Skip to contents

Select a case-study trial from an MRCT simulation

Source: R/mrct_case_study.R
mrct_select_case.Rd

Reads a mrct_region_sims result run with record = "full" and returns the trials that meet the conditions of a regional-consistency case study: the trigger (the upper 95\ limit of the all-patient hazard ratio below itt_ub and the region's hazard ratio above region_itt), a declared subgroup \(\hat H\), and a Cox hazard ratio below 1.0 in region(\(\hat G\)), the region without \(\hat H\).

Usage

mrct_select_case(
  res,
  conditions = c("trigger", "declared", "region_Ghat_lt1"),
  which = c("first", "all"),
  itt_ub = 1,
  region_itt = 1
)

Arguments

res

A mrct_region_sims result with record = "full" (it carries the "estimators" attribute; take the attribute's owner, not a subset of it).

conditions

Character. The conditions a trial must meet, any of "trigger", "declared", "region_Ghat_lt1". Default: all three.

which

"first" (default): the first trial by index meeting the conditions; "all": every such trial.

itt_ub

Numeric. The trigger's bound on the all-patient upper 95\ limit (strictly below); Inf drops the bound. Default 1.0.

region_itt

Numeric. The trigger's bound on the region's hazard ratio (strictly above); -Inf drops the bound. Default 1.0.

Value

A data.frame with one row per selected trial (zero rows, with a message, when none qualifies): sim; the flags trigger, declared, region_Ghat_lt1 (logical); and the trial's hazard ratios hr_itt_all, itt_ub (its upper 95\ hr_itt_region, hr_region_Ghat and, when the record has the region_Gstar row, hr_region_Gstar; and sg_found. The attribute "n_meeting" is the number of trials meeting the conditions.

Details

The flags are evaluated per trial from the "estimators" attribute: the trigger from the itt_all row's hi and the itt_region row's hr; declared from the trial's any_found == 1; region(\(\hat G\)) < 1.0 from the region_Ghat row's hr. A missing hazard ratio never satisfies a condition. With which = "first" the result is the first trial by index meeting every condition in conditions: with the three default conditions, the rule that chose the case example of the mrct-consistency report (replicate 5 of its S2 cell).

See also

mrct_case_study to redraw and examine the selected trial; mrct_region_sims.

Examples

if (FALSE) { # \dontrun{
res <- mrct_region_sims(dgm, n_sims = 20, n_sample = 1000,
                        region_var = "z_region", seed = 613800000,
                        record = "full")
mrct_select_case(res)                    # the first qualifying trial
mrct_select_case(res, which = "all")     # every qualifying trial
} # }