Network meta-analysis (NMA) compares the efficacy or safety of more than two treatments or treatment doses that are directly or indirectly compared through a common comparator.1 2 Treatment doses are not routinely incorporated into NMAs, which potentially limits the applicability and validity of results.3 4 In this paper, we will introduce readers to NMA incorporating dose effects, specifically, how they can implement a hierarchical random effects NMA incorporating exchangeable dose effects. In particular, we share a motivating clinical example describing the comparative risk of vomiting associated with different cholinesterase inhibitor doses (ie, donepezil, galantamine and rivastigmine) and use this example to illustrate (1) key similarities and differences between the standard NMA and NMA incorporating exchangeable dose effects; (2) why incorporating dose effects in NMA is important; (3) steps to follow when completing a systematic review with NMA incorporating dose effects; (4) results derived from a hierarchical random effects NMA incorporating exchangeable dose effects and how they will facilitate clinical decision-making and (5) how to consider potential dose effects when evaluating NMA relevance and credibility. Please refer to our companion publication for a more in depth theoretical discussion of hierarchical NMA models incorporating dose effects and alternative NMA models incorporating dose effects.5

Clinicians prescribe cholinesterase inhibitors (ie, donepezil, galantamine and rivastigmine) to slow cognitive decline in people with dementia.4 Cholinesterase inhibitors are associated with potential side effects, including headaches, nausea and vomiting.4 5 Understanding if certain cholinesterase inhibitor doses are associated with a higher risk of side effects than others could inform, or even change, clinical decision-making. To describe potential dose effects, we used a subset of data from a published systematic review with NMA describing treatment level risk of vomiting associated with cholinesterase inhibitor use (see online supplemental table 1 for dataset).4 We implemented …