In my company, there are 2 main categories for our testing deviations: deviating from a standard, or deviating from the protocol. For testing our implants we follow ASTM standards for our testing procedures, but there are certain tests where we always have to deviate from the standard by creating custom fixtures for testing. For testing our instruments, the verification tests are usually specific to each instrument, and are custom designed by the development and testing engineers. Since it is a new test that has not been executed before, it is common for them to have to deviate from the protocol in order to complete the test. I agree that it is not ideal to deviate from a test protocol, but in some cases the test cannot be completed without deviating, and as long as a justification is provided, I think it is acceptable to have deviations.