The purpose of the Trial Disease Assessments (TD) domain is to provide information on planned scheduling of disease assessments when the scheduling of disease assessments is not necessarily tied to the scheduling of visits. In oncology studies, compliance with the disease-assessment schedule is essential to reduce the risk of "assessment time bias." The TD domain makes possible an evaluation of assessment time bias from the SDTM, in particular for studies with progression-free survival (PFS) endpoints. TD has limited utility within oncology and was developed specifically with RECIST in mind and where an assessment-time bias analysis is appropriate. It is understood that extending this approach to Cheson and other criteria may not be appropriate or may pose difficulties. It is also understood that this approach may not be necessary in non-oncology studies, although it is available for use if appropriate.
A planned schedule of assessments will have a defined start point; the TDANCVAR variable is used to identify the variable in the ADaM subject-level dataset (ADSL) that holds the “anchor” date. By default, the anchor variable for the first pattern is ANCH1DT. An anchor date must be provided for each pattern of assessments, and each anchor variable must exist in ADSL. TDANCVAR is therefore a Required variable. Anchor date variable names should adhere to ADaM variable naming conventions (e.g. ANCH1DT, ANCH2DT). One anchor date may be used to anchor more than 1 pattern of disease assessments. When that is the case, the appropriate offset for the start of a subsequent pattern, represented as an ISO 8601 duration value, should be provided in the TDSTOFF variable.
The TDSTOFF variable is used in conjunction with the anchor date value (from the anchor date variable identified in TDANCVAR). If the pattern of disease assessments does not start exactly on a date collected on the CRF, this variable will represent the offset between the anchor date value and the start date of the pattern of disease assessments. This may be a positive or zero interval value represent in an ISO 8601 format.
A pattern of assessments consists of a series of intervals of equal duration, each followed by an assessment. Thus, the first assessment in a pattern is planned to occur at the anchor date (given by the variable named in TDANCVAR) plus the offset (TDSTOFF) plus the target assessment interval (TDTGTPAI). A baseline evaluation is usually not preceded by an interval, and would therefore not be considered part of an assessment pattern.
This domain should not be created when the disease assessment schedule may vary for individual subjects (e.g., when completion of the first phase of a study is event-driven).