Bioequivalence and Bioavailability Forum • Statistical Model for HVD Canada Regulatory

pash413
★

India,
2016-04-04 18:01
(3276 d 20:56 ago)

Posting: # 16164
Views: 6,138

Statistical Model for HVD Canada Regulatory [RSABE / ABEL]

Dear All

For HVD drugs recently Canada regulatory has published a notice proposing to adopt an average bioequivalence approach to HVDP with expanding limits based on the within-subject variability of the reference product.

As per general guidance of Canada mixed model is recommended for conventional two way crossover design.

Please suggest Which Statistical model to be followed for highly variable drugs (Canada Regulatory)?

I recommends widening of CI approach as recommended by EU, but do the Statistical analysis needs to be done same way i.e calculate Reference variability with separate model with only reference data and then analyse data with GLM and also check for outliers for Reference variability?

Or we can calculate reference variability as well as 90% CI with Mixed model (As per FDA for replicate study) and widen the acceptance limits as per reference variability, if necessary? :confused:

Edit: Category changed. [Helmut]

d_labes
★★★

Berlin, Germany,
2016-04-05 15:04
(3275 d 23:54 ago)

@ pash413
Posting: # 16165
Views: 5,098

Statistical model for replicate designs Canada

Post reply

Dear pash,

since your mentioned notice demands the use of replicate crossover designs for the evaluation of HVD/HVDP IMHO the answer is given in the guidance document Conduct and Analysis of Comparative Bioavailability Studies.

Section 2.7.4 Statistical Analysis states:
2.7.4.2 Model Fitting

... Higher order models must be analysed with the mixed model approach in order to estimate random effects properly.

...

2.7.4.4 Estimation of Random Effects

A summary of the estimates of inter-subject and intra-subject variances should be presented. For higher order designs estimates of subject by formulation and within formulation variance estimates should be given.

These statements are not compatible with the EMA recommendations.
Thus only the FDA approach (SAS code in Appendix E of the 2001 guidance "Statistical Approaches to Establishing Bioequivalence") is feasible I think.

—
Regards,

Detlew

pash413 ★ India, 2016-04-05 15:14 (3275 d 23:44 ago) @ d_labes Posting: # 16166 Views: 5,002	Statistical model for replicate designs Canada Post reply
	Thanks Detlew

Helmut
★★★

Vienna, Austria,
2016-04-09 16:24
(3271 d 22:34 ago)

@ pash413
Posting: # 16175
Views: 4,741

Reference?

Post reply

Hi pash,

❝ For HVD drugs recently Canada regulatory has published a notice proposing to adopt an average bioequivalence approach to HVDP with expanding limits based on the within-subject variability of the reference product.

Do you have a reference? I only know the archived notice Detlew mentioned. The FDA’s model sometimes fails to converge for data of the partial replicate design. Either use one of the full replicates designs (RTRT|TRTR or RTR|TRT) instead or change the random section where the structure of the variance-covariance matrix is specified from FA0(2) to FA0(1).

—
Dif-tor heh smusma 🖖🏼 Довге життя Україна!
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes