Online Program

326346
Evaluating Community-Based Translational Interventions Using Historical Controls: Propensity Score vs. Disease Risk Score Approach


Wednesday, November 4, 2015 : 12:50 p.m. - 1:10 p.m.

Luohua Jiang, PhD, Epidemiology, University of California Irvine, Irvine, CA
Ryan Hollingsworth, MSPH, Department of Statistics, Texas A&M University, College Station, TX
Shuai Chen, MS, Department of Statistics, Texas A&M University, College Station, TX
Janette Beals, PhD, Centers for American Indian and Alaska Native Health, University of Colorado Anschutz Medical Campus, Aurora, CO
Spero Manson, PhD, Centers for American Indian and Alaska Native Health, University of Colorado, Anschutz Medical Campus, Aurora, CO
Yvette Roubideaux, Office of the Director, Indian Health Service, Rockville, MD
Due to ethical and other considerations, many community-based translational projects of evidence-based interventions were designed as one-arm studies, using pretest-posttest design without a control group. For some translational initiatives, it is possible to evaluate the translational effectiveness of the intervention using historical control data from publicly available data repositories, such as the clinical trials data stored in NIDDK Central Repository. Inference based on historical controls could be subject to potential selection bias though. We propose to use propensity scores and disease risk scores to adjust for such bias. Both propensity scores and disease risk scores can be used as data dimension reduction tools to summarize the information for a large number of covariates. Propensity score matching has also been widely used to evaluate treatment effects for quasi-experimental or observational data. Yet, little is known about the performance of matching based on disease risk scores in those kinds of situations. We compared the performance of these two methods under different scenarios using simulations, and applied it to the data from the Special Diabetes Program for Indians Diabetes Prevention (SDPI-DP) demonstration project, a translational lifestyle intervention among American Indian and Alaska Native communities.

Learning Areas:

Biostatistics, economics
Chronic disease management and prevention
Conduct evaluation related to programs, research, and other areas of practice
Epidemiology
Social and behavioral sciences

Learning Objectives:
Demonstrate the possibility of evaluating the effectiveness of one-arm translational interventions using historical control data from publicly available data repositories. Compare the performance of propensity score matching and disease risk score matching methods under different scenarios using simulations. Discuss the advantages and disadvantages of propensity score matching and disease risk score matching methods in the context of a real data application.

Keyword(s): Biostatistics, Evaluation

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: I have been the principal statistician of multiple federally funded disease prevention and treatment studies, analyzed or led the analyses of cross-sectional, longitudinal, and multilevel data. My research interests have been statistical method development and applications of longitudinal and multilevel data models as well as latent variable modeling, with a strong focus on disease prevention and health disparities research.
Any relevant financial relationships? No

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.