Outliers in study intended for RSABE [Study As­sess­ment]

Hi NK,

❝ The protocol says, plasma samples from subjects who complete all the four periods will be analyzed and will be included for PK and statistical analysis.

Sounds strange to me. IMHO, all samples should be analyzed.* Excluding subjects with incomplete data from the statistical comparison is another cup of tea.

Subject A - had BLQ (Below limit of Quantification) for all the time points for one of Test treatment and

Subject B - had BLQ for all the time points in one of Reference treatment.

❝ Can we include Subject A & B for statistical analysis

Well, the statistical model is based on $$\small{\log_{e}}$$-transformed data. You have to exclude them, since $$\small{\log_{e}(0)}$$ is undefined (for nitpickers: $$\small{\lim \log_{e}(x\rightarrow 0) = -\infty}$$).

❝ (protocol not specified about handling subject with three periods data. (ie. one T & two R and one R & Two T).

AFAIK, the FDA requires analysis of complete data (see this post).

Subject C – had very low T/R Ratio (outlier was not mentioned in the protocol)

When it comes to RSABE, the FDA is not interested in assessing ‘outliers’. See this article.

❝ To handle this kind of situation in future, Can we mention in the protocol to exclude the subject (outlier) with above scenario.

Of course, you can. But I recommend a controlled correspondence with the OGD beforehand. Otherwise, you risk an R-t-R.

• Sometimes volunteers are afraid of not being recruited in later studies and hence, don’t report AEs. Instead, they withdraw consent ‘for personal reasons’.
There is a ‘classical’ example of methylphenidate. One subject was an ultra-slow metabolizer and his concentrations were 100 (‼) times higher than usual. Since MPH is a CNS stimulant, the subject had just a great time and stayed in the study. If the subject would have dropped out and his samples not analyzed, we would never have learned anything from it.

Dif-tor heh smusma 🖖🏼 Довге життя Україна!
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes