Applying multilevel models in evaluation of bioequivalence(Ⅰ)
10.3760/cma.j.issn.0254-6450.2009.12.023
- VernacularTitle:多水平模型在生物等效性评价中的应用(Ⅰ)
- Keywords:
Multilevel model;
Bioequivalence;
Cross-over test design;
Variance components
- From:
Chinese Journal of Epidemiology
2009;30(12):1302-1306
- CountryChina
- Language:Chinese
-
Abstract:
This study aims to explore the application value of multilevel models for bioequivalence evaluation.Using a real example of 2×4 cross-over experimental design in evaluating bioequivalence of antihypertensive drug,this paper explores complex variance components corresponding to criteria statistics in existing methods recommended by FDA but obtaines in multilevel models analysis.Results are compared with those from FDA standard Method of Moments,specifically on the feasibility and applicability of multilevel models in directly assessing the bioequivalence(ABE),the population bioequivalence(PBE)and the individual bioequivalence (IBE).When measuring ln(AUC),results from all variance components of the test and reference groups such as total variance(σ_(TT)~2 and σ_(TR)~2),between-subject variance(σ_(BT)~2 and σ_(BR)~2)and within-subject variance(σ_(WT)~2 and σ_(WR)~2)estimated by simple 2-level models are very close to those that using the FDA Method of Moments.In practice,bioequivalence evaluation can be carried out directly by multilevel models,or by FDA criteria,based on variance components estimated from multilevel models.Both approaches produce consistent results.Multilevel models can be used to evaluate bioequivalence in cross-over test design.Compared to FDA methods,this one is more flexible in decomposing total variance into sub components in order to evaluate the ABE,PBE and IBE.Multilevel model provides a new way into the practice of bioequivalence evaluation.
