Bioequivalence and Bioavailability Forum

Merhaba BRN,

❝ I used the default settings of average crossover design. Details of sample collection is below.

❝ After collecting the plasma from each subject who received the same product (T or R), for each time point, plasma samples of all subjects were mixed in a tube and this constituted an average concentration of that time point. Same was done for all time points. At the end I have one concentration for each time point for each product. Same was repeated in the period 2.

Are you trying to work according to this paper?

Hundt HKL, Schall R, Luus HG, Müller FO. Considerable cost savings through the analysis of pooled plasma samples in bioequivalence studies that fail to show bioequivalence. Int J Clin Pharm Ther Toxicol. 1993;21(7):331–6. PMID: 8370632

❝ While entering the data, I considered one pool as the first subject and the second pool as the second subject and tried to run the analysis with 2 subjects 2 periods. Why I'm getting this error message?

This can’t work; don’t blame the software. Homework: How many degrees of freedom do you expect?

❝ Could you please suggest how to proceed with the pooled sample analysis?

You could only work according to the paper, i.e. calculate the ratio of the two pooled AUCs and (assum­ing a maximum CV) following Table 1 in order to decide whether it makes sense to proceed with analyz­ing subject’s samples or stop the trial. This method was fashionable for a short period of time decades ago, but was quickly abandoned for the following reasons:

• You’ll get an unbiased estimate of the AUC-ratio only if there were no drop-outs in the study.
  
• Pooling samples is equivalent to calculating the arithmetic (!) mean of concentrations. It is ques­tion­able whether assuming a normal distribution is correct.
  
• No estimate of the C_max-ratio which – due to its generally higher CV – leads the power of the study. Ratio of means ≠ mean of ratios.
  
• The method might lead to stopping a study although it still might have passed – though with lower power than planned. IMHO, not analyzing subject’s samples based on dubious assumptions is not ethical.

Forget it. Doesn’t make sense.