Recorded presentation

Carolyn Lichtenstein, PhD¹, Kristin Zempolich, MS¹, L. Joseph Sonnefeld, MA², Gregory B. Teague, PhD³, and Steven M Banks, PhD⁴. (1) Northrop Grumman Information Technology, 1700 Research Blvd., Suite 400, Rockville, MD 20850-3142, (301) 294-5471, carolyn.lichtenstein@ngc.com, (2) Westat, Inc., 1650 Research Blvd., Rockville, MD 20850, (3) Department of Mental Health Health Law & Policy, Louis de la Parte Florida Mental Health Institute, University of South Florida, 13301 Bruce B. Downs Boulevard, Tampa, FL 33612, (4) Center for Mental Health Services Research, University of Massachusetts Medical School, Department of Psychiatry, 55 Lake Ave. North, Worcester, MA 01655

This presentation describes the primary cross-site findings of the COSP-MRI study, which involved a randomized design and a common data collection protocol across all sites. Both intent-to-treat (ITT) and as-treated analyses of the short-term outcome wellbeing, constructed as a composite of several scales, were implemented. The ITT analysis tested the primary hypothesis about the randomized design, using multi-level modeling to account for the clustering of participants within sites and the repeated measurements for each participant. Analysis of the baseline, 4-month and 8-month data for all participants showed an interaction between time and experimental group that approached statistical significance at the conventional p=.05 level (p = 0.0589). Additional ITT analyses, including additional multilevel modeling, effect size computation, and examination of attrition patterns, further investigated the differences in outcomes of the 2 experimental groups of participants, focusing on variation in the outcomes across sites. Preliminary findings of these analyses indicated that the experimental group*time interaction was particularly strong among the sites with a drop-in center intervention. The wellbeing measure was refined on the basis of these additional analyses. The as-treated analyses were two-pronged: (1) assigning participants to comparison groups based on their originally assigned experimental group and their actual attendance at one or more COS sessions, and (2) estimating each participant’s amount of exposure to the COS intervention. Each examined the relationship of the participation measure to the change over time in wellbeing. As with the ITT analyses, analytic methods included multilevel modeling, effect size computation, and examination of attrition patterns.

Learning Objectives: At the conclusion of the session, participants will be able to

(
• describe the intent-to-treat cross-site findings regarding the outcome wellbeing for the COSP multisite study, and (
• describe several as-treated cross-site findings regarding the same outcome.

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.

Recorded presentation