OBJECTIVES: Despite the comparative clinical studies evaluating Bulevirtide (BLV), Pegylated interferon (PEG-IFN), and anti-viral medications for Chronic Hepatitis D (CHD), a Network Meta-Analysis (NMA) is needed to systematically synthesize and consolidate all existing evidence. This early NMA aims to elucidate the methodology, challenges encountered, and the corresponding mitigation strategies, enabling an effective comparison of treatments for CHD.

METHODS: In accordance with the guidelines outlined by the ISPOR Task Force, NICE UK, and the Cochrane Handbook, the initial stage of the NMA framework involved constructing network geometry and evidence network diagrams based on a systematic literature review. Subsequent steps encompassed assessing heterogeneity, inconsistency, statistical models, assumptions, and model fit, followed by validation. Assumptions and modelling approaches were explored in pursuit of convergence and reliable NMA estimates.

RESULTS: The first challenge in the NMA was the lack of an interconnected network. This was resolved by assuming similar efficacy of Nucleoside/nucleotide analogs (NATs) and control groups, as NATs are part of background therapy among most patients. Owing to the connected network involving closed loops, the conventional Bayesian NMA was feasible but encountered difficulties in achieving convergence for binomial outcomes due to single-study effects and rare or zero events. Specifically, convergence challenges persisted in Bayesian odds ratio models, even when employing informative priors. Furthermore, the Frequentist NMA approach produced confidence intervals with substantial width, indicative of pronounced uncertainty in the estimates and thus limiting the evidence synthesis to qualitative comparisons only. The convergence issue was resolved by employing a risk difference model, allowing for reliable NMA estimates for relative efficacy and economic model inputs.

CONCLUSIONS: Early NMA enables timely identification of data and methodological challenges, allowing proactive mitigation, and guiding prioritization. The suggested risk difference NMA approach proves valuable when conventional NMA faces challenges due to convergence issues related to rare events and single-study connections.