The TA USER GUIDE-Asthma piloted a new approach to standards development, which is being applied to all TA projects. The steps in this approach are described below. For some groups of data, all products described below will not be present because they are redundant with material available elsewhere e.g., data collection and datasets are adequately described in the CDASH and SDTM standards; they were still under development at the time this guideline was finalized e.g., controlled terminology and supplements for questionnaires); or they are determined to be out of scope for a specific project in the Project Charter.
This process will be modified in the future to extend the TA User Guides to cover implementation of standards for study design and analysis.
1) Understand the meaning and role of data in the therapeutic area and where the data come from in the clinical processes. Write text for the TA User Guide to explain this. Include references and figures as necessary.
2) Construct concept maps for inclusion in the TA User Guide to illustrate processes and the inherent structure of the data. This process uses “modules” based on BRIDG classes. The color code in Section 5.4.7.2 explains the kinds of things (instances of modules) that were used in the concept maps. Concept maps also include data items associated with these things, generally in “bubbles” with thin black outlines. Concept maps usually do not show all data items as doing so would often make the concept maps cluttered and less clear.
3) Identify the concepts in the concept map. In general a concept corresponds to a module or a pair of modules. The most common kind of concept is an observation/response pair.
4) Devise a template for creating metadata for a group of related concepts. This template will be included in the metadata workbook for the group of concepts. Note that templates created for the TA User Guide-Asthma may be re-used for other therapeutic areas, so this step will not always be needed.
a) The template is constructed by selecting among the attributes of the BRIDG classes corresponding to the modules involved (those which are important for the concepts being modeled). It includes concept variables that represent individual data items; each is derived from BRIDG and named with BRIDG class, class attribute, complex datatype, and datatype component(s). Periods separate the parts of the concept variable name. Complex datatypes often have multiple levels within them, so the number of levels of datatype specification for a concept variable varies.
b) Associate with each concept variable either a description of a simple datatype (e.g., decimal or free text) or information on the Controlled Terminology associated with the concept variable. Sometimes the Controlled Terminology can be specified completely at the concept level, sometimes a code list can be identified, but not the value list, and sometimes only that there may be controlled terminology.
c) Add to each group of closely related concept variables (usually at the class attribute level) an “attribute” description
d) Add SDTM variable mappings, as much as possible. SDTM variable mappings are not always present e.g., SDTM generally does not include codes, only decodes and sometimes the mapping between concept variables and SDTM variables is not direct.
5) Copy the template to create a sheet in the metadata workbook for each concept identified.
a) Provide metadata about the concept (Definition, TEST and TESTCD if applicable, value of PRESP if applicable)
b) Remove any concept variables not relevant for this particular concept.
c) Further specify controlled terminology, as much as possible.
d) Further specify SDTM mappings, as needed.
e) Add information about other concepts that may or must be linked to the concept. Sometimes properties of another concept will be mapped to the SDTM record for the concept being developed; if so, the SDTM variable name is included e.g., the lab that conducts a lab test is a separate concept from a lab test/result, but its name is included in the SDTM record for a lab test/result.
6) For each concept, create a bulleted list of data items to be included in the TA User Guide. The list of data items is a simplified, natural language version of the information in the metadata.
7) Create examples for inclusion in the TA User Guide
a) Write a description of the situation that produced the example data.
b) Create a CRF mock-up (not relevant if data is likely to be supplied electronically).
c) Create SDTM dataset example(s) with row captions.