Where every figure comes from

The product spine

The spine is the FDA National Drug Code Directory, served through openFDA at api.fda.gov/drug/ndc.json. It is the register of drug products listed with FDA by their labellers. Retrieved 4 September 2026. The dataset's own last-updated stamp at that point was 2 September 2026.

The query is on dosage_form.exact, taking four values: LOZENGE, PASTILLE, TROCHE and GUM, CHEWING. Results are paged in blocks of 1,000 and the pipeline fails loudly if the number of records retrieved does not match the total openFDA reports for that query, so a partial pull cannot silently become the dataset.

After deduplication this yields 1,260 products, resolving to 52 distinct active ingredients, 438 brand names, 157 labellers and 2,081 package configurations. 1,257 are over-the-counter and 3 are prescription.

The four dosage forms

The listed column is the total openFDA reports for that exact dosage form. The kept column is what remains after deduplication.

openFDA drug/ndc, dosage_form.exact, retrieved 4 September 2026. The difference of 20 between the two columns is explained in the next section.

One row per product code

A National Drug Code can appear in the directory against more than one label document: a relabelled listing, a superseded label still on file, or a bulk entry sitting alongside the retail one. 20 product codes in this dataset did.

The site keeps one record per code, chosen by a fixed rule rather than by whichever arrived first. A listing with a brand name beats one without. A finished product beats a bulk one. A retail marketing category beats one recorded as supplied for further processing. Where those are equal, the later marketing start date wins. The rule is deterministic, so two builds of the same data produce the same site.

No record was dropped for having no product code and none for having no active ingredient. Both counters exist in the pipeline and both were zero on this build.

The label text

Label content comes from openFDA's drug label endpoint at api.fda.gov/drug/label.json, retrieved 4 September 2026. It is queried by set_id, taken from the openfda.spl_set_id field on the product record, in batches of fifty codes per request. That endpoint returns the Drug Facts panel as discrete fields, so no label markup has to be parsed.

1,183 of 1,260 products carry a label extract. The remaining 77 have no matching label record, usually because the product record carries no set id at all. Those pages show the listing fields and say plainly that no label was found. They do not borrow text from a similar product.

A reference quotes a label; it does not republish one. Each field is stored as a capped extract, the specific fields used are named on the page, and every product page links to the full label on DailyMed, where the complete and current document is the authoritative one.

Strengths

openFDA records a strength as a string such as 5.8 mg/1 or 1.4 g/100g. The pipeline splits it into a number, a unit and a denominator so a column of strengths can be sorted and aligned. Where the denominator is 1, meaning one lozenge or one piece of gum, it is not printed, because it is noise to a reader.

Strings that do not fit that pattern are carried through and displayed exactly as FDA recorded them, and are excluded from any numeric range. A range mixes only strengths that share both a unit and a denominator: milligrams per lozenge and grams per hundred grams are not the same measurement, and a minimum and maximum spanning both would be meaningless. Where an ingredient has strengths in more than one unit, each unit gets its own range.

How a use is assigned

A product appears under a use only if its own label says so. Assignment is by phrase match against the label's Uses field, the Drug Facts section headed Uses, or the Indications and Usage section on a prescription label. Nothing is inferred from the active ingredient, and nothing is inferred from what a product is generally understood to be for.

The Purpose field is deliberately excluded from this match, and the reason is instructive. An earlier build matched on Purpose as well. Because the monograph purpose category for menthol is oral anesthetic, that put several hundred ordinary menthol cough drops under a heading about mouth and gum pain. Oral anesthetic is a regulatory classification of how an ingredient acts, not a description of the complaint someone has. Matching on it produced a confident, tidy and wrong answer.

Twelve uses have at least one product. Where a category was defined and no label matched it, the category does not appear on the site at all rather than appearing as an empty page.

How a purpose is assigned

The purpose is matched separately, against the label's Purpose field and its active ingredient statement only. This is the regulatory category the labeller declared, and it is the closest thing on a Drug Facts panel to a pharmacological classification. Five purposes have at least one product.

Keeping the two facets apart is the point. A use answers what the label says the product treats. A purpose answers what the label says the ingredient does. They are different questions and combining them loses both answers.

The rule layer

21 CFR 341, Cold, Cough, Allergy, Bronchodilator, and Antiasthmatic Drug Products for Over-the-Counter Human Use, is the part of the Code of Federal Regulations that still governs a cough lozenge in its own words. The text used here is the 2025 annual edition, retrieved from GovInfo at govinfo.gov on 4 September 2026. 7 sections are quoted verbatim with their paragraph citations. The regulation is never paraphrased, because a paraphrased regulation is a wrong regulation.

The definition the whole subject rests on, for example, is this:

(b) Oral antitussive drug. A drug that either is taken by mouth or is dissolved in the mouth in the form of a lozenge and acts systemically to relieve cough.21 CFR 341.3(b)

Most of these products are marketed under the over-the-counter monograph system as it was restructured by the CARES Act in 2020, and the FDA listing records a monograph identifier for each. In this dataset the identifiers are these.

Identifiers and counts from openFDA drug/ndc application_number, retrieved 4 September 2026. Titles and CFR crosswalk verified 4 September 2026 against OTC Monographs@FDA and Final Administrative Orders for Over-the-Counter Monographs; Availability, 86 FR 52474 (21 September 2021) and Rulemaking History for OTC Oral Healthcare Drug Products and 21 CFR Part 341 on eCFR.

Why the rule layer quotes one part and not two

A reference on this subject is expected to cite two monographs: cough and cold, and oral health. Only the first has a CFR part. Oral health care drug products were proposed as 21 CFR part 356 in 1988 and again in 1994 and never finalised, so the tentative final monograph became a deemed final order under the CARES Act without ever being codified. A citation to 21 CFR 356 is a citation to something that does not exist. That is why this site quotes part 341 for cough suppressants and describes the oral health rules through the monograph identifier on each listing instead.

This matters here more than it might elsewhere, because M022 is the monograph most sore throat and mouth pain lozenges are actually marketed under. Benzocaine, dyclonine, hexylresorcinol and the oral demulcents sit there rather than in the cough and cold monograph, and a reference that invented a CFR citation for them would be wrong about the majority of its own subject.

What is excluded

Orally disintegrating tablets

The dosage form TABLET, ORALLY DISINTEGRATING is not included. openFDA listed 572 products in that form at the time of this build, so the exclusion is a large one and worth stating plainly.

A different dosage form. It disintegrates on the tongue in seconds rather than dissolving slowly, and the listings are dominated by systemic prescription medicines. It is a tablet engineered to fall apart on contact with saliva in a few seconds, which is close to the opposite of the slow dissolve that defines a lozenge, and the listings are dominated by systemic prescription medicines that have nothing to do with the throat. Including them would roughly double the size of the site and change what it is about. A reader who wants those products should query the FDA directory directly.

Nothing else is excluded

Within the four dosage forms, every listing is on the site. That includes 76 products in marketing categories nobody can buy: bulk ingredients for prescription compounding, and products supplied for further processing. They are real rows in the federal register of drug products, so they are shown, flagged as not available to a consumer, and sorted below the retail listings. Deleting them would make the counts on this site disagree with FDA's for no stated reason.

Homeopathic products are included on the same terms. Twenty-one listings in this dataset carry the marketing category recorded by FDA as unapproved homeopathic. Where such a product's label states that its uses are based on traditional homeopathic practice rather than on modern evidence, that statement is reported, because it is on the label.

Three sources that mislead

Each of these was tested during this build. They are documented because the next person to build something like this will otherwise lose the same afternoon.

• DailyMed ignores the dosage form filter. The v2 service accepts spls.json?dosage_form=LOZENGE without complaint and returns unrelated products. It fails silently rather than erroring, so the result looks like data. It is not used here for anything.
• DailyMed will not serve a full label as JSON. Requesting spls/{setid}.json answers HTTP 415. The complete document is available only as XML; the JSON suffix works on subresources such as media. Label text on this site therefore comes from openFDA, which serves the same content as structured fields.
• The eCFR full-text endpoint requires response compression and returned HTTP 503 throughout this build. GovInfo's annual CFR edition serves the same text reliably and is what the pipeline uses.
• The OTC monograph register has moved. The address dps.fda.gov/omuf now redirects and should not be cited. The citable host is accessdata.fda.gov/scripts/cder/omuf/.

Known limits

A listing is not an inventory. The National Drug Code Directory records what labellers have told FDA they market. A product listed here may not be on any shelf, and a product on a shelf may have a lapsed listing. The site carries each product's listing expiration date as recorded, and does not use it to hide anything, because an expired listing is a fact about the record rather than proof the product is gone.

Labeller is not manufacturer. The labeller is the firm whose name appears on the package and which holds the listing. It is frequently not the company that made the product. The site reports the field FDA holds and does not speculate beyond it.

Ingredient names are FDA's, not tidied. Where the directory records menthol, levomenthol and menthol in unspecified form as separate substances, they appear separately here. They carry different substance identifiers in FDA's own records, so merging them would be an editorial judgment presented as a fact.

Prices are not held. This site has no price data and shows none. Any statement about what something costs would be invented.

The data is a snapshot. Every figure on the site is from the build dated 4 September 2026. FDA revises the directory continuously. The build date is printed in the footer of every page for that reason.