Bioequivalence and Bioavailability Forum

Hi Joy,

❝ We have removed “magic” time allowance window (TAW) from our protocol. So every blood sampling deviation, will be reported in CRF and statistical analysis.

I think we disagree in terminology. Samples with deviations within the TAW don’t have to be com­mented in the CRF. However, the actual time point will always be recorded and used in PK. If you mandate to comment every deviation (even if considered irrelevant) you put stress on the clinical staff. They should concentrate on handling samples – not on prose.

❝ But one of our customer, ask us, to put TAW in our protocol. Since we can't justify the deviation, we can't accept the request.

I don’t understand. Can you elaborate?

❝ What I know from training, TAW will be specific for any kind of drug molecules, it can't 5%, more or less, depend on the impact to pharmacokinetic analysis.

Maybe, maybe not. I’m not aware of any paper exploring it. Papst wrote:

[…] only those time deviations will be of pharmacokinetic relevance in which the time interval between the preceding and the affected sample or the interval between the affected and the following sample was prolonged or shortened by, for example, more than 5%. In practice it has been proven acceptable to consider only those time deviations out of this specified range when calculating the AUC.
This procedure for calculating pharmacokinetic parameters by ignoring any time deviations of less than 5% of the shorter of the two time intervals surrounding the sample affected has been shown by simulations to lead to estimates that differ at the most by 1.5% from the theo­re­tical value. In unfavourable circumstances (very few samples, all deviations in one direction) the relative error may increase to 3.6%). The simulation furthermore showed that a well-chosen design (e.g. with respect to sampling times) is more important than other factors (e.g. theo­re­tical profile, analytical accuracy).

But: He compared PK parameters calculated from scheduled times (i.e., ignoring deviations) with ones calculated from actual times. Only the latter method is recommended in current guidelines anyway. Therefore, this issue is history.

❝ But we don't how to decide the percentage of blood sampling deviation which can't make an effect to pharmacokintetic analysis.

Use the actual times.

❝ Would you mind to share how to decide the percentage of blood sampling which can't categorize as deviation, so TAW won't be "magic" again, but scientifically sound.

I would keep it at 5% (yes, arbitrarily!). The idea to comment on larger deviations is sumfink like this:

“Hhm, this concentration looks really weird. It is much lower than the ones before and after. Let’s have a look at the CRF. Ah, the venflow was blocked and it took the study nurse 10 minu­tes to rinse it with Ringer’s solution. Maybe she didn’t discard the first milliliters and the sample was diluted? I will go downstairs and ask her.”

BTW, things like this are pretty common. Train the staff not to reach for the stars. If a venflow cannot be opened quickly (say within two minutes), opt for venipuncure at this timepoint (make sure to mention it in the ICF). They can deal with the blockage later on.

Dear Joy,

What I often see in protocols is that any time deviation of more than a certain value will be considered as a protocol deviation and will be reported as such in the study report. But all time deviations are reported and are considered in the PK analysis.

I have seen strange interpretations of such protocol provisions: allowed deviation ± 2 min, actual deviation + 5 min, reported deviation + 3 min...

I agree with Helmut: there is no reason why any sample should be collected early. Meaning that any sample collected early should be considered as a protocol deviation. After all, if a subject arrives late for an ambulatory sample there is little you can do; but if he arrives early you can ask him to wait. He was informed of the protocol constraints and agreed to them when signing the consent form.