Investigators and sponsors should cover every stage of the data life cycle, concentrating on the data that matter most, proportionately and documented, through processes that protect participant confidentiality, keep computerised systems fit for purpose and properly used, safeguard randomisation, dose adjustments and blinding, and support key decisions (data finalisation, unblinding, analysis set allocation, design changes, IDMC activities). Blinding integrity matters in system design, user accounts, delegation, site data access, transfers, database review before planned unblinding and analysis; who may access unblinded information, and how, should be defined and written down (in the protocol, data management or statistical plans or delegation records), and staff interacting with sites should not see unblinding information unless the design justifies it (for example unblinded monitors), with mitigations against inadvertent unblinding of blinded site staff. The risk assessment should cover the chance of unblinding, and every planned or unplanned unblinding documented, with unplanned ones assessed for impact and acted on.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.