Gluten Screen
Gluten Screen›Data

Provenance

Where the data comes from

Every source behind this site, when it was captured, and whether we refresh it. Including the ones we deliberately don't.

Live Queried at the moment you load the page. Never cached for long, never archived.
Periodic Parsed once from a source that still changes. The date says how stale it is.
Point-in-time Deliberately not refreshed. The date is the measurement itself, not a warning.
Third-party Someone else maintains it. The date is theirs, and it may be old.

DailyMed (FDA Structured Product Labels)

Periodic 30 April 2026

Screened per request against our parsed catalogue.

76 drugs in catalogue source

EMA product information leaflets

Point-in-time 30 April 2026

A point-in-time census, deliberately not refreshed. Wheat starch appears in nationally-authorised generics, which are outside the centrally-authorised set entirely, so re-running finds the same zeros.

1,360 centrally-authorised products parsed source

Paired US/EU starch source disclosure

Point-in-time 30 April 2026

Derived from the DailyMed and EMA captures above.

86 molecules authorised in both markets source

Plogsted gluten-free drug list

Third-party 28 April 2017

Maintained by a third party and not updated since the date shown. We report its age rather than hide it.

9.4 years old source

Matson et al., top 100 pediatric medications

Third-party 9 June 2025

A published study; fixed by definition.

source

openFDA enforcement reports

Live per request

Queried per request, cached one hour. Nothing archived.

source

PubMed (NCBI E-utilities)

Live per request

Standing query run per request, cached six hours.

source

Why some of this is deliberately frozen

The EU census is the clearest case. It parsed 1,360 centrally-authorised products and found zero declaring wheat starch and zero carrying a quantitative gluten amount. That is not a parsing gap. Wheat starch as a pharmaceutical excipient sits almost entirely in nationally-authorised generics, which are outside the centrally -authorised set that the EMA publishes centrally — so the disclosure clause that would fire has very little to fire on.

Re-running that capture would find the same zeros. A census is point-in-time by definition: it measured what the labels said when they were read. Dating it honestly is the correct treatment, and pretending it is continuously current would be the dishonest one.

The Plogsted list is the opposite lesson. It is a genuinely useful third-party resource that has not been updated in years, and its age is shown above rather than hidden. We keep it because a stale list with a visible date is still useful; a stale list presented as current is not. Our decay analysis measures exactly how much has drifted since.

Reproducing any of this

Derived datasets are rebuilt with a single command, which also rewrites the dates on this page:

python3 update_datasets.py

The analysis scripts and their CC0 outputs are linked from each research paper.

By

Founder and maintainer of Gluten Screen

Updated August 18, 2026

Not clinically reviewed — verify with your pharmacist

Drug labels change without notice. No date on this page makes a screening result a guarantee — confirm the specific product you have with your pharmacist.