OBJECTIVES: Network meta-analysis (NMA) is a statistical technique used to compare multiple interventions simultaneously, including those not directly compared in head-to-head trials. NMA assumes homogeneity in the reference arm (e.g., placebo) across trials. However, pooling reference arms of varying routes of administration (ROA; e.g., oral, subcutaneous, intravenous) may challenge this assumption. The current objective was to review recently published NMAs that compared drugs with varying ROAs, to understand the frequency at which these analyses are conducted, and to identify the methods and considerations used to address this potential source of bias.

METHODS: A search strategy using MESH terms and keywords was implemented in MEDLINE on May 9^th, 2023, to capture published NMAs from January 2022 to the present. Studies of interest were NMAs that indirectly assessed pharmacological therapies with ≥1 ROA. For studies that met these criteria, information was extracted on the disease area, ROAs, methodological adjustments, and any discussion of multiple ROAs as a study limitation.

RESULTS: The search identified 1308 abstracts, 427 full text articles were reviewed, and 140 studies met final inclusion criteria. The included NMAs were conducted in a range of disease areas (rheumatology, neurology, cardiology, endocrinology, and others). Of these studies, only 6 (4.3%) adjusted for differing ROAs in their analysis. Adjustment methods included meta-regression by ROA and subgroup analysis by ROA. A larger proportion discussed multiple ROAs as a study limitation or as a consideration in their results interpretation (n=24, 17.1%).

CONCLUSIONS: In the recent literature, comparing drugs of differing ROAs using NMA is common, however most studies do not account for this as a possible source of bias. Studies that do adjust for ROA use meta-regression or subgroup analysis. Future work will aim to quantify the impact of pooling reference arms of varying ROAs on the validity and interpretation of NMA results.