臺灣博碩士論文加值系統

美國食品藥物管理局於1999年6月提出鼻藥噴劑(如 : 過敏性鼻炎之用藥)體外試驗結果評估生體相等性的規範草案 "Bioavailability and Bioequivalencestudies for Nasal Aerosols and Nasal Sprays for Local Action" (www.fda.gov/cder/)。
草案中推薦使用卡方差值(chi-square differences)測度量作為兩藥物在鼻腔中和藥罐各指定部位比較分布比例的方法,然而, 該方法的統計性質並未完全被研究。 該草案建議使用經驗抽樣分配的95上百分位數(95% upper confidence limits of the empirical sampling distribution) 作為判定生體相等性與否的方法。但是本方法欲檢定的參數形式未知，以致於無法決定其上(Bioequivalence upper limit)。
本論文擬針對該草案中卡方差值測度量提供統計方面的理論評估。
例如: 若在多項式分配假設下, 該統計量的近似期望值, 漸進變異數和漸進分配雖可推導,但非常複雜，故提出另一些測度量作為判定生體相等性與否的方法。 最後,利用一模擬研究來評估這些方法的檢力和型I誤差, 並與FDA所建議之方法和Tsong(2000)建議之方法比較。

The Food and Drug Administration(FDA) draft guidance " Bioavailability and Bioequivalence Studies for Nasal Aerosols and Nasal Sprays for Local Action" (www.fda.gov/cder/) has recommended chi-square differences for in vitro profile comparison procedure. However, statistical properties of the method based on chi-square differences are not clear and not thoroughly investigated. Furthermore, the proposal in the draft guidance suggests the use of 95% upper confidence limits of the empirical sampling distribution for the conclusion of bioequivalence. The upper limit of this method can not determined since the function form of parameters in the hypothesis is unknown.
The objective of this paper is to provide a statistical assessment of in vitro profile bioequivalence and refers the drawback of FDA draft guidance.
Since the form of the criterion is complex, some statistical properties can be derived such as the asymptotic expectation and the asymptotic variance through the Taylor's expansion of chi-square difference at population proportions. But they are very complicated.
Then, some alternative procedures are proposed to assess the in vitro profile bioequivalence. Finally, a simulation study is conducted to investigate its empirical distribution. The FDA recommended criterion and several criteria proposed by Tsong(2000) are also compared.

Chapter 1 Introduction 1
Chapter 2 Current Statistical Procedures 5
2.1 FDA Criterion for Comparison 5
2.2 FDA Chi-square Differences 8
2.3 Weighted Mean Absolute Distance 14
2.4 Weighted Mean Square Distance 15
2.5 F2 Factor 17
2.6 LOGNORMAL DISTRIBUTION 19
Chapter 3 Proposed method 22
3.1 Procedures based on multinomial distribution 22
3.2 Procedures Based on Dirichlet distribution 26
3.3 Procedures Based on Compound distribution 27
3.4 Other Proposed statistical procedures 28
Chapter 4 Simulation Study 33
4.1 The Simulation processes 33
4.2 The drawback of lognormal distribution 35
4.3 The simulation study under the other distributions 40
4.4 Simulation results 42
Chapter 5 Conclusion 46
Reference 50
Appendix A 51
Appendix B 53

1.FDA. Guidance for Industry: In vivo bioequivalence studies on population and individual bioequivalence approaches. Center for Drug Evaluation and Research ( CDER), Food and Drug Administration, US Department of Health and Human Services, October, 1997.
2.FDA. Guidance for Industry: Bioavailability and Bioequivalence studies for Nasal Aerosols and Nasal Sprays for Local Action. CDER, FDA, US Department of Health and Human Services, June, 1999.
3.Tsong, "Comparative Statistics for Assessing In-vitro Equivalence Based on Profile measures", Technical report of Orally Inhaled and Nasal Delivery Drug Products Subcommittee of the Advisory Committee for Pharmaceutical Science, 2000.
4.Hyslop T., Hsuan F., Holder D.J., "A small sample confidence interval approach to assess individual bioequivalence", Statistics in Medicine, 19:2885-2897, 2000.
5.EDWIN L. CROW & KUNIO SHIMIZU., Lognormal distributions: theory and applications, MARCEL DEKKER, INC., 1988.
6.Agresti, A., "Categorical data analysis", Wiley, Inc., New York, 1990.