Watch this Ask the Expert session to see how SAS® Viya® enables drug developers to accelerate evidence generation and create integrated evidence packages with raw data.
Watch the Webinar
You will learn how to:
- Efficiently and accurately define patient cohorts without the need to harmonize your data first.
- Accelerate evidence generation through cross-collaboration among study team members.
- Use raw data to create cohorts and generate evidence to support integrated evidence packages.
The questions from the Q&A segment held at the end of the webinar are listed below and the slides from the webinar are attached.
Q&A
The code flows involve procedures such as PROC FEDSQL. Do we know if any level of validation is required for these procedures if they are used for regulatory submission?
The advantage of SAS code is that it is well-documented, which regulators appreciate because it provides transparency, thorough documentation, and governance for any updates related to validation in accordance with your SOPs. Whether validation is required depends on your SOPs, including those governing submission or integrated package submission. For many clients, macro libraries are validated. SOPs may also allow the use of out-of-the-box procedures, provided you can demonstrate that the procedure is standard and supply the corresponding SAS documentation. While this may not be a definitive answer, the requirement ultimately depends on your SOPs.
Can I replicate my cohort definition in multiple datasets?
Absolutely. One of the greatest things about what Mary has built for me is that I do not have to first standardize my data, then harmonize it, and then ensure that I have a common data model. I can take the raw data—whether it is Optum, my own registry, Meredith, or even Purple Lab—and use my existing code and macro library to read and interrogate the data, so that metadata is already available to me. Then, I can apply that definition across multiple datasets. The only limitation is that it will not combine datasets together, especially if you have legal or contractual obligations to keep them separate, which is the case for the majority of data vendors. We will honour that, and we expect you to honour your own contracts with your data vendors. You simply cannot do this simultaneously. However, you can take the same custom step, as well as the Visual Analytics report, and apply it by uploading new data. You will then have two, three, or even ten reports to compare, right?
Do I need a certain level of expertise to build cohorts to use this in this way?
No, as long as you have someone who can help you build the code asset when creating a custom step, you’re good to go. Building the interface is drag-and-drop, so just about anyone can do that now. You just need backend code support to assist you. There are also out-of-the-box SAS custom steps that cover a wide range of processes, and those will often get you quite far. Ultimately, it depends on how you want the user interface to look. The system is accessible to a wide range of experience levels and supports various coding languages—it’s really a ‘choose your own adventure’ approach.
Can I use R or Python to build my cohorts or must it be SAS?
Within custom steps, you can incorporate your open-source code. For example, you can include your Python code within PROC Python. There are additional methods for integrating R, with new options emerging, and you can also insert Python code directly into your flows. By adding a Python code node, you can code in Python alongside all the custom and out-of-the-box steps that SAS provides. This offers significant flexibility.
In the life sciences industry, there is increasing interest in using multiple programming languages, and this approach makes it easy to work creatively with various languages. It also allows you to validate your code and packages against others written in different languages. For instance, if there is an R or HADES package from OMOP or OHDSI that you would like to use, you can do so within SAS. SAS provides the general capabilities, governance, and transparency needed to trace data lineage—from its origin to its endpoint—while evaluating the impact of Python, R, SAS, or any other package on your data and results.
Want more tips? Be sure to subscribe to the Ask the Expert board to receive follow up Q&A, slides and recordings from other SAS Ask the Expert webinars.
