Periodontal disease estimates in NHANES III: clinical measurement and complex sample design issues

This paper evaluates the possibility that examiner bias or other factors contributed to an observed decline in pocket depth and gingivitis between the two three-year sequential periods of time (or phases) covered by the Third National Health and Nutrition Examination Survey (NHANES III).

Prevalences of periodontal conditions were analyzed using data from two sets of repeat oral health examinations by examining dentists of NHANES III sample persons. The first set includes sample persons who were examined twice by the same examining dentist at an interval of one to six weeks. The second set includes sample persons who were assessed on the same day by both an examining dentist and a reference dentist. Other possible sources of error also were evaluated.

Overall kappa statistics measuring agreement between or within dental examiners were within the range observed for other periodontal disease surveys. While differences were found among dentists in the prevalence of pocket depth of 4 mm or more, for each group of sample persons assessed by a reference examiner-examining dentist pair, the reference examiner's periodontal measurements closely corresponded to measurements made by the examining dentists.

Differences between dental examiners in prevalences of periodontal conditions may be due in part to the fact that examinees were not randomly assigned to examiners. As a result, the sample persons examined by each dentist may not have been alike in characteristics thought to affect periodontal disease status. These findings suggest that the observed declines in periodontal health status between phases is not due to examiner bias. This unexplained decline may be the result of sampling variation. It is recommended that combined six-year survey results be presented whenever possible.