Clinical data collected in SDTM, what can be left out and how to document

Hi Hannes, …

Hi Hannes, 

System generated data like CRF name or page numbers etc., which is not actively collected

Usually this is not considered as relevant clinical information. So, it’s not mapped to SDTM tabulation data. There are always protentional exceptions. A general rule is to store only information related to planned analysis.

NOT DONE records

It depends on particular data type and data collection process. Again, decisions are usually made based on planned use/analysis of collected data. Here are some examples:

There is a selected check field on CRF that Subject did not have any AEs –> No mapping to SDTM; no records in AE domain. In case if Sponsor need this info, then typical implementation is to create a record in SUPPDM dataset.

Not done schedule Lab assessment –> A record in LB dataset with LBSTAT=’NOT DONE’. Explanation for Reason Not Done is expected in LBREASND variable. If there are only few Lab tests have missing results for scheduled assessments, then you need to create separate records for each test. If all Lab tests were not done, then a recommended approach is to create a single record with  a special Test Code/Name like “LBALL”/”All Lab Tests”  

Any AE, or Any Medical History information

does not need to be tabulated in SDTM, right?

I am not sure about your question and assume that it’s a case of “No AEs” described in above example.

But what if a CRF system as well as the eDiary is collecting for example demographic data like weight, age etc.? Can one source be left out from SDTM (they can be different)? My understanding of SDTM IG 3.2 is that all collected data must be tabulated. In this example AGE from eCRF would for example be in DM and the source from eDiary in SUPPDM.

Note, that Subject Weight is Vital Signs info, not Demographics and should be mapped to VS domain. There is a special ADSL (Analysis Subject Level) dataset in ADaM data which keeps such analysis related info.

Yes, you can store two different Age values from different data collection systems for the same subjects. Use DM domain for Age you will utilize in your analysis according to the study SAP. Document all details in the study metadata like define.xml and Reviewer’s Guide to avoid potential confusions and misinterpretation of study data.    

Another example, which might be more interesting, is if SAE data is collected twice, on time for clinical data base (tabulated in SDTM) and one time for safety evaluation (ignore the fact that it would be possible to collect this one time only). So the same data is collected twice on different pages and can then obviously be different (is probably not the best solution). Is there any guidance on how to deal with this? Can the SAEs not intended for the Clinical Database be left out and the aCRF annotated with [NOT SUBMITTED] for those pages or must this data be added to SUPPAE

A guidance is to avoid such cases because they can compromise an overall quality of the whole study data. This is serious data management problem. If it’s too late (data is already locked), you need to provide a lot of explanations in different documents like Clinical Study Report (CSR), both tabulation and analysis Reviewer’s Guides, etc.

Kind Regards, 

Sergiy