Editorial: The Devil Is in the Detail—Interpreting the Findings of Network Meta‐Analyses

The success of infliximab, the original anti-tumour necrosis factor (TNF) α agent targeting cytokine-mediated inflammation [1], has led to a surge of biological drugs and small molecules with proven efficacy in treating Crohn's disease (CD) [2]. An expanding therapeutic arsenal offers more choice, but poses challenges when considering appropriate drug selection. Factors including cost, mode of administration, side effect profile, patient preference, and associated co-morbidities exert influence, but drug efficacy is arguably the most important clinical consideration [3, 4]. Randomised controlled trials (RCT) comparing the efficacy of medical therapies against placebo are commonplace; but with a paucity of head-to-head trials of available therapies, uncertainty in the sequencing of drugs in CD persists [5-7]. Network meta-analysis (NMA) indirectly assesses the efficacy of individual therapies via a common comparator, often placebo, to determine the superiority of one drug over another [2, 4, 8]. The NMA by Shehab et al. [2] compared the efficacy of biologics and small molecules in moderate-to-severe CD, re-demonstrating the efficacy of anti-TNF agents and risankizumab in inducing clinical remission [4]. In a rapidly evolving field, this study also demonstrated the comparable efficacy of novel interleukin-23 inhibitors in biologic-exposed patients, where drug selection is typically more nuanced [2]. Furthermore, upadacitinib, a Janus kinase inhibitor, demonstrated superiority in inducing endoscopic remission [2, 4]. Ranking the efficacy of drugs by indirect comparison relies on the assumption that individuals within the network would be eligible to receive any of the active treatments. In reality, participants of more recently published studies may have already failed several therapies prior to enrolment and therefore be less likely to respond to newer therapies [3]. Shehab et al. addressed this issue by performing sub-group analyses of biologic-exposed and -naïve patients, which is a strength [2]. Another issue of NMA is that study findings may be confounded by differences in RCT design, such as permitted prior use of advanced therapies or concurrent medications, dosing schedules, and maintenance design; scrutiny of the methodology and a degree of caution when interpreting the findings is therefore needed [9]. Although the authors performed sub-group maintenance analyses according to treat-through and re-randomisation design in Supporting Information, the headline maintenance findings are pooled from data extracted from studies of differing maintenance trial design [2]. This is likely to have introduced bias and may affect the validity of the study findings. These results should be interpreted with caution and should be considered in combination with the supplementary maintenance analyses. Shebab et al. have provided a contemporary analysis of the efficacy of advanced therapies in the induction and maintenance of remission in CD. More head-to-head trials will improve the quality of NMA such as this and will enable a better estimate of optimal sequencing of drugs used in the management of CD. Christy Riggott: writing – original draft. David J. Gracie: writing – review and editing. This article is linked to Shehab et al. papers. To view these articles, visit https://doi.org/10.1111/apt.70295 and https://doi.org/10.1111/apt.70342. Data sharing not applicable to this article as no datasets were generated or analysed during the current study.