Rate Differences for Stacked Hierarchical ARDs

Calculate the difference in event rates between two groups of a stacked hierarchical ARD created with ard_stack_hierarchical(). For every node in the hierarchy (e.g. each system organ class and each preferred term), the rate (p statistic) of a second group is subtracted from the rate of a first group and returned under a new statistic named "estimate".

The input ARD must contain the p statistic and at least one by variable. The by (grouping) dimension is collapsed in the result—only the estimate rows are returned.

diff_ard_hierarchical(x, levels = NULL)

Arguments

x

(card)
a stacked hierarchical ARD of class 'card' created using ard_stack_hierarchical(). Must contain the p statistic and at least one by variable.

levels

(list)
a list specifying the two groups being differenced, p(group 1) - p(group 2). Required unless x has a single by variable with exactly two levels, in which case the difference defaults to the first level minus the second level (in the order the levels appear in the source data). Two forms are accepted:

• a flat named list keyed by a single by variable whose name is repeated to give its two levels in order, e.g. list(TRTA = "Xanomeline High Dose", TRTA = "Placebo"). When x has more than one by variable, every other by variable must also be named once to pin it to a single level.

• a list of two named lists, each fully specifying one group's combination of by variable levels, e.g. list(list(TRTA = "Xanomeline High Dose"), list(TRTA = "Placebo")) or, with two by variables, list(list(TRTA = "Xanomeline High Dose", SEX = "F"), list(TRTA = "Placebo", SEX = "F")). This form is required when x has more than one by variable.

Defaults to NULL.

Value

an ARD data frame of class 'card'

Details

Rates are stored on the [0, 1] scale (as p is), so the returned estimate values lie in [-1, 1]. A hierarchy node observed in only one of the two groups is treated as a rate of 0 in the group where it is absent.

See also

sort_ard_hierarchical(), filter_ard_hierarchical()

Examples

ard <- ard_stack_hierarchical(
  ADAE,
  variables = c(AESOC, AEDECOD),
  by = TRTA,
  denominator = ADSL,
  id = USUBJID
)

# difference between two of the three treatment arms
diff_ard_hierarchical(ard, levels = list(TRTA = "Xanomeline High Dose", TRTA = "Placebo"))
#> # An ARD data frame: 265 × 11
#>    group1 group1_level      variable variable_level    context stat_name    stat
#>    <chr>  <list>            <chr>    <list>            <chr>   <chr>      <list>
#>  1 NA     <NULL>            AESOC    CARDIAC DISORDERS diff_h… estimate   0.0631
#>  2 AESOC  CARDIAC DISORDERS AEDECOD  ATRIAL FIBRILLAT… diff_h… estimate   0.0241
#>  3 AESOC  CARDIAC DISORDERS AEDECOD  ATRIAL FLUTTER    diff_h… estimate   0.0119
#>  4 AESOC  CARDIAC DISORDERS AEDECOD  ATRIAL HYPERTROP… diff_h… estimate  -0.0116
#>  5 AESOC  CARDIAC DISORDERS AEDECOD  ATRIOVENTRICULAR… diff_h… estimate  -0.0116
#>  6 AESOC  CARDIAC DISORDERS AEDECOD  ATRIOVENTRICULAR… diff_h… estimate   0.0125
#>  7 AESOC  CARDIAC DISORDERS AEDECOD  BRADYCARDIA       diff_h… estimate  -0.0116
#>  8 AESOC  CARDIAC DISORDERS AEDECOD  BUNDLE BRANCH BL… diff_h… estimate  -0.0116
#>  9 AESOC  CARDIAC DISORDERS AEDECOD  BUNDLE BRANCH BL… diff_h… estimate  -0.0116
#> 10 AESOC  CARDIAC DISORDERS AEDECOD  CARDIAC DISORDER  diff_h… estimate   0.0119
#> # ℹ 255 more rows
#> # ℹ 4 more variables: stat_label <chr>, fmt_fun <list>, warning <list>,
#> #   error <list>

# equivalent, using the two-cell form
diff_ard_hierarchical(
  ard,
  levels = list(list(TRTA = "Xanomeline High Dose"), list(TRTA = "Placebo"))
)
#> # An ARD data frame: 265 × 11
#>    group1 group1_level      variable variable_level    context stat_name    stat
#>    <chr>  <list>            <chr>    <list>            <chr>   <chr>      <list>
#>  1 NA     <NULL>            AESOC    CARDIAC DISORDERS diff_h… estimate   0.0631
#>  2 AESOC  CARDIAC DISORDERS AEDECOD  ATRIAL FIBRILLAT… diff_h… estimate   0.0241
#>  3 AESOC  CARDIAC DISORDERS AEDECOD  ATRIAL FLUTTER    diff_h… estimate   0.0119
#>  4 AESOC  CARDIAC DISORDERS AEDECOD  ATRIAL HYPERTROP… diff_h… estimate  -0.0116
#>  5 AESOC  CARDIAC DISORDERS AEDECOD  ATRIOVENTRICULAR… diff_h… estimate  -0.0116
#>  6 AESOC  CARDIAC DISORDERS AEDECOD  ATRIOVENTRICULAR… diff_h… estimate   0.0125
#>  7 AESOC  CARDIAC DISORDERS AEDECOD  BRADYCARDIA       diff_h… estimate  -0.0116
#>  8 AESOC  CARDIAC DISORDERS AEDECOD  BUNDLE BRANCH BL… diff_h… estimate  -0.0116
#>  9 AESOC  CARDIAC DISORDERS AEDECOD  BUNDLE BRANCH BL… diff_h… estimate  -0.0116
#> 10 AESOC  CARDIAC DISORDERS AEDECOD  CARDIAC DISORDER  diff_h… estimate   0.0119
#> # ℹ 255 more rows
#> # ℹ 4 more variables: stat_label <chr>, fmt_fun <list>, warning <list>,
#> #   error <list>