Skip to main content Skip to navigation

Advancing Longitudinal Statistical Methods for Substance Use Disorder Clinical Trials

Washington State University Spokane Faculty Seed Grant
7/11/2012 – 7/10/2013

Abstract

Data collected through resource-intensive, time-consuming, substance use disorder (SUD) clinical trials is perhaps the most important modality for producing externally valid treatments to the larger research/clinician community in order to improve public health. However, the capability of these data to produce an ongoing stream of evidence that could improve public health and inform future clinical practice is not always fully realized. Often, after the primary analyses have been conducted and published, secondary analyses are conducted on a limited scale. Moreover, such secondary analyses rarely utilize the potential of modern longitudinal analyses. There are a variety of reasons for this, but one of the most likely reasons for this lack of full utilization is that research teams have not had the opportunity or access to advanced methods of longitudinal statistical analyses. Thus, it remains largely unknown whether advanced longitudinal statistical methods of analysis can offer significant benefit to the evaluation of treatment effectiveness for many SUD clinical trials.

Having access to cutting edge analytic techniques along with real data examples could significantly increase WSU’s capability to push the boundaries for various types of research questions we would be capable of answering. As a result, design of future treatment protocols will greatly benefit, and so will public health. Multilevel modeling (MLM) and latent growth curve modeling (LGCM) within the generalized latent variable (GLVM) framework offer multiple advantages that go beyond common, quasi-longitudinal models (e.g., survival modeling) or endpoint analyses (e.g., chi-square).

Our long-term goal for this project is to disseminate and advance modern, longitudinal statistical methods in the area of SUD treatment research and beyond to wherever clinical trial data are collected. While such methods are commonly used in the broader substance abuse literature, these methods have not been put to work on clinical trial data. The overall objective of this project is to demonstrate the potential of modern, longitudinal latent variable methodologies through examples that will also serve to investigate scientific questions of interest, yet previously unexplored.

Primary Investigator

Sterling McPherson
Sterling McPherson, PhD

Co-Investigators