Good Stone and Provenance (not Providence)

Provenance not Providence in validation.

All validations whether applied to process, method, software or equipment rely on some unimpeachable component. This component may be a data matrix, dye and diluted dye, radioactive standard, calibration plate, etc. As in architecture, validations rely on “good stone” for the first story. Subsequent stories or operations are only as good as that first story.  If for example a dye is used, the validation is only as the provenance supporting that dye; if a thermo probe is used to map a freezer or heater, the validation is only as good as the provenance and calibration of that probe.

When using dilutions of a dye or authentic material to emulate results from a curve and QC set, the documentation from the original preparation must be as complete as possible. Subsequent dilutions must be documented to complete the provenance.  The National Institute of Standards and Technology (NIST) provides some well documented dyes and the US Pharmacopeia provides limited standard materials to meet some validation needs.

Provenance of a data matrix to validate a software’s performance is a bit more circular but no less important. The data set must be constructed, validated using an approach orthogonal to that intended to be tested and verified by another laboratorian and approved before use.  It is usually the case that orthogonal approaches may not be available and here is where the validation of the data set becomes circular with the validation of the software.

This should not be insurmountable.  The data matrix can be constructed and verified in a pre-validation exercise, certified through a protocol and fixed such that it cannot be modified.  With that provenance, the data matrix can then be moved into testing of the software.

Again “Good Stone” is as important to validation as to architecture, without it buildings and bridges fail and validations are rejected.