The Study of Sampling Bias in an Oral Health Survey of Elderly : Comparison of Oral and General Health Condition between Respondents and Non-Respondents to a Questionnaire and between Participants and Non-Participants in an Examination

Abstract

The purpose of this study was to evaluate sampling bias caused by present teeth or general health condition in an oral health survey of elderly subjects. A mailed questionnaire survey was carried out for residents of Niigata City who were 70 or 80 years old, investigating the number of present teeth, general health condition and intention to participate in an oral examination. Follow-up mailings were sent once to non-respondents. The response rate was increased from 60.4% to 79.5% by the follow-up mailing. A telephone survey was carried out for subjects who didn't answer the questionnaire. The invitation to an examination was targeted toward subjects who answered that they would participate in an examination in the questionnaire, and 763 elderly participated in the examination. We compared the number of present teeth, general health status and other items whether the subjects answered the questionnaire or not and whether subjects participated in an examination or not. We also estimated the number of present teeth in the population. Results showed that the number of present teeth of participants in the examinations was higher than the estimated value in the population (difference: 2.6-3.9 in 70-year-old, 0.8-1.8 in 80-year-old). The general health condition of early responders to the questionnaire was better than that of the later responders, but the degree of difference was smaller than that of the number of present teeth. Therefore, bias caused by the number of present teeth was larger than that caused by general health condition under which the dental examination was carried out for subjects, according to their wishes to participate in an examination.