You may recall that recently the FDA asked for feedback on potential adoption of the new CDISC standard: Dataset-JSON v1.1 as a replacement for XPT files. Well, the feedback was public and I read through all the comments and I’ll summarize here, so you don’t need to sift through them all yourself.
40 different individuals, companies or cross-industry bodies replied to the agency. Although most didn’t explicitly state it, it seems that all except a few were exclusively from the clinical side of our industry.
Most, 35, were positive and very much in favour of FDA adopting Dataset-JSON sooner rather than later. To be honest, I came away with the feeling that the vast majority of respondents thought that a move such as this was long overdue. Only two respondents were against this move. The remaining three gave feedback, but didn’t seem to indicate a preference in either direction.
Of those in favour, many balanced their enthusiasm with some concerns. I’ve grouped them into broad categories for us to discuss, but here’s a simple chart visualizing the feedback, ranked by frequency:
- Tool concerns
- Eleven pieces of feedback voiced concern over the fact that tools are not available yet, and vendors will need to update their existing tools, and that such updates come at a price. Pinnacle 21 was specifically called out in the feedback as not being currently compatible with Dataset-JSON v1.1.
- Transition concerns
- Seven respondents either voiced concern over the speed at which such adoption may take place or recommended that the FDA accept either XPT or JSON for a period of time.
- Implementation concerns
- Six gave feedback concerning the cost and impact of implementing Dataset-JSON, with the need for additional training frequently being called out as a potential barrier.
- define.xml
- Five discussed the future of the define.xml file in light of moving to Dataset-JSON.
- Tools exist
- While the need for tools was the single biggest concern, it was interesting to note that five respondents from the clinical space stated that their tools had already been updated. There were several mentions of the fact that CDISC CORE has already been made compatible with Dataset-JSON.
- JSON vs NDJSON
- As Dataset-JSON supports both .json and .ndjson file formats, four pieces of feedback mentioned this with several calls for the agency to simply standardize on .ndjson.
- Other health authorities
- Four also voiced concerns how other health authorities, as well as the various consortia and other innovation initiatives, are using CDISC standards, and that there would be potential difficulty if FDA moved to Dataset-JSON while the others remained on XPT.
- Floating point values
- There were also four pieces of feedback describing a very technical concern regarding the handling of floating-point values.
- Ease of editing
- Though only occurring twice, an interesting concern that was raised spoke to the fact that it’s quite difficult to edit XPT files, and that this adds an extra layer of security. Whereas Dataset-JSON files are essentially text that can be opened and modified very easily.
So where does that leave us regarding FDA adoption?
Obviously, I can’t speak for the agency, so I’m just reading the tea leaves here and giving it my best guess, but the Federal Register Notice seemed to be focused on determining two things:
- Does the industry want to move forward with Dataset-JSON v1.1?
- Are tools available, or would this be a barrier to adoption?
On the first question, the overwhelming feedback was positive.
On the second question, clearly there’s some divided opinion here with eleven organisations expressing a concern for the lack of available tools, while five were enthusiastically stating that tools were already available. So, does that send the message to the agency that industry is ready to move ahead already, or not?
For Instem, we have just finished up a Beta test of Dataset-JSON v1.1 in Submit™ with a number of our customers, and I’ll make a subsequent post about that, but Dataset-JSON has been implemented in Submit™ already, and we are now ready to deploy this at a moment’s notice. Just as soon as the agency adds it to their Data Standards Catalogue and/or the study data Technical Conformance Guide.
‘til next time
Marc
