It’s crunch time for the pharmaceutical industry with CDISC’s Study Data Tabulation Model (SDTM) standards. While the standards aren’t new, as of this year they will be required by at least two regulatory authorities: The U.S. Food and Drug Administration (FDA) and Japan’s Pharmaceuticals and Medical Devices Agency (PDMA).
By Mark Tumelty, Manager Biometrics & CDISC, CSC Life Sciences Business Process Services
In theory, this shouldn’t pose too big a challenge to most large or mid-tier companies. However, my perception from working with clients is that there is a lack of understanding about SDTM’s wider intended use.
Not only is SDTM a database specification, but it’s also a submission specification, which means you have to be able to describe those datasets and submit them in accordance with the guidelines, as well as be able to understand and explain when you run into exceptions.
For example, you can create a SDTM dataset that looks exactly like the specification for data definitions written by CDISC, with all the variables, the labels, etc. However, while the standard is pretty flexible and accounts for a lot of variables, there are situations in a submission where certain datasets don’t quite fit to the recommendation in the SDTM dictionaries. CDISC recognizes that this can – and will – occur, so what should you do?
To begin with, you need to understand that SDTM is a guideline, not a black and white standard. There is leeway, and it’s okay if a certain set of data doesn’t fit the standard – this is where the exceptions rule comes in. When that situation does occur you need to explain in the reviewer’s guide exactly why you are making an exception and what steps you have taken to try to make the datasets fit the specifications. This is in addition to case report data tabulation data definition in xml formats and annotated CRFs as required per CDISC’s guideline to describe the datasets for each study, specific data analysis and integrated summaries.
Where and When
The timelines for making SDTM mandatory have also created a bit of a stir. Most companies had expected FDA to require SDTM first, but in fact the PDMA will be the first agency to make SDTM mandatory.
While this shouldn’t create too much disruption, it does mean that if you were planning to file to PDMA later this year without using SDTM you’ll either have to shift your dates or ensure you are SDTM-ready across the board. Where this might be a challenge is with a legacy study or older data that was a piece of a filing that a company deemed not worth the time to translate into SDTM.
The advantage of SDTM is it enables true global standardization of study datasets. So in the future, when an agency gets data from other parts of the world, all of that data will be submitted in the same format, following the same rules about deriving data points.
When you are dealing with data from statisticians and clinicians, there are thousands of different ways to break down the data or to conduct analysis, or even how to group data together. Using SDTM standards ensures all the data – no matter where in the world it comes from – is derived in same way with the same mindset behind it by defining what’s in the dataset and what it will look like.
Companies also benefit because it removes uncertainty over data definition and helps them streamline processes. Before SDTM, companies ended up with raw data and had to try to decide the best way to present it to the agencies, such as how to group it. SDTM enables companies to write standardized programs that make the development of datasets more efficient and streamlined, so a dataset on Study One will look the same as in Study 50.
This is particularly relevant when filing to different agencies. In the past, companies might have different programs for FDA, PDMA and the European Medicines Agency (EMA). A common challenge was a company might start off filing in one region, for example the United States, then extend worldwide, which would involve finding the submission standard for other regions, seeing if those match your U.S. filing, and if not, fixing the submission before sending it.
But the dress rehearsal is now over and as of October 2016 for the PDMA and December 2016 for the FDA every company regardless of size will have to the SDTM ready.
I hope to see you at the PhUSE Computational Science Symposium on March 13-15, 2016 in Silver Spring, MD or contact us.