Skip to contents

Compare the Fit of Several Cox Models in One Table

Source: R/tbl_cox_glance_compare.R
tbl_cox_glance_compare.Rd

One row per model with the number of subjects and events, log-likelihood, AIC, BIC and concordance (with its standard error), taken from [broom::glance()]. Use it to line up candidate survival specifications.

Usage

tbl_cox_glance_compare(
  models,
  names = NULL,
  sort_by = c("AIC", "BIC", "concordance"),
  digits = 2
)

Arguments

models

A list of [survival::coxph()] fits, or of [cox_single()] / [cox_multi()] results (their `model` is used). A single model is accepted.

names

Optional model names (one per model). By default the names of `models`, otherwise `"Model 1"`, `"Model 2"`, ...

sort_by

`"AIC"` (default) or `"BIC"` (smallest first), `"concordance"` (largest first) or `NULL` to keep the order of `models`.

digits

Decimals for the log-likelihood, AIC, BIC and concordance.

Value

A tibble of class `cbe_cox_glance` with text columns `Model`, `N`, `Events`, `Log-Likelihood`, `AIC`, `BIC` and `Concordance (SE)`, and the attribute `"numeric"` holding the unformatted values (`model`, `n`, `events`, `loglik`, `AIC`, `BIC`, `concordance`, `se_concordance`).

Comparability

AIC and BIC are only comparable between models fitted to the **same rows**. The function warns when the models were fitted to different numbers of subjects or events (usually because a predictor had missing values in one model and not in another).

See also

[cox_table()], [cox_single()], [cox_multi()]

Examples

lung <- survival::lung
m1 <- survival::coxph(survival::Surv(time, status) ~ age, data = lung)
m2 <- survival::coxph(survival::Surv(time, status) ~ age + sex, data = lung)
tbl_cox_glance_compare(list(age = m1, `age + sex` = m2))
#> # A tibble: 2 × 7
#>   Model     N     Events `Log-Likelihood` AIC     BIC     `Concordance (SE)`
#>   <chr>     <chr> <chr>  <chr>            <chr>   <chr>   <chr>             
#> 1 age + sex 228   165    -742.85          1489.70 1495.91 0.60 (0.03)       
#> 2 age       228   165    -747.79          1497.58 1500.68 0.55 (0.03)