| 王杨,胡泊,陈涛,李卫.抽样调查法和单组目标值法对诊断试验样本量计算差异的分析[J].Chinese journal of Epidemiology,2010,31(12):1403-1405 |
| 抽样调查法和单组目标值法对诊断试验样本量计算差异的分析 |
| Comparison of calculation methods for diagnostic trials under different sample size |
| Received:June 10, 2010 |
| DOI：10.3760/cma.j.issn.0254-6450.2010.12.018 |
| KeyWord: 诊断试验 样本量计算 Monte Carlo随机模拟 |
| English Key Word: Diagnostic trial Sample size calculation Monte Carlo random simulation |
| FundProject: |
| Author Name | Affiliation | E-mail | | Wang Yang | Department of Biometrics,National Cerare for Cardiovascular Disease,Cardiovascular Institute＆Fu Wai Hospital,CAPM&PUMC,Beijing 100037,China | | | Hu Bo | Department of Biometrics,National Cerare for Cardiovascular Disease,Cardiovascular Institute＆Fu Wai Hospital,CAPM&PUMC,Beijing 100037,China | | | Chen Tao | Department of Biometrics,National Cerare for Cardiovascular Disease,Cardiovascular Institute＆Fu Wai Hospital,CAPM&PUMC,Beijing 100037,China | | | Li Wei | Department of Biometrics,National Cerare for Cardiovascular Disease,Cardiovascular Institute＆Fu Wai Hospital,CAPM&PUMC,Beijing 100037,China | liwei0325@yahoo.com.cn |
|
| Hits: 4053 |
| Download times: 2873 |
| Abstract: |
| 探讨设计以验证灵敏度和特异度为目的的诊断试验时,不同样本量计算方法间的区别.通过直观的样本量公式与计算结果比较,分析不同样本量计算方法间的差异,进一步通过随机模拟方法,验证所得结果的正确性.抽样调查法计算所需的样本量明显低于单组目标值法,随机模拟显示,在相同的参数设置下,单组目标值法给出的样本量能够提供更高的研究把握度.两种样本量设计方法的适用条件不同,存在本质区别,研究者必须根据研究目的设置相应参数.如果在以检验某种诊断方法的诊断能力是否不低于某个临床认可的标准时,按照单组目标值法设计的样本量,才能提供足够的检验把握度,证明新诊断方法的有效性. |
| English Abstract: |
| To discuss the calculation methods under different sample size，used for diagnostic trials．The purpose of the diagnostic trial was to demonstrate the sensitivity and specificity of the new method．Equations and results were directly compared．Monte Cado random simulation was used to validate the results．Sample size obtained from the sampling method was always smaller than from the target value method．Results from simulation showed that the target value method could offer more and larger power．The two sample size determination method showed essential differences of the results，suggesting that the investigator should choose appropriate method in accordance with the study design．If the hypothesis of study was to demonstrate the new diagnostic method which could meet the clinical requirements，only if the target value method provides enough statistical power． |
|
View Fulltext
Html FullText
View/Add Comment Download reader |
| Close |
|
|
|