Members of the Adaptive Design Working Group (AD WG) are involved in a number of trials of potential COVID-19 treatments. Amongst them is the national flagship platform RECOVERY trial which seeks to provide definitive evidence about treatments for hospitalised patients. RECOVERY uses a pragmatic efficient multi-arm multi-stage platform approach.
The AGILE platform trial on the other hand undertakes a dose-finding and proof of concept evaluation of COVID-19 treatments of patients, both in and out of hospital, and utilises state of the art model-based dose-finding methods alongside a sequential Bayesian design. In addition to these platform trials, the AD WG is also contributing to individual trials such as DECISIVE. In all cases a member of the AD WG has been responsible for the design of the studies and remains closely involved with their conduct and analysis.
Members of the Statistical Analysis Working Group (SA WG), led by Darren Dahly, have been providing rapid peer-reviews of publications, preprints and protocols from international clinical trials of COVID-19 treatments independent of journal specific review processes. The aim is to provide timely, constructive, clear expert advice aimed at improving both the research outputs under review, as well as future studies. Thus far, the reviews focus on the designs of the trials and other statistical content (methods, presentation and accuracy of results, inferences) to aid interpretation for general scientific readers. The reviews are publicly available here. This current work has already been cited and used in real-world decision making by Independent Data Monitoring Committees (IDMCs), and a summary article highlighting the main findings is in preparation.
SA WG members have also begun to explore methodological issues arising from the first wave of funded COVID-19 trials. For example, Brennan Kahan and colleagues have considered the importance of ensuring these trials are asking the right questions through the selection of appropriate treatment effects (estimands) to be estimated in COVID-19 trials.
Additionally, we have been considering the impact of the pandemic on existing clinical trials, and how the data from these studies will be affected. For example, Suzie Cro and colleagues have outlined a four-step strategy for handling missing outcome data in trials affected by a pandemic.
Members of the TMRP Trial Conduct Working Group Inclusivity Subgroup have contributed to the development of the NIHR INCLUDE Ethnicity Framework, which aims to help trial teams think carefully about which ethnic groups should be included in their trial for its results to be widely applicable, and what challenges there may be to making this possible. This is relevant to all trials, including COVID-19 trials.
NIHR has included a link to the Ethnicity Framework resource in its Standard Guidance for Applicants and a number of other funders are considering doing the same.
In February 2020, three COVID-19 COS projects were registered in the COMET database [1,2,3]. A fourth was registered in March , with the aim of engaging patients in the COS development process, and is therefore complementary to the existing initiatives. Each COS is slightly different in its scope or methods, including the range of stakeholders involved, and they are complementary to one another. Teleconferences were convened with representatives of the four COVID-19 COS for studies of in hospital patients and the COMET Initiative. These discussions will continue, in order to maintain awareness of work on this topic, determine what information to disseminate more widely, and promote the uptake of COS in relevant research.
In April 2020, three of the four COS were completed and found to overlap. Thus, a ‘meta-COS’ for research in adult hospitalised patients (Table 1) was agreed by the leads of the COS development projects [1,2,3]. The results of the fourth COS  also agreed with the meta-COS. The meta-COS is composed of two domains: mortality and respiratory support. Adverse events, relevant to the particular research question, should also be measured and reported. Each of the COS include other core outcome domains for hospitalised patients, and also core outcome domains for those patients with asymptomatic or mild disease who are not hospitalised. These other outcomes deemed core but not in all COS are described in Table 2. See here for further details.
The COS COVID-P study was set up to develop a COS for studies evaluating public health, primary and secondary care interventions for prevention of COVID-19 transmission . This rapid project began with examining outcomes measured in COVID-19 prevention intervention studies, followed by online scoping workshops with key international stakeholders to define the scope of the COS. Findings from the workshop were fed back to the Steering Group, who reached agreement on the inclusion of two outcome domains: infection and intervention-specific harms. To date, the COS-COVID-P team have identified key issues for designing COVID-19 prevention studies, including that SARS-CoV-2 infection is an essential outcome to measure. However, there has been much discussion of challenges around how to measure infection. Understanding of the mechanisms by which COVID-19 prevention interventions may work is also developing. This requires continuing engagement with specialist disease prevention and clinical experts, to inform how researchers may optimally measure infection and, where relevant, develop intervention-specific COS modules. The COS-COVID-P team are on the way, therefore, to developing guidance for those designing COVID-19 prevention studies around how infection may be optimally defined and measured so that all these studies can make the maximum contribution to meeting this global challenge.
COS-COVID-P team: Kerry Avery, Jane Blazeby, Susie Dodd, Sarah Gorst, Nicola Harman, Rhiannon Macefield, Paula Williamson.