大规模流行病学现场调查和编码的质量控制

目的 以目前开展的一项前瞻性研究中随访调查为例，对现场调查和编码质量控制方法进行评价，为大规模流行病学调查研究的质量控制提供参考。方法 上海市肿瘤研究所在1996～2002年完成了“上海市女性健康队列”调查及第1次随访。从现场准备工作、人员培训和质量控制几个方面采取了一系列措施来保证质量。以差错率为指标，对调查员和编码员每月的工作进行考核，并采用线性回归评价总体质量变化趋势。结果 调查初期及新调查员的差错率较高，质量相对较差。当去除新调查员第1个月差错率时，平均差错率随着调查的进行呈下降趋势。新调查员经过6个月的实践，差错率明显下降，基本达到成熟调查员的水平。通过随机抽查和计算机输入检查编码质量，两者反映出的编码差错均随着编码工作的进行呈明显下降趋势。结论 影响调查和编码质量的因素是多方面的，需要通过实践不断总结，制定合适的管理方法，其中全面的培训、合理的质控方法及定期考核制度对提高调查和编码质量是有效可行的。

Quality control of survey and coding in large-scale epidemiological study

Objective To evaluate the quality control methods of survey and coding in an ongoing large-scale cohort study, and introduce our experience in the epidemiological studies. Methods The Shanghai Women's Health Study was established during 1996 to 2000 in urban Shanghai and the first round of follow-up was finished in 2002. A series of approaches for quality control were carried out in the preparation, staff's training and interview methods. Error rate (ER) was used to assess the work of interviewers and coders. The outcomes were evaluated by means of linear regression. Results The interviewers' ER was higher in the early time of the investigation and the quality was poorer, especially for new interviewers. Excluding the influence of new interviewers' ERs in first working month, the ER tended to go down over time. The new interviewers' quality can reach to the average level by 6 months practice. Quality of coding was examined by computer input and checking randomly. The errors are decreased gradually when the coders were familiar with all items of the questionnaire. Conclusion There are many factors to affect the quality of survey and coding, so it needs to establish relevant rules and reinforce supervision especially in the early stage of investigation. All of the results showed that comprehensive training, suitable methods of quality control and periodical assessment are effective and feasible for enhancing the quality of survey and coding.