队列研究资料发病率的贝叶斯估计  

Estimation on the incidence of cohort study by Bayse method

作　　者：张艳芳[1] 魏永越[1] 陈峰[1] 张肖敏[1,2] 

机构地区：[1]南京医科大学公共卫生学院流行病与卫生统计学系,江苏南京210029 [2]江苏省人口和计划生育委员会,江苏南京210008

出　　处：《中华疾病控制杂志》2012年第8期721-723,共3页Chinese Journal of Disease Control & Prevention

基　　金：江苏省科技支撑计划(社会发展项目)(BS2007080)

摘　　要：目的以某一模拟队列资料的宫颈上皮内瘤变Ⅱ级(cervical intraepithelial neoplasia,CINⅡ°)及以上(CINⅢ°、宫颈癌)的发病率估计为例,探讨Bayse法在队列研究资料发病率估计中的应用。方法假设某地区进行宫颈癌基线筛查8万人,对其中5万基线筛查阴性者进行隔年随访,总计随访10万人年。其中1 000人细胞学阳性,需要做病理学检查,最终收到900份病理报告,检出CINⅡ°及以上者120例,通过Bayse法估计该地区CINⅡ°及以上的发病情况。结果通过Bayse法考虑病理检查随访人群中的失访病例以及细胞学检查阴性人群中的漏诊病例,估计该队列CINⅡ°及以上的发病率为180.23/10万,95%CI为(146.59/10万～226.06/10万)。结论基于Bayse法,考虑了病理检查随访人群中的失访病例以及细胞学检查阴性人群中的漏诊病例,估计得到的该地区CINⅡ°及以上的发病率更合理、更准确。Objective Take estimating the incidence of the cervical intraepithelial neoplasca （ CIN Ⅱ° ） and above （CIN Ⅲ °, cervical cancer） of one simulated cohort for example, to explore the application of the Bayse method in the estimation of the incidence. Methods Suppose that cervical cancer screening was proceed to 80 000 women in a region, 50 000 of the normal women were followed-up every other year, accumulated 100 000 person-years, 1 000 women were required for pathological examination, and 900 copies of the pathology results were received finally. 120 cases of CIN H o and above were detected. Bayesian method was used to estimate the incidence of CIN Ⅱ° and above in the region. Results The cases of lost to follow up for following-up crowd diagnosised by pathology and the missed cases for cytology-negative population was considerated by bayesian, the incidence of the cohort cervical cancer at CIN Ⅱ° or above was 180.23/ 100 000, 95% confidence interval（CI） was （ 146.59/100 000-226.06/100 000）. Conclusions Based on the bayesian method, considering the pathologic examination of the people lost to cases and cytology negative diagnosis in the crowd, the incidence of CIN Ⅱ° and above will be more reasonable and accurate.

关 键 词：队列研究 发病率 贝叶斯定理 

分 类 号：R195.1[医药卫生—卫生统计学]