Validity of the self-reported number of teeth in independent older people in Japan.

Background: In questionnaire surveys, questions about self-reported number of teeth (SRNT) are often used as a measure oral health. This study investigated the validity of SRNT in older Japanese people.

Methods: In total, 4984 75- and 80-year-old patients who underwent dental examinations were enrolled. A self-administered questionnaire that asked about the number of teeth was used in the analysis. The percentage agreement and kappa value were calculated for the agreement between SRNT and observed numbers of teeth. To identify factors that affect the reliability of SRNT, a logistic regression analysis was performed using correctness of SRNT as the dependent variable.

Results: Among the 3950 participants who responded as to whether they had ≥ 20 teeth, the degree of agreement was 92.9% (kappa value 0.856, p < 0.001) in an objective evaluation. Of the 2621 participants who reported their numbers of teeth, the SRNT and observed number of teeth matched in 57.5% (kappa value 0.559; p < 0.001). Observed number of teeth and annual dental checkup had a significant effect on the accuracy of SRNT. Multivariate logistic regression analysis, with the agreement between SRNT and the observed number of teeth (i.e. whether the number of teeth exceeded 20) as the dependent variable, showed that the observed number of teeth, use of interdental cleaning tools, and annual dental checkup were significantly associated with the agreement between SRNT and the actual number of teeth. In multivariate analysis with tooth number agreement (± 1 tooth) as the dependent variable, the observed number of teeth and use of interdental cleaning tools were significantly associated with the agreement between SRNT and the observed number of teeth.

Conclusion: Although SRNT did not perfectly match the observed numbers of teeth, the results of this study imply that the SRNT of older people is reliable and useful in epidemiological studies.