Process validation is documented evidence that the process, run within set parameters, reproducibly makes material meeting its specifications and quality attributes. Prospective validation is preferred and should normally be done for all API processes within scope and completed before commercial distribution of the medicine made from the API. Concurrent validation can be used where replicate data are lacking because few batches exist, batches are made infrequently, or a validated process was modified; batches may be released before it completes on the basis of thorough monitoring and testing. Retrospective validation is an exception for well-established, unchanged processes where critical attributes and parameters are identified, in-process criteria and controls exist, there have been no significant failures other than operator or unrelated equipment errors, and impurity profiles exist; the batches reviewed should represent all batches in the period, including failures, and retained samples may be tested. The number of runs should reflect complexity and the size of any change: three consecutive successful batches as a guide for prospective and concurrent validation, more where needed, and usually ten to thirty consecutive batches for retrospective review unless fewer are justified. During validation the critical parameters should be kept under control and monitored (parameters unrelated to quality need not be), and validation should confirm each API's impurity profile is within limits and comparable to or better than historical or pivotal-study batches.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.