PASS软件使用指南

2013/12/2812:07:3711.第一种情况：随访为6月，入组时间12月，总时间18个月，治疗组生存率从0.8-0.85时，计算对应所需样本量。参数设置如图：LogrankTestPowerAnalysisNumericResultsinTermsofSampleSizewhentheTestisTwo-SidedandT0is6Acc-CtrlTrtrualHazPropPropAcc-Time/CtrlTrtRatioSurvSurvrualTotalCtrlTrttotoPowerN1N2N(HR)(S1)(S2)Pat'nTimeLossLossTrtCtrlAlphaBeta0.80022202214410.62560.70000.8000Equal12/180.05000.05000.00000.00000.05000.19980.80051791793580.59080.70000.8100Equal12/180.05000.05000.00000.00000.05000.19950.80071471482950.55640.70000.8200Equal12/180.05000.05000.00000.00000.05000.19930.80051231232460.52240.70000.8300Equal12/180.05000.05000.00000.00000.05000.19950.80161041042080.48880.70000.8400Equal12/180.05000.05000.00000.00000.05000.19840.801888891770.45560.70000.8500Equal12/180.05000.05000.00000.00000.05000.1982ReferencesLakatos,Edward.1988.'SampleSizesBasedontheLog-RankStatisticinComplexClinicalTrials',Biometrics,Volume44,March,pages229-241.Lakatos,Edward.2002.'DesigningComplexGroupSequentialSurvivalTrials',StatisticsinMedicine,Volume批注[i1]:T0=6是随访时间6个月，此时入组时间为12月，总时间为18月。批注[i2]:关键是对治疗组的生存率的估计，也就是1-出血率，可以看到在对照组生存率为0.70，随访6月，入组时间为12月，总时间为18月，把握度为0.8时，随着治疗组生存率的增加，所需样本量下降。请你们确定哪个生存率最合适？21,pages1969-1989.ReportDefinitionsPoweristheprobabilityofrejectingafalsenullhypothesis.Powershouldbeclosetoone.N1|N2|Narethesamplesizesofthecontrolgroup,treatmentgroup,andbothgroups,respectively.E1|E2|Earethenumberofeventsinthecontrolgroup,treatmentgroup,andbothgroups,respectively.HazardRatio(HR)isthetreatmentgroup'shazardratedividedbythecontrolgroup'shazardrate.ProportionSurvivingistheproportionsurvivingpasttimeT0.AccrualTimeisthenumberoftimeperiods(yearsormonths)duringwhichaccrualtakesplace.TotalTimeisthetotalnumberoftimeperiodsinthestudy.Follow-uptime=(TotalTime)-(AccrualTime).CtrlLossistheproportionofthecontrolgroupthatislost(dropout)duringasingletimeperiod(yearormonth).TrtLossistheproportionofthetreatmentgroupthatislost(dropout)duringasingletimeperiod(yearormonth).CtrltoTrt(dropin)istheproportionofthecontrolgroupthatswitchtoagroupwithahazardrateequaltothetreatmentgroup.TrttoCtrl(noncompliance)istheproportionofthetreatmentgroupthatswitchtoagroupwithahazardrateequaltothecontrolgroup.Alphaistheprobabilityofrejectingatruenullhypothesis.Itshouldbesmall.Betaistheprobabilityofacceptingafalsenullhypothesis.Itshouldbesmall.2013/12/2812:05:312LogrankTestPowerAnalysisNumericResultsinTermsofEventswhentheTestisTwo-SidedandT0is6Acc-CtrlTrtrualCtrlTrtTotalHazPropPropAcc-Time/CtrlTrtEvtsEvtsEvtsRatioSurvSurvrualTotalCtrlTrttotoPowerE1E2E(HR)(S1)(S2)Pat'nTimeLossLossTrtCtrlAlphaBeta0.800284.759.1143.80.62560.70000.8000Equal12/180.05000.05000.00000.00000.05000.19980.800568.745.8114.50.59080.70000.8100Equal12/180.05000.05000.00000.00000.05000.19950.800756.635.992.50.55640.70000.8200Equal12/180.05000.05000.00000.00000.05000.19930.800547.228.475.60.52240.70000.8300Equal12/180.05000.05000.00000.00000.05000.19950.801639.922.762.60.48880.70000.8400Equal12/180.05000.05000.00000.00000.05000.19840.801834.018.252.20.45560.70000.8500Equal12/180.05000.05000.00000.00000.05000.1982SummaryStatementsAtwo-sidedlogranktestwithanoverallsamplesizeof441subjects(220inthecontrolgroupand221inthetreatmentgroup)achieves80.0%powerata0.050significanceleveltodetectahazardratioof00.6256whentheproportionsurvivinginthecontrolgroupis00.7000.Thestudylastsfor18timeperiodsofwhichsubjectaccrual(entry)occursinthefirst12timeperiods.Theaccrualpatternacrosstimeperiodsisuniform(allperiodsequal).Theproportiondroppingoutofthecontrolgroupis0.0500.Theproportiondroppingoutofthetreatmentgroupis0.0500.Theproportionswitchingfromthecontrolgrouptoanothergroupwithahazardratioequaltothatofthetreatmentgroupis0.0000.Theproportionswitchingfromthetreatmentgrouptoanothergroupwithahazardequaltothatofthecontrolgroupisis0.0000.2013/12/2812:05:323LogrankTestPowerAnalysisChartSection2.第二种情况：随访为6月，入组时间24月，总时间30个月，治疗组生存率从0.8-0.85时，对应所需样本量：参数设置：LogrankTestPowerAnalysisNumericResultsinTermsofSampleSizewhentheTestisTwo-SidedandT0is6Acc-CtrlTrtrualHazPropPropAcc-Time/CtrlTrtRatioSurvSurvrualTotalCtrlTrttotoPowerN1N2N(HR)(S1)(S2)Pat'nTimeLossLossTrtCtrlAlphaBeta0.80041891903790.62560.70000.8000Equal24/300.05000.05000.00000.00000.05000.19960.80051531543070.59080.70000.8100Equal24/300.05000.05000.00000.00000.05000.19950.80131261272530.55640.70000.8200Equal24/300.05000.05000.00000.00000.05000.19870.80181051062110.52240.70000.8300Equal24/300.05000.05000.00000.00000.05000.19820.800388891770.48880.70000.8400Equal24/300.05000.05000.00000.00000.05000.19970.802175761510.45560.70000.8500Equal24/300.05000.05000.00000.00000.05000.1979ReferencesLakatos,Edward.1988.'SampleSizesBasedontheLog-RankStatisticinComplexClinicalTrials',Biometrics,Volume44,March,pages229-241.Lakatos,Edward.2002.'DesigningComplexGroupSequentialSurvivalTrials',StatisticsinMedicine,Volume批注[i3]:随访时间为6个月，入组时间为24个月，总时间为30个月。批注[i4]:此时的治疗组生存率分别从0.8-0.85，可见样本量是减少的。入组时间由12月延长到24个月，可以看到样本量变化并不是太大，所以建议选择第一种情况的12个月入组时间为宜。21,pages1969-1989.ReportDefinitionsPoweristheprobabilityofrejectingafalsenullhypothesis.Po