Pooling C-index of Logistic and Cox Regression Models

Introduction

The miceafter package includes the function pool_cindex, to pool c-index values from logistic and Cox regression models. This vignette shows you how to use this function.

Examples

Pooling the C-index After the mice function and Logistic Regression

The lbp_orig is a dataset as part of the miceafter package with missing values. So we first impute them with the mice function. Than we use the mids2milist function to turn the mids object with multiply imputed datasets, as a result of using mice, into a milist object. Than we use the with function to apply repeated analyses with the cindex function across the multiply imputed datasets. Finally, we pool the results by using the pool_cindex function. We do that in one pipe.

  lbp_orig %>%
      mice(m=5, seed=3025, printFlag = FALSE) %>%
        mids2milist() %>% 
          with(expr = cindex(glm(Chronic ~ Gender + Radiation, family=binomial))) %>% 
            pool_cindex()
#>        C-index Critical value 95 CI low 95 CI high
#> [1,] 0.6553774        1.97818  0.567203   0.734012
#> attr(,"class")
#> [1] "mipool"

Pooling the C-index after Multiply Imputed datasets are stored in a dataframe and with Logistic Regression

The dataset lbpmilr as part of the miceafter package is a long dataset that contains 10 multiply imputed datasets. The datasets are distinguished by the Impnr variable. First we convert the dataset into a milist object by using the df2milist function. Than we use the with function to apply repeated analyses with the cindex function across the multiply imputed datasets. Finally, we pool the results by using the pool_cindex function.

  imp_data <- df2milist(lbpmilr, impvar = "Impnr") 

  ra <- with(data=imp_data,
   expr = cindex(glm(Chronic ~ Gender + Radiation, family=binomial)))

  res <- pool_cindex(ra)
  res
#>        C-index Critical value 95 CI low 95 CI high
#> [1,] 0.6638267       1.976656 0.5764274  0.7412861
#> attr(,"class")
#> [1] "mipool"

Pooling the C-index after Multiply Imputed datasets are stored in a dataframe and with Cox Regression

The dataset lbpmicox as part of the miceafter package is a long dataset that contains 10 multiply imputed datasets. The datasets are distinguished by the Impnr variable. First we convert the dataset into a milist object by using the df2milist function. Than we use the with function to apply repeated analyses with the cindex function across the list of multiply imputed datasets. Finally, we pool the results by using the pool_cindex function.

  library(survival)
  lbpmicox %>% 
    df2milist(impvar = "Impnr") %>%
      with(expr = cindex(coxph(Surv(Time, Status) ~ Radiation + Age))) %>%
        pool_cindex()
#>        C-index Critical value 95 CI low 95 CI high
#> [1,] 0.5413464       1.959964 0.4952103  0.5867842
#> attr(,"class")
#> [1] "mipool"