Bayesian mixed treatment comparisons meta-analysis for correlated outcomes subject to reporting bias.
ABSTRACT: Many randomized controlled trials (RCTs) report more than one primary outcome. As a result, multivariate meta-analytic methods for the assimilation of treatment effects in systematic reviews of RCTs have received increasing attention in the literature. These methods show promise with respect to bias reduction and efficiency gain compared to univariate meta-analysis. However, most methods for multivariate meta-analysis have focused on pairwise treatment comparisons (i.e., when the number of treatments is two). Current methods for mixed treatment comparisons (MTC) meta-analysis (i.e., when the number of treatments is more than two) have focused on univariate or very recently, bivariate outcomes. To broaden their application, we propose a framework for MTC meta-analysis of multivariate (? 2) outcomes where the correlations among multivariate outcomes within- and between-studies are accounted for through copulas, and the joint modeling of multivariate random effects, respectively. We consider a Bayesian hierarchical model using Markov Chain Monte Carlo methods for estimation. An important feature of the proposed framework is that it allows for borrowing of information across correlated outcomes. We show via simulation that our approach reduces the impact of outcome reporting bias (ORB) in a variety of missing outcome scenarios. We apply the method to a systematic review of RCTs of pharmacological treatments for alcohol dependence, which tends to report multiple outcomes potentially subject to ORB.
Project description:Outcome reporting bias (ORB) is recognized as a threat to the validity of both pairwise and network meta-analysis (NMA). In recent years, multivariate meta-analytic methods have been proposed to reduce the impact of ORB in the pairwise setting. These methods have shown that multivariate meta-analysis can reduce bias and increase efficiency of pooled effect sizes. However, it is unknown whether multivariate NMA (MNMA) can similarly reduce the impact of ORB. Additionally, it is quite challenging to implement MNMA due to the fact that correlation between treatments and outcomes must be modeled; thus, the dimension of the covariance matrix and number of components to estimate grows quickly with the number of treatments and number of outcomes. To determine whether MNMA can reduce the effects of ORB on pooled treatment effect sizes, we present an extensive simulation study of Bayesian MNMA. Via simulation studies, we show that MNMA reduces the bias of pooled effect sizes under a variety of outcome missingness scenarios, including missing at random and missing not at random. Further, MNMA improves the precision of estimates, producing narrower credible intervals. We demonstrate the applicability of the approach via application of MNMA to a multi-treatment systematic review of randomized controlled trials of anti-depressants for the treatment of depression in older adults.
Project description:We aimed to integrate evidence from all randomized controlled trials (RCTs) and assess the impact of different doses of exenatide or liraglutide on major gastrointestinal adverse events (GIAEs) in type 2 diabetes (T2DM).RCTs evaluating different doses of exenatide and liraglutide against placebo or an active comparator with treatment duration ?4 weeks were searched and reviewed. A total of 35, 32 and 28 RCTs met the selection criteria evaluated for nausea, vomiting, and diarrhea, respectively. Pairwise random-effects meta-analyses and mixed treatment comparisons (MTC) of all RCTs were performed.All GLP-1 dose groups significantly increased the probability of nausea, vomiting and diarrhea relative to placebo and conventional treatment. MTC meta-analysis showed that there was 99.2% and 85.0% probability, respectively, that people with exenatide 10??g twice daily (EX10BID) was more vulnerable to nausea and vomiting than those with other treatments. There was a 78.90% probability that liraglutide 1.2?mg once daily (LIR1.2) has a higher risk of diarrhea than other groups. A dose-dependent relationship of exenatide and liraglutide on GIAEs was observed.Our MTC meta-analysis suggests that patients should be warned about these GIAEs in early stage of treatment by GLP-1s, especially by EX10BID and LIR1.2, to promote treatment compliance.
Project description:Bayesian statistical approaches to mixed treatment comparisons (MTCs) are becoming more popular because of their flexibility and interpretability. Many randomized clinical trials report multiple outcomes with possible inherent correlations. Moreover, MTC data are typically sparse (although richer than standard meta-analysis, comparing only two treatments), and researchers often choose study arms based upon which treatments emerge as superior in previous trials. In this paper, we summarize existing hierarchical Bayesian methods for MTCs with a single outcome and introduce novel Bayesian approaches for multiple outcomes simultaneously, rather than in separate MTC analyses. We do this by incorporating partially observed data and its correlation structure between outcomes through contrast-based and arm-based parameterizations that consider any unobserved treatment arms as missing data to be imputed. We also extend the model to apply to all types of generalized linear model outcomes, such as count or continuous responses. We offer a simulation study under various missingness mechanisms (e.g., missing completely at random, missing at random, and missing not at random) providing evidence that our models outperform existing models in terms of bias, mean squared error, and coverage probability then illustrate our methods with a real MTC dataset. We close with a discussion of our results, several contentious issues in MTC analysis, and a few avenues for future methodological development.
Project description:In randomized controlled trials (RCTs) with time-to-event outcomes, the difference in restricted mean survival times (RMSTD) offers an absolute measure of the treatment effect on the time scale. Computation of the RMSTD relies on the choice of a time horizon, $\tau$. In a meta-analysis, varying follow-up durations may lead to the exclusion of RCTs with follow-up shorter than $\tau$. We introduce an individual patient data multivariate meta-analysis model for RMSTD estimated at multiple time horizons. We derived the within-trial covariance for the RMSTD enabling the synthesis of all data by borrowing strength from multiple time points. In a simulation study covering 60 scenarios, we compared the statistical performance of the proposed method to that of two univariate meta-analysis models, based on available data at each time point and based on predictions from flexible parametric models. Our multivariate model yields smaller mean squared error over univariate methods at all time points. We illustrate the method with a meta-analysis of five RCTs comparing transcatheter aortic valve replacement (TAVR) with surgical replacement in patients with aortic stenosis. Over 12, 24, and 36 months of follow-up, those treated by TAVR live 0.28 [95% confidence interval (CI) 0.01 to 0.56], 0.46 (95% CI $-$0.08 to 1.01), and 0.79 (95% CI $-$0.43 to 2.02) months longer on average compared to those treated by surgery, respectively.
Project description:BACKGROUND:Depression is common in patients with coronary artery disease (CAD) and is associated with poorer outcomes and higher costs. Several randomised controlled trials (RCTs) targeting depression, of various modalities (including pharmacological, psychotherapeutic and other approaches), have been conducted and summarised in pairwise meta-analytic reviews. However, no study has considered the cumulative evidence within a network, which can provide valuable indirect comparisons and information about the relative efficacy of interventions. Therefore, we will adopt a review of review methodology to develop a network meta-analysis (NMA) of depression interventions for depression in CAD. METHODS:We will search relevant databases from inception for systematic reviews of RCTs of depression treatments for people with CAD, supplementing this with comprehensive searches for recent or ongoing studies. We will extract data from and summarise characteristics of individual RCTs, including participants, study characteristics, outcome measures and adverse events. Cochrane risk of bias ratings will also be extracted or if not present will be conducted by the authors. RCTs that compare depression treatments (grouped as pharmacological, psychotherapeutic, combined pharmacological/psychotherapeutic, exercise, collaborative care) to placebo, usual care, waitlist control or attention controls, or directly in head-to-head comparisons, will be included. Primary outcomes will be the change in depressive symptoms (summarised with a standardised mean difference) and treatment acceptability (treatment discontinuation: % of people who withdrew). Secondary outcomes will include change in 6-month depression outcomes, health-related quality of life (HRQoL), mortality, cardiovascular morbidity, health services use and adverse events. Secondary analyses will form further networks with individual anti-depressants and psychotherapies. We will use frequentist, random effects multivariate network meta-analysis to synthesise the evidence for depression intervention and to achieve a ranking of treatments, using Stata. Rankograms and surface under the cumulative ranking curves will be used for treatment ranking. Local and global methods will evaluate consistency. GRADE will be used to assess evidence quality for primary outcomes. DISCUSSION:The present review will address uncertainties about the evidence in terms of depression management in CAD and may allow for a ranking of treatments, including providing important information for future research efforts. SYSTEMATIC REVIEW REGISTRATION:PROSPERO CRD42018108293.
Project description:Small study effects occur when smaller studies show different, often larger, treatment effects than large ones, which may threaten the validity of systematic reviews and meta-analyses. The most well-known reasons for small study effects include publication bias, outcome reporting bias, and clinical heterogeneity. Methods to account for small study effects in univariate meta-analysis have been extensively studied. However, detecting small study effects in a multivariate meta-analysis setting remains an untouched research area. One of the complications is that different types of selection processes can be involved in the reporting of multivariate outcomes. For example, some studies may be completely unpublished while others may selectively report multiple outcomes. In this paper, we propose a score test as an overall test of small study effects in multivariate meta-analysis. Two detailed case studies are given to demonstrate the advantage of the proposed test over various naive applications of univariate tests in practice. Through simulation studies, the proposed test is found to retain nominal Type I error rates with considerable power in moderate sample size settings. Finally, we also evaluate the concordance between the proposed tests with the naive application of univariate tests by evaluating 44 systematic reviews with multiple outcomes from the Cochrane Database.
Project description:Multiple outcomes multivariate meta-analysis (MOMA) is gaining in popularity as a tool for jointly synthesizing evidence coming from studies that report effect estimates for multiple correlated outcomes. Models for MOMA are available for the case of the pairwise meta-analysis of two treatments for multiple outcomes. Network meta-analysis (NMA) can be used for handling studies that compare more than two treatments; however, there is currently little guidance on how to perform an MOMA for the case of a network of interventions with multiple outcomes. The aim of this paper is to address this issue by proposing two models for synthesizing evidence from multi-arm studies reporting on multiple correlated outcomes for networks of competing treatments. Our models can handle continuous, binary, time-to-event or mixed outcomes, with or without availability of within-study correlations. They are set in a Bayesian framework to allow flexibility in fitting and assigning prior distributions to the parameters of interest while fully accounting for parameter uncertainty. As an illustrative example, we use a network of interventions for acute mania, which contains multi-arm studies reporting on two correlated binary outcomes: response rate and dropout rate. Both multiple-outcomes NMA models produce narrower confidence intervals compared with independent, univariate network meta-analyses for each outcome and have an impact on the relative ranking of the treatments.
Project description:BACKGROUND: A variety of pharmacological and surgical treatments have been developed for heavy menstrual bleeding (HMB), which can have negative physical, social, psychological, and economic consequences. We conducted a systematic literature review and mixed-treatment-comparison (MTC) meta-analysis of available data from randomized controlled trials (RCTs) to derive estimates of efficacy for 8 classes of treatments for HMB, to inform health-economic analysis and future studies. METHODS: A systematic review identified RCTs that reported data on menstrual blood loss (MBL) at baseline and one or more follow-up times. Eight treatment classes were considered: COCs, danazol, endometrial ablation, LNG-IUS, placebo, progestogens given for less than 2 weeks out of 4 during the menstrual cycle, progestogens given for close to 3 weeks out of 4, and TXA. The primary measure of efficacy was the proportion of women who achieved MBL < 80 mL per cycle (month), as measured by the alkaline hematin method. A score less than 100 on an established pictorial blood-loss assessment chart (PBAC) was considered an acceptable substitute for MBL < 80 mL. Estimates of efficacy by treatment class and time were obtained from a Bayesian MTC model. The model also included effects for treatment class, study, and the combination of treatment class and study and an adjustment for baseline mean MBL. Several methodological challenges complicated the analysis. Some trials reported various summary statistics for MBL or PBAC, requiring estimation (with less precision) of % MBL < 80 mL or % PBAC < 100. Also, reported follow-up times varied substantially. RESULTS: The evidence network involved 34 RCTs, with follow-up times from 1 to 36 months. Efficacy at 3 months of follow-up (estimated as the posterior median) ranged from 87.5% for the levonorgestrel-releasing intrauterine system (LNG-IUS) to 14.2% for progestogens administered for less than 2 weeks out of 4 in the menstrual cycle. The 95% credible intervals for most estimates were quite wide, mainly because of the limited evidence for many combinations of treatment class and follow-up time and the uncertainty from estimating % MBL < 80 mL or % PBAC < 100 from summary statistics. CONCLUSIONS: LNG-IUS and endometrial ablation are very efficacious in treating HMB. The study yielded useful insights on using MTC in sparse evidence networks. Diversity of outcome measures and follow-up times in the HMB literature presented considerable challenges. The Bayesian credible intervals reflected the various sources of uncertainty.
Project description:BACKGROUND: Outcome reporting bias (ORB) occurs when variables are selected for publication based on their results. This can impact upon the results of a meta-analysis, biasing the pooled treatment effect estimate.The aim of this paper is to show how to assess a systematic review and corresponding trial reports for ORB using an example review of intravenous and nebulised magnesium in the treatment of asthma. METHODS: The review was assessed for ORB by 1) checking the reasons, when available, for excluding studies to ensure that no studies were excluded because they did not report the outcomes of interest in the review; 2) assessing the eligible studies as to whether the review outcomes of interest were reported. Each study was classified using a system developed in the ORBIT (Outcome Reporting Bias In Trials) project to indicate whether ORB was suspected and a reason for the suspicion. Authors of trials that did not report the outcomes of interest were contacted for information. A sensitivity analysis was performed to assess the robustness of the conclusions of the review to this potential source of bias. RESULTS: Twenty-four studies were included in the review; two studies had been excluded for not reporting either of the two outcomes of interest. Six included studies did not report hospital admission and two did not report pulmonary function. There was high suspicion of outcome reporting bias in four studies. Results from the sensitivity analysis indicate that review conclusions were not overturned. CONCLUSION: This paper demonstrates, with the example of the magnesium review, how to assess a review for outcome reporting bias. A review should not exclude studies if they have not reported the outcomes of interest and should consider the potential for outcome reporting bias in all included studies.
Project description:When combining results across related studies, a multivariate meta-analysis allows the joint synthesis of correlated effect estimates from multiple outcomes. Joint synthesis can improve efficiency over separate univariate syntheses, may reduce selective outcome reporting biases, and enables joint inferences across the outcomes. A common issue is that within-study correlations needed to fit the multivariate model are unknown from published reports. However, provision of individual participant data (IPD) allows them to be calculated directly. Here, we illustrate how to use IPD to estimate within-study correlations, using a joint linear regression for multiple continuous outcomes and bootstrapping methods for binary, survival and mixed outcomes. In a meta-analysis of 10 hypertension trials, we then show how these methods enable multivariate meta-analysis to address novel clinical questions about continuous, survival and binary outcomes; treatment-covariate interactions; adjusted risk/prognostic factor effects; longitudinal data; prognostic and multiparameter models; and multiple treatment comparisons. Both frequentist and Bayesian approaches are applied, with example software code provided to derive within-study correlations and to fit the models.