Register-based study among employees showed small nonparticipation bias in health surveys and check-ups
Register-based study among employees showed small nonparticipation bias in health surveys and check-ups
复制标题
DOI：
10.1016/j.jclinepi.2007.09.010
复制
发表时间：
2008-09-01
影响因子：
7.2
通讯作者：
Lahelma, Eero
中科院分区：
文献类型：
--
作者：
Laaksonen, Mikko;Aittomaki, Akseli;Lahelma, Eero
目的：通过社会人口学变量和健康状况(以医学确认的疾病缺勤来衡量)来检验不参与问卷调查和职业健康检查，以及其他研究变量和参与之间的关联是否受健康状况的影响。研究设计和地点：对赫尔辛基市的员工进行问卷调查和健康检查。样本信息来自用人单位的人事登记，并通过参与和同意将数据链接到外部行政登记进行分析。结果：参加问卷调查的人群在年龄较大、职业级别较高、收入较高的人群中更常见。永久就业，以及因医学确认的疾病而未缺勤的人。尤其是在女性中，差异很小。给予同意通常遵循与调查回应相似的模式。不参加体检与经济收入低、签订临时雇佣合同有关。在问卷调查和健康体检中，其他研究变量与参与度之间的关系不受健康状况的影响。结论：问卷调查和健康体检具有广泛的目标人群代表性。其他研究变量和参与度之间的关联并不因健康状况而异。这表明，即使数据不是完全具有代表性，研究变量之间的关联也不需要有偏见。(C)2008 Elsevier Inc.保留所有权利。
Objectives: To examine nonparticipation to a questionnaire survey and occupational health check-ups by sociodemographic variables and health status, measured by medically confirmed sickness absence, and whether the associations between other study variables and participation were affected by health status.Study Design and Setting: Questionnaire surveys and health check-LIPS were conducted among the City of Helsinki employees. Sample information was derived from the employer's personnel register and analyzed by participation and giving consent to link the data to external administrative registers.Results: Participation to the questionnaire survey was more common among the older, higher occupational classes, those with higher income. permanent employment, and those with no absence due to medically confirmed sickness. Among women in particular, the differences were small. Consent giving followed generally similar patterns than survey response. Nonparticipation to health check-LIPS was related to low income and temporary employment contract. In both questionnaire survey and health check-ups, associations between other study variables and participation were not affected by health status.Conclusions: Questionnaire surveys and health check-ups were broadly representative of the target population. Associations between other study variables and participation did not differ by health status. This sugggests that even when the data are not fully representative associations between the study variables need not to be biased. (C) 2008 Elsevier Inc. All rights reserved.