The goal of JAGS (Just Another Gibbs Sampler) software is to remedy the short of BUGS software that unable to running on a system besides Microsoft Windows, such as Unix or Linux. JAGS owns independent computing function and formula of Bayesian theory; it is mischaracterized with simple user interface, good system compatibility, smoother operation, and good interactivity with other programming software. However, due to the limitations of lacking function for results data reading and unscrambling and graph plotting, the popularization and application of JAGS software is restricted. Calling JAGS software from R software through R2jags package, rjags package, or runjags package can overcome these limitations. The operating principle of these three packages is calling JAGS software in the framework of the R software, they have similar functional structure and all have easy maneuverability, concise command, perfect function of data reading and unscrambling and graph drawing; however, there are some differences among them in practice. This article introduces how to performing network meta-analysis by calling JAGS software from R through these three packages.
The nlme package is developed based on the generalized least squares (gls) and linear mixed-effects model (lme). It can perform meta-analysis based on linear and nonlinear mixed effects models in R language. When conducting meta-analysis using nlme package in R language, the first step is to translate the data into its logarithm estimation. In this article, we introduce how to perform network meta-analysis using R language nlme package and show the core step of data translation in detail.
R language could call OpenBUGS software for performing network meta-analysis using R2OpenBUGS package, BRugs package, and rbugs package. In this paper, we introduced how to implement network meta-analysis using these three packages. The results show that the computed results are similar for the three packages; however, the rbugs package could not draw the plot, only R2OpenBUGS package could draw forest plot.
Dose-response meta-analysis, as a subset of meta-analysis, plays an important role in dealing with the relationship between exposure level and risk of diseases. Traditional models limited in linear regression between the independent variables and the dependent variable. With the development of methodology and functional model, Nonlinear regression method was applied to dose-response meta-analysis, such as restricted cubic spline regression, quadratic B-spline regression. However, in these methods, the term and order of the independent variables have been assigned that may not suit for any trend distribution and it may lead to over fitting. Flexible fraction polynomial regression is a good method to solve this problem, which modelling a flexible fraction polynomial and choosing the best fitting model by using the likelihood-ratio test for a more accurate evaluation. In this article, we will discuss how to conduct a dose-response meta-analysis by flexible fraction polynomial.
ObjectiveTo evaluate the efficacy and safety of bisphosphonates in preventing and treating glucocorticoid induced osteoporosis. MethodsDatabases including PubMed, EMbase, The Cochrane Library (Issue 1, 2016), CNKI, WanFang Data and VIP were searched to collect randomized controlled trials (RCTs) related bisphosphonates for the prevention and treatment of glucocorticoid induced osteoporosis from inception to January 2016. Two reviewers independently screened literature, extracted data, and evaluated the risk of bias of included studies. Meta-analysis was performed using RevMan 5.3 software. ResultsA total of 20 RCTs were included, which involved 2 330 patients. The results of meta-analysis showed that, compared with the placebo group, the bisphosphonates group could significantly increase the bone mineral density (BMD) at lumbar and femoral neck (MD=3.70, 95%CI 2.65 to 4.75, P<0.000 01; MD=2.18, 95%CI 1.30 to 3.06, P<0.000 01), but the bisphosphonates group could not decrease the incidence rates of vertebral fracture or non-vertebral fracture (OR=0.66, 95%CI 0.38 to 1.16, P=0.15; OR=0.73, 95%CI 0.42 to 1.28, P=0.28). There were no significant differences in the incidence rates of total adverse reactions and total severe adverse reactions between the two groups (OR=0.89, 95%CI 0.62 to 1.28, P=0.53; OR=0.93, 95%CI 0.62 to 1.39, P=0.72). ConclusionCurrent evidence shows that, compared with placebo, bisphosphonates canld effectively prevent and treat the decrease of bone mineral density of glucocorticoid induced osteoporosis, not decrease the incidence of fracture, but not increase the incidence of adverse reactions.
when we conducted a meta-analysis, it is often an annoying thing to deal with the data of discrete exposure and multiple outcomes. Conventional "high VS low" approach abandoned the information of middle category, and led to the loss of statistical power. In this paper, we introduced a method and software to combine the groups of discrete exposure and multiple outcomes in the meta-analysis of epidemiological studies. Firstly, we introduced the transforming and combination theory and method, and then, we conducted the combination using EXCEL macro software. The result was consistent with the results of the original data in the combination of discrete exposure and multiple outcome data. Therefore, in the case of the original research data cannot be acquired, EXCEL macro software can be a good solution.
Stata is statistical software that combines programming and un-programming, which is easy to operate, of high efficiency and good expansibility. In performing meta-analysis, Stata software also presents powerful function. The mvmeta package of Stata software is based on a multiple regression model to conduct network meta-analysis, and it also processes "multiple outcomes-multivariate" data. Currently, the disadvantages of mvmeta package include relatively cumbersome process, poor interest-risk sorting, and lack of drawing function in the process of conducting network meta-analysis. In this article, we introduce how to implement network meta-analysis using this package based on cases.
BugsXLA is a Microsoft Excel add-in that facilitates Bayesian analysis of GLMMs and other complex model types by providing an easy to use interface for the BUGS package. BugsXLA macro is of good compatibility, ease to operation, smoothly running, low memory cost, and ease for data entry, extraction, and storage compared with other software which can calling BUGS to perform network meta-analysis currently. BugsXLA macro also integrates data storage and calculation. However, its function of plot drawing is very simple and only draws the density plot nowadays; Moreover, the function for calling WinBUGS is mature while is premature for calling OpenBUGS.
The netmeta package is specialized for implementing network meta-analysis. This package was developed based on the theories of classical frequentist under R language framework. The netmeta package overcomes some difficulties of the software and/or packages based on the theories of Bayesian, for these software and/or packages need to set prior value when conducting network meta-analysis. The netmeta package also has the advantages of simple operation process and ease to operate. Moreover, this package can calculate and present the individual matched and pooled results based on the random and fixed effect model at the same time. It also can draw forest plots. This article gives a briefly introduction to show the process to conduct network meta-analysis using netmeta package.
ObjectiveTo compare the characteristics and functions of the network meta-analysis software and for providing references for users. MethodsPubMed, CNKI, official website of Stata and R, and Google were searched to collect the software and packages that can perform network meta-analysis up to July 2014. After downloading the software, packages, and their user guides, we used the software and packages to calculate a typical example. The characteristics, functions, and computed results were compared and analyzed. ResultsFinally, 11 types of software were included, including programming and non-programming software. They were developed mainly based on Bayesian or Frequentist. Most types of software have the characteristics of easy to operate, easy to master, exactitude calculation, or good graphing; however, there is no software that has the exactitude calculation and good graphing at the same time, which needs two or more kinds of software combined to achieve. ConclusionWe suggest the user to choose the software at least according to personal programming basis and custom; and the user can consider to choose two or more kinds of software combined to finish the objective network meta-analysis. We also suggest to develop a kind of software which is characterized of fully function, easy operation, and free.