Cumulative Numerator Event Count per Denominator Event Count

Calculate a subject level cumulative numerator event count per denominator event count.

This function takes in a list of data frames including dfSUBJ, dfNumerator, and dfDenominator, and calculates a subject level cumulative numerator count per cumulative denominator count. Numerator events between two denominator events are assigned to the later denominator event.

Numerator events before first denominator event will be assigned to the first denominator event. Numerator events after last denominator event will be assigned to last denominator event.

When numerator events fall on the same day as a denominator event they will be assigned to that event event if the date time columns imply that numerator event occurred after the denominator.

The data requirements for the function are as follows:

dfSubjects: A data frame with columns for SubjectID and any other relevant subject information
dfNumerator: A data frame with a column for SubjectID and strNumeratorCol if strNumeratorMethod is "Sum"
dfDenominator: A data frame with a column for SubjectID and strDenominatorCol if strDenominatorMethod is "Sum"

All other columns are dropped from the output data frame. GroupIDs will be added via dfSubjects if not present events without GroupID and subjects without denominator or numerator event will be dropped.

Numerator events with no SubjectID but with GroupID can get assigned to a random enrolled subject of the same GroupID

For terminal events such as discontinuations pass planned visits instead of actual visits as dfDenominator to avoid survivor-bias. strInstanceNameCol, vLikePatternInstanceName, and nMinSubjectRatioInstance can be used to attempt to extrapolate planned visits. See ExtrapolateDenominator() for details.

Usage

Input_CumCount(
  dfSubjects,
  dfNumerator,
  dfDenominator,
  strGroupCol,
  strGroupLevel = NULL,
  strSubjectCol,
  strNumeratorCol = NULL,
  strDenominatorCol = NULL,
  strNumeratorDateCol,
  strDenominatorDateCol,
  strOrphanedMethod = c("filter", "assign"),
  strInstanceNameCol = NULL,
  vLikePatternInstanceName = c("%unsch%", "%disc%"),
  nMinSubjectRatioInstance = 0.7
)

Arguments

dfSubjects: data.frame with columns for SubjectID and any other relevant subject information
dfNumerator: data.frame with a column for SubjectID and strNumeratorDateCol
dfDenominator: data.frame with a column for SubjectID and strDenominatorDateCol
strGroupCol: character Column name in dfSubjects to use for grouping.
strGroupLevel: character value for the group level. Default: NULL defaults to strGroupCol
strSubjectCol: character Column name in dfSubjects to use for subject ID.
strNumeratorCol: character Column name in dfNumerator to use for numerator ID. Default: NULL
strDenominatorCol: character Column name in dfDenominator to use for denominator ID. Default: NULL
strNumeratorDateCol: character Column name in dfNumerator to use for numerator calculation.
strDenominatorDateCol: character Column name in dfDenominator to use for denominator calculation.
strOrphanedMethod: character one_of("filter", "assign") filter orphaned numerator events or assign to random patient enrolled at site the time of event. Default: "filter"
strInstanceNameCol: character When provided will extrapolate planned Numerator events e.g. visits. Use for terminal binary events such as patient discontinuations to avoid survival bias. Default: NULL
vLikePatternInstanceName: character vector vector of sql like patterns to filter instance names for extrapolation. Default: c("%unsch%", "%disc%")
nMinSubjectRatioInstance: numeric, minimum subject ratio per instance name to consider for extrapolation. Default: 0.7

Value

data.frame with the following specification:

Column Name	Description	Type
SubjectID	The subject ID	Character
GroupID	The group ID	Character
GroupLevel	The group type	Character
Numerator	Cumulative Count Numerator	Numeric
Denominator	Cumulative Count Denominator	Numeric

Examples


# dfSubjects tibble with one subject and one site
dfSubjects <- tibble::tibble(
  SubjectID = 1,
  GroupID = 1
)

dfNumerator <- tibble::tibble(
  SubjectID = rep(1, 10),
  num_dt = as.Date("2000-01-01") + c(months(0:4), rep(months(7), 2), months(9:11)),
) %>%
  dplyr::mutate(
    num_dt = num_dt + lubridate::hours(12)
  )

# dfNumerator tibble with one subject 10 AEs, two of which on same day
dfNumerator
#> # A tibble: 10 × 2
#>    SubjectID num_dt             
#>        <dbl> <dttm>             
#>  1         1 2000-01-01 12:00:00
#>  2         1 2000-02-01 12:00:00
#>  3         1 2000-03-01 12:00:00
#>  4         1 2000-04-01 12:00:00
#>  5         1 2000-05-01 12:00:00
#>  6         1 2000-08-01 12:00:00
#>  7         1 2000-08-01 12:00:00
#>  8         1 2000-10-01 12:00:00
#>  9         1 2000-11-01 12:00:00
#> 10         1 2000-12-01 12:00:00

dfDenominator <- tibble::tibble(
  SubjectID = rep(1, 4),
  denom_dt = c(as.Date(c("2000-01-03", "2000-04-12", "2000-08-01", "2000-11-12")))
) %>%
  dplyr::mutate(
    denom_dt = denom_dt + lubridate::hours(1)
  )

# dfDenominator tibble with one subject 4 visits, one on same day as two Numerator events
# Denominator time indicates that they occurr before Numerator events
dfDenominator
#> # A tibble: 4 × 2
#>   SubjectID denom_dt           
#>       <dbl> <dttm>             
#> 1         1 2000-01-03 01:00:00
#> 2         1 2000-04-12 01:00:00
#> 3         1 2000-08-01 01:00:00
#> 4         1 2000-11-12 01:00:00


# numerator before first denominator rolls up to first denominator
# numerator after last denominator rolls back to last denominator
# two numerator events on 08-01 add to the 3rd denominator event on same day

Input_CumCount(
  dfSubjects = dfSubjects,
  dfNumerator = dfNumerator,
  dfDenominator = dfDenominator,
  strSubjectCol = "SubjectID",
  strGroupCol = "GroupID",
  strGroupLevel = "Site",
  strNumeratorDateCol = "num_dt",
  strDenominatorDateCol = "denom_dt"
)
#> # A tibble: 4 × 5
#>   SubjectID GroupID GroupLevel Numerator Denominator
#>       <dbl>   <dbl> <chr>          <dbl>       <dbl>
#> 1         1       1 Site               1           1
#> 2         1       1 Site               4           2
#> 3         1       1 Site               7           3
#> 4         1       1 Site              10           4



# {clindata} Example for cumulative AE per Visit Count
Input_CumCount(
  dfSubjects = clindata::rawplus_dm,
  dfNumerator = clindata::rawplus_ae,
  dfDenominator = clindata::rawplus_visdt %>% dplyr::mutate(visit_dt = lubridate::ymd(visit_dt)),
  strSubjectCol = "subjid",
  strGroupCol = "siteid",
  strGroupLevel = "Site",
  strNumeratorDateCol = "aest_dt",
  strDenominatorDateCol = "visit_dt"
)
#> # A tibble: 28,074 × 5
#>    SubjectID GroupID GroupLevel Numerator Denominator
#>    <chr>     <chr>   <chr>          <dbl>       <dbl>
#>  1 0025      10      Site               0           1
#>  2 0025      10      Site               0           2
#>  3 0025      10      Site               0           3
#>  4 0025      10      Site               0           4
#>  5 0025      10      Site               0           5
#>  6 0025      10      Site               0           6
#>  7 0025      10      Site               0           7
#>  8 0025      10      Site               0           8
#>  9 0025      10      Site               0           9
#> 10 0025      10      Site               0          10
#> # ℹ 28,064 more rows