top of page
Search

SEND Errors Are Not Going Away What Three Years of FDA Data Tells Us and What Teams Should Do About It

Regulatory teams are under more pressure than ever to submit clean, review-ready data. SEND has been mandatory for CDER nonclinical studies since 2017. Enforcement of Technical Rejection Criteria has been active since September 2021. By now, you would expect most submissions to be arriving at FDA without structural errors.


The data says otherwise.


Three years into active TRC enforcement, SEND continues to account for the majority of errors across every rejection code FDA tracks. The errors are not complex. They are not obscure edge cases in the SENDIG. They are missing files, wrong file tags, and mismatched study IDs the kind of issues that a structured pre-submission review would catch in an hour.


This is not a standard compliance problem. It is a process problem. And the cost is real not just in rework hours, but in the one thing drug development cannot afford to lose: time.


What FDA’s data actually shows

FDA has been tracking TRC conformance since enforcement began and presents the data at public CDISC SEND face-to-face meetings. Looking at the numbers across 2021–2023 tells a story that should concern any team managing SEND submissions.


Before enforcement took hold (2020–2021): error rates were staggering.


In the quarters leading up to September 2021, FDA’s own data showed that IND nonclinical SEND submissions were arriving with TRC errors at rates of 51–61%. For NDA nonclinical submissions, error rates ran between 32–47%. For BLA nonclinical submissions, the error rate peaked at 85%.


To be clear about what this means: in some quarters, the majority of SEND packages arriving at FDA had a structural problem serious enough to trigger a TRC flag. This was before any scientific review had begun or a reviewer had opened a single dataset.


After enforcement began: measurable improvement, persistent problems


Once TRC enforcement began in September 2021, error rates dropped sharply to the 5–12% range for IND nonclinical studies. While that improvement was significant, the errors did not disappear, and the pattern of underlying issues remained largely unchanged.


In 2023: SEND still dominates the error picture.


FDA’s April 2024 presentation covering the full calendar year 2023 shows that across 331,647 eCTD submissions to CDER, roughly 1% were rejected. Within that, SEND nonclinical data accounted for a disproportionate share of TRC errors across every error code:

  • 65 submissions failed Rule 1734 due to a missing Trial Summary (ts.xpt) dataset. Of those failures, 66% involved SEND nonclinical data in Module 4.

  • 21 submissions failed Rule 1735 due to incorrect file tags. Of those failures, 62% involved SEND nonclinical data

  • 17 submissions failed Rule 1736 due to missing DM datasets or define.xml files. Of those failures, 41% involved SEND nonclinical data.

  • 123 submissions received Rule 1738 errors due to STUDYID mismatches between ts.xpt and the STF file. Of these, 61 involved nonclinical submissions, and 89% were INDs, with repeat-dose and single-dose toxicology representing the most commonly affected study sections.


Every one of these error codes maps to something structural and checkable. These are not interpretation failures or gaps in SENDIG knowledge. They are submission packaging failures the kind that a pre-submission review, even a basic one would identify before the file reaches FDA.

What these errors actually cost

When a submission fails a TRC check, the rejection happens automatically, often within 24 hours of submission. The submission is not uploaded into FDA’s system, and the review clock does not begin. The package must then be corrected and resubmitted, with the review clock restarting only after the corrected submission clears validation.


That correction and resubmission cycle adds a minimum of two to four weeks to the review timeline in practice. For a repeat-dose toxicology INDs, which FDA data shows is the most common study type caught by Rule 1734, those weeks matter. They can delay the planned start of first-in-human clinical studies, put pressure on development programmes operating against competitive timelines, and postpone filings that entire organisations have been working toward.


And the rework cost is not just calendar time. It is internal resource time diverted from the next study. It means coordinating with the CRO to retrieve corrected files. It is QC cycles that should never have been needed.


The harder question is: how many of these delays were avoidable? Based on what the errors actually are a missing ts.xpt, a mismatched STUDYID, a wrong file tag, the answer is almost all of them.

Why the same errors keep recurring

Three years of enforcement data showing the same error types is worth pausing on. It is not a sign that SEND is too technically difficult to get right. It is a sign that something in how SEND is being prepared and checked is not working.


The most likely explanation is structural: in most organisations and CROs, SEND is still largely a back-end activity. It is prepared after the study report is finalised, often by a separate team working from a different process. The submission is packaged late, checked quickly, and filed.


In that model, there is no natural point where someone stops and asks: does this package pass the same checks FDA will run the moment it arrives? Does the ts.xpt exist? Do the STUDYID values match across the STF and the ts.xpt? Are all files in the right directory?

Those checks are not complicated. They take time, rather than deep technical expertise. What they require is a deliberate, structured step in the workflow one that happens before submission, not after rejection.


FDA has now formalised exactly this expectation in Appendix I of TCG v6.2, published in March 2026. The guidance describes a pre-submission SEND data review as a best practice, and lists the categories of errors submission issues, SENDIG conformance failures, study design representation problems that sponsors are expected to check for before filing. The FDA data makes clear why: these are not hypothetical failure modes. They are the real errors FDA keeps receiving.

What a structured pre-submission check looks like

The TRC Self-Check Worksheet published by FDA is a useful starting point. It covers the four high-level validation rules (1789, 1734, 1735, 1736) and can be completed per study before a submission is assembled. However, these checks alone are not sufficient to ensure a review-ready submission.


A complete pre-submission SEND review should cover two distinct areas, both described in Appendix I of TCG v6.2:


Verification of the submission package:

  • All SEND datasets and supporting files (nSDRG.pdf, define.xml) should be placed correctly within m4\datasets\[study]\tabulations\send.

  • File names and STUDYID values should be consistent across all datasets and define.xml.

  • No missing files, no empty (0 kb) files, no duplicate dataset names.

  • Direct folder-level verification not just through eCTD tooling, which can mask physical file locations.

 

Verification of the dataset content:

  • SEND datasets should comply with FDA Business Rules and accurately represent the study design.

  • Groupings, counts, and summary statistics in SEND should be consistent with the study report.

  • Individual animal-level data should align with the study report for representative endpoints, days, and timepoints.

  • Any discrepancies or special grouping logic are clearly described in the nSDRG.

 

This is not a light-touch final check. It is a structured, documented review that mirrors what FDA reviewers will do when they open the package. The point is to find problems before they do.

The Takeaway

FDA’s conformance data across 2021–2023 makes a consistent point: SEND errors are not a knowledge problem. Teams know what SEND requires. The errors that continue to trigger TRC failures - missing ts.xpt files, wrong file tags, mismatched study IDs are process failures, not technical ones.


The fix is not more SENDIG training. It is a formal, repeatable pre-submission review step that treats SEND packaging with the same discipline as the data itself.


TCG v6.2’s Appendix I puts that expectation in writing and FDA’s own error data makes clear why.


The question is no longer whether your SEND dataset is compliant.

It is whether it will clear FDA’s first check before a single reviewer opens it.

References

  • FDA SEND F2F Spring 2023 — TRC Conformance Statistics and Trends (Data period: January 2020–February 2023)

  • FDA CDER — Study Data Rejections in eCTD Submissions, April 24, 2024 (Data period: January–December 2023)

  • FDA Study Data Technical Conformance Guide (TCG) v6.2, March 2026 — Appendix I

Comments

bottom of page