Effect of sampling strategy on the false-negative rate for detection of selected subgingival species.

Subgingival plaque samples were obtained from the mesial surface of each tooth (maximum 28 samples per subject) in 62 subjects with prior evidence of destructive periodontal disease. The resulting 1596 samples were evaluated for their content of 14 selected taxa using a colony lift method and DNA probes. The present investigation compared the ability of 6 sampling strategies to detect a species known to be present in a subject as determined by the 28-site sampling procedure. On average, a species was not detected in 68% of the positive subjects if only the upper right first molar was sampled; 55% of subjects if both upper first molars were sampled; 36% of subjects if the 4 first molars were sampled; 28% of subjects if the 6 Ramfjord teeth were sampled; 60% of subjects if the deepest pocket was sampled and 25% of subjects if the 4 deepest pockets were sampled. The error rate was greatest for species that were infrequently detected in plaque samples such as Actinobacillus actinomycetemcomitans serotype b. This species was not detected in 49% of positive subjects, when the 6 Ramfjord teeth were sampled and 38% of subjects when the 4 deepest pockets were sampled. The data indicated that multiple plaque samples are needed to minimize false-negative rates.