What this sample contains
On October 18, 2026, we requested the FDA warning-letter table ordered by posting date, read a 25-record source window and saved five records. All five saved records showed a posting date of October 13, 2026. Their issue dates preceded posting by 7 to 32 calendar days.
This difference matters when building a publication-monitoring queue. A posting date can indicate when a record appeared in the public table, while the issue date describes the letter itself. Neither should silently replace the other. The calculation is postedDate minus letterIssueDate in UTC calendar days.
Read missing links narrowly
The five saved table records had no posted response or closeout link in this observation. That is a statement about the fields captured from this table window. It does not establish that a company never responded, that corrective action has not occurred, or that a present regulatory concern remains unresolved.
Open the original FDA letter and any later updates before making decisions. The FDA warning-letter index is the source; every saved record retains its official letter URL. We do not reproduce letter bodies or assign supplier risk scores.
A better review-queue design
Use letterUrl as the record identity. Preserve MARCS-CMS ID, posting date, issue date, office and observedAt, then add a human reviewer and a reviewed-on date. If you link a record to a supplier master, verify the entity match rather than treating similar names as identical companies.
A repeated observation can surface a newly posted response or closeout link. An absent row from a later capped window cannot establish that a letter was removed or a matter resolved.
Method and boundaries
The output cap stopped this run after five saved records, while the source table reported more available records. Rows tied on the same posting date can rotate at the boundary. This is not a census of letters issued or posted in a week, nor a sample suitable for estimating industry-wide enforcement trends. The Actor captured table metadata only.
Download the exact input, selected records and coverage summary, or use the linked workflow to build a small review sheet. The example provides a research queue, not a legal or product-safety conclusion.