Plan a GCC multicentre study with a separate record for each site

1. Identify the actual institutions and activities

List the participating legal institutions, clinical units, principal investigators and proposed tasks. A collaborator’s university affiliation does not establish authority to release a hospital’s records. Distinguish who recruits participants, extracts information, holds identifiers, conducts analysis and receives an output. Mark a proposed collaborator as proposed until the relevant role is accepted.

In the fictional example, site S is a Saudi hospital and site O is a hospital within Oman’s Ministry of Health. The team is discussing a study of return encounters after an index visit. Only a concept note and draft data dictionary exist. There is no ethics decision, institutional data-access authorization, transfer agreement or collected dataset. The site letters are invented and do not identify real institutions.

2. Find the review route for each setting

Saudi NCBE describes local ethics committees as responsible for ethical evaluation and participant protection. Oman’s Ministry of Health describes a protocol-review service for investigators conducting research within that ministry, with review and approval before the study starts. These source pages identify relevant routes within their stated settings; they do not establish a single GCC approval or automatic recognition of another institution’s decision.

Ask each site which committee and administrative offices must consider this exact activity, and which documents they require. Record the protocol version, application status and actual written determination. Where a team believes an activity is exempt or a consent waiver is appropriate, leave that as a question for the authorized body; do not generate a decision from the study’s label or its retrospective design.

Reference: Saudi NCBE: local ethics committees · Oman Ministry of Health: National Center for Health Research

3. Track distinct permissions without a single green tick

A governance worksheet should show what each decision permits, who issued it and any conditions. Training completion, an ethics opinion, permission to use a clinical system and permission to send data elsewhere address different questions. Keep the underlying document and its version available to the people authorized to inspect it. Do not attach participant information to a general coordination email.

For this proposed collaboration, the following entries are all pending. That is a useful planning record: it prevents a friendly meeting or a drafted agreement from appearing as authorization. The institution’s responsible offices must decide the applicable requirements, including whether a transfer is allowed and under what safeguards; this table does not prescribe a universal legal route.

On small screens, scroll the table horizontally, or focus it and use the arrow keys.

4. Reconcile meaning before combining fields

Take the intended outcome apart: the index event, start of follow-up, type of return, observation window and source of ascertainment. Compare actual data definitions and collection processes with the intended question. A field name alone cannot establish that the sites measure the same event. Preserve the local source name and write any proposed mapping beside it.

The fictional draft proposes return encounters within 30 days. At S, a field called return_visit means the first scheduled return; at O it means any repeat encounter. The definitions are not equivalent. Neither should be relabelled as an unplanned return. The team must determine whether suitable source information exists and whether a common outcome is supportable. No mapping or pooled analysis has been approved or performed.

On small screens, scroll the table horizontally, or focus it and use the arrow keys.

5. Agree custody, contributions and changes

Write who holds the source data, who may inspect a derived file, who can amend the dictionary and who resolves conflicting versions. An analysis conducted separately at each site can be considered as an option, but it is not automatically equivalent to a pooled analysis or automatically exempt from approval. Methods and governance reviewers should assess the chosen approach together.

Discuss contribution roles early and confirm what each person actually does. NISO’s CRediT guidance supports agreeing roles and distinguishes contribution description from determining authorship. In this exercise no completed investigation or formal analysis can be credited, because neither has occurred. Keep publication plans and author-order discussions separate from permission to access data, and provide a route for contributors to challenge an inaccurate record.

Reference: NISO CRediT: agreed contribution roles and authorship limits

6. Choose the next step that the record supports

Bring a one-page status sheet to the coordinating meeting. Identify the exact protocol and dictionary versions under discussion, one owner for each unresolved permission and the evidence needed to close it. Record conditions and unresolved questions, rather than copying the status of one site into every column. A later amendment should preserve the earlier decision trail and prompt a check of affected sites.

For S and O, the next proposed task is to ask the appropriate institutional offices to confirm the routes and to clarify the encounter definitions using permitted specification documents. Recruiting participants, extracting identifiable records and sending a pooled file are not supported by the fictional record. A draft manuscript or a journal’s expression of interest would not resolve those missing decisions.

Questions for a practical final check

• The institutions and proposed activities are named separately.
• Every decision has a documented status, version and issuing authority.
• A draft or meeting is not recorded as an approval.
• Local field meanings and unresolved mappings remain visible.
• Data custody and contribution roles have distinct records.
• The next activity follows the actual permissions obtained.