Most low-carnitine screens normalized; the environmental model cannot confirm PCD

In 104,451 newborns, 782 had initial free carnitine below 10 µmol/L and 735 normalized on repeat. A four-factor preanalytic model reached AUC 0.75, but the study did not report molecular confirmation or follow screen-negative infants—so it is a quality signal, not a PCD diagnostic shortcut.

THE HOT TAKE

The strongest result is not 94% false positives. It is 94% repeat normalization—and those are not the same claim.

A 2026 regional multicenter cohort screened 104,451 newborns for primary carnitine deficiency. Initial dried-blood-spot free carnitine was below 10 µmol/L in 782 infants (0.75%). On repeat testing, C0 normalized in 735 and remained below 10 µmol/L in 47.

That large separation makes preanalytic quality worth attention. It does not show that temperature, humidity, season, or spot quality caused every initial low result, and it does not establish which infants did or did not have PCD.

WHAT THE PAPER ACTUALLY TESTED

The retrospective study covered births in Mianyang City from February 2019 through June 2025. Sampling occurred after 72 hours and at least six to eight feeds. Infants under 1500 g, those with severe birth asphyxia or severe diseases linked to secondary carnitine deficiency, and infants without repeat testing were excluded—an important boundary for NICU readers.