Ghani A C, Donnelly C A, Garnett G P
Wellcome Trust Centre for the Epidemiology of Infectious Disease, Department of Zoology, University of Oxford, U.K.
Stat Med. 1998 Sep 30;17(18):2079-97. doi: 10.1002/(sici)1097-0258(19980930)17:18<2079::aid-sim902>3.0.co;2-h.
The structures of sexual partner networks are important in determining patterns of transmission of STDs including HIV. Empirical data on sexual partnerships and sexual partner networks collected through sampling individuals are a non-random sample of partnerships and network structures even if individuals are sampled randomly. This has the potential to bias estimates of measures describing the sexual partner network. In addition, biases may be introduced through non-response and missing data. Using Monte Carlo simulation, we investigate the biases that are introduced in estimates measures of the sexual partner network through three common sampling methods. The results indicate that substantial systematic biases are introduced. The direction and magnitude of these biases suggest that, by ignoring them, the risk for the establishment and persistence of infection in a population may be underestimated.
性伴侣网络结构对于确定包括艾滋病毒在内的性传播感染的传播模式至关重要。通过对个体进行抽样收集到的性伴侣关系和性伴侣网络的实证数据，即使个体是随机抽样的，也是伴侣关系和网络结构的非随机样本。这有可能使描述性伴侣网络的测量指标估计产生偏差。此外，无应答和数据缺失也可能引入偏差。我们使用蒙特卡洛模拟，研究了通过三种常见抽样方法在性伴侣网络测量指标估计中引入的偏差。结果表明，会引入大量的系统偏差。这些偏差的方向和大小表明，若忽略它们，可能会低估人群中感染建立和持续存在的风险。
