Trial Sequential Analysis (TSA), one kind of cumulative meta-analysis, is a method which introduces sequential analysis into traditional meta-analysis to avoid random errors (false positive or false negative outcomes) that occurred during repeated updates when traditional meta-analysis is performing. It is also applied to calculate required information size (RIS) of a firm conclusion. This study aims to summarize the proposal, fundamental theory, application software, and current limitation of TSA, and to clarify the advantages of TSA on the basis of detailed examples, in order to attract more attention of researchers and promote the methodological development of meta-analysis in China.
The theoretical foundation of relevant packages of R software for network meta-analysis is mainly based on Bayesian statistical model and a few of them use generalized linear model. Network meta-analysis is performed using GeMTC R package through calling the corresponding rjags package, BRugs package, or R2WinBUGS package (namely, JAGS, OpenBUGS, and WinBUGS software, respectively). Meanwhile, GeMTC R package can generate data storage files for GeMTC software. Techonically, network meta-analysis is performed through calling the software based on Markov Chain Monte Carlo method. In this article, we briefly introduce how to use GeMTC R package to perform network meta-analysis through calling the OpenBUGS software.
Network plots can clearly present the relationships among the direct comparisons of various interventions in a network meta-analysis. Currently, there are some methods of drawing network plots. However, the information provided by a network plot and the interface-friendly degree to a user differ in the kinds of software. This article briefly introduces how to draw network plots using the network package and gemtc package that base on R Software, Stata software, and ADDIS software, and it also compares the similarities and differences among them.
The WinBUGS software can be called from either R (provided R2WinBUGS as an R package) or Stata software for network meta-analysis. Unlike R, Stata software needs to create relevant ADO scripts at first which simplify operation process greatly. Similar with R, Stata software also needs to load another package when drawing network plots. This article briefly introduces how to implement network meta-analysis using Stata software by calling WinBUGS software.
R Software is an open, free of use and charge statistical software which has a powerful graphic capability; however, it requires more complex codes and commands to perform network meta-analysis, which causes errors and difficulties in operation. WinBUGS software is based on Bayesian theory, which has a powerful data processing capability, and especially its codes are simple and easy to operate for dealing with network meta-analysis. However, its function of illustrating statistical results is very poor. In order to fully integrate the advantages of R software and WinBUGS software, an R2WinBUGS package based on R software has been developed which builds a “bridge” across two of them, making network meta-analysis process conveniently, quickly and result illustration more beautiful. In this article, we introduced how to use the R2WinBUGS package for performing network meta-analysis using examples.
Objective To assess the efficacy and safety of fibfinogen-depleting agents (snake venom extracts) in the treatment of acute ischemic stroke. Method A systematic review of all the relevant randomized controlled trails (RCTs) was performed. RCTs were identified from the Cochrane Stroke Group’s Specialized Trials Register, additional electronic and handsearching, and personal contract with pharmaceutical companies. We included all completed and unconfounded truly or quasi-randomized trials in patients with ischemic stroke comparing fibrinogen depleting agents for analysis. Results Ten completed and one ongoing RCTs have been identified so far. Up to 1998, only three trials using ancrod (182 patients) met the inclusion criteria. Ancrod was associated with a significant reduction in early deaths (5.6% vs. 16%; odds ratio [OR], 0.33; 95% confidence interval [CI], 0.13 to 0.85; 2P=0.02) suggesting that treatment of 100 patients would avoid about 10 early deaths. The frequency of asymptomatic intracranial hemorrhage shown by computed tomography was similar between ancrod-treated and control groups (7.6% vs. 9.6%; OR 0.78; 95%CI 0.26 to 2.33; 2P=0.65). No major intracranial or extracranial hemorrhages or recurrent ischemic strokes occurred in the ancord-allocated patients. There were nonsignificant trends in favor of ancrod in death from any cause (OR 0.57; 95%CI 0.27 to 1.23; 2P=0.15) and death or disability (OR 0.52; 95%CI 0.26 to 1.03; 2P=0.06) at the end of trial follow-up. Up to 2000, other two trials published results. This review will be updated with new trial results soon, which will provide more data. Conclusions There were too few patients and outcome events to draw reliable conclusions from the present data. Although ancrod-like agents appeared promising, their routine use cannot be recommended at the moment. Future trials should test simpler fixed-dose regimens to allow better generalizability.
Objective To assess the effect of different thrombolytic agents, and different regimens in acute ischaemic stroke. Methods A systematic review of all the relevant randomized controlled trials (RCTs) was performed. RCTs were identified from the Cochrane Stroke Group trials register, Embase (1980 to 1997), handsearching Japanese and Chinese journals, and personal contact with pharmaceutical companies. We included randomised and quasi-randomised trials in patients with confirmed acute ischaemic stroke comparing different doses of a thrombolytic agent, or different thrombolytic agent, or the same agent given by different routes. Results Eight trials involving 1 334 patients were included. Concealment of allocation was generally adequate. All the trials were conducted in Japan. Different doses (of tissue plasminogen activator or urokinase) were compared in six trials. Different agents (tissue plasminogen activator versus urokinase,or tissue-cultured urokinase versus conventional urokinase) were compared in three trials. Few data were available for functional outcomes. A higher dose of thrombolytic therapy was associated with a five-fold increase in fatal intracranial haernorrhages (odds ratio 5.02, 95% confidence interval 1.56 to 16.18). There was a non-significant trend towards more early deaths or clinically significant intracranial haemorrhages in higher dose group. No difference in late deaths or extra-cranial haemorrhages was shown between low and higher doses. However, very few of these events occurred. No difference was shown between the different thrombolytic agents tested. Conclusions There is not enough evidence to conclude whether lower doses of thrombolytic agents might be safer or more effective than higher doses in acute ischaemic stroke. It is not possible to conclude whether one agent might be better than another, or which route of administration might be best.
Objective To assess the effects and safety of vasodilators for sudden sensorineurial hearing loss (SSHL). Search strategy Electronic databases: MEDLINE from 1966, EMBASE from 1974, the Cochrane Controlled Trials Register, Chinese Bio-medicine Database from 1989. Hand search: Five kinds of Chinese otolaryngology journals were searched. Literature references were checked intensively. Selection criteria Randomized controlled trials comparing vasodilators with placebo or other drugs in patients with SSHL. Data collection and analysis At least two reviewers independently assessed trials quality and extracted data. Main results Thirteen trials with 1 155 patients were eligible and included in the systematic review. Ten of the trials were from developed countries and them were from P. R. China. None of the four trials showed that the effects of vasodilators were better than placebo for SSHL. None of the seven trials showed that the effects of one kind of vasodilators were better than that of the other vasodilators. Two trials showed that other drugs, such as batroxobin and hypaque,were probably better than some vasodilators (dextran, papaverine, 654-2, danshen). Eight trials reported the side effects of vasodilators, such as pruritus, allergy, etc. Reviewers’ conclusions Base on the systematic review of current eligible randomized controlled trials, there is no evidence to prove that vasodilator therapy is better than placebo or other therapies for SSHL, or the effects of one kind of vasodilator are better than that of the other vasodilators. We can’t draw a reliable conclusion about the effects of vasodilators for SSHL at the moment. And we must pay attention to their potential adverse reactions.
In the absence of large trials, it is important for us to discuss whether a well-conducted meta-analysis of smaller randomized controlled trial (RCT) can replace large trials or not. We have evaluated the quality of original literature and methodological quality. The difference between meta-analysis of smaller RCT and the largest randomized trials have also discussed.
目的 为避免选择和发表偏倚,系统评价者应采用多种查询技术,并尽力获得未发表的研究.本文试图探讨,英特网检索对鉴定未发表和正在进行的临床试验是否有用.研究设计 利用七个Cochrane系统评价的查询策略回顾性地在英特网上检索未纳入的随机对照试验.方法 检索策略 以普通检索式"研究方法学 NEAR干预措施NERA 条件"、用AltaVista在英特网上搜索.测量指标包括搜索时间、英特网搜索已发表研究的回溯率、精确度(已发表和未发表的随机临床试验链接的网页比例)、英特网检索到的未纳入的未发表和正在进行的研究数.结果 用21小时查询了429个网页,找到14个链接到未发表的、正在进行的或最近完成的试验,至少有9个与4篇系统评价相关.英特网检索已发表研究文献的回溯率在0～43.6%,其链接已发表和未发表研究的精确度在0～20.2%.结论 未发表尤其是正在进行的试验的信息可在英特网上找到.潜在的问题是如何评价未经同行评审的电子出版物的质量.急需更强的搜索工具.建议用"Open Trial Initiative"定义英特网发表试验的语法,以加强试验登记的共同操作性.因此,专门的搜索引擎可找到更多有关正在进行和已完成的临床试验信息.