Bioequivalence and Bioavailability Forum

Dear Bear inventors,

I would suggest a change in your statistical summary output.
Here a piece:
         Statistical Summaries for Bioequivalence Study (N= 14 )
--------------------------------------------------------------------------

    parameters Test_Mean  Test_SD  Ref_Mean   Ref_SD Ratio
1         Cmax  1630.786  197.522  1539.643  220.015 1.079
2       AUC0-t 13975.284 1997.852 13158.790 1971.053 1.096
3     AUC0-inf 14659.822 2092.321 13776.930 2033.362 1.097
4     ln(Cmax)     7.390    0.122     7.329    0.147 1.009
5   ln(AUC0-t)     9.535    0.145     9.473    0.161 1.007
6 ln(AUC0-inf)     9.583    0.144     9.520    0.159 1.007

The ratio of mean_Test/mean_Ref. in case of log-transformed values does not make sense.
It is misleading as it may be interpreted as point estimator of the bioavailability ratio by some 'naive' users. This is especially true because the following part, the 90% confidence intervals, does nor give any point estimator.

My suggestion therefore is:

1. Do not give the ratio in case of log-transformed values. Include the point estimator of bioavailability ratio as exp(LSMean_T-LSMean_R) in the part with the 90% CIs.

or

2. Give the ratio in case of log-transformed values as exp(mean_T-mean_R) as an approximation to the biovailability ratio. Include the point estimator of bioavailability ratio as exp(LSMean_T-LSMean_R) in the part with the 90% CIs.
Both estimates are identical only in case of balanced studies.
But giving both may confuse the 'naive' users or even regulators.

Dear Yung-jin,
at all great job.