Teal Module: Cox Regression Model — tm_t_coxreg • teal.modules.clinical

Teal module to fit Cox univariate or multivariate models consistent with COXT01 and COXT02 standard outputs, respectively.

Usage

tm_t_coxreg(
  label,
  dataname,
  parentname = ifelse(inherits(arm_var, "data_extract_spec"),
    teal.transform::datanames_input(arm_var), "ADSL"),
  arm_var,
  arm_ref_comp = NULL,
  paramcd,
  cov_var,
  strata_var,
  aval_var = teal.transform::choices_selected(teal.transform::variable_choices(dataname,
    "AVAL"), "AVAL", fixed = TRUE),
  cnsr_var = teal.transform::choices_selected(teal.transform::variable_choices(dataname,
    "CNSR"), "CNSR", fixed = TRUE),
  multivariate = TRUE,
  conf_level = teal.transform::choices_selected(c(0.95, 0.9, 0.8), 0.95, keep_order =
    TRUE),
  pre_output = NULL,
  post_output = NULL,
  basic_table_args = teal.widgets::basic_table_args()
)

Arguments

label

(character)
menu item label of the module in the teal app.

dataname

(character)
analysis data used in teal module.

parentname

(character)
parent analysis data used in teal module, usually this refers to ADSL.

arm_var

(choices_selected or data_extract_spec)
object with all available choices and preselected option for variable names that can be used as arm_var. It defines the grouping variable(s) in the results table. If there are two elements selected for arm_var, second variable will be nested under the first variable.

arm_ref_comp

optional, (list)
If specified it must be a named list with each element corresponding to an arm variable in ADSL and the element must be another list (possibly with delayed teal.transform::variable_choices() or delayed teal.transform::value_choices() with the elements named ref and comp that the defined the default reference and comparison arms when the arm variable is changed.

paramcd

(choices_selected or data_extract_spec)
variable value designating the studied parameter.

cov_var

(choices_selected or data_extract_spec)
object with all available choices and preselected option for the covariates variables.

strata_var

(choices_selected or data_extract_spec)
names of the variables for stratified analysis.

aval_var

(choices_selected or data_extract_spec)
object with all available choices and preselected option for the analysis variable.

cnsr_var

(choices_selected or data_extract_spec)
object with all available choices and preselected option for the censoring variable.

multivariate

(logical)
If FALSE, the univariate approach is used (equivalent to COXT01 standard) instead of the multivariate model (equivalent to COXT02 standard).

conf_level

(choices_selected)
object with all available choices and preselected option for the confidence level, each within range of (0, 1).

pre_output

optional, (shiny.tag)
with text placed before the output to put the output into context. For example a title.

post_output

optional, (shiny.tag)
with text placed after the output to put the output into context. For example the shiny::helpText() elements are useful.

basic_table_args

optional, (basic_table_args)
object created by teal.widgets::basic_table_args() with settings for the module table. The argument is merged with option teal.basic_table_args and with default module arguments (hard coded in the module body).

For more details, see the vignette: vignette("custom-basic-table-arguments", package = "teal.widgets").

Details

The Cox Proportional Hazards (PH) model is the most commonly used method to estimate the magnitude of the effect in survival analysis. It assumes proportional hazards: the ratio of the hazards between groups (e.g., two arms) is constant over time. This ratio is referred to as the "hazard ratio" (HR) and is one of the most commonly reported metrics to describe the effect size in survival analysis. For further information about the Cox Proportional Hazards Model, check "Statistical Analysis of Clinical Trials Data with R", NEST team.

This modules expects that the analysis data has the following variables:

`AVAL`	time to event
`CNSR`	boolean or 0,1 is element in `AVAL` censored
`PARAMCD`	variable used to filter for endpoint (e.g. OS), after filtering for `paramcd` one observation per patient is expected

The arm variables, stratification and covariate variables are taken from the ADSL data.

Note

The likelihood ratio test is not supported for model including strata, Wald test will be substituted.
Multivariate is the default choice for backward compatibility.

Examples

## First example
## =============
## The example below is based on the usual approach involving creation of
## a random CDISC dataset and then running the application.

library(scda)

synthetic_cdisc_data_latest <- synthetic_cdisc_data("latest")
ADSL <- synthetic_cdisc_data_latest$adsl
ADTTE <- synthetic_cdisc_data_latest$adtte
arm_ref_comp <- list(
  ACTARMCD = list(
    ref = "ARM B",
    comp = c("ARM A", "ARM C")
  ),
  ARM = list(
    ref = "B: Placebo",
    comp = c("A: Drug X", "C: Combination")
  )
)

app <- init(
  data = cdisc_data(
    cdisc_dataset("ADSL", ADSL,
      code = "synthetic_cdisc_data_latest <- synthetic_cdisc_data('latest')
        ADSL <- synthetic_cdisc_data_latest$adsl"
    ),
    cdisc_dataset("ADTTE", ADTTE,
      code = "synthetic_cdisc_data_latest <- synthetic_cdisc_data('latest')
        ADTTE <- synthetic_cdisc_data_latest$adtte"
    )
  ),
  modules = modules(
    tm_t_coxreg(
      label = "Cox Reg.",
      dataname = "ADTTE",
      arm_var = choices_selected(c("ARM", "ARMCD", "ACTARMCD"), "ARM"),
      arm_ref_comp = arm_ref_comp,
      paramcd = choices_selected(
        value_choices(ADTTE, "PARAMCD", "PARAM"), "OS"
      ),
      strata_var = choices_selected(
        c("COUNTRY", "STRATA1", "STRATA2"), "STRATA1"
      ),
      cov_var = choices_selected(
        c("AGE", "BMRKR1", "BMRKR2", "REGION1"), "AGE"
      ),
      multivariate = TRUE
    )
  )
)
#> [INFO] 2022-10-14 09:10:28.9758 pid:3139 token:[] teal.modules.clinical Initializing tm_t_coxreg
if (FALSE) {
shinyApp(app$ui, app$server)
}


## Second example
## ==============
## This time, a synthetic pair of ADTTE/ADSL is fabricated for a Cox regression
## where ties and pval_method matter.

## Dataset fabrication
## -------------------

ADTTE <- data.frame(
  STUDYID = "LUNG",
  AVAL = c(4, 3, 1, 1, 2, 2, 3, 1, 2),
  CNSR = c(1, 1, 1, 0, 1, 1, 0, 0, 0),
  ARMCD = factor(
    c(0, 1, 1, 1, 1, 0, 0, 0, 0),
    labels = c("ARM A", "ARM B")
  ),
  SEX = factor(
    c(0, 0, 0, 0, 1, 1, 1, 1, 1),
    labels = c("F", "M")
  ),
  INST = factor(c("A", "A", "B", "B", "A", "B", "A", "B", "A")),
  stringsAsFactors = FALSE
)
ADTTE <- base::rbind(ADTTE, ADTTE, ADTTE, ADTTE)
ADTTE <- dplyr::as_tibble(ADTTE)
set.seed(1)
ADTTE$INST <- sample(ADTTE$INST)
ADTTE$AGE <- sample(seq(5, 75, 5), size = nrow(ADTTE), replace = TRUE)
ADTTE$USUBJID <- paste("sub", 1:nrow(ADTTE), ADTTE$INST, sep = "-")
ADTTE$PARAM <- ADTTE$PARAMCD <- "OS"
ADSL <- subset(
  ADTTE,
  select = c("USUBJID", "STUDYID", "ARMCD", "SEX", "INST", "AGE")
)

## Teal application
## ================
## Note that the R code exported by `Show R Code` does not include the data
## preprocessing. You will need to create the dataset as above before
## running the exported R code.

arm_ref_comp <- list(ARMCD = list(ref = "ARM A", comp = c("ARM B")))
app <- init(
  data = cdisc_data(
    cdisc_dataset(
      dataname = "ADSL",
      x = ADSL
    ),
    cdisc_dataset(
      dataname = "ADTTE",
      x = ADTTE
    )
  ),
  modules = modules(
    tm_t_coxreg(
      label = "Cox Reg.",
      dataname = "ADTTE",
      arm_var = choices_selected(c("ARMCD"), "ARMCD"),
      arm_ref_comp = arm_ref_comp,
      paramcd = choices_selected(
        value_choices(ADTTE, "PARAMCD", "PARAM"), "OS"
      ),
      strata_var = choices_selected(c("INST"), NULL),
      cov_var = choices_selected(c("SEX", "AGE"), "SEX"),
      multivariate = TRUE
    )
  )
)
#> [INFO] 2022-10-14 09:10:29.1329 pid:3139 token:[] teal.modules.clinical Initializing tm_t_coxreg
if (FALSE) {
shinyApp(app$ui, app$server)
}