Overlapping network meta-analyses on psoriasis systemic treatments, an overview: quantity does not make quality

Background: Network meta-analyses (NMAs) have become successful in addressing gaps in the comparative effectiveness of systemic treatments in moderate-to-severe psoriasis. However, their increasing number carries both a risk of overlap and reproducibility issues that can hamper clinical decision-making.

Objectives: In this overview, we aimed to assess redundancy across these NMAs and to describe their characteristics.

Materials and methods: We considered all systematic reviews with NMAs of randomized controlled trials that included adult patients with moderate-to-severe psoriasis and that evaluated the efficacy and/or safety of systemic treatments compared with placebo or with an active comparator. PubMed/MEDLINE, Epistemonikos, PROSPERO and the Evidence update of the Centre of Evidence-Based Dermatology of the University of Nottingham were searched up to 25 February 2021. Our main outcome was the number per year of redundant NMAs and the extent of their overlap. We also described their features, especially, the confidence in the results of the reviews, the funding of the studies and the presence of spin (a description that overstates efficacy and/or understates harm), reporting issues and methodological characteristics.

Results: In total, 47 redundant NMAs were included. Only two of 47 (4%) included all available treatments. Both efficacy and safety were evaluated in 14 of 47 (30%) NMAs and both short- and long-term evaluations were assessed in five of 47 (11%). Confidence in the results was critically low for 39 of 47 (83%) NMAs and only 10 of 47 (21·3%) registered a protocol. Twenty-six of 47 NMAs (55%) received pharmaceutical funding. Contract research organizations were involved in 19 of 47 (40%) NMAs. Reporting was poor across most of the NMA abstracts and spin was present in all of the abstracts. Almost half of the NMAs failed to consider important limitations such as heterogeneity (considered in 32%) or consistency (considered in 66%).

Conclusions: In addition to a duplication of efforts, our overview showed heterogeneous methods and poor confidence in the results in a majority of the included NMAs, further distorted by reporting issues and spin. Clinicians need to interpret NMAs with caution when looking for the most reliable and comprehensive evidence.