ller and Riten Mitra. Bayesian Anal 8:342-345|
|Paddock, Susan M; Hunter, Sarah B; Watkins, Katherine E et al. (2011) ANALYSIS OF ROLLING GROUP THERAPY DATA USING CONDITIONALLY AUTOREGRESSIVE PRIORS. Ann Appl Stat 5:605-627|