Table 2.

Summary of recommendations

Area in results section	Recommendation
1. Communication	Trial descriptions should be tailored to each audience and each occasion: this is to reduce the (perceived) complexity of a platform protocol where a patient may contribute to only one part of the trial
	Clear diagrams are essential to express: (a) Each individual comparison (b) The currently-recruiting comparisons (c) The changes to the open comparisons over time
	All patient-facing and ethics committee text should clearly describe: (a) The current trial (b) The information which is relevant for the specific participant group
	Avoid overloading participants and recruitment teams by using a staged consent process
	A trial with a big central team should provide one person as a dedicated point of contact to each site
2. Funding	Funding is likely (to need) to come from multiple sources
	Each contribution to funding should aim to include a contribution to the overall infrastructure and the common delivery of the platform
	Express to funders both the savings (time, patients, cost) and/or gains (additional scientifically important questions) of using a platform protocol over separate trials
3. Protocol	Choose the most future-proofed option between modular or single approach to protocol development
	When choosing whether patients not meeting the eligibility criteria for one comparison could be randomised to other comparisons, consider: (a) Implications on recruitment (b) Generalisability of findings (c) Practical implementation at sites
	Explicitly state in the protocol from the outset that future comparisons will be incorporated into the protocol if appropriate
	Early engagement and ongoing communication with the regulator is essential
	Aim for review by an ethics committee with previous experience and training in platform protocols
4. Database and randomisation system	The database needs to be flexible and scalable
	Modular database design with shared elements is preferable if the current or future arms may differ in terms of the information required
	Allow for sufficient data management time in each grant. Platform protocols are more efficient in real-time results and input may be required over a shorter time than for any one trial
	Ensure choice of randomisation system can incorporate any necessary future amendment (e.g. to eligibility, weightings and stratification factors)
5. Patient and public involvement	Early PPI input improves the design
	Support PPI to understand design implications, particularly in adding new comparisons
	PPI participation helps the trial team with explanations to ethics committees
	Comparison-specific PPI representation can give a more manageable workload for PPI members and enable trial teams to better support PPI members
6. Contracts	Contracts must allow for the longevity of platform trials
	Platform protocols are likely to have more external collaborators so allow time for set-up and agreement
7. External trial oversight	Trial oversight committees must expect greater longevity and a considerable workload over time and per meeting, particularly if a platform protocol has many comparisons
	Members must be experienced, and any handover should aim to include an overlap period
8. Trial Management Group	The responsibilities of large TMGs may usefully be delegated to specific sub-committees, each responsible for components of the platform
	A dedicated lead for each comparison could better support the chief investigator and trial team in development, conduct, and reporting, e.g. comparison CI and comparison co-CI
9. Trial staffing	Flexible staffing allows for more staff at time of higher need
	Allow more senior time to manage a bigger team
10. Data management	Data cleaning and checking must be an ongoing process rather than analysis driven, so that all arms are updated fairly
	Create a dedicated site advisory team including site representatives
11. Statistical considerations	Choice of single or separate SAPs is driven by which comparisons will be analysed and reported and when they are to be reported (contemporaneously or at different times)
	Need to have a senior statistician unblinded and involved in analyses during the trial and another senior statistician who is blinded and unaware of the accumulating data analysis
12. Safety	SAEs must be assessed against the expected events for each of the treatments. Multiple research treatments require additional time at sites and during safety review
	Careful management of reference safety information is required with multiple treatments
	Multiple groups often need to be notified of SAEs
13. Training	Regularly update training materials and documentation
	Make training materials simple and inspiring to avoid site fatigue in long-term protocols with multiple new comparisons and amendments
	Make clear to staff that recruitment need not be paused around intermediate analysis
14. Reporting	Aim to give results from across the protocol to all patients, with a contextualising preface specific to their allocation/comparison
	Ideally ask whether participants want findings from only “their” comparison, from all comparisons, or no results
	CONSORT extensions for multi-arm randomised controlled trials and adaptive trials provide pertinent guidance
	Write a publication plan to minimise scheduling clashes for limited staff time
	Discuss authorship principles at comparison set-up. Authorship need not be the same for each primary comparison but all relevant names, including all relevant funders, must be noted
15. Adding and closing comparisons	Adding comparisons requires agreement from oversight committees, regulators, and assent from sites
	Consider using a checklist in deciding whether to add to the current trial or start a new trial [7]
	Aim to close down elements of each comparison as soon as practicable rather than leaving all comparisons open
16. Maintaining relevant control arm treatment	Be prepared for the standard-of-care to change over the lifetime of a platform trial
17. Onward data sharing and re-use	Consider whether data from reported comparisons can be shared on appropriate data release request without compromising ongoing comparisons or planned analyses