艾滋病高危人群基数调查估计的统计方法及应用

Currently in conducting population-based surveys on HIV high risk groups at home and abroad，several major problems are inappropriate methods, low confidence level data and inaccurate results which have seriously influenced the prevention and control of HIV and AIDS. According to the literatures in research works and China's status quo, 120 types of survey methodologies are selected, which are comprised of varied permutations and combinations of three RRT models and ten sampling methods and two sample sizes and two CMR schemes. For each kind of survey methodology, which is employing the latest mathematical statistic theories and methods and which is based on long-term, systematic study and research achievements in three national projects obtained by our team in recent years, relevant formulae for estimators, variance, interval estimate, optimum sample size of population-based surveys on HIV high risk groups will be deduced; corresponding software that can perform formulae calculation and object identification will be designed. Expert view of all survey methodologies, statistical formulae and software will be needed. Monte Carlo simulations on large sampling will be employed to examine the reliability and validity. In these ways, the size of the population can be accurately estimated in China's major cities, such as prostitutes, whoremasters, MSM and drug addicts. This would be expected to judge a trend in the AIDS epidemic properly, but also to implement and evaluate the AIDS community intervention effectively and correctly. This would be expected to not only allow local policy makers to better formulate public health policy and guide efficient allocation of resources, but also provide the scientific basis for effective prevention and control of HIV/AIDS among high risk groups. At the same time, our research results will fill the research blank for the sampling survey methods.

当前国内外对艾滋病高危人群基数的调查估计，方法不科学、数据不可信、结果不准确,严重影响艾滋病的预防控制！本项目根据大量文献研究与我国实际，优选3种RRT模型10种抽样方法2种样本数2种CMR方案组合成120种调查方法。对每种调查方法，在申请人近年主持的3项国家级抽样研究成果基础上，采用当今数理统计最新理论方法，拟推导出艾滋病高危人群基数的估计量及其方差、区间估计、最优样本大小的统计公式；拟编制对象识别、公式计算软件；对每种调查方法及其估计公式、统计软件进行专家评审，蒙特卡洛仿真模拟大样本抽样以评价信度与效度；准确可靠地调查估计中国大城市妓女、嫖客、MSM、吸毒人群的基数。对正确判断艾滋病流行趋势，合理分配艾滋病防治资源，有效实施艾滋病社区干预，正确评价艾滋病干预效果，为有关部门科学制定艾滋病预防控制政策、法规、规划及有关技术策略与措施具有重要意义。研究结果也将填补国内外抽样研究方法的空白。

本项目根据大量文献研究与我国实际，优选3种适合于敏感问题对象总人数调查、实施较方便、计算较简单的RRT模型（二分类敏感问题Simmons模型、二分类敏感问题改进的RRT模型、多分类敏感问题RRT模型）；挑选出10种常用的或适用于大范围调查的抽样方法（简单随机抽样、整群抽样、二阶段抽样、二阶段整群抽样、三阶段抽样、分层简单随机抽样、分层整群抽样、分层二阶段抽样、分层二阶段整群抽样、分层三阶段抽样）；根据艾滋病高危人群资料的来源种类，确定2种样本个数（2来源样本、多来源样本）；精选出2种适合中国国情与实际，信度、效度较高,可作统计推断的调查方案（抽样-标记-再抽样（CMR）调查方案、抽样-替代标记-再抽样方案）。3种RRT模型、10种抽样方法、2种样本数、2种调查方案共组合成120种调查方法。对每种调查方法，在申请人近年主持的3项国家级抽样研究成果基础上，采用当今数理统计最新理论方法，已推导出艾滋病高危人群基数的估计量及其方差、区间估计、最优样本大小的统计公式；编制出对象识别、公式计算软件；对每种调查方法及其统计公式、统计软件进行了专家评审及蒙特卡洛仿真模拟大样本抽样，获得了高信度与高效度的评价结果。采用本项目给出的敏感问题分层多阶段抽样调查方法，从中国12个（超）大城市中随机抽取4个城市，按本项目研究的调查方法、统计公式与统计软件，准确可靠地调查估计出这4个城市妓女、嫖客、MSM、吸毒人群的基数，进一步估计出中国（特）大城市妓女、嫖客、MSM、吸毒人群的基数。. 已发表标注本项目编号的论文34篇，其中：SCI（E）源期刊18篇，EI源期刊1篇，EI收录国际会议论文3篇，中文核心期刊12 篇。. 对正确判断艾滋病流行趋势，合理分配艾滋病防治资源，有效实施艾滋病社区干预，正确评价艾滋病干预效果，为有关部门科学制定艾滋病预防控制政策、法规、规划及有关技术策略与措施具有重要意义。研究结果也将填补国内外抽样研究方法的空白。