发帖
雷达卡
最近投了一篇文章，文章内容是写A指标可以作为一个biomarker预测临床结局B。文章中纳入一个患者队列检测这个指标，然后根据临床结局B分组发现这个指标有差异。基于此，做了一个ROC曲线评估A指标对于临床结局B的预测价值，最后得出结论。文章现在大修，3个审稿人的意见中有审稿人提到的患者样本量的问题，审稿人意见“Howdo you justify the small sample size? Please provide a power calculation.”，我学习了一下，写了如下的response，请各位看看能不能符合要求啊？
repose:We calculated the required sample size for the comparison of the area under a ROC curve with a null hypothesis value using Medcalc software (version 12.0.0.0; MedCalcSoftware, Mariakerke, Belgium). Since there is no study reporting on the role of A on predicting B, we first of all did a preliminary study at first to estimate the possible AUC of A. We retrospectively enrolled patients who admitted to our hospital with the diagnosis of B. A had been measured in some of them. According to our data,we found that A had a potential possibility to be a marker in B and the AUC is 0.863. There is no doubt that the AUC in preliminary study is not accurate because our preliminary study is a retrospective study and not all patients had A value. Taking all of this into account, we use AUC of 0.8 for RP% to calculate the sample size. For α-level we select 0.05 and for β-level we select 0.20 (power is 80%). The estimated total patient number is 26.If we select 0.10 for β-level (power is 90%), the estimated total patient numberis 34. If we use AUC of 0.75 for RP%,α-level of 0.05 and β-level of 0.20 (poweris 80%) to calculate the sample size, The estimated total patient number is 50.After consulting with our statistician (mentioned in the acknowledgement section), we decided to set our patient size of 60 in this study.
顺便还有一个问题，有个审稿人问“Do you think enrollment timing is similar for your patients? How do you convince this issue?”应该如何回答啊？
谢谢各位
京公网安备 11010802022788号
