This FIRST application is a resubmission of a similar application that was reviewed in (Subcommittee 1). This revised application addresses a statistical problem common in periodontal research: the analysis of data when the observations fall into clusters. Typically, dental investigators record data on multiple sites in the mouth of a single subject. This statistical design incurs correlation between sites rendering standard analytic techniques invalid. A major objective of this application is to develop simple statistical techniques for assessing the association between a site-specific outcome and a covariable of interest, while controlling for potential confounders, when either the covariable of interest or the secondary covariable, or both, vary by site as well as by subject.
Specific aims of this project are: 1) To study the properties of new methods for analyzing several four-fold tables that can handle clustered correlated data when the covariates are site- specific in nature. Methods include an extended Mantel-Haenszel test, an estimator of the standard error of the common odds ratio and a test of homogeneity across strata. 2) To develop modified analysis of variance-type statistics for clustered correlated continuous outcomes when the covariables vary by site. 3) To propose and compare several procedures for evaluating pre- and post-treatment binary responses in the setting of a longitudinal clinical trial in periodontal research. 4) To evaluate the consequences of errors in diagnosing active periodontal disease in assessing the therapeutic effects of new agents in periodontitis. 5) To conduct extensive small sample simulations in order to investigate the properties of the proposed methods in aims 1, 2, and 3. Particular attention will be paid to situations in which the number of sites is high relative to the number of subjects, as is the case in many dental research studies. 6) To compose a user-friendly computer program written in a commonly used software package (SAS) that can tabulate the raw data and compute the statistics proposed in specific aims 1, 2, and 3.
| Panageas, K S; Begg, M D; Grbic, J T et al. (2003) Analysis of multiple 2x2 tables with site-specific periodontal data. J Dent Res 82:514-7 |
| Begg, M D (1999) Analyzing k (2 x 2) tables under cluster sampling. Biometrics 55:302-7 |
| Begg, M D; Panageas, K S (1999) Interval estimation of the common odds ratio from k(2 x 2) tables under cluster sampling. Stat Med 18:1087-100 |
