Skip to contents

Find the samples missing from each subject's nominal sampling schedule

Source: R/missing_samples.R
pknca_missing_samples.Rd

The schedule of a group is every nominal time that has a row in the concentration data of that group, where the group is every grouping variable of the concentration data except the subject (for example, the study, treatment, and analyte). Each subject of the group is expected to have a usable concentration at every time of its group's schedule, and a time without one is reported. Missing samples are therefore found only when some subject of the group has a row at that nominal time; a time that no subject has a row for is not part of the schedule. Recording a missed sample as a row with an NA concentration keeps the time in the schedule even when every subject missed it.

Usage

pknca_missing_samples(object)

Arguments

object

A PKNCAconc, PKNCAdata, or PKNCAresults object (the concentration data of a PKNCAdata or PKNCAresults object are checked). Its concentration data must have a nominal time (the time.nominal argument of PKNCAconc()).

Value

A data.frame with one row per missing sample and the columns: the grouping columns of the concentration data other than the subject, the subject column, the nominal time column (with the names they have in the concentration data), and reason. It is sorted by those columns and has no rows when no sample is missing.

Details

A nominal time of a subject's schedule is reported with one of these reasons:

  • "no row": the subject has no row at that nominal time.

  • "NA concentration": every row of the subject at that nominal time has an NA concentration.

  • "excluded": the subject has a concentration at that nominal time, but every row with one is excluded (with the exclude column of PKNCAconc() or with exclude()), so the calculations do not use it.

A concentration below the limit of quantification (zero) is a usable sample. Rows with an NA nominal time (unscheduled samples) are not part of any schedule, and they neither add a time to it nor fill one. Nominal times are compared as they are, so they may be in any unit.

The nominal times must share one origin for each subject (for example, the time since the first dose). Nominal times that restart at each dose repeat for each dose, so one dose's samples would fill another dose's missing times. They are found when a subject's nominal times go back to the start of its schedule (its first nominal time after zero, or earlier) while its actual times increase; two samples drawn out of order are not a restart. The groups with such a subject are not reported, and a pknca_warning_missing_samples_nominal_restart warning names them.

Sparse data (see the sparse argument of PKNCAconc()) have no schedule that each subject follows: each subject gives a few of the group's samples by design. For sparse data, only the rows that exist are checked (reasons "NA concentration" and "excluded"), and a message says that absent rows are not reported.

See also

exclude_nca_tmax_coverage(), which excludes the results of a subject whose samples miss the Tmax range of its group

Other Result exclusions: exclude(), exclude_nca, pknca_exclude_rules()

Examples

d_conc <-
  data.frame(
    subject = rep(1:3, each = 4),
    time_nominal = rep(c(0, 1, 2, 4), 3),
    time = rep(c(0, 1, 2, 4), 3),
    conc = c(0, 5, 3, 1, 0, 4, NA, 1, 0, 6, 2, 1)
  )
# Subject 3 has no sample at 4 hours
d_conc <- d_conc[-12, ]
o_conc <- PKNCAconc(d_conc, conc ~ time | subject, time.nominal = "time_nominal")
pknca_missing_samples(o_conc)
#>   subject time_nominal           reason
#> 1       2            2 NA concentration
#> 2       3            4           no row