Biometrics teams can generate hundreds, if not thousands, of SAS programs throughout the drug development process. So how do you know which ones to submit to the FDA? And how do you go about submitting the programs?

Why

Picture this. You submit your marketing application, and everything seems to go smoothly. But a few months later, you get that dreaded notification from your regulatory team. It’s a request for information, the FDA is asking for 25 additional adverse event tables. Due in 3 days! A medical reviewer disagreed with how you calculated your incidence rates, and they didn’t have the source code to verify your method. So the easiest path forward was for them to ask you to submit new tables, using the method that they specify in their request. That’s a lot of work that could have been avoided if the reviewers had the code to confirm the methods used in the initial tables.

What

According to the FDA’s Study Data Technical Conformance Guide, sponsors should provide the source code used to create all ADaM datasets, tables, and figures associated with primary and secondary efficacy analyses. This gives the agency everything they would need to reproduce the critical efficacy results. But why stop there? Submitting the source code for all ADaM datasets, tables, listings, and figures can help mitigate FDA follow-up down the road. You already did the hard part (writing the code). All that’s left to do is pass the files on to the publishing team.

Where

Within your ADaM define package (m5\datasets\[study]\analysis\adam), the eCTD structure allows for both a “datasets” folder, and a separate “programs” folder. Your SAS transport files and define.xml are placed in the “datasets” folder, and your source code files for datasets and displays can all be placed in the “programs” folder. No need to split them into subfolders to separate dataset programs from display programs. Just make sure there are no duplicate file names. Naming the programs in a way that would allow someone to easily identify what the code is used for is also a bonus. For example, the code that creates the primary efficacy table can simply be named “t-primeff”.

How

Although .sas files are technically an acceptable file format for submission to FDA, the guidance documents do say to submit non-executable file formats. While some argue that .sas is not an executable file format per se, our experience is that the FDA prefers .txt files. You can even write a quick SAS program that reads in all of the .sas files and exports them as .txt files.

In your Analysis Dataset Reviewers Guide, consider creating a table that lists each program, what datasets are used as input for the program, and what dataset/display is generated from the program. This table can hyperlink to the individual .txt files and to the statistical displays for ease of navigation.

Need additional clarification on submission of software programs? Contact us to speak with one of our Biometrics Regulatory Experts.

Evan Zucker, Principal Biostatistician, has 8 years of experience providing statistical support for clinical trials, with a specific focus on late-stage regulatory submissions. Mr. Zucker has supported upwards of 20 marketing applications and over a dozen FDA meetings, often working as the primary author of integrated statistical analysis plans and CDISC-compliant integrated analysis dataset (ADaM) specifications for the ISS and ISE. In supporting marketing application submissions, he also acts as a data standards expert, reviewing and updating study data packages to ensure compliance with the latest FDA requirements, technical rejection criteria, and CDISC/legacy guidelines. Mr. Zucker has collaborated with clinical researchers and medical authors across various therapeutic areas, including pain/anesthesia, psychiatry, otolaryngology, oncology, ophthalmology, rare/genetic diseases, and vaccines/infectious diseases.