Bayesian multivariate meta‐analysis of multiple factors

Lin, Lifeng; Chu, Haitao

doi:10.1002/jrsm.1293

Cited by 17 publications

(22 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Technical details and applications to multivariate meta-analyses are in Lu and Ades, 4 Wei and Higgins, 31 and Lin and Chu. 32 However, using this approach, the marginal distributions of the ij depend strongly on the indexes i and j; see figure 3 in Wei and Higgins. 31 The different marginal distributions lack statistical or clinical interpretations and meta-analysts may reasonably be concerned about the potential impact of using different marginal priors for different correlations.…”

Section: Summary and Discussionmentioning

confidence: 99%

“…Then we can place a weakly informative prior on L through a spherical parameterization. Technical details and applications to multivariate meta‐analyses are in Lu and Ades, 4 Wei and Higgins, 31 and Lin and Chu 32 . However, using this approach, the marginal distributions of the ρ ij depend strongly on the indexes i and j ; see figure 3 in Wei and Higgins 31 .…”

Section: Summary and Discussionmentioning

confidence: 99%

See 1 more Smart Citation

The impact of covariance priors on arm‐based Bayesian network meta‐analyses with binary outcomes

Wang

Lin

Hodges

et al. 2020

Statistics in Medicine

Self Cite

View full text Add to dashboard Cite

Bayesian analyses with the arm‐based (AB) network meta‐analysis (NMA) model require researchers to specify a prior distribution for the covariance matrix of the treatment‐specific event rates in a transformed scale, for example, the treatment‐specific log‐odds when a logit transformation is used. The commonly used conjugate prior for the covariance matrix, the inverse‐Wishart (IW) distribution, has several limitations. For example, although the IW distribution is often described as noninformative or weakly informative, it may in fact provide strong information when some variance components are small (eg, when the standard deviation of study‐specific log‐odds of a treatment is smaller than 1/2), as is common in NMAs with binary outcomes. In addition, the IW prior generally leads to underestimation of correlations between treatment‐specific log‐odds, which are critical for borrowing strength across treatment arms to estimate treatment effects efficiently and to reduce potential bias. Alternatively, several separation strategies (ie, separate priors on variances and correlations) can be considered. To study the IW prior's impact on NMA results and compare it with separation strategies, we did simulation studies under different missing‐treatment mechanisms. A separation strategy with appropriate priors for the correlation matrix and variances performs better than the IW prior, and should be recommended as the default vague prior in the AB NMA approach. Finally, we reanalyzed three case studies and illustrated the importance, when performing AB‐NMA, of sensitivity analyses with different prior specifications on variances.

show abstract

Section: Summary and Discussionmentioning

confidence: 99%

Section: Summary and Discussionmentioning

confidence: 99%

The impact of covariance priors on arm‐based Bayesian network meta‐analyses with binary outcomes

Wang

Lin

Hodges

et al. 2020

Statistics in Medicine

Self Cite

View full text Add to dashboard Cite

show abstract

“…If HRs were not provided, they were estimated as relevant effect measures from the given median survival times or survival rates at time points. The Bayesian hybrid model for random effect multivariate meta‐analysis 17 was used to evaluate surrogacy between the HRs for OS and each endpoint. A frequentist hybrid model for random‐effects multivariate meta‐analysis was used for sensitivity analysis.…”

Section: Methodsmentioning

confidence: 99%

Searching for potential surrogate endpoints of overall survival in clinical trials for patients with prostate cancer

et al. 2021

View full text Add to dashboard Cite

Background The purpose of this study was to investigate the correlation between overall survival (OS) and other clinical outcomes in patients with prostate cancer. Further, we conducted subgroup analysis in the correlation of OS. Aim This study intended to investigate potential surrogate endpoints of OS for prostate cancer by examining the correlation between OS and the other endpoints. Methods We performed a systematic review through a literature search by computer‐based searches of the Medline database (January 1965 and May 2014). Results The contents of 115 studies with endpoint as OS were analyzed in our study. Our results showed that 47.8% (55/115) of the studies used progression‐free survival as an endpoint besides OS, followed by time to progression (43.5% [50/115]) and PSA response (40.9% [47/115]). Also, the relationship between OS and each surrogate endpoint was examined using the hazard ratio (HR) by a Bayesian hybrid model for random effect multivariate meta‐analysis. Our results showed that the endpoint that had the highest correlation with OS was progression‐free survival (PFS) with an estimated marginal correlation of 0.939 (95%CI: 0.900, 0.967). Furthermore, our stratified analysis identified PFS in castration‐resistant prostate cancer patients (0.937), in sensitive patients (0.932), in none of chemotherapy patients (0.929), in first line of the chemotherapy (0.948), in patients who received no Docetaxel previously (0.942), in both symptomatic and asymptomatic patients (0.950), in patients who received only chemotherapy (0.956), and in phase III (0.960), time to progression (TTP) in castration‐resistant prostate cancer (CRPC) patients (0.942), in metastasis patients (0.948), in both symptomatic and asymptomatic patients (0.953), in patients who received only chemotherapy (0.938), and in Phase III (0.927) as endpoints, which showed a lower limit for 95% CI of estimated marginal correlation ≥0.850 with overall survival. Conclusions Our study suggests that PFS is a potential surrogate endpoint of OS in clinical trials for patients with prostate cancer. It also suggests potential surrogate endpoints for CRPC and locally advanced prostate cancer.

show abstract

“…Another major benefit of Bayesian methods is that prior information can be explicitly incorporated in meta-analytic models and, thus, have an impact on the results [ 25 ]. In the literature of multivariate meta-analysis of multiple outcomes and/or multiple treatments, Bayesian methods are a standard approach for effectively modeling complicated variance-covariance structures [ 26 , 27 ]. However, frequentist methods still dominate conventional univariate meta-analyses that compare each pair of treatments for each outcome separately [ 28 ].…”

Section: Introductionmentioning

confidence: 99%

Bayesian Methods for Meta-Analyses of Binary Outcomes: Implementations, Examples, and Impact of Priors

Amer

Thompson

Lin

2021

IJERPH

Self Cite

View full text Add to dashboard Cite

Bayesian methods are an important set of tools for performing meta-analyses. They avoid some potentially unrealistic assumptions that are required by conventional frequentist methods. More importantly, meta-analysts can incorporate prior information from many sources, including experts’ opinions and prior meta-analyses. Nevertheless, Bayesian methods are used less frequently than conventional frequentist methods, primarily because of the need for nontrivial statistical coding, while frequentist approaches can be implemented via many user-friendly software packages. This article aims at providing a practical review of implementations for Bayesian meta-analyses with various prior distributions. We present Bayesian methods for meta-analyses with the focus on odds ratio for binary outcomes. We summarize various commonly used prior distribution choices for the between-studies heterogeneity variance, a critical parameter in meta-analyses. They include the inverse-gamma, uniform, and half-normal distributions, as well as evidence-based informative log-normal priors. Five real-world examples are presented to illustrate their performance. We provide all of the statistical code for future use by practitioners. Under certain circumstances, Bayesian methods can produce markedly different results from those by frequentist methods, including a change in decision on statistical significance. When data information is limited, the choice of priors may have a large impact on meta-analytic results, in which case sensitivity analyses are recommended. Moreover, the algorithm for implementing Bayesian analyses may not converge for extremely sparse data; caution is needed in interpreting respective results. As such, convergence should be routinely examined. When select statistical assumptions that are made by conventional frequentist methods are violated, Bayesian methods provide a reliable alternative to perform a meta-analysis.

show abstract

Bayesian multivariate meta‐analysis of multiple factors

Cited by 17 publications

References 49 publications

The impact of covariance priors on arm‐based Bayesian network meta‐analyses with binary outcomes

The impact of covariance priors on arm‐based Bayesian network meta‐analyses with binary outcomes

Searching for potential surrogate endpoints of overall survival in clinical trials for patients with prostate cancer

Bayesian Methods for Meta-Analyses of Binary Outcomes: Implementations, Examples, and Impact of Priors

Contact Info

Product

Resources

About