Statistical methods in medical research bivariate random. Bivariate random effects models for meta analysis of comparative studies with binary outcomes. Multivariate and network metaanalysis of multiple outcomes. These bivariate random effects models use all available data without ad hoc continuity corrections, and accounts for the potential correlation between treatment or exposure and control groups within studies naturally. I 2 29%, which indicates that a 1 gl increase in fibrinogen levels is associated, on average, with a 31% relative increase in the hazard of cardiovascular. Fixed and random effects models have been the most popular techniques for pooling effects in meta analysis. This use of the bivariate random effects model for metaanalysis is perhaps especially appealing due to the pioneering work of harbord et al. Bivariate random effect models are currently one of the main methods recommended to synthesize diagnostic test accuracy studies. These bivariate random effects models use all available data without ad hoc.
The withinstudy correlations are assumed known, but they are usually unavailable, which limits the multivariate approach in practice. Simply select your manager software from the list below and click on download. The existing bivariate randomeffects models, which jointly model bivariate accuracy indices e. Multivariate randomeffects metaanalysis researchgate. An r package for copulabased bivariate betabinomial models for.
The analyses were carried out using the software package sas proc nlmixed. Bayesian randomeffects metaanalysis using the bayesmeta r. These parameters can, for example, refer to multiple outcomes or comparisons between more than two groups. Previously, we showed how to perform a fixedeffect model metaanalysis using the metagen and metacont functions.
Inference for correlated effect sizes using multiple univariate metaa. Multivariate metaanalysis is becoming more commonly used and the techniques and related computer software, although continually under development, are now in place. Let n ki be the number of subjects, and p ki be the probability of success for the i th study i 1, 2, n in the k th treatment or exposure group with k 1 denoting the placebo or unexposed group and k 2 denoting the treated or exposed group. This page contains the necessary material for a handson tutorial on multivariate meta analysis using stata. Bivariate random effects metaanalysis of diagnostic studies using generalized linear mixed models show all authors. Conventional metaanalysis methods either exclude such studies or. Bivariate random effects models for metaanalysis of comparative studies. Imputing covariance matrices for metaanalysis of correlated.
Fiestas navarretea and robert benamouzigb background current literature evidences higher accuracy. Metaanalyses of diagnostic accuracy in imaging journals. Overview one goal of a metaanalysis will often be to estimate the overall, or combined effect. Modeling sensitivity and specificity using the bivariate beta distribution. Whitea the multivariate random effects model is a generalization of the standard univariate model. Consider a metaanalysis of n diagnostic accuracy studies, each of them providing information as a twobytwo table reporting the number of true positives, true negatives, false positives and false negatives, denoted by n 11i,n 00i,n 10i and n 01i, respectively. A multivariate randomeffects metaanalysis was performed using the mvmeta command white, 2009. Bivariate random effects formulation for meta analysis. Bivariate randomeffects formulation for metaanalysis. When doing so it is important to maintain the randomisation and clustering of patients within trials30 and to incorporate random effects to allow for between trial heterogeneity in the magnitude of treatment effects. Consider a meta analysis of n diagnostic accuracy studies, each of them providing information as a twobytwo table reporting the number of true positives, true negatives, false positives and false negatives, denoted by n 11i,n 00i,n 10i and n 01i, respectively. In this paper we assess maximum likelihood estimation of a general normal model and a generalised model for bivariate randomeffects metaanalysis brma. Metaanalysis via multivariatemultilevel linear mixed.
A random effects model assumes variance both within and between studies, explaining the. Bivariate random effects model was applied to the pooled data in meta analysis. Methods for meta analysis of diagnostic test accuracy studies 145 the hierarchical summary receiver operator characteristic hsroc and bivariate random 146 effects techniques are considered the most appropriate methods for pooling sensitivity and 147 specificity from multiple diagnostic test accuracy studies. This copy is for your personal, noncommercial use only. This kind of metaanalysis models the betweenstudy heterogeneity with normally distributed random effects and accounts for a possible correlation between the primary. The fixed effects model, usually in conjunction with an underlying assumption of homogeneity, uses the inverse variance of each within study effect as a weighting factor to derive an overall average effect size and confidence interval. Multivariate metaanalysis is increasingly utilised in biomedical research to combine data of multiple comparative clinical studies for evaluating. Metaanalysis with linear and nonlinear multilevel models. A multivariate random effects meta analysis was performed using the mvmeta command white, 2009. Bivariate random effects model was applied to the pooled data in metaanalysis.
Bivariate random effects metaanalysis of roc curves l. Kalaian and raudenbush 1996 introduced a multivariate random effects model, which can be used to perform a joint meta analysis of studies that contribute effect sizes on distinct, related outcome constructs. Bivariate randomeffects metaanalysis of sensitivityandspecificitywiththebayesian sas proc mcmc. Metaanalysis of diagnostic test accuracy often involves mixture of casecontrol and cohort studies.
Bayesian random e ects meta analysis using the bayesmeta r package christian r over university medical center g ottingen abstract the random e ects or normalnormal hierarchical model is commonly utilized in a wide range of meta analysis applications. The use of the bivariate random effects model has been advocated for meta analyses of diagnostic accuracy studies chu et al. Bivariate random effects models for metaanalysis of. We have shown that the hsroc model and the bivariate random effects model for meta analysis of diagnostic accuracy studies are very closely related, and in common situations identical.
This approach could be implemented using the software stata. The umeta command performs ustatisticsbased random effects meta analysis on a dataset of univariate, bivariate or trivariate point estimates, sampling variances, and for bivariate or trivariate data, withinstudy correlations or covariances. Metaanalysis of receiver operating characteristic roccurve data is often done with fixedeffects models, which suffer many shortcomings. Sep 10, 2011 the multivariate random effects model is a generalization of the standard univariate model. In this paper, we consider synthesis of 2 correlated endpoints and propose an alternative model for bivariate random effects meta analysis brma. If all studies in the analysis were equally precise we could simply compute the mean of the effect sizes. A double simex approach for bivariate randomeffects meta. Sep, 2017 a standard univariate random effects meta analysis of the direct evidence from 14 trials gives a summary fully adjusted hazard ratio of 1. Nov 19, 2012 point estimates from univariate and bivariate random effects meta analyses were similar when performing pairwise univariate vs. Because of this heterogeneity, random effects models including the hierarchical summary receiver operating characteristic model and bivariate random effects meta analysis on sensitivities and specificities 6,8,10, which are identical in some situations, have been recommended 11,12,15. Pdf bivariate random effects metaanalysis of roc curves. Performance measures of the bivariate random effects model. The two make different assumptions about the nature of the studies, and these assumptions lead to different definitions for the combined effect, and different mechanisms for assigning weights. From bivariate random effects meta analyses fit using fully bayesian versus mle estimation using the exact binomial likelihood to represent withinstudy variability 24 figure 15.
Bivariate randomeffects metaanalysis and the estimation of. Jan 12, 2007 a multivariate randomeffects metaanalysis must incorporate and estimate the betweenstudy correlation. Methods for the absolute risk difference and relative risk haitao chu,1 lei nie,2, yong chen,3 yi huang4 and wei sun5 abstract multivariate meta analysis is increasingly utilised in biomedical research to combine data of multiple. In the absence of studylevel covariates, they are different parameterizations of the same model.
Subgroup analysis was used to investigate and resolve the source of heterogeneity between studies. Random effects of this form are useful to model clustering and hence nonindependence induced by a multilevel structure in the data e. Bivariate random effects metaanalysis of diagnostic studies using. Bivariate random effects metaanalysis of diagnostic.
Thereby the authors show that the bivariate random effects approach not only extends the sroc approach but also provides a unifying framework for other approaches. Multivariate metaanalysis is becoming more commonly used and the techniques and related computer software, although. An empirical assessment of bivariate methods for meta. If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. A new stata command, mvmeta, performs maximum likelihood, restricted maximum likelihood, or methodofmoments estimation of random effects multivariate meta analysis models. Bivariate random effects metaanalysis of roc curves. We then utilise the bivariate random effects models to reanalyse two recent metaanalysis data sets. Methods for meta analysis of diagnostic accuracy data were classified as traditional univariate fixed or random effects models or summary roc or recommended bivariate random effects, hierarchic summary roc, or traditional univariate fixed or random effects pooling if only a single diagnostic parameter was assessed. The effortless implementation with standard software is an interesting feature of the. Methodology and empirical evaluation in 50 metaanalyses.
Tutorial on multivariate metaanalysis computational. A simple and robust method for multivariate metaanalysis of. Ben dwamena asked earlier this month about using bivariate random effects modeling in meta analysis of accuracy indexes of diagnostic tests. I was interested in using bivariate random effects regression to meta analyze simultaneously sensitivity and specificity logit transforms as correlated heterogeneous outcomes versus a number of covariates such as study quality, sample size. They demonstrate the model using data from a synthesis on the effects of sat coaching, where many studies reported effects on both the. The authors bring the statistical meta analysis of roccurve data back into a framework of relatively standard multivariate meta analysis with random effects. Efficient twostep multivariate random effects metaanalysis of individual. However, only the logit transformation on sensitivity and specificit. However, only the logit transformation on sensitivity and specificity has been previously considered in the literature. Bivariate random effects metaanalysis of diagnostic studies. Differences of estimated logit sensitivity and specificity between models fit with bayesian methods versus mle both from bivariate random effects meta analyses. The methodology is described in ma and mazumdar, statistics in medicine 2011.
1085 1463 1472 595 657 1524 446 771 278 939 661 1504 1524 1527 1528 827 467 678 757 1041 961 486 3 506 43 1471 1273 961 1569 48 287 875 422 1362 1092 349 1122 232 610 736