Back to Annual Meeting Page		American Public Health Association 133rd Annual Meeting & Exposition December 10-14, 2005 Philadelphia, PA

Abstract #119468

William K. Pan, DrPH, MS, MPH, Department of International Health, The Johns Hopkins University Bloomberg School of Public Health, 615 N Wolfe Street, Rm W5508, Baltimore, MD 21224, 410-502-2141, wpan@jhsph.edu

The objective of this paper will be to examine the effect of spatial dependency of clusters in the design and analysis of Community-randomized trials (CRT). CRT's typically involve clusters of patients that exhibit moderate to high intra-class correlation due to similarities of people within a defined geographic region, but an added problem in some CRT settings is when clusters exhibit spatial autocorrelation due to spatial spillover of the characteristics of variables of interest into neighboring clusters. Examples can be found in CRT's examining treatments for infectious disease (e.g., sexually transmitted infections, smoking cessation, tuberculosis, malaria, etc.), crime, and nutritional supplementation, among others. Van Es et.al. (1993) demonstrated in agricultural field trials that randomization does not ensure spatial balance, such that short-distance treatment contrasts may increase Type II error and long-distance treatment contrasts may increase Type I error. Whether similar problems exist in clinical trial settings has not been examined. Therefore, the primary goals for this paper are: (i) to review findings in agricultural trials related to spatial variation and compare these results to a variety of clinical trial settings; and (ii) to perform simulations that will assess inference from CRT's that exhibit spatially correlated clusters, focusing on the influence of the randomization process (including constrained randomization) and analytical techniques that might reduce spatial dependency.

Learning Objectives: Findings from this paper will help the participant

• Assess issues associated with spatial autocorrelation in the design and analysis of experiments
• Identify ways to reduce spatial autocorrelation prior to the design of a community-randomized trial
• Discuss techniques used to analyze spatial data

Keywords: Community-Based Public Health, Biostatistics

Presenting author's disclosure statement:

I wish to disclose that I have NO financial interests or other relationship with the manufactures of commercial products, suppliers of commercial services or commercial supporters.

Recorded presentation