Table 1 Practical things to consider when planning a Study Within A Trial (SWAT)

Cost

 • SWATs need not be expensive; our experience is that many are likely to cost between £5000 and £10,000. Ideally, they should be built into the host trial from the start and the associated costs can be included in the budget for the host trial. If the findings of the SWAT will be reported in a standalone publication in an author-pays open-access journal, the costs of this will be need to be budgeted for

Randomisation

 • Whether randomisation is needed depends on the question being asked. If the intention is to evaluate the effect of alternative ways of doing a trial process, then the alternatives being compared should be allocated at random. This may not always be possible and another allocation method (e.g. before and after the new alternative) can be used but in most cases this will weaken confidence in the results. However, if the question being asked is not focused on measuring effect sizes (e.g. it could be concerned with understanding why something is done the way it is) then randomisation is likely to be inappropriate and other qualitative methods would be required. Randomisation is not a defining feature of a SWAT

 • Randomisation can be by a separate process to that used for the host trial randomisation

Ethics

 • Ethical approval guidelines and regulations for conducting research in humans vary between countries. Depending on the specific SWAT protocol being evaluated, it is advised that the researcher checks national guidance and discusses whether ethical approval is required with their institutional or local ethical committee

 • It is likely that some, but not all, SWATs will need ethical and other approvals. Clinical trials of medicinal products in the EU are provided for in Directive 2001/20/EC of the European Parliament. Such trials require research ethical approval and it is likely that any SWAT within a host trial subject to the EU directive will require ethical review

 • Ethical governance of clinical trials outside of the directive, i.e. non-medicinal products for human use vary between countries. In the UK for instance, SWATs within non-medicinal product trials that involve only trial staff will not normally need UK NHS ethical approval (but may need institutional review), while it is likely that those that involve NHS patients will. In the Republic of Ireland, there is a system of national approval for trials of medicinal products but not for non-medicinal products and, therefore, for the latter ethical approval is usually sought from sites conducting the host trial and/or from the SWAT principal investigator’s host institution. If the SWAT was planned at the same time as the host trial, then it could be included in the application for ethical approval of the host trial. Trial Forge has a collection of material that has been used before to obtain ethical approval for a SWAT, which it adds to its own SWAT packages. Contact info@trialforge.org for more details

 • SWATs are generally low risk and it is rare for them to impose additional burden or risk on participants and consequently it will not usually be necessary to get individual consent from participants. Indeed, in many cases individual consent may not be appropriate. It may confuse patients as to what they are consenting to, and may impact on their behaviour if they are aware that different recruitment methods are being tested, confounding the evaluation

 • SWATs aimed at staff, but which directly affect patients/participants, may need NHS or other ethical review (e.g. studies that change what recruiters say to potential participants, or who says it to them). Where there is any doubt researchers should contact the Health Research Authority (HRA)/Devolved Nation’s REC administrative body to check whether NHS ethical review is required. In the Republic of Ireland, researchers should check with the ethics committee approving the host trial and with any site in which the trial will be conducted

 • In the UK NHS Research Ethics Committee approval is not normally required for research only involving staff who are recruited by virtue of their professional role. However, where such studies are led from England and involves the NHS in England ‘HRA approval’ may be required (see http://www.hra.nhs.uk/about-the-hra/our-plans-and-projects/assessment-approval/ for further information)

Analysis

 • The analysis of SWATs might be simple (such as the comparison of two proportions) and might be done by members of the trial team other than a senior statistician

 • Sample size calculations for SWATs can be done in the usual way using estimates of minimum important differences that the investigators or others consider appropriate. The size of a SWAT is constrained by the host trial. The size of a recruitment SWAT will generally be larger than the host trial sample size (the constraint is the size of the patient population approached, not recruited). Other SWATs (such as those on retention) will be limited to the actual host trial sample. It is highly unlikely that the size of the host trial will be changed for the benefit of a SWAT. SWATs are designed for future meta-analysis. In other words, while an individual SWAT may be underpowered, a meta-analysis of several well-done SWATs evaluating the same intervention and following the same protocol can provide compelling evidence for trial process decision making. As with all meta-analysis, judgements need to be made about whether it is sensible to combine studies done in different populations, disease areas and settings. This issue will be the topic of future Trial Forge Guidance

 • SWATs exploring qualitative questions about how a trial process is delivered, organised or perceived will be analysed using a suitable qualitative analysis method

Implementing the SWAT

 • Some of the extra work needed for a SWAT (e.g. putting additional materials or incentives into envelopes along with information leaflets) might be done by temporary staff, or existing staff who have a lull in the work for their own trial. For other SWATs there might be little additional work involved (for example, using mail merge software to generate different invitation letters). However, confidentiality/data protection issues may limit who can do the work, depending on its content and the potential for identifying participants to those who would not otherwise have lawful access to personal identifiable information

 • They need not run for the whole duration of the host trial so any extra work may be both modest and short-term

Publication

 • The findings of the SWAT should be put into the public domain and should be accessible by others. This might be possible through inclusion in the report of the host trial (with appropriate signposting, perhaps in the abstract, to highlight its presence), in a standalone dedicated publication or through inclusion in a relevant systematic review