An alternative model for bivariate random-effects meta-analysis when the within-study correlations are unknown.

Multivariate meta-analysis models can be used to synthesize multiple, correlated endpoints such as overall and disease-free survival. A hierarchical framework for multivariate random-effects meta-analysis includes both within-study and between-study correlation. The within-study correlations are assumed known, but they are usually unavailable, which limits the multivariate approach in practice. In this paper, we consider synthesis of 2 correlated endpoints and propose an alternative model for bivariate random-effects meta-analysis (BRMA). This model maintains the individual weighting of each study in the analysis but includes only one overall correlation parameter, rho, which removes the need to know the within-study correlations. Further, the only data needed to fit the model are those required for a separate univariate random-effects meta-analysis (URMA) of each endpoint, currently the common approach in practice. This makes the alternative model immediately applicable to a wide variety of evidence synthesis situations, including studies of prognosis and surrogate outcomes. We examine the performance of the alternative model through analytic assessment, a realistic simulation study, and application to data sets from the literature. Our results show that, unless rho is very close to 1 or -1, the alternative model produces appropriate pooled estimates with little bias that (i) are very similar to those from a fully hierarchical BRMA model where the within-study correlations are known and (ii) have better statistical properties than those from separate URMAs, especially given missing data. The alternative model is also less prone to estimation at parameter space boundaries than the fully hierarchical model and thus may be preferred even when the within-study correlations are known. It also suitably estimates a function of the pooled estimates and their correlation; however, it only provides an approximate indication of the between-study variation. The alternative model greatly facilitates the utilization of correlation in meta-analysis and should allow an increased application of BRMA in practice.

[1]  Haitao Chu,et al.  Bivariate meta-analysis of sensitivity and specificity with sparse data: a generalized linear mixed model approach. , 2006, Journal of clinical epidemiology.

[2]  Roger M Harbord,et al.  A unification of models for meta-analysis of diagnostic accuracy studies. , 2007, Biostatistics.

[3]  P C Lambert,et al.  An evaluation of bivariate random‐effects meta‐analysis for the joint synthesis of two correlated outcomes , 2007, Statistics in medicine.

[4]  Theo Stijnen,et al.  Combining multiple outcome measures in a meta‐analysis: an application , 2003, Statistics in medicine.

[5]  D. Cox,et al.  Generalized least squares for the synthesis of correlated information. , 2003, Biostatistics.

[6]  F Mosteller,et al.  Meta-analysis of multiple outcomes by regression with random effects. , 1998, Statistics in medicine.

[7]  P C Lambert,et al.  A comparison of summary patient-level covariates in meta-regression with individual patient data meta-analysis. , 2002, Journal of clinical epidemiology.

[8]  D. Altman,et al.  CONSORT statement: extension to cluster randomised trials , 2004, BMJ : British Medical Journal.

[9]  Richard D Riley,et al.  A Systematic Review of Molecular and Biological Tumor Markers in Neuroblastoma , 2004, Clinical Cancer Research.

[10]  Richard D Riley,et al.  Sensitivity analyses allowed more appropriate and reliable meta-analysis conclusions for multiple outcomes when missing data was present. , 2004, Journal of clinical epidemiology.

[11]  M J Daniels,et al.  Meta-analysis for the evaluation of potential surrogate markers. , 1997, Statistics in medicine.

[12]  Richard D Riley,et al.  Evidence synthesis combining individual patient data and aggregate data: a systematic review identified current practice and possible methods. , 2007, Journal of clinical epidemiology.

[13]  K. Dear,et al.  Iterative generalized least squares for meta-analysis of survival data at multiple times. , 1994, Biometrics.

[14]  D. Rubin INFERENCE AND MISSING DATA , 1975 .

[15]  Mark C Simmonds,et al.  Meta-analysis of individual patient data from randomized trials: a review of methods used in practice , 2005, Clinical trials.

[16]  J. Gentle Numerical Linear Algebra for Applications in Statistics , 1998 .

[17]  Betsy Jane Becker,et al.  Modeling multivariate effect sizes. , 1988 .

[18]  Kerrie Mengersen,et al.  Multivariate meta‐analysis , 2003, Statistics in medicine.

[19]  Richard D Riley,et al.  Beyond the Bench: Hunting Down Fugitive Literature , 2004, Environmental Health Perspectives.

[20]  Theo Stijnen,et al.  Advanced methods in meta‐analysis: multivariate approach and meta‐regression , 2002, Statistics in medicine.

[21]  Johannes B Reitsma,et al.  Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. , 2005, Journal of clinical epidemiology.

[22]  V. Hasselblad,et al.  Meta-analysis of Multitreatment Studies , 1998, Medical decision making : an international journal of the Society for Medical Decision Making.

[23]  D. Rubin INFERENCE AND MISSING DATA , 1975 .

[24]  Richard D Riley,et al.  Meta‐analysis of genetic studies using Mendelian randomization—a multivariate approach , 2005, Statistics in medicine.

[25]  S G Thompson,et al.  Systematic Review: Why sources of heterogeneity in meta-analysis should be investigated , 1994, BMJ.

[26]  Paul S Albert,et al.  Assessing surrogates as trial endpoints using mixed models , 2005, Statistics in medicine.

[27]  C S Berkey,et al.  Multiple-outcome meta-analysis of clinical trials. , 1996, Statistics in medicine.

[28]  D. Hoaglin,et al.  Multiple-outcomes Meta-analysis of Treatments for Periodontal Disease , 1995, Journal of dental research.