How can you find a document if you don't know it's missing? This is no Zen koan, but a problem that faces every study team that attempts a quality control review of their Trial Master File. Ask this question in an audit, and you get a variety of approaches, ranging from the thoroughly redundant ("We print an inventory and compare it to the TMF") to the intuitive ("Each department checks their own sections, because they know what's supposed to be there"); to the quantitative ("We run a missing-documents report from the eTMF system that calculates the number of empty placeholders").
The most useful analogy for QCing a TMF is cleaning a database. Every clinical trial database is cleaned through a series of clearly-defined manual and electronic checks performed by a cross-functional team, defined in a number of plans and specifications. Site monitors perform source data verification; medical monitors perform medical reviews; the EDC system runs automated edit checks for missing and inconsistent data; data managers reconcile against safety and other external data; and biostatisticians look for inconsistencies via statistical deliverables.
A good TMF QC plan should approach the same level of definition and complexity. Roles and responsibilities should be clearly defined. Both "internal" (eTMF) and "external" documents (those stored outside the TMF) should be taken into account. Cross-checks should be built in, with each section compared to at least one other source of information to gauge completeness. Checks should be performed at both the micro level - looking at individual documents for completeness, legibility, and accuracy of filing - and at the macro level - tracing issues through the file to determine if the file as a whole tells the "story" of the study.
In this next series of blog posts, we'll look at specific approaches for "cleaning" different TMF sections.