找回密码
 立即注册

QQ登录

只需一步,快速开始

搜索

SPSS中Cox回归问题,请求确认几项统计方法

[复制链接]
发表于 2017-2-26 06:40:18
陈老师:
非常感谢您之前帖子的回复!还有几个问题望您不吝赐教:
1, 有一篇文章,统计方法和我要写的文章非常一致,他这么写的Receiver operating characteristic curves using the C-statistic were used to compare accuracy in predicting 30-day survival according to nadir platelet count and DPC degree and were compared by the bootstrap method. Best estimate had highest summation of sensitivity and specificity. The estimated 30-day survival curve was calculated and displayed by a hazard function for the probability of death for each platelet nadir, based on the current data set. Survival curves were constructed using Kaplan–Meier estimates, while comparisons relied on。the log-rank test. The multivariable model was built by selecting baseline variables of clinical interest and/or satisfaction of the entry criterion of P , 0.05 in the univariable analysis: Society of Thoracic Surgeons (STS) score, baseline renal failure, major vascular complication, major or lifethreatening bleeding, major stroke, sepsis, acute kidney injury (stage II/ III), and severe thrombocytopenia. A two-sided alpha level of 0.05 was used for all superiority testing. All statistical analyses were performed using SAS software version 9.2 (SAS Institute, Cary, NC, USA).
1,其中的加黑部分是不是使用COX生存回归?,
2,划线部分是不是使用二元Logistic回归?
3,
cox回归.png 能帮我确认一下这幅图片是是使用什么统计方法做出来的吗?
4, 在浏览 论坛的帖子(http://www.spssbbs.com/forum.php ... =549&extra=page%3D1)时我看到您回复,将“高血压”选入协变量,不要进行“分类”设置,但高血压本身就是二分类变量,我在统计时也有很多很多同样的变量,比如高血压、糖尿病、是否输血等,也遇到了和那个网友一模一样的问题,就是想请教:什么时候需要将这些变量进行“分类”设置?什么时候不需要?
5,我在二元Logistic回归时,先进行单因素Logistic回归,筛选出有Sig小于0.05的,然后进行多因素分析。在多因素分析时发现选入不同个数的协变量,出来的结果完全不一样?想请教:单因素分析出来的协变量是否需要全部入选多因素分析,还是根据临床判断,选入几个临床关系比较大的就可以了?
拜谢!
使用道具 举报 回复
发表于 2017-2-26 06:49:46
问题2的划线部分是:The multivariable model was built by selecting baseline variables of clinical interest and/or satisfaction of the entry criterion of P , 0.05 in the univariable analysis: Society of Thoracic Surgeons (STS) score, baseline renal failure, major vascular complication, major or lifethreatening bleeding, major stroke, sepsis, acute kidney injury (stage II/ III), and severe thrombocytopenia.
使用道具 举报 回复 支持 反对
发表于 2017-2-27 11:54:53
1、黑线部分是累计生存曲线,
2、虚线是累计生存曲线的95%置信区间
3、这个图很多方法可以画,如果没有自变量,KM生存曲线就可以画,如果有自变量,就是COX回归分析画
4、二分类的自变量,可以设置分类也可以不设置分析,如果设置了分类,SPSS软件会重新编码,分析起来有些麻烦,因此建议二分类自变量不设置分析,3类或者3类以上的自变量必须设置分类
5、自变量的个数不同,运算出来的结果肯定不同,所以要严格遵守单因素显著的自变量全部纳入多因素中这个规则
6、画线部分是COX生存回归,不是二元logistic回归
6、
使用道具 举报 回复 支持 反对
您需要登录后才可以回帖 登录 | 立即注册

本版积分规则

 
QQ在线咨询