Can someone explain how the analysis datasets are created once you have the raw dataset? Is there a specification for that? Is there a raw demog dataset from which me need to create the demog analysis datatset (this is just an eg)? Who defines the derived variables or the method for defining them?
Also, are all the studies that are going on now follow the CDISC standards? I guess for creating analysis datasets we need to follow the ADaM model, right? Then what is this SDTM do? Can a particular dataset exist in both SDTM and ADaM model?
I know its a lot of questions. I am trying to get as much info as possible, and I spend a lot of time reading, but these basic things still remain unanswered!!
Every study is different from other and creating the analysis datasets is entirely dependent on the table shells. Depending upon the statistical analysis to be done on a given study the statisticians prepare the table shells , its a dummy table that shows you how the final table should look like and what variables it should contain. From these shells we do the programming and create analysis dataset for each table. Then reports are created as per the design of table shell.
If complex stats are being used the analysis datasets are sent to the biostatisticians and they create the report. So your analysis dataset will have variables from many tables and also calculation that have to be done to get the final variable.
The CDISC and STDM models are not used by all the companies, although FDA is in a process of standardizing the reporting.