中文摘要：

目的研究无金标准（无法获得患病人数）条件下患病率与阳性检出率、灵敏度、特异度的关系。方法根据患病率和阳性检出率、试验灵敏度、特异度间的定量关系，采用计算机模拟和实际调查数据研究不同灵敏度、特异度、患病率下，患病率和阳性检出率的关系。结果阳性检出率πT与患病率πD间存在一个界值点π0,当πT〈π0时，πT将高估πD，表现为正偏倚；反之，πT将低估πD，表现为负偏倚；当πT=π0时，丌丁将正确估计πD。负偏倚主要受灵敏度影响，正偏倚主要受特异度影响。当阳性检出率接近试验假阳性率（1-Sp）时，可产生严重正偏倚，医学实例表现规律与模拟研究相一致。结论在低患病率时，即便使用高灵敏度、高特异度的筛检试验，阳性检出率也可显著高估实际患病率。

英文摘要：

Objective To estimate prevalence in the absence of a gold standard, study the relationship between prevalence and positive detective rate, sensitivity, specificity. Methods The relationship was studied by a medical example and simulating different sensitivities and specificities based on the formulation of positive detective rate and prevalence. Results There isa cutoff point（π0） to estimate prevalence（πD） by positive detective rate（πT）. When πT〈 π0, πT will overestimate πD （ positive bias）, when πT 〉 π0, πwill underestimate πD negative bias）, when πT =π0, πT will estimate πD correctly. When positive detective rate is near to false positive rate （ 1 - Sp ）, serious positive bias occurs. Conclusion Positive detective rate overestimates prevalence significantly in the population of low prevalence even though using screening tests with excellent sensitivity and specificity.