优化疫苗的安全性警戒 疫苗上市后安全性评价.ppt

　　Optimal Vaccine Safety PharmacovigilanceAssessing safety of vaccines post introduction 优化疫苗的安全性警戒疫苗上市后安全性评价

　　Steven Black, MD Professor of Pediatrics Cincinnati Children’s Hospital Cincinnati, Ohio USA

　　Monitoring the effects of vaccines or drugs after they have been licensed for use, especially in order to identify and evaluate previously unreported adverse reactions. Key points Builds on what is known from pre-licensure studies Is likely to be dealing with events that were to rare to detect in pre-licensure studies or events in special populations. 疫苗或药物上市使用后监测其效果， 特别是要识别和评估以前未报告的不良反应 关键点 基于注册前研究所获得的数据 很可能会涉及那些罕见的、在注册研究中未发现的事件，或在特殊人群中发生的事件。

　　什么是药物警戒?What is pharmacovigilance?

　　2

　　Why is this important?

　　3

　　为什么药物警戒很重要?

　　不安全的疫苗可带来严重后果：安全危机会毁掉免疫规划项目Unsafe vaccines can have serious consequences: Safety crises derail immunization programs

　　Real incidents: Tuberculosis following oral BCG Polio following IPV (Cutter Incident) Swine flu campaign and GBS Real safety issues: Program errors Anaphylaxis VAPP following OPV Disseminated BCG disease in HIV infected Rumours, poor science and over-reaction: Pertussis vaccine coverage in the UK MS and hepatitis B vaccine in France OPV and chronic diseases in Nigeria Thimerisal and neuro-developmental disorders Pentavalent vaccine in Asian countries

　　真实案例: 口服卡介苗后罹患结核 接种IPV(Cutter事件)后发生脊髓灰质炎 猪流感疫苗接种运动与GBS 真正的安全性问题: 项目出错 过敏反应 OPV后发生VAPP HIV感染者接种卡介苗后发生的播散性疾病 谣言、低质量科研和过度反应: 英国百日咳疫苗的接种率 法国MS(多发性硬化)和乙肝疫苗 尼日利亚OPV与慢性病 柳柳汞与神经发育障碍 亚洲国家五价疫苗问题

　　Confidence in Vaccines is Facing New Challenges对疫苗的信心面临新挑战

　　在另一个讲座中会介绍更多关于这方面的信息 儿童感染下降导致对疾病后果的认识缺乏，疫苗需求的认知下降 谣言和错误信息降低对疫苗的信心，降低疫苗接种率 优化疫苗-药物警戒可以及时对关注点进行应对

　　You will hear more about this in another session Decline of childhood infections leads to lack of awareness of disease consequences and less perceived need for vaccines. Rumors and Misinformation lowers confidence and lowers vaccine coverage Optimal vaccine pharmacovigilance can respond to concerns in a timely manner.

　　Diagram adapted from Chen RT et al. The Vaccine Adverse Event Reporting System (VAERS). Vaccine, 1994: 12(6): 542-550

　　疫苗接种项目的阶段

　　第一阶段 疫苗前时期

　　第二阶段 疫苗接种率增加

　　第三阶段 对疫苗信心丧失

　　第四阶段 对疫苗信心恢复

　　第五阶段 消灭

　　疾病

　　疫苗接种率增加

　　发病

　　暴发

　　停止疫苗

　　消灭

　　免疫接种项目成熟

　　不良事件

　　数目或感知的不良事件

　　Real Example: Pertussis in the U.K.真实案例：英国百日咳

　　Gay N, Miller E, Lancet 2000;355:1553

　　So how do you do optimal pharmacovigilance?那么，怎样优化药物警戒?

　　被动报告系统: 以VAERS为例 基于人群的安全性评估: 四期研究、数据链接、登记和快速循环研究 事件背景发生率与疫苗相关的发生率 被动与主动监测 全球合作努力

　　Passive Reporting Systems: VAERS as an example Population based evaluations of safety: Phase Four studies, Datalink, registry and rapid cycle studies Background rates of events versus vaccine related events Passive versus Active Surveillance Global collaborative efforts

　　Passive Reporting: VAERS as an example被动报告系统: 以VAERS为例

　　VAERS(疫苗不良事件报告系统)依赖于医生或患者的被动报告。没有真正的分母数据 已经开发出统计技术，用“数据挖掘”进行信号检测 依赖贝叶斯统计量比较预期的疫苗事件数与观察的事件数 在评估这项技术的回顾性分析中，早期就检出了接种轮状病毒疫苗RotaShield后发生肠套叠的事件 下述事件的EBGM (经验贝叶斯几何平均)得分提高 肠套叠 腹泻，消化道出血

　　VAERS (Vaccine Adverse Event Reporting System) relies on passive reporting by physicians or patients. No real denominator data. Statistical techniques have been developed for signal detection using “data mining”). Relies on comparing expected versus observed numbers of events following vaccine using Bayesian statistics. In a retrospective analysis evaluating this technique, intussusception following Rotashield vaccine identified early. EBGM ( Empirical Bayesian Geometric Mean) score elevated for Intussusception Diarrhea, GI hemorrhage

　　10

　　Passive Reporting: Intussusception Rotashield vaccine Cases Reported to VAERS被动报告:肠套叠与Rotashield疫苗报至VAERS的肠套叠事件

　　MMWR

　　病例数

　　11

　　Empirical Bayesian Geometric Mean Score following Rotashield接种Rotashield后的经验贝叶斯几何平均得分(EBGM)

　　1

　　Passive Reporting: Intussusception and VAERS被动报告: 肠套叠与VAERS

　　VAERS在检出肠套叠信号方面发挥了良好的作用 统计技术可以帮助早期检出疫苗的安全性信号 但被动报告系统： 往往优先报告时间上最接近疫苗的事件 漏报为一个主要问题 可能发生事件的错分 事件的信息可能有限 公认的不良事件更可能被报告 不能够评估相对或归因风险

　　VAERS was excellent at signal detection for intussusception. Statistical techniques can lead to earlier detection of vaccine safety signals. Passive reporting systems, however: Are prone to bias in that events closest to vaccine are reported preferentially Underreporting is a major issue. Misclassification of events can occur. Information on cases may be limited. Recognized adverse events are more likely to be reported. Do not allow assessment of relative or attributable risk.

　　Active Post Marketing Surveillance上市后主动监测

　　生产商/监管机构要求的四期研究 数据链接研究 快速循环分析 登记研究

　　Phase Four studies by manufacturers/mandated by regulators Datalink Studies Rapid cycle analyses Registry Studies

　　Vaccine Data Link studies vs Passive Reporting Systems.疫苗数据链接研究与被动报告系统的对比

　　与被动报告系统不同，数据链接研究有下述潜力 以无偏方式检出事件 可以获得病历，更好地描述和理解病例 可计算事件发生率、相对危险度和归因危险度 评估疫苗影响和疾病流行病学的变化

　　Unlike passive reporting systems, data link studies have the potential to Identify cases of events in an unbiased way. Allow access to medical records to better characterize and understand cases. Calculate rates of events, relative risk and attributable risk. Evaluate vaccine impact and changes in disease epidemiology

　　Why use Computerized Data to Evaluate Vaccine Safety Concerns?为什么要用计算机化数据来评估疫苗的安全性?

　　计算机化的信息可以用大量人群的数据对安全性顾虑快速反应 迄今为止的应用经验主要集中在一个国家内部：丹麦、英国、美国、新西兰、越南 欧洲的VESCO合作是跨国性的 有效使用跨国计算机化临床数据需要建立基础设施、建立数据管理规则 中国的许多地区，已有计算机化的临床信息，复杂程度各不相同

　　Computerized information offers the potential of a relatively rapid response to a safety concern using data on a large number of people. Experience to date has been mostly within one country: Denmark, UK, USA, New Zealand, Vietnam In Europe, the VAESCO collaboration is multinational Efficient use of computerized clinical data across borders requires a priori development of infrastructure and data management rules. Computerized clinical information exists, with varying degrees of complexity, in many areas of China.

　　What Computerized Data?什么是计算机化数据?

　　Computerized Vaccine Data on an individual level 计算机化疫苗接种数据 个人层面的数据

　　Computerized Hospital and/or Clinic Diagnoses On an individual level 计算机化医院 和/或临床诊断 个人层面数据

　　Demographic Data on a Population Adjust for confounders 人群的人口统计数据 调整混杂因素

　　Exposure 暴露

　　NOTE: Can do case series with Outcome Data alone to assess risk and hence causality 注：可以单独用结果的案例数据来评估风险和因果关系

　　Need a unique identifier to link 需要唯一识别码 链接

　　Outcome or Possible “Adverse Event”结果或可能的“不良事件”

　　What about comparison with another group?与另一组相比较结果如何?

　　如果需要单独的对照组，确定适当的比较组通常存在很多问题 未接种疫苗的儿童通常在健康状况、父母的医疗素养或护理的可及性方面都存在差异 延迟接种疫苗的儿童也存在差异 对于常规推荐的高接种率的疫苗，可能不存在便利的对照 与其他人群或历史对照相比也存在问题 由于疾病诊断的长期趋势、人口特征的变化或病例确诊方法的变化而容易引起偏倚

　　Where a separate control group is required, defining an appropriate comparison group is often problematic Unvaccinated children are usually different either in the health status, parental medical sophistication or access to care Children who are vaccinated at a later age are different as well. For routinely recommended vaccines with high uptake, no convenient control group may exist Comparing to another population or historical controls is also problematic. Prone to bias due to secular trends in disease diagnosis, changes in population characteristics or changes in case ascertainment

　　最初由Farrington设计，由其他人应用和调整 仅使用感兴趣的结局和疫苗信息就可以计算相对风险 在接种前的特定时间窗内接种疫苗的概率与研究的病例在其他时间接种疫苗的概率进行比较 如果没有关联，那么这两个概率应该相等，得出的比值比为1

　　Case Series病例系列

　　Developed by Farrington initially and since used and adapted by others. Allows calculation of relative risk using only cases of the outcome of interest and vaccine information. Probability of being vaccinated within a specified time window prior to vaccination is compared to probability of being vaccinated at other times in the cases with the outcome under study. If there is no association, then these two probabilities should be equal yielding an odds ratio of one.

　　Evaluation of Uncommon Events Resulting in Medical Utilization: Case Series评估导致利用医疗资源的罕见事件：病例系列

　　对于自我对照来说，暴露和未暴露的随访时间要对同一个人进行定义 要解决组间存在的差异问题 暴露前一段时间间隔内“健康儿童效应”导致的偏倚问题 如果事件会妨碍接种疫苗，如SIDS或隐匿性发作或长时间延迟发病的事件，则无效

　　疫苗

　　疫苗

　　前

　　暴露

　　1剂后 /2剂前

　　暴露2

　　后

　　For self control comparison, exposed and unexposed follow up time are defined for the same individual. Resolves issue of differences between groups Issue of “well child effect” bias for pre-exposure time interval Does not work if event prevents vaccination – eg SIDS or for events of insidious onset or with long delay to onset.

　　Rapid Cycle Techniques快速循环技术

　　需要快速访问自动数据源，确定背景发生率，并随访新疫苗引入后的发生率。目前正在使用VSD数据，每周进行分析 使用序贯概率检验，分析与预期数相比，出现观察到的病例数的可能性 建立一个“停止值”，根据预期的事件数和风险水平接受或排除关联 对于肠套叠，在接种2589剂疫苗后确定了风险……类似于VAERS数据挖掘产生的时间框 与DTwP相比，接种DTaP 12周内也检测到癫痫发作、发热和其他异常神经事件的风险降低

　　Requires rapid access to automated data sources to establish rates of events in baseline and following a new vaccine. Currently in use in VSD with weekly data pulls. Uses sequential probability testing to test the likelihood of an observed number of cases versus expected. A “stopping value” is established to either rule in or rule out an association based upon the expected number of cases and risk level. For intussusception, a risk established after 2589 doses of vaccine given… a similar time frame as VAERS data mining. Also able to detect a decrease in risk of seizures, fever and other abnormal neurologic events within 12 weeks of introduction of DTaP as compared to DTwP.

　　What do these analyses look like?这些分析是什么样的分析?

　　Case Study Rapid Cycle analysis of MMR-V 案例研究 MMR-V的快速循环分析

　　A Case Study: MMR-V Pre-licensure vs Post Licensure Studies案例研究: MMR-V 注册前与注册后研究的比较

　　注册前的研究规模通常很小 关注常见的局部和全身事件 按照预先定义的时间窗进行分析 注册后的研究规模通常要大得多，并且有能力更灵活地分析事件

　　Pre-licensure studies of safety are usually small Focus on common local and systemic events Analyses done within predefined windows Post licensure studies are usually much larger and have ability to look at events more flexibly

　　23

　　MMR-V Pre-licensure SafetyBlack et al. PIDJ 24:8-12, 2005MMR-V注册前的安全性研究Black et al. PIDJ 24:8-12, 2005

　　VSD Rapid Cycle Evaluation of MMRV ： Outpatient Visits for Fever by Day after Vaccineat Northern California Kaiser Permanente: 1995-2008MMRV的VSD快速评价：接种后因发热门诊就诊的时间分布1995-2008年北加州Kaiser Permanente疫

　　(Gp:) Age 12-23 months 12-23月

　　(Gp:) 6241 total fever visits after 302,670 MMR+V, 147,762 MMR, 46,390 MMRV, 38,251 VZV 接种302670剂MMR+V，147762剂MMR，46390剂MMRV，38251剂VZV，共有6241例发热门诊就诊

　　0

　　50

　　100

　　150

　　200

　　250

　　300

　　350

　　0

　　5

　　10

　　15

　　20

　　25

　　30

　　35

　　40

　　接种后天数

　　事件数 / 100,000剂

　　MMR

　　MMR+V

　　MMRV

　　V

　　VSD Rapid Cycle Evaluation of MMRVTemporal distribution of seizures after MMRV vaccination vs after simultaneous MMR and varicella vaccination疫苗的VSD快速循环评价：接种MMRV后或MMR+V后癫痫的时间分布

　　接种MMRV后的时间

　　(2/06-9/07)

　　接种MMR+V后的时间

　　(1/00-8/07)

　　5

　　10

　　15

　　20

　　25

　　10

　　20

　　30

　　40

　　50

　　60

　　1

　　2

　　3

　　4

　　5

　　6

　　7

　　8

　　9

　　10

　　11

　　12

　　13

　　14

　　15

　　16

　　17

　　18

　　19

　　20

　　21

　　22

　　23

　　24

　　25

　　26

　　27

　　28

　　29

　　30

　　31

　　32

　　33

　　34

　　35

　　36

　　37

　　38

　　39

　　40

　　41

　　42

　　1

　　2

　　3

　　4

　　5

　　6

　　7

　　8

　　9

　　10

　　11

　　12

　　13

　　14

　　15

　　16

　　17

　　18

　　19

　　20

　　21

　　22

　　23

　　24

　　25

　　26

　　27

　　28

　　29

　　30

　　31

　　32

　　33

　　34

　　35

　　36

　　37

　　38

　　39

　　40

　　41

　　42

　　Nicola Klein as presented at PAS/SPR in Honolulu May 2008

　　26

　　VSD Rapid Cycle Evaluation of MMRV Evaluating for the risk of seizure 7-10 days post-vaccination using chart verified febrile seizuresVSD快速循环评估MMRV与接种后7～10天内热性惊厥风险

　　N= 42 for MMR-V out of 43,353 MMR-V and 124 for 314,599 doses of MMR + V for chart review confirmed seizures. Adjusted for age and influenza season. Increased risk with MMRV cannot be explained by concomitant vaccines, temporal trends in seizure, VSD site, age or influenza season.

　　Summary of Advantages of Population Based Data Studies人群为基础的数据研究的优势小结

　　Allow calculation of incidence on AEFI and background rates of disease without vaccination Allow calculation of relative risk Allow calculation of attributable risk. Allow adjustment for confounders Allow assessment of trends including vaccine impact on disease for risk-benefit analyses 可计算AEFI发病率和无疫苗接种时的背景发病率 可计算相对风险 可计算归因风险 可调整混杂因素 可评估疫苗对疾病影响的趋势，以进行风险收益分析

　　AN EXAMPLE:A Comparison of Active versus Passive Surveillance in Burkino Faso实例：布基诺法索的主动与被动监测的比较

　　Men A vaccine was introduced into the Sub-Saharan African meningitis belt in December 2010 and almost 12 million people vaccinated in one campaign in Burkino Faso Two safety surveillance systems established Nation wide passive surveillance Active Surveillance in one district 2010年12月，撒哈拉以南非洲脑膜炎带地区引入了MenA疫苗，在布基诺法索的一次接种运动中接种了近1200万人 建立了两套安全性监测系统 全国范围的被动监测 一个地区的主动监测

　　VACCINE in press

　　MenA Introduction in Burkino Faso: Passive Surveillance Data布基诺法索引入MenA：被动监测数据

　　Per 100,000

　　MenA Introduction in Burkino Faso:Active Surveillance Data布基诺法索引入MenA：主动监测数据

　　30

　　Per 100,000

　　Per 100,000

　　MenA Introduction in Burkino Faso:Active Surveillance Data布基诺法索引入MenA：主动监测数据

　　Passive surveillance can be useful for signal detection Under-reporting can severly limit the usefullness of passive reporting Active surveillance in a defined sub-population can be more useful than passive surveillance in a larger population. Both types of surveillance are useful Passive for signal detection Active for signal evaluation 被动监测可用于信号检测 漏报严重制约了被动报道的价值 在人群亚组中，主动监测可能比更大人群中的被动监测更有价值 两种监测都有价值 被动监测用于检出信号 主动监测用于评估信号

　　Summary: Active vs Passive Surveillance小结：主动与被动监测

　　32

　　What to do if you observe a possible signal?如果你观察到一个可疑信号，该如何处理?

　　Evaluation of a possible consistent time association of the event with vaccination Evaluation in a different analytic framework: self-control analysis or other reference group Possible associations can serve as a source of hypothesis generation for further studies – ie case-control study conducted for intussusception. 事件与疫苗接种时间关联性评价 不同分析框架下的评价：自我对照分析或参考其他参照群体 可能的关联性可作为进一步研究的假说来源——如，肠套叠的病例对照研究

　　34

　　What happens without vaccines?不接种疫苗会发生什么?

　　The importance of knowing background rates and epidemiology 了解背景发病率和流行病学的重要性

　　What happens without vaccines?不接种疫苗会发生什么? “AEFI” without Vaccines: 不接种疫苗时发生的“AEFI”：Autoimmune < 30days Outpatient Events in Teens青少年<30天内自身免疫事件引起的门诊就诊事件

　　Claire Anne Seigrist PIDJ

　　Pandemic Flu Safety The importance of background rates of disease in assessment of vaccine safety during mass immunization with H1N1 influenza vaccines流感大流行疫苗的安全性在H1N1流感大流行疫苗突击接种期间，疾病背景发生率对疫苗安全性评估至关重要

　　37

　　Considerations of Statistical PowerTotal Study Population Size Required to Detect a Selected Increased Levels of Risk统计功效的总体考虑检测出选定的风险水平增加所需的研究样本规模

　　Other Vaccine Safety Assessment Infrastructure其他疫苗安全性评估的基础设施

　　Special study populations: Maternal immunization Identifying subsets at risk and treating them differently HIV infected children and receipt of BCG 特殊研究人群：母体免疫 识别风险人群并对其进行不同处理 HIV感染儿童与接种卡介苗

　　Importance of maternal immunization母亲免疫的重要性

　　孕妇接种疫苗对母亲或婴儿或两者都有重要益处 正在考虑的新的预防新生儿和婴儿感染的方法 世卫组织战略咨询专家组(SAGE) 接种注册的疫苗，例如流感、百日咳、破伤风 突击接种黄热病疫苗、风疹或脑膜炎A群结合疫苗 新疫苗或即将上市的疫苗 戊肝疫苗，B族链球菌，呼吸道合胞病毒，重组百日咳

　　Vaccination of pregnant women may provide important benefits to mother or infant or both New approaches to prevent newborn and infant infections are being considered WHO SAGE (Strategic Advisory Group of Experts) Use of licensed vaccines, e.g., for influenza, pertussis, tetanus Mass campaigns YF, rubella or MenA conjugate New or upcoming vaccines Hepatitis E, group B Streptococcus, respiratory syncitial virus, recombinant pertussis

　　Challenges to pregnancy vigilance在孕妇开展药物警戒面临的挑战

　　Limited data available about safety in pregnant women Theoretical concerns about fetal risk Manufacturers’ liability concerns Methodological issues in assessing vaccine safety 在孕妇的安全性数据有限 理论上的对胎儿风险的顾虑 生产商的责任顾虑 疫苗安全性评价的方法学问题

　　Picture from Financial Technologies Forum http://blog.ftfnews.com/2011/06/21/the-challenges-of-corporate-actions/

　　You need a plan需要制定一个计划

　　Next steps 下一步

　　41

　　Pharmacovigilance Plan (PVP)药物警戒计划(PVP)

　　42

　　PVP为在使用的文件 内容和组成部分符合监管当局的要求 随着研究的完成和上市后经验的积累，会不断更新 可能出现新的问题 药物警戒计划的组成部分： 常规药物警戒(PV) 针对特定事件的强化PV 针对性安全性研究(TSS)

　　PVP is a living document Content and components agree with Regulatory Authority requirements. Evolves as studies are completed and post-marketing experience accrues New issues may arise Components of a pharmacovigilance plan: Routine pharmacovigilance (PV) Enhanced PV for specified events Targeted Safety Study (TSS)

　　强化药物警戒

　　针对性安全性研究

　　Enhanced Pharmacovigilance Background rates (pre-defined and ad hoc) Observed vs expected analyses Targeted follow-up procedure 强化药物警戒 背景率(事先确定和临时确定) 分析观察值与期望值 针对性随访程序

　　Prospective cohort study Large-scale databases (defined populations) 前瞻性队列研究 大规模数据库(定义的人群)

　　药物警戒

　　Data sources Spontaneous reports Vaccine exposure Nonclinical/Clinical data 数据来源 自发报告 疫苗暴露 非临床/临床数据

　　Methods Weekly individual case review Monthly aggregate reviews Monthly literature review 方法 每周个案回顾 每月汇总回顾 每月文献综述

　　Pharmacovigilance Plan Components药物警戒计划组成部分

　　Routine Pharmacovigilance 常规药物警戒

　　What are global trends?全球趋势如何?

　　增加样本量和地理范围 区域和全球财团 对仅在发展中国家使用的疫苗进行评估 最大化统计功效 例如：建立了全球疫苗安全数据网络，共同研究疫苗安全性问题。复旦大学是会员之一 世界卫生组织全球蓝图

　　Increasing the sample size and geographic reach Regional and global consortiums Offers potential to evaluate vaccines only used in developing world. Maximizes statistical power EXAMPLE: A Global Vaccine Safety Data Net has been established to collaboratively look at vaccine safety issues. Fudan University is a member. WHO Global Blueprint

　　Global Vaccine Safety Data Network全球疫苗安全数据网络：概念证明

　　共有18个国家

　　Summary小结

　　注册后有多种方式来评估安全性 信号检出系统，如VAERS 基于人群的使用数据链接的系统 企业申办/政府要求的四期效果、安全性、效果持久性研究 主动监测研究可行，在中国也在开展 公众的期望是对可能的关注要快速回应 否则信心就会降低，“反疫苗”组织就有了机会 即使是机会性事件，因为事件可以聚集性地在疫苗接种者和非疫苗接种者中出现，也会出现顾虑 全球合作可以提高效率和能力

　　There are multiple modalities to evaluate safety in the post-licensure setting Signal detection systems such as VAERS Population based systems using data linkage Industry sponsored/ government mandated phase four studies of effectiveness, safety, duration of effect. Active surveillance studies are feasible and being conducted in China Public expectation is for a rapid response to possible concerns. Without this confidence is undermined and “anti-vaccine” groups get a head start. Concerns will arise because events occur and occur in clusters in both vaccinees and non vaccinees due to chance alone. Global collaborations can increase efficiency and capacity