Legacy data and SDTM terminology

Legacy data and SDTM terminology

Legacy data conversion to SDTM is always difficult ...

CDISC occasionally organizes "Legacy Data Conversion courses", usually in combination with a CDISC Interchange - see  http://www.cdisc.org/Education. Also if you google for "CDISC legacy data conversion" you will find a multitude of articles and presentations about this topic, usually companies sharing their experience with this difficult topic. There will surely also be a good information about codelist mapping in these articles and presentations.

My own "quick thought" about your question is that in case you cannot map your terminology to the CDISC one, you simply develop your own codelist and put it in the define.xml and reference it in your variable declaration (ItemDef) in the define.xml.

Hi Victoria,The actual

Hi Victoria,

The actual problem in your example is that collected data does not tell if a drug dose was actually increased or reduced?

In some cases a reference to study protocol may help. For example, if the only option per protocol is reducing dosage, then you can use this info to adjust your mapping specifications ("DOSE ADJUSTED"=>"DOSE REDUCED"). Good and detailed documentation of such decisions is extremely important.

Kind Regards,

Sergiy  

unfortunatley on some studies

unfortunatley on some studies we don't have access to the protocol.  I am working on 17 legacy studies that were dropped by the original company and are now being picked up by my client.  We are looking at data that is 15+ years old.  Most of the data was collected before global participation in CDISC happend, some before CDISC was even around on a voluntary basis.

So I am still stuck with no way to use the standard codelists if I can't find a way to use what is in the existing codelists where they are not extendable.

thank you for your reply.

I know how to work with

I know how to work with legacy data as I have done quite a bit of migrating these type data to SDTM.  This was a specific issue I ran into that I had not encountered in the past (by luck).

I will look at the link and see if anything listed meets my needs

legacy studies

Hi Victoria,

As I already stated, just use your own codelist, when no mapping to the CDISC one is possible, and put it into the define.xml, and reference it in the ItemDef for AEACN or any other appropriate variable where you use your own codelist for.
Essentially define.xml is "leading" i.e. what is in define.xml is "your truth". Be sure to explain why you created your own codelists and did not use the CDISC ones in the reviewers guide. Also explain that any errors thrown by validator tools are a result of this decision.