Ralph Waldo Emerson wrote, "A foolish consistency is the hobgoblin of little minds." Ralph, however, was not developing drugs (although the Transparent Eyeball suggests he may have sampled them). We in clinical research know that "consistency" is one of the criteria that make up the "plus" in ALCOA+. But whereas attributability, legibility, originality, and contemporaneousness are all conditions that increase the likelihood of accuracy, consistency is integral to accuracy. When documentation is inconsistent, it presents two conflicting versions of events, which means that one version must be false. For example, let's say that a site's Delegation of Authority log lists a sub-investigator, but that person is not listed on the 1572. If this person is a sub-investigator, they must be on the 1572; if they are not a sub-investigator, they must not be listed as such on a delegation log. We can't simply assume that one source is correct and another is incorrect;  BOTH sources of information are suspect until we investigate the discrepancy and identify the likely course of events. An essential part of Good Documentation Practice, then, requires interrogation of all potential inconsistencies and resolution of all identified discrepancies - in other words, adequate quality control.  If we don't interrogate all potential inconsistencies as part of our quality control efforts, and a discrepancy is identified later (say, by an auditor or regulatory authority inspector), then the effectiveness of our entire quality control process is called into question. This is why our TMF audit process involves over 150 cross-checks - comparisons of two or more sources of documentation for consistency.  In past blog posts we've discussed how these checks help us identify missing documents, but it also helps us identify inconsistencies in documentation, such as the following:

• Team member roles and start and stop dates that differ on contact lists, training records, and meeting minutes
• Study-specific plan versions and dates that are not consistent in headers/footers, title pages, tracking lists, and version histories
• Names, versions, and dates of IRB-approved documents that are inconsistent on documents vs. IRB approvals
• Clinical site staff, roles, and start and stop dates that differ between the 1572 and delegation log
• Delegated tasks that are not consistent with roles on delegation log or with training documentation
• Study-specific training assignments on the training matrix that differ from training documentation, or that are not consistent between staff with the same roles
• Financial disclosure forms that are filled out differently by the same investigator across multiple sponsor studies
• Vendor contracts in which the scope does not match the TORO or activities suggested by documentation
• Vendor lists that differ in vendor management plans, Data Management Plans, and meeting minutes
• Inconsistencies in IP accountability

Inconsistent documentation is invalid documentation until enough corroborating evidence is found to support one version of events or another. Sometimes the "corroborating" evidence produces a third version of events, causing another mad scramble. When planning a clinical study, we need to be conscious of how many different documents are capturing information about the same topic.