Bias of vaccination coverage in a household questionnaire survey in Japan

Although a household questionnaire survey is important for estimating vaccination coverage, it raises several problematic issues.

A household survey was conducted on 900 subjects aged 2, 4, and 6 years living in Obu City, Japan, and a second survey for non-respondents to the first survey was then conducted. Questionnaires bearing a subject's name were used for half of the subjects, while the others were anonymous (the named and nameless groups, respectively). The vaccination dates of six kinds of vaccines, including poliovirus and measles vaccine, for those in the named group were reviewed using the administrative records at the Obu City Health Center.

The response rate was 70.1% in the first survey and 84.1% in the first and second surveys combined. The response rate for both groups was nearly equal. Based on administrative records in the named group, the vaccination coverage in the respondents was 0.9-2.9% higher than that in total subjects, and that in the respondents to the first survey was 0.8-4.9% higher. There were very few inconsistencies in the vaccination status between responses to the questionnaire and data of administrative records among respondents in the named group.

These results suggested that vaccination coverage from a household questionnaire survey in Japan might not be extremely biased by either non-responses or incorrect answers.