A multifaceted graphical display, including treatment ranking, was developed to aid interpretation of network meta-analysis.

OBJECTIVES: Network meta-analysis (NMA) is becoming a popular statistical tool for analyzing a network of evidence comparing more than two interventions. A particular advantage of NMA over pairwise meta-analysis is its ability to simultaneously compare multiple interventions including comparisons not previously trialed together, permitting intervention hierarchies to be created. Our aim was to develop a novel graphical display to aid interpretation of NMA to clinicians and decision-makers that incorporates ranking of interventions. STUDY DESIGN AND SETTING: Current literature was searched, scrutinized, and provided direction for developing the novel graphical display. Ranking results were often found to be misinterpreted when presented alone and, to aid interpretation and effective communication to inform optimal decision-making, need to be displayed alongside other important aspects of the analysis including the evidence networks and relative intervention effect estimates. RESULTS: Two new ranking visualizations were developed-the 'Litmus Rank-O-Gram' and the 'Radial SUCRA' plot-and embedded within a novel multipanel graphical display programmed within the MetaInsight application, with user feedback gained. CONCLUSION: This display was designed to improve the reporting, and facilitate a holistic understanding, of NMA results. We believe uptake of the display would lead to better understanding of complex results and improve future decision-making.

PALM: PATIENT-CENTERED TREATMENT RANKING VIA LARGE-SCALE MULTIVARIATE NETWORK META-ANALYSIS.

The growing number of available treatment options has led to urgent needs for reliable answers when choosing the best course of treatment for a patient. As it is often infeasible to compare a large number of treatments in a single randomized controlled trial, multivariate network meta-analyses (NMAs) are used to synthesize evidence from trials of a subset of the treatments, where both efficacy and safety related outcomes are considered simultaneously. However, these large-scale multiple-outcome NMAs have created challenges to existing methods due to the increasing complexity of the unknown correlations between outcomes and treatment comparisons. In this paper, we proposed a new framework for PAtient-centered treatment ranking via Large-scale Multivariate network meta-analysis, termed as PALM, which includes a parsimonious modeling approach, a fast algorithm for parameter estimation and inference, a novel visualization tool for presenting multivariate outcomes, termed as the origami plot, as well as personalized treatment ranking procedures taking into account the individual's considerations on multiple outcomes. In application to an NMA that compares 14 treatment options for labor induction, we provided a comprehensive illustration of the proposed framework and demonstrated its computational efficiency and practicality, and we obtained new insights and evidence to support patient-centered clinical decision making.

Effectiveness and Cost-Effectiveness of 360 Disease-Modifying Treatment Escalation Sequences in Multiple Sclerosis.

OBJECTIVES: The rapid expansion in treatment options for relapsing-remitting multiple sclerosis (RRMS) of the past decade requires clinical decision making on the sequential prescription of these treatments. Here, we compare 360 treatment escalation sequences for patients with RRMS in terms of health outcomes and societal costs in The Netherlands. METHODS: We use a microsimulation model with a societal perspective, developed in collaboration with MS neurologists, to estimate the effectiveness and cost-effectiveness of 360 treatment sequences starting with first-line therapies in RRMS. This model integrated data on disease progression, disease-modifying treatment efficacy, clinical decision rules, age-dependent relapse rates, quality of life, healthcare, and societal costs. RESULTS: Costs and health outcomes were overlapping among different treatment escalation sequences. In our model for RRMS treatment, optimal lifetime health outcomes (20.24 ± 1.43 quality-adjusted life-years [QALYs], 6.11 ± 0.30 relapses) were achieved with the sequence peginterferon-dimethyl fumarate-ocrelizumab-natalizumab-alemtuzumab. The most cost-effective sequence (peginterferon-glatiramer acetate-ocrelizumab-cladribine-alemtuzumab) yielded numerically worse health outcomes per patient (19.59 ± 1.43 QALYs, 6.64 ± 0.43 relapses), but resulted in €98 127 ± €19 134 less costs than the most effective treatment sequence. CONCLUSIONS: Effectiveness estimates of treatments have overlapping confidence intervals but the treatment sequence that yields most QALYs is not the most cost-effective option, also when taking uncertainty into account. It is important that neurologists are aware of cost constraints and its relationship with prescription behavior, but treatment decisions should be individually tailored.

Multinomial network meta-analysis using response rates: relapsed/refractory multiple myeloma treatment rankings differ depending on the choice of outcome.

BACKGROUND: Due to the fast growing relapsed/refractory multiple myeloma (RRMM) treatment landscape, a comparison of all the available treatments was warranted. For clinical practice it is important to consider both immediate effects such as response quality and prolonged benefits such as progression-free survival (PFS) in a meta-analysis. The objective of this study was to assess the impact of the choice of outcome on the treatment rankings in RRMM. METHODS: A multinomial logistic network meta-analysis was conducted to estimate the ranking of sixteen treatments based on both complete and objective response rates (CRR and ORR). Seventeen phase III randomized controlled trials from a previously performed systematic literature review were included. Treatment ranking was based on the surface under the cumulative ranking curve (SUCRA). Sensitivity analysis was conducted. RESULTS: The ranking of treatments differed when comparing PFS hazard ratios rankings with rankings based on CRR. Pomalidomide, bortezomib and dexamethasone ranked highest, while a substantial lower ranking was observed for the triplet elotuzumab, lenalidomide, dexamethasone. The ranking of treatments did not differ when comparing PFS hazard ratios and ORR. The scenario analyses showed that the results were robust. In all scenarios the top three was dominated by the same triplets. The treatment with the highest probability of having the best PFS and ORR was the triplet daratumumab, lenalidomide plus dexamethasone in the base case. CONCLUSION: This analysis shows that depending on the chosen outcome treatment rankings in RRMM may differ. When conducting NMAs, the response rate, a clinically recognized outcome, should therefore be more frequently considered.

Predictive P-score for treatment ranking in Bayesian network meta-analysis.

BACKGROUND: Network meta-analysis (NMA) is a widely used tool to compare multiple treatments by synthesizing different sources of evidence. Measures such as the surface under the cumulative ranking curve (SUCRA) and the P-score are increasingly used to quantify treatment ranking. They provide summary scores of treatments among the existing studies in an NMA. Clinicians are frequently interested in applying such evidence from the NMA to decision-making in the future. This prediction process needs to account for the heterogeneity between the existing studies in the NMA and a future study. METHODS: This article introduces the predictive P-score for informing treatment ranking in a future study via Bayesian models. Two NMAs were used to illustrate the proposed measure; the first assessed 4 treatment strategies for smoking cessation, and the second assessed treatments for all-grade treatment-related adverse events. For all treatments in both NMAs, we obtained their conventional frequentist P-scores, Bayesian P-scores, and predictive P-scores. RESULTS: In the two examples, the Bayesian P-scores were nearly identical to the corresponding frequentist P-scores for most treatments, while noticeable differences existed for some treatments, likely owing to the different assumptions made by the frequentist and Bayesian NMA models. Compared with the P-scores, the predictive P-scores generally had a trend to converge toward a common value of 0.5 due to the heterogeneity. The predictive P-scores' numerical estimates and the associated plots of posterior distributions provided an intuitive way for clinicians to appraise treatments for new patients in a future study. CONCLUSIONS: The proposed approach adapts the existing frequentist P-score to the Bayesian framework. The predictive P-score can help inform medical decision-making in future studies.