搜索
生存分析生存分析的主要内容生存过程的描述生存时间的分布特点估计生存率估计平均生存时间绘制生存率曲线生存过程的统计推断生存时间的影响因素研究生存分析需具备的资料生存时间或称时间变量观察结果或称截尾变量暴露变量及混杂因素肾移植例一般手术(A组):391520202630414664+64135223365450596+680+900+900+改进手术(B组):1070+70+120225366390+475+518+647+801+1001+1045+1045+手术方式生存时间观察结果treattimeoutcome1311911411146116411640………………190002101………………28010210010210450210450生存资料的定义(1)将数据定义为生存资料格式stset时间变量[，failure(截尾变量)]failure(截尾变量)stata以所有不等于0的非缺失值为预期的出现结果；若failure选择项缺省，则stata默认所有纪录均出现结果生存资料的定义(2)四个变量---定义数据库后，stata系统自动产生的_st/*数据库中该条记录是否被定义为生存资料_d/*该条记录是否出现预期结果_t/*观察对象被随访的时间_t0/*观察对象第一次被观察到的时间(开始过程的时间为0)stsettime,failure(outcome)failureevent:outcome!=0&outcome.obs.timeinterval:(0,time]exitonorbefore:failure------------------------------------------------------------------33totalobs.0exclusions------------------------------------------------------------------33obs.remaining,representing18failuresinsinglerecord/singlefailuredata11370totalanalysistimeatrisk,atriskfromt=0earliestobservedentryt=0lastobservedexitt=1045生存资料的描述stci,[by(分组变量)选择项]p(#)/*生存时间的百分位数level(#)/*可信区间的可信度media/*计算中位生存时间，为默认选项rmean/*计算平均生存时间.stci,by(treat)failure_d:outcomeanalysistime_t:time|no.oftreat|subjects50%Std.Err.[95%Conf.Interval]-------+-------------------------------------------------------------1|196446.8995204502|14..225.-------+-------------------------------------------------------------total|33365143.400164..stci,rmeanby(treat)failure_d:outcomeanalysistime_t:time|no.ofrestrictedtreat|subjectsmeanStd.Err.[95%Conf.Interval]-------+-------------------------------------------------------------1|19297.0329(*)83.96867132.457461.602|14766.4481(*)116.2397538.622994.274-------+-------------------------------------------------------------total|33512.9716(*)83.84662348.635677.308(*)largestobservedanalysistimeiscensored,meanisunderestimated.生存率的估计输出生存率、生存概率的标准误等统计量stslist[,by(分组变量)选择项]failure/*输出死亡函数(1-S(t))level/*规定所输出可信区间的可信度.stslist,by(treat)failure_d:outcomeanalysistime_t:timeBeg.NetSurvivorStd.TimeTotalFailLostFunctionError[95%Conf.Int.]-------------------------------------------------------------------------------treat=1319100.94740.05120.68120.9924…...6803010.23680.10150.07580.44759002020.23680.10150.07580.4475treat=21014100.92860.06880.59080.98967013020.92860.06880.59080.9896…...10013010.67530.13440.34640.864710452020.67530.13440.34640.8647-------------------------------------------------------------------------------绘制生存（死亡）曲线(1)绘制Kaplan-Meier生存(死亡)曲线stsgraph[,by(分组变量)separate选择项]failure/*指定绘制“死亡”曲线，缺省时绘制生存曲线Gwood/*绘制生存(或死亡)曲线及其Greenwood可信区间lost/*在曲线上标出该时间点截尾值例数Kaplan-Meiersurvivalestimates,bytreatanalysistime050010000.000.250.500.751.00treat1treat2212622绘制生存（死亡）曲线(2)绘制Kaplan-Meier生存曲线与Cox预测曲线stcoxkm[,by(分组变量)separate绘图命令选择项]Observedvs.PredictedSurvivalProbabilitiesByCategoriesoftreatanalysistimeObserved:treat=1Observed:treat=2Predicted:treat=1Predicted:treat=2310450.000.250.500.751.00生存率的比较ststest分组变量[,选择项]logrank/*进行log-rank检验，为缺省值.ststesttreatfailure_d:outcomeanalysistime_t:timeLog-ranktestforequalityofsurvivorfunctions|EventsEventstreat|observedexpected------+-------------------------1|148.572|49.43------+-------------------------Total|1818.00chi2(1)=6.71Prchi2=0.0096例8.8某临床试验比较A，B两治疗方案对某病的治疗效果，A组(group=0)12人，B组(group=1)13人。病人分组后检验其肾功能(kidney)，功能正常者记为0，不正常者记为1；治疗后生存时间为stime(天)；问不同治疗方案及肾功能对病人的生存时间是否有影响?编号NO.治疗生存观察肾功能kidney编号NO.治疗生存观察肾功能kidney方案时间结果方案时间结果Groupstimecensorgroupstimecensor108111311801020852001416321030521115122400040220101611951050631117176106081018170107019760019113118012960020123119014600021112961010063112212101011013280023170010120365002411811251199000例8.8Cox比例风险模型(1)估计Cox比例风险模型的命令格式为stcox[协变量][,nohr选择项]nohr/*指定输出回归系数b而不是危险比exp(b)level(#)/*可信区间的可信度例8.8Stsetstime,failure(censor)Stcoxgroupkidney,nolognohrfailure_d:censoranalysistime_t:stimeCoxregression--BreslowmethodfortiesNo.ofsubjects=25Numberofobs=25No.offailures=17Timeatrisk=15334LRchi2(2)=22.67Loglikelihood=-35.707534Probchi2=0.0000------------------------------------------------------------------------------_t|_d|Coef.Std.Err.zP|z|[95%Conf.Interval]-------------+----------------------------------------------------------------group|1.243078.59931772.070.038.06843652.417719kidney|4.1054551.1645333.530.0001.8230126.387898------------------------------------------------------------------------------Cox比例风险函数:(t)=0(t)exp(1.243group+4.105kidney)肾功能正常者接受B治疗方案比接受A治疗方案在某时刻的相对危险度为：RR=(t|group=1,kidney=0)/(t|group=0,kidney=0)=0(t)exp(1.243*1+4.105*0)/0(t)exp(1.243*0+4.105*1)=3.466肾功能不正常者接受B治疗方案比接受A治疗方案在某时刻的相对危险度为3.466肾功能不正常者接受B治疗方案比肾功能正常者接受A治疗方案在某时刻的相对危险度为210.300Weibull回归和指数回归streg[协变量]，[选项]dist（分布类型）/*拟合模型的分布类型。exponential（指数分布）weibull（weibull分布）nohr/*指定输出回归系数b而不是危险比exp(b)level(#)/*可信区间的可信度