Objective: The objective of the study was to evaluate the relationship
between the magnitude of the sample means for a bioavailability measu
re and the ratio of the sample means in bioequivalence evaluations. Me
thods: We assumed that bioavailability data were obtained from crossov
er trials on a test and a reference drug products. The true test-refer
ence ratio was constructed as a function of dosage strength under cert
ain dose-response relationship assumptions, and was used for deriving
analytical results of assessing the scale effect due to dosage strengt
h. In addition, the test-reference ratio of sample means was construct
ed as a function of the true test-reference ratio and the correlated s
ampling deviation in the sample means, and was used in a deterministic
simulation for quantifying the scale effect due to the correlated sam
pling deviation, Results: The major systematic factors that influence
the magnitude of the sample means include the dosage strength, dose-re
sponse relationship, and the positively correlated sampling deviation
in the sample means. Results of this study showed that the true test-r
eference ratios can be improved with increasing dosage strength in mos
t of the situations studied, but at a diminishing rate. The correlated
sampling deviation can alter the test-reference ratio of the sample m
eans through changing magnitude of the sample means, especially for dr
ug products with low bioavailability. Conclusions: For extended releas
e drug products with multiple dosage strengths, the development effort
s should focus on the lowest strength if there is little or Ilo formul
ation by dose interaction effect. Pilot studies should be conducted fo
r both the highest and the lowest strength instead for possible formul
ation by dose interaction effect. The magnitude of observed means for
bioavailability measures on the reference drug product can indicate th
e degree of systematic variation in the ratio of means. The expected m
ean values of the reference drug product can be derived from previous
trial(s) and/or from other information sources. If the sample mean for
the reference drug product in a failed bioequivalence trial is lower
than expected, the investigator should consider repeating the trial.