临床真实世界研究中的实验性研究设计

真实世界研究作为解释性随机对照试验在医疗实践中评价干预措施效果的进一步验证和补充已成为医疗卫生领域关注的焦点。但是也存在错误将真实世界研究等同于观察性研究,认为真实世界研究不能实施人为干预,更不能采取随机化。实际上,真实世界研究的基本设计既可以是观察性的,也可以是实验性的。其中真实世界研究的实验性研究设计主要是指实用性随机对照试验和基于注册登记研究的随机对照试验,也可采用非随机对照、自适应设计等其他研究设计方案。     

PRECIS研究进展

随机对照试验(RCT)可分为评估效力的解释性试验和评估临床效果的实用性临床试验(PCT)。PCT反映了临床实际条件下的干预效应,具有一定的外推性,但其内部真实性相对较差。与此相反,解释性试验是在理想条件下进行,其内部真实性较好,而外推性较差。但在RCT实际设计中,PCT和解释性试验并不是截然分离的两个极端,有许多RCT同时兼有两种设计的属性。PRECIS通过评价RCT设计的解释和实用两方面,指导如何实行干预和试验设计,使RCT在内部和外部真实性之间达到平衡。目前国内对PRECIS介绍甚少,更未见应用的报道,基于此以下介绍PRECIS基本原理、特点及应用,以期为临床试验设计提供参考。     

108篇系统评价或Meta分析的方法学质量评价——基于AMSTAR-2

目的 利用AMSTAR-2评价同时纳入随机和非随机对照试验或仅纳入非随机对照试验的系统评价或Meta分析的方法学质量,探讨影响其方法学质量的显著因素.方法 计算机检索PubMed、EMbase、Cochrane Li-brary数据库,搜集国内外于2015—2017年发表的系统评价或Meta分析的文献,分别由2名研究人...     

基于大数据的随机对照试验

我国医疗领域目前已积累了海量数据,如何利用大数据开展随机对照试验日益得到关注。本研究结合国外利用大数据实施随机对照试验的成功经验,从数据来源、研究对象和研究结局确定、干预措施、随机化方法、知情同意的实施等方面进行梳理总结,以期为国内未来开展相关研究提供借鉴。     

真实世界研究与随机对照试验、单病例随机对照试验在临床治疗性研究中的关系比较

真实世界研究、随机对照试验及单病例随机对照试验在设计及具体的实施环节上存在明显不同.随机对照试验属于新治疗措施实施前的研究,真实世界研究属于新治疗措施实施后的研究.两者不是对同一个问题的平行论证,而是承启关系.精心设计的随机对照试验是临床上任何干预措施效果评价的基础,其结果需要真实世界研究的进一步验证及拓展补充,综合考虑二者才是最佳的选择.单病例随机对照试验更易在短时间内获得一些特殊病例的信息,是随机对照试验结果的良好补充,也是一定条件下最经济的真实世界研究.临床工作及其研究是十分复杂的过程.不同个体虽患同种疾病,但临床表现互有差异,且临床反应的变化也不尽相同.因此,无法获得同一干预措施下不同个体的相同治疗效果;加之有的治疗措施缺乏真实性和实用价值,从而使得疗效评价成为一个难题.近些年来,普遍采用试验性的研究结果作为证据指导临床实践活动,其中以随机对照试验(RCT)最为受到重视,但由于RCT属于药物面市前研究,对研究对象的选择、治疗措施的应用等均有严格的限定.     

真实世界研究与随机对照试验、单病例随机对照试验在临床治疗性研究中的关系比较

真实世界研究、随机对照试验及单病例随机对照试验在设计及具体的实施环节上存在明显不同.随机对照试验属于新治疗措施实施前的研究,真实世界研究属于新治疗措施实施后的研究.两者不是对同一个问题的平行论证,而是承启关系.精心设计的随机对照试验是临床上任何干预措施效果评价的基础,其结果需要真实世界研究的进一步验证及拓展补充,综合考虑二者才是最佳的选择.单病例随机对照试验更易在短时间内获得一些特殊病例的信息,是随机对照试验结果的良好补充,也是一定条件下最经济的真实世界研究.临床工作及其研究是十分复杂的过程.不同个体虽患同种疾病,但临床表现互有差异,且临床反应的变化也不尽相同.因此,无法获得同一干预措施下不同个体的相同治疗效果;加之有的治疗措施缺乏真实性和实用价值,从而使得疗效评价成为一个难题.近些年来,普遍采用试验性的研究结果作为证据指导临床实践活动,其中以随机对照试验(RCT)最为受到重视,但由于RCT属于药物面市前研究,对研究对象的选择、治疗措施的应用等均有严格的限定.     

大型随机对照试验:精准流行病学研究的典范与陷阱

现代流行病学是医学应用型研究的方法论,是在人群中定量地研究有关健康、疾病和医疗服务实践问题一般规律的科学和艺术。流行病学研究结果的准确度主要取决于研究的设计类型,精确度主要取决于样本量大小。大型随机对照试验是最精、最准的流行病学研究设计类型,但是由于伦理的限制,只能用于评估医学干预效果。一项研究需要的设计严谨性和样本量与预期效果的大小成反比：效果越小,所需的研究设计就越严谨,需要的样本量就越大。因此,只有当疗效比较小时,才需要大型随机对照试验,当疗效十分明显时,中小型随机对照试验甚至观察性研究就足以证明其有效性。从研究阶段上看,它是确认性、终结性研究,而不是提出假设的原创性研究。然而,研究的价值最终取决于研究问题的意义和原创性,而不是研究方法和P值的大小。过度推崇大型随机对照试验会引发：①对中、小疗效干预的过度强调;②对确认性研究的过度重视,以及对项目大小和经费多少而不是科学问题的追逐,进而弱化原创性研究工作;③增加研究资源、医学活动和患者利益被制药公司绑架的风险。     

偏倚风险评估系列:(三)交叉设计随机对照试验

针对交叉设计随机对照试验Cochrane偏倚评估工具2.0版本（RoB2.0）的主要内容进行详细介绍,主要阐述了与平行设计RoB2.0的不同之处,并举例说明交叉设计RoB2.0的使用方法和注意事项。交叉设计RoB2.0针对交叉设计的自身特点,设置了相应的信号问题,为交叉设计试验纳入系统综述进行证据整合提供偏倚风险信息。     

偏倚风险评估系列:(四)整群设计随机对照试验

介绍整群设计随机对照试验Cochrane偏倚评估工具2.0版本（RoB2.0）的主要内容,阐述与平行设计RoB2.0的不同之处,并举例说明整群设计RoB2.0的使用方法和注意事项。RoB2.0针对整群设计的自身特点,设置了相应的信号问题,为将整群设计试验纳入系统综述进行证据整合提供偏倚风险信息。     

甲型肝炎疫苗H2减毒株的流行病学效果考核

目的 考核病毒性甲型肝炎（简称甲肝）减毒活疫苗（H2减毒株）的流行病学效果。方法 以正定县的3个乡和石家庄郊区全部10个乡为研究现场,以其适龄儿童约3.7万人为观察对象,进行随机、对照试验。观察对象按其出生的单、双月份随机分为疫苗组和空白对照组,疫苗组接种1ml 10^6.5/ml半数细胞培养感染量（TCID50）的甲肝疫苗,空白对照组不做任何处理。接种后严格随访观察对象中甲肝病例的发生情况。结果 在3.5年的随访中,易感的对照组出现甲型肝炎20例,发病率为136.4/10万;疫苗组出现1例,发病率8.1/10万。2组发病率差异有显性。疫苗的保护率为94.1%,95%可信限：67.1%～100%。结论 每毫升10^6.5TCID50的甲肝减毒活疫苗保护效果良好,可以满足人群免疫的需要。     

Pragmatic vs. explanatory: an adaptation of the PRECIS tool helps to judge the applicability of systematic reviews for daily practice

Objective

The Pragmatic-Explanatory Continuum Indicator Summary (PRECIS) tool was designed to classify randomized clinical trials (RCT) as being more pragmatic or explanatory. We modified the PRECIS tool (called PRECIS-Review tool [PR-tool]) to grade individual trials and systematic reviews of trials. This should help policy makers, clinicians, researchers, and guideline developers to judge the applicability of individual trials and systematic reviews.

Study Design and Setting

To illustrate the usefulness and applicability of the PR-tool, we applied it to two systematic reviews. Each included RCT was scored on the 10 PRECIS domains on a scale of 1-5. After this scoring, a 10-domain average for each individual trial and for the systematic review a single domain average and an overall average was calculated.

Results

One review was more pragmatic with an average score of 3.7 (range, 2.9-4.6) on our PR-tool, whereas the other review was more explanatory with an average score of 1.9 (range, 1.1-3.3). The results also suggest that the included studies within each systematic review were rather uniform in their approach, although some domains seemed more prone to heterogeneity.

Conclusion

The PR-tool provides a useful estimate that gives insight by estimating quantitatively how pragmatic each RCT in the review is, which methodological domains are pragmatic or explanatory, and how pragmatic the review is.     

偏倚风险评估系列:(二)平行设计随机对照试验偏倚评估工具2.0介绍

本文针对平行设计随机对照试验（RCT）Cochrane偏倚评估工具2.0版本（RoB2.0）的主要内容进行介绍,同时对RoB2.0与前一个版本RoB1.0的区别与联系进行阐述,并举例说明RoB2.0的使用方法和注意事项。RoB2.0内容丰富,覆盖全面,为RCT的证据整合与评价提供了更多的偏倚风险信息。RoB2.0仍处于开发与进一步完善中,使用者可持续关注其后续更新与进展。     

真实世界证据与随机对照试验：RCT DUPLICATE项目启动、实施、进展解读与启示（二）

随着医疗大数据的推广与应用,监管机构开始使用非随机化的真实世界证据（RWE）来评估医疗产品的有效性。本文简要介绍了哈佛大学研究团队于2018年发起RCT DUPLICATE项目的最新进展与研究成果,并结合我国医疗服务特点总结了相关研究经验,为国内学者今后开展相关RWE研究提供借鉴。     

Cancer randomized trials showed that dissemination bias is still a problem to be solved

ObjectiveThe objective of the present study was to determine the publication rate of cancer randomized controlled trial (RCTs) and to analyze the determinants of the publication, as well as to estimate the possible existence of a location and time lag bias. We also described the bibliometric characteristics of the publications.Study design and SettingWe conducted an observational study that identified publications resulting from RCTs involving cancer-related drug products. These studies were authorized and registered by the Spanish Agency of Medicines and Medical Devices between 1999 and 2003.ResultsWe identified 168 publications of 303 RCTs, resulting in a publication rate of 55.4% after a mean follow-up of 12 years. The only factor associated to the likelihood of nonpublication was the study setting favoring only national RCTs (odds ratio 2.7; 95% confidence interval 1.5–4.8). Type of sponsor did not seem to be associated, although the largest volume of nonpublished trials is international, industry-sponsored. Positive results seemed to be associated to a publication in a higher impact factor journal and a shorter time-to-publication.ConclusionsAbout half of the cancer RCTs during the target period have not been published. The national setting is a factor associated to nonpublication, whereas the direction of results determines its dissemination (impact factor and timely publication).     

The balance of positive and negative effects of specific messages in the evaluation of interventions for preventing HIV infection

The study objectives were to determine the effects of a large- scale school-based HIV prevention campaign, in terms of both positive and negative effects of each single message, and to identify sub- populations more at risk. Forty-six schools, randomly sampled from all schools in the Lazio region, were randomized to either an intervention or a control group. The study population consisted of 3866 students. Questionnaires on AIDS-related knowledge and risk perceptions were administered to students before and after the intervention. Odds ratios were calculated to represent the extent to which the intervention was associated with an improvement (OR+), and the extent to which it prevented a worsening (OR–). Overall, the intervention was successful in communicating important messages, such as the impossibility of transmitting HIV through social contacts (OR+ all significantly >1 and OR– always<1), the meaning of seropositivity (OR+: 1.28, 95% CI: 0.99–1.64; OR–: 0.73, 95% CI: 0.58–0.91), and the lack of a resolutive cure for AIDS (OR+: 1.61, 95% CI: 1.10–2.36; OR–: 0.76, 95% CI: 0.57–1.02). The worst results were observed in vocational and art schools, where OR+>1 were observed for only three questions and OR– values usually exceeded 1. This study highlights the necessity of remodelling the intervention, indicating which messages need to be modified. The low impact of educational programmes among students of vocational and art schools makes them a population more at risk, that should be considered as a priority target population for interventions.     

An observational study found that authors of randomized controlled trials frequently use concealment of randomization and blinding, despite the failure to report these methods

BACKGROUND AND OBJECTIVE: Readers of randomized controlled trials (RCTs) commonly assume that what was not reported did not occur. We undertook an observational study to determine whether concealment of randomization or blinding was used in RCTs that failed to report these bias-reducing strategies. METHODS: We recorded the reporting of concealment of randomization and blinding in 105 RCTs. We subsequently contacted the authors and determined if they had used these methodological safeguards. RESULTS: We successfully obtained data from 98 authors. The authors in the full-text publications of these 98 RCTs failed to report the presence or absence of concealment of randomization in 55%, and the blinding status of participants in 26%, health care providers in 64%, data collectors in 84%, outcome assessors in 83%, and data analysts in 96%. In direct contact, authors frequently reported concealing randomization (96%; 95% confidence interval CI=87-100%), blinding participants (20%; 95% CI=7-41%), blinding health care providers (65%; 95% CI=52-77%), blinding data collectors (65%; 95% CI=53-75%), blinding outcome assessors (79%; 95% CI=69-87%), and blinding data analysts (50%; 95% CI=40-60%), despite not reporting the use of these methodological safeguards in their publications. CONCLUSIONS: Readers should not assume that bias-reducing procedures not reported in an RCT did not occur.     

The 2 × 2 cluster randomized controlled factorial trial design is mainly used for efficiency and to explore intervention interactions: a systematic review

ObjectivesTo describe the health care settings, purposes, and study reporting quality of the 2 × 2 cluster randomized controlled factorial trial design.Study Design and SettingThis study is a systematic review. We searched Medline, Embase, Cochrane Library, and Web of Knowledge for articles published up to May 2012. Cluster randomized controlled 2 × 2 factorial trials in health, evaluating at least one complex intervention, were included. Two authors independently reviewed and extracted data from the studies.ResultsTwenty-nine studies covering a wide range of clinical areas and health care settings were included. The cluster design was mostly used to minimize contamination. The factorial design was mostly used to assess the effects of two interventions in the same study and to explore interactions between interventions. However, although most studies explored the presence or absence of intervention interactions, they were often either not sufficiently powered to detect any interactions or did not provide information on whether the study was sufficiently powered to detect any interactions. There was a considerable variability in the reporting of a number of study characteristics and methodological aspects. Study quality was also variable within and across studies.ConclusionThe design has been used in a wide range of health care settings and clinical areas to minimize contamination, assess the effects of two interventions in the same study, and explore intervention interactions. There is need for improvement on and guidelines for the reporting of factorial trials.     

多水平模型在社区随机对照试验分析中的应用

目的探讨多水平模型在社区随机对照试验数据分析中的适用性,为该领域统计分析方法提供新的思路。方法于2014年7—11月,以北京市房山区16个社区(村)510例2型糖尿病患者为研究对象,将患者随机分为干预组(260例)和对照组(250例)。对照组接受国家基本公共卫生服务,干预组在对照组的基础上进行糖尿病自我管理小组活动干预。采用调查问卷收集患者人口学信息、患病情况,采用糖尿病患者自我效能量表收集患者自我效能信息,采集患者空腹静脉血测量糖化血红蛋白。分别计算干预前后两组糖化血红蛋白差值及自我效能得分差值。采用SAS 9.4软件进行秩和检验,采用多水平模型对差值进行分析并与多重线性回归模型分析结果进行比较。结果本研究共纳入500名研究对象进行分析。干预后,干预组糖化血红蛋白差值中位数[0.40%(P25~P75:-0.40%~1.70%)]与对照组差值中位数[0.70%(P25~P75:-0.30%~1.50%)]比较,差异无统计学意义(P>0.05);干预组自我效能得分差值中位数[17.00分(P25~P75:3.00~28.00分)]与对照组差值中位数[8.00分(P25~P75:0.00~21.00分)]比较,差异有统计学意义(P<0.01)。多水平零模型拟合结果显示,同一社区(村)内患者结局指标存在聚集性(糖化血红蛋白组内相关系数=0.12,P<0.05;自我效能得分组内相关系数=0.20,P<0.05),适宜采用多水平模型进行分析。进一步分析显示,控制协变量前后,干预组糖化血红蛋白差值与对照组比较,差异均无统计学意义(P>0.05);控制协变量前后,干预组自我效能得分较对照组均有升高,差异均有统计学意义(P<0.01)。多水平模型与多重线性回归模型分析结果比较,多水平模型组别回归系数标准误较多重线性回归模型下降。以糖化血红蛋白为因变量,控制协变量后,医疗费用支出方式在多水平模型中差异无统计学意义(P>0.05),在多重线性回归模型中差异有统计学意义(P<0.05);以自我效能得分为因变量,控制协变量后,多水平模型赤池信息准则(AIC)较控制前下降91.30,多重线性回归模型AIC较控制前上升9.50。结论多水平模型适用于分析具有层次结构的随机对照试验数据。与多重线性回归模型相比,多水平模型可以减少假阳性错误,改善模型拟合效果,今后应进一步推广应用。