ObjectiveTo compare the effects of treatments for coronavirus disease 2019 (covid-19).DesignLiving systematic review and network meta-analysis.Data sourcesUS Centers for Disease Control and Prevention COVID-19 Research Articles Downloadable Database, which includes 25 electronic databases and six additional Chinese databases to 20 July 2020.Study selectionRandomised clinical trials in which people with suspected, probable, or confirmed covid-19 were randomised to drug treatment or to standard care or placebo. Pairs of reviewers independently screened potentially eligible articles.MethodsAfter duplicate data abstraction, a bayesian random effects network meta-analysis was conducted. Risk of bias of the included studies was assessed using a modification of the Cochrane risk of bias 2.0 tool, and the certainty of the evidence using the grading of recommendations assessment, development and evaluation (GRADE) approach. For each outcome, interventions were classified in groups from the most to the least beneficial or harmful following GRADE guidance.Results23 randomised controlled trials were included in the analysis performed on 26 June 2020. The certainty of the evidence for most comparisons was very low because of risk of bias (lack of blinding) and serious imprecision. Glucocorticoids were the only intervention with evidence for a reduction in death compared with standard care (risk difference 37 fewer per 1000 patients, 95% credible interval 63 fewer to 11 fewer, moderate certainty) and mechanical ventilation (31 fewer per 1000 patients, 47 fewer to 9 fewer, moderate certainty). These estimates are based on direct evidence; network estimates for glucocorticoids compared with standard care were less precise because of network heterogeneity. Three drugs might reduce symptom duration compared with standard care: hydroxychloroquine (mean difference −4.5 days, low certainty), remdesivir (−2.6 days, moderate certainty), and lopinavir-ritonavir (−1.2 days, low certainty). Hydroxychloroquine might increase the risk of adverse events compared with the other interventions, and remdesivir probably does not substantially increase the risk of adverse effects leading to drug discontinuation. No other interventions included enough patients to meaningfully interpret adverse effects leading to drug discontinuation.ConclusionGlucocorticoids probably reduce mortality and mechanical ventilation in patients with covid-19 compared with standard care. The effectiveness of most interventions is uncertain because most of the randomised controlled trials so far have been small and have important study limitations.Systematic review registrationThis review was not registered. The protocol is included as a supplement.Readers’ noteThis article is a living systematic review that will be updated to reflect emerging evidence. Updates may occur for up to two years from the date of original publication.

Clinical questionWhat is the role of drug interventions in the treatment and prevention of covid-19?RecommendationsThe first version on this living guidance focuses on corticosteroids. It contains a strong recommendation for systemic corticosteroids in patients with severe and critical covid-19, and a weak or conditional recommendation against systemic corticosteroids in patients with non-severe covid-19. Corticosteroids are inexpensive and are on the World Health Organisation list of essential medicines.Howthis guideline was created This guideline reflects an innovative collaboration between the WHO and the MAGIC Evidence Ecosystem Foundation, driven by an urgent need for global collaboration to provide trustworthy and living covid-19 guidance. A standing international panel of content experts, patients, clinicians, and methodologists, free from relevant conflicts of interest, produce recommendations for clinical practice. The panel follows standards, methods, processes, and platforms for trustworthy guideline development using the GRADE approach. We apply an individual patient perspective while considering contextual factors (that is, resources, feasibility, acceptability, equity) for countries and healthcare systems.The evidenceA living systematic review and network meta-analysis, supported by a prospective meta-analysis, with data from eight randomised trials (7184 participants) found that systemic corticosteroids probably reduce 28 day mortality in patients with critical covid-19 (moderate certainty evidence; 87 fewer deaths per 1000 patients (95% confidence interval 124 fewer to 41 fewer)), and also in those with severe disease (moderate certainty evidence; 67 fewer deaths per 1000 patients (100 fewer to 27 fewer)). In contrast, systemic corticosteroids may increase the risk of death in patients without severe covid-19 (low certainty evidence; absolute effect estimate 39 more per 1000 patients, (12 fewer to 107 more)). Systemic corticosteroids probably reduce the need for invasive mechanical ventilation, and harms are likely to be minor (indirect evidence).Understanding the recommendationsThe panel made a strong recommendation for use of corticosteroids in severe and critical covid-19 because there is a lower risk of death among people treated with systemic corticosteroids (moderate certainty evidence), and they believe that all or almost all fully informed patients with severe and critical covid-19 would choose this treatment. In contrast, the panel concluded that patients with non-severe covid-19 would decline this treatment because they would be unlikely to benefit and may be harmed. Moreover, taking both a public health and a patient perspective, the panel warned that indiscriminate use of any therapy for covid-19 would potentially rapidly deplete global resources and deprive patients who may benefit from it most as potentially lifesaving therapy.UpdatesThis is a living guideline. Work is under way to evaluate other interventions. New recommendations will be published as updates to this guideline.Readers noteThis is version 1 of the living guideline, published on 4 September (BMJ 2020;370:m3379) version 1. Updates will be labelled as version 2, 3 etc. When citing this article, please cite the version number.SubmittedAugust 28AcceptedAugust 31