Custom PKNCA_impute_method_start: does start need to be a vector?

[Gero1999]

Regarding the PKNCA_impute_method I just wanted to make sure if there is any potential case/reason on why these functions define start argument as a possible vector. Could it be instead of time %in% start just time == start? I guess that if the function is performed on each interval separately during pk.nca() should be fine right?

Also I did not manage to know what does the variable name ret apply-for/refer-to. I think for future deploy this could be always helpful to know!

PKNCA_impute_method_start_conc0 <- function(conc, time, start=0, ..., options = list()) {
  ret <- data.frame(conc = conc, time = time)
  mask_start <- time %in% start
  if (any(mask_start)) {
    ret$conc[mask_start] <- 0
  } else {
    ret <- rbind(ret, data.frame(time = start, conc = 0))
    ret <- ret[order(ret$time), ]
  }
  ret
} 

[billdenney]

Good questions.

I tend to use %in% instead of == reflexively to be safe for NA values. time should never be NA at this point in the code, though. So, that's is also my habit more than being necessary for that reason here.

(In theory, start could be a vector if you had a full washout and many times for starting single dose administration. For example, an animal study with a 2 hour half-life drug and dosing separated by 7 days could have start as c(0, 168). That said looking at the code, it wouldn't work if there were one time that matched start and one that did not.)

And, ret is the name that I typically give for the intended return value of any function. There's nothing more in its name. It helps me track that "this will be returned" and visually confirm that the return value is managed in each part of if blocks, for instance. A name like d_conc_time_impute could also have been used (I tend to prefix data frame variable names with "d_", but that practice is newer than most PKNCA code.)

[Gero1999]

That makes sense, thanks Bill!