Bayesian or frequentist: there is no question when comparing single-inhaler triple therapies via network meta-analysis. Focus on fluticasone furoate/umeclidinium/vilanterol fixed-dose combination in chronic obstructive pulmonary disease

ABSTRACT

Objectives

Single-inhaler triple therapies (SITTs) have never been directly compared in randomized controlled trials (RCTs) in chronic obstructive pulmonary disease (COPD). Cochrane recommends the Bayesian approach for indirect comparisons but a frequentist network meta-analysis (NMA) reported superiority of fluticasone furoate/umeclidinium/vilanterol (FF/UMEC/VI) over other SITT. We assessed the most appropriate inference method for NMA characterized by between-study heterogeneity on SITT in COPD.

Methods

Bayesian and frequentist NMA were performed on RCTs investigating the effect of SITT on exacerbations and trough forced expiratory volume in the 1^st second (FEV₁) in COPD.

Results

The included RCTs (ETHOS, FULFIL, IMPACT, KRONOS 200812) reported significant between-study heterogeneity (I² > 99%, p < 0.001). The Bayesian random-effect NMA provided unbiased evidence that FF/UMEC/VI was not superior to other SITT on exacerbations and trough FEV₁. The frequentist fixed-effect NMA indicated that FF/UMEC/VI was significantly (p < 0.05) more effective than other SITT, although results were affected by dispersion, asymmetry, and significant risk of bias. Frequentist random-effect NMA provided effect estimates rather similar but not equal to those of Bayesian approach.

Conclusion

Indirect comparison should be performed via Bayesian approach instead of frequentist inference with a fixed-effect model. Claiming the superiority of a specific medication over other therapies should be confirmed by findings originating from well-designed RCTs.

KEYWORDS:

• Bayesian
• chronic obstructive pulmonary disease
• frequentist
• network meta-analysis
• single-inhaler triple therapies

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/17476348.2024.2316167

Additional information

Funding

This paper was not funded.