Establishing the Functions to Be Validated

The functions of a clinical data management system (CDMS) are to ensure that: the database reflects the case report form (CRF) data, as an exact representation or with defined transformations (for example, coding); the data are complete; and that access to the data are controlled. Each of these has implications for the validation of the CDMS. Formal rigorous and complete validation of any software package as complex as a CDMS is not possible. A pragmatic approach to selecting the functions to be validated must therefore be followed.

This paper discusses which functions are critical and must be included in any validation. These include basic functions such as data representation but also: data transformation, for example, coding or export; query management functions; and project- or study-specific software, such as plausibility tests. Some of these functions require ongoing validation of the CDMS.

Another important aspect of validation of a CDMS is the relationship between the CDMS, the people who use it, and the procedures by which it is used. Those human and procedural functions which affect the core functions of a CDMS must also be validated.

This paper describes a methodology for describing which functions of a clinical data management system must be validated. The practical impossibility of detailed validation of all functions of such a system dictates that a pragmatic approach is required. Such an approach is presented and explained.