Bioequivalence and Bioavailability Forum

Dear MGR!

❝ Thanks for your reply.

You are welcome!

❝ And a small doubt, you told that bartlett test is not recommended by FDA.

OK, maybe my wording was ambigious (English is not my native language)...
Maybe better would have been:
'Bartlett's Test' tests for equal variances (which is not recommended by the FDA).
Have a look at FDA's Guideline on http://www.fda.gov/cder/guidance/3616fnl.pdf Statistical Approaches Establishing Bioequivalence (VI. STATISTICAL ANALYSIS, B. Data Analysis, 1. Average Bioequivalence, d. Parallel Designs):
'For parallel designs, the confidence interval for the difference of means in the log scale can be computed using the total between-subject variance. As in the analysis for replicated designs (section VI. B.1.b), equal variances should not be assumed.'
In other words you should not test the assumption of equal variances - what Bartlett would do - but simply not assume equal variances.

❝ So which test is suitable for the Analysis if it is "An open label, balanced, randomized, single treatment, single period, single dose, parallel, bioequivalence study".

One-way ANOVA on log-transformed data, confidence intervals for treatment difference(s) (see above).
From the title it's still not clear to me how many treatments you will have in your study. Only if you will have more than two treatments (e.g. 2 tests and 1 reference), you should go with Duncan's Test (multiple comparisons).

❝ Since it is a pivotal study, which test is recommended by the FDA for this type of study.

If you go with the FDA, they will have to approve your protocol anyway

Good luck,
Hermann

---

Edit: Link corrected for FDA’s new site. [Helmut]