All PMAs should have a publicly available protocol. Developing a protocol for a PMA is similar, conceptually, as doing so for a single trial. The essential elements of a PMA are detailed as follows and summarized in .
As in any protocol, the first important step is to define the hypotheses and then to establish eligibility criteria for studies. For example, studies to be included in the PMA may be required to use random assignment of participants to interventions, although it is possible to include other study designs in a PMA. If randomized, the individual trials may choose to share a common randomization method, or at least to use the same stratification factors. The required attributes of the participating population need to be specified, as do the minimum requirements for each of the interventions and the comparator arms. The protocol should also specify what endpoints need to be measured, and when and how they should be measured, which are primary and which are secondary, as well as other features of study design as necessary. If a PMA is established de novo, it may be possible for each trial in the PMA to share exactly the same trial protocol.
The protocol should describe in detail the efforts made to identify ongoing trials, including how potential collaborators have been (or will be) located and approached to participate.
Details of trials already identified for inclusion (if relevant) should be listed in the protocol. The listing might include the anticipated number of participants and timelines for each participating trial. The protocol should include a statement outlining if, at the time of submission for registration, any trial results were known (to anyone outside the trial's own data monitoring committee). Trials should be included only if their results were unknown at the time they were identified and added to the PMA. If eligible trials are identified but not included in the PMA because their results are already known, the PMA protocol should outline how these data will be dealt with. For example, secondary sensitivity analyses, using aggregate or individual patient data from these trials might be undertaken. The protocol should describe actions to be taken if subsequent trials are located while the PMA is in progress.
The protocol should outline the plans for the collection and analyses of data in a similar manner to that of an IPD meta-analyses (see Chapter ). This would include details of sample size and power calculation (for the PMA), any interim analyses to be undertaken, and details of planned subgroup analyses. Strategies for addressing additional questions beyond the main hypothesis of interest can also be incorporated in a PMA. These additional questions can be added as long as the results of studies to be included in the analysis are not known, i.e. they not ‘data driven’ research questions. Of note, there may be analyses that are unique to the PMA, that are not done within the individual trials, such as subgroup analyses.
The investigators of trials to be included in a PMA should generally be asked to agree to provide individual patient data. The protocol should describe what will occur if the investigators of some studies within the PMA are unable (or unwilling) to provide patient-level data, perhaps because of concerns about confidentiality or informed consent. Would the PMA secretariat, for example, accept appropriate summary data? (A two-stage analysis could be performed, in which the effect estimate of interest is calculated separately within each study, using the patient-level data, and those within-study estimates are then combined across studies using standard meta-analytic methods.) The protocol should specify whether it is intended to update the PMA data at regular intervals via ongoing cycles of data collection (e.g. 5 yearly), and hence when trialists would be expected to supply updated, long term outcome data.
The PMA protocol should outline details of project management structure (including any Committees, see Section 19.2.1), the procedures for data management (how data are to be collected, the format required, when data will be required to be submitted, quality assurance procedures, etc; see Chapter 18, Section 18.3), and who will be responsible for the statistical analyses.
A key element of the PMA protocol is the publication policy. It essential to have a policy regarding authorship (e.g. specifying that publication will be in the group name, but also include a list of individual authors). A policy regarding manuscript preparation is also important. For example, it might be specified that drafts of papers be circulated to all trialists for comment, prior to submission for publication. There might be a writing committee, like those that are often formed within cooperative study groups.
A unique issue that arises in the context of the PMA (which would generally not arise for a multicentre study or an IPD meta-analysis) is whether or not individual studies should publish on their own and the timing of those publications. Most investigators would want to publish their own studies individually in addition to contributing to the PMA, and it is likely that the investigators would want these publications to appear before the PMA is published, so as to avoid issues related to duplicate publication of the same data. In a similar spirit, though, any PMA publication(s) should clearly indicate the sources of the included data and refer to prior publications of the same data. The PMA protocol should also state what will occur if any of the participating trials fail to publish their individual results within a specified timeframe. This may occur if a trial is not completed due to insufficient funds, is terminated prematurely or the trial simply remains unpublished after a pre-specified date. The protocol should also address how to deal with trials that renege on their agreement to participate in the PMA.