confidence intervals of brand vs generics

For example, the ratio of test drug to reference drug AUC (test/reference AUC) for Drug A in six subjects results in the following values:

120%, 120%, 110%, 110%, 90%, 110%

The mean test/reference AUC was 110% with 90% CI of 102 to 117. This means that there is 90% probability that the true mean lies between 102% and 117%. -ME-IT EITHER DOES OR DOES'T LIE IN THIS INTERVAL=> 9 OUT OF TEN TIMES IT WILL-Therefore, the drug passes the bioequivalence test and is considered bioequivalent to the reference drug.

In another example, the test/reference AUC for Drug B in six subjects results in the following values:

110%, 190%, 30%, 130%, 90%, 110%

The mean test/reference AUC was 110% with 90% CI of 75 to 145. Although the average test/AUC ratio of the drug (110%) is the same as drug A, there is a 90% probability that the true mean lies between 75% and 145%. The confidence intervals are too wide and therefore, Drug B fails to establish bioequivalence with the reference drug.

As illustrated above, the confidence interval range of 80% to 125% does not mean that drug levels can vary up to 20% between referenced (brand) drug and the test (generic) drug. If drug levels vary by more than 10%, the range of possible values within the 90% confidence interval will likely become too broad and the drug will fail to establish bioequivalence with the reference drug. 12 In fact, in a study using the FDA bioequivalence criteria, the first 224 post-1962 drugs approved over the two year period after the Waxman Hatch amendments were passed, including some narrow therapeutic index drugs, showed a mean bioavailability variation between the generic and brand products of only 3.5%.