Nakai S, Hashimoto S, Murakami Y, Hayashi M, Manabe K, Noda H
Department of Epidemiology and Biostatistics, School of Health Sciences and Nursing, University of Tokyo.
Nihon Koshu Eisei Zasshi. 1997 Mar;44(3):184-91.
A health-related mailed survey was conducted to investigate the effect of follow-up mailings to response rates. In addition, the answer distributions among early and late respondents were compared to check non-response bias. Approximately 3,000 persons aged 40-64 were randomized into two groups; a questionnaire with four pages (twenty-two questions) was assigned to the first group, and a questionnaire with eight pages (thirty-five questions) to the second group. Both questionnaires contained questions of current health status, health-related practice, smoking status, etc. Follow-up mailings were sent twice to non-respondents. Response rates were increased from 38% to 62% by the first follow-up mail, and to 71% by the second follow-up mail. Although the length of the questionnaire did not affect response rates, response rates among the older subjects was higher than the younger subjects. Positive response to current smoking status was 10% lower among early respondents than late respondents, collected after the second follow-up mailing, and response to regular participation in physical/medical checkup was 15% higher among early respondents, whereas there were few differences for answers to other questions. Odds ratios between current health status and several health-related questions may not be biased by late response. However, increased response rates are needed to prevent non-response bias, because there were some differences in responses from follow-up mailings.
开展了一项与健康相关的邮寄调查，以研究后续邮寄对回复率的影响。此外，还比较了早期和晚期回复者的答案分布情况，以检查无回复偏差。约3000名40至64岁的人被随机分为两组；第一组发放一份四页（22个问题）的问卷，第二组发放一份八页（35个问题）的问卷。两份问卷都包含当前健康状况、与健康相关的行为、吸烟状况等问题。对未回复者进行了两次后续邮寄。第一次后续邮寄使回复率从38%提高到62%，第二次后续邮寄使其提高到71%。虽然问卷长度不影响回复率，但老年受试者的回复率高于年轻受试者。在第二次后续邮寄后收集的早期回复者中，对当前吸烟状况的肯定回答比晚期回复者低10%，而早期回复者中定期参加身体/体检的回答比晚期回复者高15%，而对其他问题的回答差异不大。当前健康状况与几个与健康相关问题之间的优势比可能不会因晚期回复而产生偏差。然而，由于后续邮寄的回复存在一些差异，需要提高回复率以防止无回复偏差。
