Documentation of silent alterations: roughly half of U.S. federal health datasets changed without notice

Data/communications

Date
Actor
Multiple health agencies (documented by academic investigators in The Lancet)
Type
Data/communications
Section
DATA ERASURE, TERMINOLOGY MANDATES, AND STATISTICS

Summary

A July 2025 investigation published in The Lancet found that approximately half of U.S. federal health datasets examined were substantially altered in the two months after EO 14168 — predominantly silent word substitutions such as "gender" changed to "sex" in variable names, codebooks, and documentation — without version notes or change logs, corrupting the scientific record's integrity because the altered files remained presented as originals. This is the principal peer-reviewed quantification of the "gender"-to-"sex" live-dataset edits.

Status

Alterations documented remain in the affected files except where court-ordered restoration applied; archival mirrors (RestoredCDC.org, Data Rescue Project, End of Term Archive) preserve pre-alteration versions.

As recorded in the data of . Check the sources for anything newer.

Sources