2-病例对照研究的分析

叶小华观察性研究病例对照研究的分析分析方法定量资料均数比较：t检验、方差分析、线性回归定性资料率的比较(卡方检验，趋势卡方检验，二分类logistic回归)等级资料中位数比较(秩和检验，有序logistic回归)实例分析1：某地肝癌成组病例对照研究分析框架1基本情况的描述（均衡性检验）2对影响因素进行单因素分析3对影响因素进行多因素分析4交互作用分析1基本情况的描述：性别分布1基本情况的描述：年龄分布1基本情况的描述：年龄分布Reportage58.7717612.694309357.6217012.680279058.2034612.6822793case12TotalMeanNStd.DeviationMinimumMaximum1基本情况的描述本次调查共收集肝癌病例、对照例。其中病例组男例，女例，年龄～岁，平均(±)岁。对照组男例，女例，年龄～岁，平均(±)岁。经均衡性检验，病例与对照在年龄、性别差异无统计学意义(P0.05)2对影响因素进行单因素分析根据case变量，产生因变量y赋值：病例=1对照=02对影响因素进行单因素分析VariablesintheEquation1.709.30132.2721.0005.5223.0629.956-2.092.37231.5851.000.123ganyanConstantStep1aBS.E.WalddfSig.Exp(B)LowerUpper95.0%C.I.forEXP(B)Variable(s)enteredonstep1:ganyan.a.VariablesintheEquation.574.2177.0011.0081.7751.1602.714-.850.3515.8581.016.427drinkConstantStep1aBS.E.WalddfSig.Exp(B)LowerUpper95.0%C.I.forEXP(B)Variable(s)enteredonstep1:drink.a.VariablesintheEquation.404.2253.2351.0721.498.9642.327-.631.3852.6811.102.532smokeConstantStep1aBS.E.WalddfSig.Exp(B)LowerUpper95.0%C.I.forEXP(B)Variable(s)enteredonstep1:smoke.a.VariablesintheEquation-.046.311.0221.882.955.5191.756.092.369.0631.8021.097sexConstantStep1aBS.E.WalddfSig.Exp(B)LowerUpper95.0%C.I.forEXP(B)Variable(s)enteredonstep1:sex.a.Edu2您的教育程度是：1=小学及以下2=初中3=高中及以上Yusheng2您吃鱼生（或虾生，螺生等）吗？0：从不吃或小于1次/月1：1～5次/月或6～10次/月或11次/月及以上VariablesintheEquation1.6952.429.450.3461.6931.1931.569.7963.090.386.3761.0521.3051.471.7043.072-.345.3171.1831.277.708edu2edu2(1)edu2(2)ConstantStep1aBS.E.WalddfSig.Exp(B)LowerUpper95.0%C.I.forEXP(B)Variable(s)enteredonstep1:edu2.a.VariablesintheEquation-.301.360.6971.404.740.3661.499.288.342.7091.4001.333yusheng2ConstantStep1aBS.E.WalddfSig.Exp(B)LowerUpper95.0%C.I.forEXP(B)Variable(s)enteredonstep1:yusheng2.a.2对影响因素进行单因素分析表1肝癌影响因素的单因素logistic分析结果影响因素回归系数标准误Wald2POR(95%CI)性别男1.00女-0.0460.3110.0220.8820.955（0.519-1.756）文化程度小学及以下1.00初中高中及以上吸烟否1.00是饮酒否1.00是鱼生1次/月1.00≥1次/月肝炎史无1.00有卡方检验卡方检验卡方检验Chi-SquareTests.022b1.882.00011.000.0221.8821.000.502.0221.882351PearsonChi-SquareContinuityCorrectionaLikelihoodRatioFisher'sExactTestLinear-by-LinearAssociationNofValidCasesValuedfAsymp.Sig.(2-sided)ExactSig.(2-sided)ExactSig.(1-sided)Computedonlyfora2x2tablea.0cells(.0%)haveexpectedcountlessthan5.Theminimumexpectedcountis23.52.b.Crosstab14815530386.0%86.6%86.3%24244814.0%13.4%13.7%172179351100.0%100.0%100.0%Count%withinyCount%withinyCount%withiny12sexTotal.001.00yTotal趋势卡方检验Chi-SquareTests1.712a2.4251.7182.4241.3471.246348PearsonChi-SquareLikelihoodRatioLinear-by-LinearAssociationNofValidCasesValuedfAsymp.Sig.(2-sided)0cells(.0%)haveexpectedcountlessthan5.Theminimumexpectedcountis20.15.a.卡方检验趋势卡方检验Crosstab9911020957.9%62.1%60.1%48509828.1%28.2%28.2%24174114.0%9.6%11.8%171177348100.0%100.0%100.0%Count%withinyCount%withinyCount%withinyCount%withiny1.002.003.00edu2Total.001.00yTotal2对影响因素进行单因素分析表1肝癌影响因素的单因素分析结果影响因素对照病例2POR(95%CI)性别男148(86.0)155(86.6)女24(14.0)24(13.4)0.020.8820.96（0.52-1.76）文化程度小学及以下初中高中及以上吸烟否是饮酒否是鱼生1次/月≥1次/月肝炎史无有3对影响因素进行多因素分析3对影响因素进行多因素分析VariablesintheEquation-.497.3841.6761.196.609.2871.291.594.2366.3651.0121.8121.1422.8761.712.30431.7331.0005.5383.05310.045-2.577.60218.3071.000.076yusheng2drinkganyanConstantStep1aBS.E.WalddfSig.Exp(B)LowerUpper95.0%C.I.forEXP(B)Variable(s)enteredonstep1:yusheng2,drink,ganyan.a.3对影响因素进行多因素分析表4肝癌影响因素的多因素logistic分析结果影响因素回归系数标准误Wald2POR(95%CI)饮酒否1.00是鱼生1次/月1.00≥1次/月肝炎史无1.00有4交互作用产生交互作用项：drink-ganyan=drink*ganyan4交互作用交互作用无统计学意义（P=0.118）.VariablesintheEquation1.726.7695.0291.0255.6151.24325.3723.2361.0589.3491.00225.4233.195202.293-.976.6242.4441.118.377.1111.281-4.7781.29113.7041.000.008drinkganyandrink_ganyanConstantStep1aBS.E.WalddfSig.Exp(B)LowerUpper95.0%C.I.forEXP(B)Variable(s)enteredonstep1:drink,ganyan,drink_ganyan.a.4交互作用若存在交互作用，可进一步分层比较：饮酒肝炎对照病例OR(95%CI)--84381.00+-70670.13(0.06-0.28)-+6260.28(0.13-0.59)++11381.25(0.41-3.82)变量赋值drink：无=1有=2ganyan2：无=1有=3Sum=drink+ganyu2sum*yCrosstabulation843812249.1%22.5%35.9%706713740.9%39.6%40.3%626323.5%15.4%9.4%1138496.4%22.5%14.4%171169340100.0%100.0%100.0%Count%withinyCount%withinyCount%withinyCount%withinyCount%withiny2.003.004.005.00sumTotal.001.00yTotal回归结果：sum以分类变量进入模型VariablesintheEquation39.8423.000-2.033.39426.5871.000.131.060.284-1.283.38311.2501.001.277.131.587.227.568.1591.6901.254.4123.8171.240.34213.1101.0003.455sumsum(1)sum(2)sum(3)ConstantStep1aBS.E.WalddfSig.Exp(B)LowerUpper95.0%C.I.forEXP(B)Variable(s)enteredonstep1:sum.a.实例分析2：某地低出生体重的1:1匹配病例对照研究具体步骤Analyze→Survival→CoxRegressionTime框：time(指定虚拟的生存变量)Status框：low(选入生存状态变量)Defineevent：singlevalue：1：continue（“1表示出现观察结局”）Strata框：pairid（指定分层变量）Options框→CIforexp(B)OK分析结果母亲妊娠期间吸烟（OR=3.05，95%CI1.26-7.40）和既往早产史(OR=4.19，95%CI1.22-14.39)均是新生儿低出生体重的危险因素。VariablesintheEquation1.114.4536.0561.0143.0461.2557.3961.433.6295.1931.0234.1931.22214.387-.009.0071.4371.231.991.9781.005smokeptdlwtBSEWalddfSig.Exp(B)LowerUpper95.0%CIforExp(B)