我试图用CPH回归对R进行亚组分析，我有很多糖尿病患者（是/否）和一定的尿量（〉/〈200 ml）。希望测量这种相互作用对生存率的影响。我已经设法运行了糖尿病组的CPH我还创建了糖尿病/尿量组合的子集（为隐私起见重命名了数据集）。

# create new variable splitting above and below 200 ml urine amount
    xxxx$uacr200 <- ifelse(xxxx$uacr>=200,1,0)

# create a new categorical variable that combines diabetes status AND urine amount above and below 200
xxxx$dmuacrgr <- NA
xxxx$dmuacrgr[xxxx$uacr<200 & xxxx$diabfl=="N"] <- 1
xxxx$dmuacrgr[xxxx$uacr>=200 & xxxxl$diabfl=="N"] <- 2
xxxx$dmuacrgr[xxxx$uacr<200 & xxxx$diabfl=="Y"] <- 3
xxxx$dmuacrgr[xxxx$uacr>=200 & xxxx$diabfl=="Y"] <- 4

# create version of treatment variable that is coded as 1 and 0
xxxx$trt <- NA
xxxx$trt[xxxx$trt02pn==2] <- 1
xxxx$trt[xxxx$trt02pn==1] <- 0

xxxx_sub <- xxxx[,c("usubjid", "mageg1n","mgfrg1n","msexg1n","muacrg1n", "trt")]

#Diab subgroup analysis (across all urine amounts)

# fit cox model to estimate HRs in diabetes subgroups
cox_mod_diab <- coxph(Surv(aval, cnsr==0) ~ trt + trt*diabfl + diabfl +
                            mageg1n + mgfrg1n + msexg1n + muacrg1n + sitegr2n, 
                         data = zzzz, subset=(paramcd=="KDPDVASC"), ties="breslow")

    subg_cox_mod_diab <-            subgroupAnalysis(cox_mod_diab_int_KDPDVASC,zzzz[zzzz$paramcd=="KDPDVASC",],treatment="trt",
                                 subgroups=("diabfl"),confint.method="default")

我现在只需要使用我上面创建的组合子集来找出糖尿病的影响是什么，在每个尿量....任何帮助都会真正保存我！！