《BayesianHierarchicalModelsforDetectingSafetySignalsin检测安全信号的贝叶斯层次模型》由会员分享，可在线阅读，更多相关《BayesianHierarchicalModelsforDetectingSafetySignalsin检测安全信号的贝叶斯层次模型（40页珍藏版）》请在装配图网上搜索。
1、Click to edit Master title style,Click to edit Master text styles,Second level,Third level,Fourth level,Fifth level,*,Bayesian Hierarchical Models for Detecting Safety Signals in Clinical Trials,H.Amy Xia and Haijun Ma,Amgen,Inc.,MBSW 2021,Muncie,IN,March 20,2021,Disclaimer:The views expressed in th
2、is presentation represent personal views and do not necessarily represent the views or practices of Amgen.,Outline,Introduction,A motivating example,Bayesian Hierarchical Models,Meta analysis of Adverse Events data from multiple studies incorporating MedDRA structure,Incorporate patient level data,E
3、ffective graphics,Closing Remarks,Three-Tier System for Analyzing Adverse Events in Clinical Trials,Tier 1:Pre-specified Detailed Analysis and Hypothesis Testing,Tier 1 AEs are events for which a hypothesis has been defined,Tier 2:Signal Detection among Common Events,Tier 2 AEs are those that are no
4、t pre-specified and“common,Tier 3:Descriptive Analysis of Infrequent AEs,Tier 3 AEs are those that are not pre-specified and infrequent,Gould 2002&Mehrotra 2004,SPERT White Paper 2021,Multiplicity Issue in Detecting Signals Is Challenging,Detection of safety signals from routinely collected,not pre-
5、specified AE data in clinical trials is a critical task in drug development,Multiplicity issue in such a setting is a challenging statistical problem,Without multiplicity considerations,there is a potential for an excess of false positive signals,Traditional ways of adjusting for multiplicity such a
6、s Bonferroni may lead to an excessive rate of false negatives,The challenge is to develop a procedure for flagging safety signals which provides a proper balance between no adjustment versus too much adjustment,Considerations Regarding Whether Flagging an Event,Actual significance levels,Total numbe
7、r of types of AEs,Rates for those AEs not considered for flagging,Biologic relationships among various AEs,1,st,two are standard considerations in the frequentist approach.The 2,nd,two are not,but relevant in the Bayesian approach,-Berry and Berry,2004,Bayesian Work in Signal Detection,Spontaneous a
8、dverse drug reaction reports,Gamma Poisson Shrinker(GPS)on FDA AERS database(DuMouchel,1999),Bayesian Confidence Propagation Neural Network(BCPNN)on WHO database(Bate,et al.1998),Clinical trial safety(AE)data,Bayesian hierarchical mixture modeling(Berry and Berry,2004),Meta Analysis,Glass(1976),Meta
9、-analysis refers to a statistical analysis that combines the results of some collection of related studies to arrive a single conclusion to the question at hand,Meta-analysis based on,aggregate patient data(APD meta-analysis),Individual patient data(IPD)meta-analysis,Bayesian modeling is a natural c
10、hoice to incorporate the complex hierarchical structure of the data,George Chi,H.M.James Hung,Robert ONeill(FDA CDER),“Safety assessment is one area where frequentist strategies have been less applicable.Perhaps Bayesian approaches in this area have more promise.,(Pharmaceutical Report,2002),An Exam
11、ple,Data from four double-blind placebo-controlled studies on drug X.Study populations are similar.,Sample sizes:,After converting all AEs into same MedDRA version,reported AEs are coded to 464 PTs under 23 SOCs and 233 HLTs,Study,Drug X,N,Drug X,Subj-yr,Placebo,N,Placebo,Subj-yr,Study A,57,28.25,55
12、,19.02,Study B,486,104.75,166,34.93,Study C,390,85.44,193,40.97,Study D,312,68.78,306,65.91,N_0:sample size in placebo arm;N_1:sample size in treatment arm,n_0:#subject with AE in placebo arm;n_1:#subject with AE in treatment arm,rt_0:subject incidence in placebo arm;rt_1:subject incidence in treatm
13、ent arm,Proposed Bayesian Approach,Hierarchical mixture models for aggregated binary responses was constructed based on the work by Berry&Berry(2004),Explore impact of using different MedDRA hierarchy,Inclusion of study effects,Further extended to a hierarchical Poisson mixture model,to account for
14、different exposure/follow-up times between patients,Individual patient level models are discussed,Implemented the above models with available software,WinBUGS for model implementation,S-Plus graphics for inference,MedDRA,MedDRA(the Medical Dictionary for Regulatory Activities Terminology)is a contro
15、lled vocabulary widely used as a medical coding scheme.,MedDRA Definition(MSSO):,MedDRA is a clinically-validated international medical terminology used by regulatory authorities and the regulated biopharmaceutical industry.The terminology is used through the entire regulatory process,from pre-marke
16、ting to post-marketing,and for data entry,retrieval,evaluation,and presentation.,MSSO:Introduction to MedDRA,MedDRA and Pharmacovigilance-The Way Forward,7/8/99,MedDRA and Pharmacovigilance-The Way Forward,7/8/99,SOC=Respiratory,thoracic and,mediastinal disorders,HLGT=Respiratory tract,infections,HLT=Viral upper respiratory,tract infections,HLT=Influenza viral,infections,HLGT=Viral infectious,disorders,SOC=Infections and,infestations,PT=Influenza,Example of MedDRA Hierarchy,MSSO:Introduction to
BayesianHierarchicalModelsforDetectingSafetySignalsin检测安全信号的贝叶斯层次模型
