中文摘要：

目的 通过构建传染病动力学模型,估计1995-2014年我国吸毒人群中人类免疫缺陷病毒（human immunodeficiency virus,HIV）的年发病率。方法 构建HIV在我国吸毒人群中的传播动力学模型,通过查阅文献收集模型运行所需输入参数。基于国家监测的吸毒者HIV患病率数据进行模型的训练与测试后,利用蒙特卡罗模拟实现贝叶斯合并方法（bayesian melding approach）对发病率的估计。结果 包含易感者、HIV感染者和艾滋病患者三个状态的模型经训练后对HIV患病率的平均相对预测误差为4.37%,模型输出患病率与实际患病率拟合优度较高（R^2=0.89,P〈0.001）。模型估计的HIV年发病率在1996年达到峰值（4.06%）,其后逐渐下降,2000年低至1.25%,2002年发病率回升至1.95%,随后发病率逐年降低,2008年之后,发病率稳定在0.50%-0.90%之间。结论应用传染病动力学模型较好地模拟了我国吸毒人群HIV患病率,进而对发病率做出了准确估计。从降低发病率来看,我国采取的HIV综合防控措施对控制HIV在吸毒人群中的传播取得了一定效果。

英文摘要：

Objective To estimate annual human immunodeficiency virus （HIV） incidence rate among drug users in China from 1995 to 2014 based on a transmission dynamic model. Methods An HIV transmission dynamic model was developed to describe the HIV epidemic among drug users in China. The model was parameterized using data from literature available. The incidence rate was estimated through Bayesian melding approach achieved by Monte Carlo simulation after training and testing the model based on HIV prevalence data from national sentinel sites. Results The trained model con- sisting of three compartments （ susceptible, HIV infected and AIDS patients） predicted HIV prevalence precisely with small average relative error（4. 37% ）. The model-predicted HIV prevalence fitted well to HIV prevalence data （R^2 =0. 89, P 〈 0. 001 ）. HIV incidence rate predicted by the model peaked in 1996（4. 06% ） , then decreased to 1.25% in 2000. It rose again to 1.95% in 2002 and continued to decrease until around 2008 to 2014 when it leveled off between 0. 50% and 0. 90%. Conclusions The model was able to attain a good fit to HIV prevalence data, allowing for reasonably precise es- timate of the incidence rate, which showed that HIV comprehensive prevention and control measures in China had positive impact on reducing HIV incidence rate among drug users.