Graphs of study contributions and covariate distributions for network meta-regression

Methodologies for network meta-analysis of randomised controlled trials in pain, anaesthesia, and perioperative medicine: a narrative review

Network meta-analysis has emerged as a method for analysing clinical trials, with a large increase in the number of publications over the past decade. Network meta-analysis offers advantages over traditional pairwise meta-analysis, including increased power, the ability to compare treatments not compared in the original trials, and the ability to rank treatments. However, network meta-analyses are inherently more complex than pairwise meta-analyses, requiring additional statistical expertise and assumptions. Many factors can affect the certainty of evidence from pairwise meta-analysis and can often lead to unreliable results. Network meta-analysis is prone to all these issues, although it has the additional assumption of transitivity. Here we review network meta-analyses, problems with their conduct and reporting, and methodological strategies that can be used by those conducting reviews to help improve the reliability of their findings. We provide evidence that violation of the assumption of transitivity is relatively common and inadequately considered in published network meta-analyses. We explain key concepts with clinically relevant examples for those unfamiliar with network meta-analysis to facilitate their appraisal and application of their results to clinical practice.

Arthroscopic procedures for degenerative rotator cuff disease: a systematic review and network meta-analysis

BACKGROUND

Treatment of rotator cuff diseases often involves various arthroscopic procedures but their combined effectiveness remains contentious, especially in complex cases.

METHODS

We focused on patients with degenerative shoulder cuff diseases requiring arthroscopic rotator cuff repair. Searches covered multiple databases (Medline, Embase, Web of Science, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, and Cochrane Clinical Answers) up to April 1, 2024. Bias risk was assessed using RevMan (v 5.4), and a network meta-analysis was conducted with netmeta (v 2.8).

RESULT

From 16 studies, 1232 patients (average age, 56.2 years; balanced sex ratio) were included. Arthroscopic rotator cuff repair ranked highest in functional score networks, surpassing other interventions. Physiotherapy was superior for pain relief compared to arthroscopic procedures combined with platelet-rich plasma (mean, 2.5; 95% confidence interval, 4.48-0.52). Arthroscopic rotator cuff repair and subacromial decompression were significantly superior to arthroscopic rotator cuff repair and subacromial decompression combined with platelet-rich plasma (MD, 1.80; 95% CI, 3.39-0.21).

DISCUSSION

Moderate bias risks were noted in both networks due to blinding issues and methodological quality reporting. Arthroscopic rotator cuff repair is favored for improving shoulder function, while other procedures or intra-articular treatments offer no significant benefits. Regarding pain management, physiotherapy is preferred; however, more evidence is needed to support this recommendation and caution is advised.

OTHER

Systematic review registration PROSPERO CRD42023450150.

Using individual participant data to improve network meta-analysis projects

A network meta-analysis combines the evidence from existing randomised trials about the comparative efficacy of multiple treatments. It allows direct and indirect evidence about each comparison to be included in the same analysis, and provides a coherent framework to compare and rank treatments. A traditional network meta-analysis uses aggregate data (eg, treatment effect estimates and standard errors) obtained from publications or trial investigators. An alternative approach is to obtain, check, harmonise and meta-analyse the individual participant data (IPD) from each trial. In this article, we describe potential advantages of IPD for network meta-analysis projects, emphasising five key benefits: (1) improving the quality and scope of information available for inclusion in the meta-analysis, (2) examining and plotting distributions of covariates across trials (eg, for potential effect modifiers), (3) standardising and improving the analysis of each trial, (4) adjusting for prognostic factors to allow a network meta-analysis of conditional treatment effects and (5) including treatment-covariate interactions (effect modifiers) to allow relative treatment effects to vary by participant-level covariate values (eg, age, baseline depression score). A running theme of all these benefits is that they help examine and reduce heterogeneity (differences in the true treatment effect between trials) and inconsistency (differences in the true treatment effect between direct and indirect evidence) in the network. As a consequence, an IPD network meta-analysis has the potential for more precise, reliable and informative results for clinical practice and even allows treatment comparisons to be made for individual patients and targeted populations conditional on their particular characteristics.

A two-stage prediction model for heterogeneous effects of treatments

Treatment effects vary across different patients, and estimation of this variability is essential for clinical decision-making. We aimed to develop a model estimating the benefit of alternative treatment options for individual patients, extending a risk modeling approach in a network meta-analysis framework. We propose a two-stage prediction model for heterogeneous treatment effects by combining prognosis research and network meta-analysis methods where individual patient data are available. In the first stage, a prognostic model to predict the baseline risk of the outcome. In the second stage, we use the baseline risk score from the first stage as a single prognostic factor and effect modifier in a network meta-regression model. We apply the approach to a network meta-analysis of three randomized clinical trials comparing the relapses in Natalizumab, Glatiramer Acetate, and Dimethyl Fumarate, including 3590 patients diagnosed with relapsing-remitting multiple sclerosis. We find that the baseline risk score modifies the relative and absolute treatment effects. Several patient characteristics, such as age and disability status, impact the baseline risk of relapse, which in turn moderates the benefit expected for each of the treatments. For high-risk patients, the treatment that minimizes the risk of relapse in 2 years is Natalizumab, whereas Dimethyl Fumarate might be a better option for low-risk patients. Our approach can be easily extended to all outcomes of interest and has the potential to inform a personalized treatment approach.

Bayesian network meta-regression hierarchical models using heavy-tailed multivariate random effects with covariate-dependent variances

Network meta-analysis (NMA) is gaining popularity in evidence synthesis and network meta-regression allows us to incorporate potentially important covariates into network meta-analysis. In this article, we propose a Bayesian network meta-regression hierarchical model and assume a general multivariate t distribution for the random treatment effects. The multivariate t distribution is desired for heavy-tailed random effects and converges to the multivariate normal distribution when the degrees of freedom go to infinity. Moreover, in NMA, some treatments are compared only in a single study. To overcome such sparsity, we propose a log-linear regression model for the variances of the random effects and incorporate aggregate covariates into modeling the variance components. We develop a Markov chain Monte Carlo sampling algorithm to sample from the posterior distribution via the collapsed Gibbs technique. We further use the deviance information criterion and the logarithm of the pseudo-marginal likelihood for model comparison. A simulation study is conducted and a detailed analysis from our motivating case study is carried out to further demonstrate the proposed methodology.

Lipid-modifying effects of krill oil vs fish oil: a network meta-analysis

CONTEXT

Krill oil is a good source of n-3 phospholipids and has greater bioavailability than fish oil, which contains n-3 triglycerides. However, it is unclear whether krill oil affects circulating lipid concentrations more beneficially than fish oil.

OBJECTIVE

A network meta-analysis was conducted to compare the lipid-modifying effects of krill oil and fish oil.

DATA SOURCES

PubMed and Embase databases were searched.

STUDY SELECTION

A total of 64 randomized controlled trials that determined the lipid-modifying effects of krill oil or fish oil were selected.

DATA EXTRACTION

The MetaXL program was used for meta-analysis. A subgroup analysis and a network meta-regression were conducted to investigate the dose-response effect of the n-3 fatty acid content of fish oil and krill oil.

RESULTS

Krill oil was associated with significantly lower triglyceride levels than control supplements (weighted mean difference [WMD] -23.26 [95%CI, -38.84 to -7.69]). However, the net differences in triglycerides (WMD -4.07 [95%CI, -15.22 to 7.08]), low-density lipoprotein cholesterol (WMD 3.01 [95%CI, -5.49 to 11.51]), high-density lipoprotein cholesterol (WMD 1.37 [95%CI, -3.73 to 6.48]), and total cholesterol (WMD 1.69 [95%CI, -6.62 to 10.01]) were not significantly different between the krill oil and fish oil groups. One gram of n-3 fatty acids contained in fish oil and krill oil lowered median triglycerides by 8.971 mg/dL (95% credible interval [CrI], 2.27 to 14.04) and 9.838 mg/dL (95%CrI, 0.72 to 19.40), respectively.

CONCLUSIONS

The lipid-modifying effects of krill oil and fish oil do not differ. The reduction in triglycerides depends on the dose of n-3 fatty acids consumed.

Analgesic medicines for adults with low back pain: protocol for a systematic review and network meta-analysis

BACKGROUND

There is limited evidence for the comparative effectiveness of analgesic medicines for adults with low back pain. This systematic review and network meta-analysis aims to determine the analgesic effect, safety, acceptability, effect on function, and relative rank according to analgesic effect, safety, acceptability, and effect on function of a single course of [an] analgesic medicine(s) or combination of these medicines for people with low back pain.

METHODS

We will include published and unpublished randomised trials written in any language that compare an analgesic medicine to either another medicine, placebo/sham, or no intervention in adults with low back pain, grouped according to pain duration: acute (fewer than 6 weeks), sub-acute (6 to 12 weeks), and chronic (greater than 12 weeks). The co-primary outcomes are pain intensity following treatment and safety (adverse events). The secondary outcomes are function and acceptability (all-cause dropouts). We will perform a network meta-analysis to compare and rank analgesic medicines. We will form judgements of confidence in the results using the Confidence in Network Meta-Analysis (CINeMA) methodology.

DISCUSSION

This network meta-analysis will establish which medicine, or combination of medicines, is most effective for reducing pain and safest for adults with low back pain.

SYSTEMATIC REVIEW REGISTRATION

PROSPERO CRD42019145257.

Charting the landscape of graphical displays for meta-analysis and systematic reviews: a comprehensive review, taxonomy, and feature analysis

BACKGROUND

Data-visualization methods are essential to explore and communicate meta-analytic data and results. With a large number of novel graphs proposed quite recently, a comprehensive, up-to-date overview of available graphing options for meta-analysis is unavailable.

METHODS

We applied a multi-tiered search strategy to find the meta-analytic graphs proposed and introduced so far. We checked more than 150 retrievable textbooks on research synthesis methodology cover to cover, six different software programs regularly used for meta-analysis, and the entire content of two leading journals on research synthesis. In addition, we conducted Google Scholar and Google image searches and cited-reference searches of prior reviews of the topic. Retrieved graphs were categorized into a taxonomy encompassing 11 main classes, evaluated according to 24 graph-functionality features, and individually presented and described with explanatory vignettes.

RESULTS

We ascertained more than 200 different graphs and graph variants used to visualize meta-analytic data. One half of these have accrued within the past 10 years alone. The most prevalent classes were graphs for network meta-analysis (45 displays), graphs showing combined effect(s) only (26), funnel plot-like displays (24), displays showing more than one outcome per study (19), robustness, outlier and influence diagnostics (15), study selection and p-value based displays (15), and forest plot-like displays (14). The majority of graphs (130, 62.5%) possessed a unique combination of graph features.

CONCLUSIONS

The rich and diverse set of available meta-analytic graphs offers a variety of options to display many different aspects of meta-analyses. This comprehensive overview of available graphs allows researchers to make better-informed decisions on which graphs suit their needs and therefore facilitates using the meta-analytic tool kit of graphs to its full potential. It also constitutes a roadmap for a goal-driven development of further graphical displays for research synthesis.

Optimal doses of antidepressants in dependence on age: Combined covariate actions in Bayesian network meta-analysis

Background: The meta-analysis by Furukawa et al. (The Lancet Psychiatry 2019, 6(7)) reported optimal doses for antidepressants in adult major depressive disorder (MDD). The present reanalysis aimed to adjust optimal doses in dependence on age. Methods: Analysis was based on the same dataset by Cipriani et al. (The Lancet 2018, 391(10128)) comparing 21 antidepressants in MDD. Random-effects Bayesian network meta-analysis was implemented to estimate the combined covariate action using restricted cubic splines (RCS). Balanced treatment recommendations were derived for the outcomes efficacy (response), acceptability (dropouts for any reason), and tolerability (dropouts due to adverse events). Findings: The combined covariate action of dose and age suggested agomelatine and escitalopram as the best-balanced antidepressants in terms of efficacy and tolerability that may be escalated until 40 and 60 mg/day fluoxetine equivalents (mg/day FE ), respectively, for ages 30-65 years. Desvenlafaxine, duloxetine, fluoxetine, milnacipran, and vortioxetine may be escalated until 20-40 mg/day FE , whereas bupropion, citalopram, mirtazapine, paroxetine, and venlafaxine may not be given in doses  > 20 mg/day FE . Amitriptyline, clomipramine, fluvoxamine, levomilnacipran, reboxetine, sertraline, and trazodone revealed no relevant balanced benefits and may therefore not be recommended for antidepressant treatment. None of the antidepressants was observed to provide balanced benefits in patients >70 years because of adverse events exceeding efficacy. Interpretation: Findings suggest that the combined covariate action of dose and age provides a better basis for judging antidepressant clinical benefits than considering dose or age separately, and may thus inform decision makers to accurately guide antidepressant dosing recommendations in MDD. Funding: No funding.

Interventions to prevent spontaneous preterm birth in high‐risk women with singleton pregnancy: a systematic review and network meta‐analysis

This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:

To compare the efficacy of current, relevant interventions to prevent preterm birth in women with singleton pregnancy and high individual risk of spontaneous preterm birth. We will consider interventions for women with a history of spontaneous preterm birth or short cervical length and women with asymptomatic vaginal infections.

Modelling time-course relationships with multiple treatments: Model-based network meta-analysis for continuous summary outcomes

BACKGROUND

Model-based meta-analysis (MBMA) is increasingly used to inform drug-development decisions by synthesising results from multiple studies to estimate treatment, dose-response, and time-course characteristics. Network meta-analysis (NMA) is used in Health Technology Appraisals for simultaneously comparing effects of multiple treatments, to inform reimbursement decisions. Recently, a framework for dose-response model-based network meta-analysis (MBNMA) has been proposed that combines, often nonlinear, MBMA modelling with the statistically robust properties of NMA. Here, we aim to extend this framework to time-course models.

METHODS

We propose a Bayesian time-course MBNMA modelling framework for continuous summary outcomes that allows for nonlinear modelling of multiparameter time-course functions, accounts for residual correlation between observations, preserves randomisation by modelling relative effects, and allows for testing of inconsistency between direct and indirect evidence on the time-course parameters. We demonstrate our modelling framework using an illustrative dataset of 23 trials investigating treatments for pain in osteoarthritis.

RESULTS

Of the time-course functions that we explored, the E_max model gave the best fit to the data and has biological plausibility. Some simplifying assumptions were needed to identify the ET₅₀ , due to few observations at early follow-up times. Treatment estimates were robust to the inclusion of correlations in the likelihood.

CONCLUSIONS

Time-course MBNMA provides a statistically robust framework for synthesising evidence on multiple treatments at multiple time points. The use of placebo-controlled studies in drug-development means there is limited potential for inconsistency. The methods can inform drug-development decisions and provide the rigour needed in the reimbursement decision-making process.

Graphs of study contributions and covariate distributions for network meta-regression

BACKGROUND

Meta-regression results must be interpreted taking into account the range of covariate values of the contributing studies. Results based on interpolation or extrapolation may be unreliable. In network meta-regression (NMR) models, which include covariates in network meta-analyses, results are estimated using direct and indirect evidence; therefore, it may be unclear which studies and covariate values contribute to which result. We propose graphs to help understand which trials and covariate values contribute to each NMR result and to highlight extrapolation or interpolation.

METHODS

We introduce methods to calculate the contribution that each trial and covariate value makes to each result and compare them with existing methods. We show how to construct graphs including a network covariate distribution diagram, covariate-contribution plot, heat plot, contribution-NMR plot, and heat-NMR plot. We demonstrate the methods using a dataset with treatments for malaria using the covariate average age and a dataset of topical fluoride interventions for preventing dental caries using the covariate randomisation year.

RESULTS

For the malaria dataset, no contributing trials had an average age between 7-25 years and therefore results were interpolated within this range. For the fluoride dataset, there are no contributing trials randomised between 1954-1959 for most comparisons therefore, within this range, results would be extrapolated.

CONCLUSIONS

Even in a fully connected network, an NMR result may be estimated from trials with a narrower covariate range than the range of the whole dataset. Calculating contributions and graphically displaying them aids interpretation of NMR result by highlighting extrapolated or interpolated results.