cox回归生存分析在stata中实现

为了评价各种指标对透析患者生存时间的影响，资料如下，为避免受过大值的影响，实际拟合模型中，ntprobnp数据取对数，记为ntprobnp1genntprobnp1=log(ntprobnp)资料如下：timediecreabunalbagetherapyntprobnpntprobnp112110323428450202.99573226113244626661343.52636138115204931560564.02535245187028326211124.718499991990343935010206.9275581200785234229019207.56008112004561943350453.80666312005702339270884.4773371200102033414202115.351858120078029402304556.1202971200670223919021007.6496931200932283523013207.18538712006892744560443.784193516703328661664.18965567112103434721774.3438051.预后因素筛选：logranktimedie,by(crea)检测肌酐是否影响生存率chi2(13)=29.07Prchi2=0.0064P0.05，说明肌酐对预后影响大logranktimedie,by(bun)logranktimedie,by(alb)logranktimedie,by(age)logranktimedie,by(therapy)logranktimedie,by(ntprobnp1)由于bunP0.05，在COX模型中去除改因素2.用COX比例风险模型分析coxtimecreaalbagetherapyntprobnp1,dead(die)Coxregression--notiesEntrytime0Numberofobs=15①LRchi2(5)=26.21②Probchi2=0.0001③Loglikelihood=-4.1883032④PseudoR2=0.7578------------------------------------------------------------------------------time|⑤Coef.⑥Std.Err.⑦z⑧P|z|⑨[95%Conf.Interval]-------------+----------------------------------------------------------------crea|-.0032338.0031113-1.040.299-.0093319.0028643alb|-1.260323.8536905-1.480.140-2.933525.4128797age|.2428877.25971760.940.350-.2661495.7519248therapy|-7.6858767.077233-1.090.277-21.5576.185245ntprobnp1|-.22935861.013952-0.230.821-2.2166691.757952①为模型无效假设(即：所有协变量的回归系数为0)所对应的似然比检验(自由度为协变量个数的卡方);②模型无效假设检验对应的p值；③对数似然比；④伪决定系数；⑤回归系数；⑥回归系数的标准误；⑦单个回归系数检验(Ho:该回归系数为0)的Z统计量；⑧单个回归系数验的p值；⑨回归系数的95%可信限。本例结果表明：在平衡诸多混杂因素后，模型中的因素与死亡风险无明显相关性。3.作图stsettime,failure(die)数据库定义为生存资料数据库；stslist为计算命令。stsliststsgraph生存概率曲线stsgraph,by(therapy)生存概率曲线，不同治疗措施对生存率的影响