Here’s an interesting test: go to your favorite search engine, search once for “clinical database validation” and a second time for “clinical data validation” (leave the quotes off). Many of the same links will be returned and in a similar order. The problem is that these two terms are for very different processes, so the results should also be very different … but aren’t. The search engines are not the only contributor to the confusion – people in clinical data management have been known to use the two terms interchangeably also.
Validation of a clinical database is a series of tests of the software.
Validation of clinical data is a series of tests of the data in the software.
So the difference is between testing the container and testing the contents.
The core of the problem seems to be the word “validation”. It is possible to validate a process, software, hardware, or data. The goals of the different types of validation are similar: to ensure the quality and consistency of the product. However, the procedures, documentation, and timing for process validation, software validation, and hardware validation are similar, but the procedures, documentation, and timing for data validation are notably different from the other three. For this reason, they are different skills – people who are good at one type of validation are not necessarily good at the other, although in practice there is some overlap. In clinical data management, software validation and data validation are the two most common forms of validation, so those will be the two types of validation discussed here.
The biggest difference between the two types of validation is in the timing:
Validation of clinical databases should be completely finished before any real data are entered.
Validation of clinical data occurs only once the data have started being entered and usually cannot be completed until all data have been entered.
Another difference between the two types of validation is in the documentation.
Software validation has a specific set of terms and expected documents that are generally consistent across GxP systems, which includes clinical databases. References to those terms and documents can be found in FDA guides, in industry standards references, and on Ofni Systems website.
Clinical data validation does have some common terms, but since few of the terms are defined by regulations the way the terms are used varies a lot more. Also, since validation of clinical data is generally a software-specific function, the documentation practices vary substantially.
Another significant difference in documentation is that validation of clinical databases is expected to include a full list and description of all test results, whether they passed or failed, but the product for a clinical data validation test that passes is the absence of results, i.e. zero problems. In other words, clinical data validation tests that pass require very little documentation; database validation tests that pass require quite a bit.
Since the expected documentation and timing of clinical software validation and clinical data validation are different, the procedures to perform the tests and generate the results are also different. The differences in timing and documentation are the primary sources of differences in procedures.
The phrases “clinical data validation” and “clinical database validation” are commonly confused even though they refer to different processes. Clinical database validation is a documented series of tests on the database while clinical data validation is a documented series of tests on the contents of the database. Despite the common goal of both processes, that is ensuring the quality and consistency of the product, the two processes differ substantially in timing, in documentation, and in the procedures used.