Configural frequency analysis for exploring bacterial sets in periodontal health and disease

Periodontal research has been concerned with identifying bacterial sets in oral health and disease. Configural frequency analysis (CFA) is a taxonomical method used to test the statistical significance of patterns of elementary cell frequencies in cross-tabulated data, and may therefore be used to evaluate interactions among occurrences and non-occurrences of periodontal microorganisms. CFA patterns may exist as configural types or anti-types depending on whether observed cell frequencies are greater or less than expected frequencies, respectively, based on table marginals. CFA was applied to two published data sets. In the first, the occurrence of Bacteroides (Porphyromonas) gingivalis (Bg) and Treponema denticola (Td) was studied in military subjects. Analysis of individual cells of the resultant two-by-two contingency table revealed the existence of a single anti-type (Td in the absence of Bg) suggesting a sequential phenomenon. The second, more complex, data set was based on reported occurrence of Actinobacillus actinomycetemcomitans (Aa), Bg, and Bacteroides (Prevotella) intermedius (Bi) in sites with and without progressive periodontitis. CFA based on all data identified three statistically significant configural types and two anti-types. However, a somewhat different configurational picture emerged when disease progression status was considered a stratification variable for separate analysis. All these analyses, which were undertaken for expository purposes, assumed independence of observations. Methodological procedures are suggested for CFA when observations are not independent.