Explore openFDA
One search box across four live FDA drug datasets. Pick a source, type a drug, firm or ingredient, and page through the real records. Every number on this screen comes from the API response you just made — nothing is estimated.
Informational only — this is not medical advice. Pharma Watch is a read-only viewer for public FDA data published through openFDA. Records are submitted by manufacturers and reporters, are not verified by the FDA in all cases, and an adverse-event report does not prove a drug caused the event. Never start, stop or change a medicine based on this page. Talk to your prescriber or pharmacist, and check the official FDA drug safety pages.
No query run yet
Recall command centre
Every drug recall record in openFDA's enforcement dataset, filterable by hazard class, status, date and recalling firm. Counts are read straight from meta.results.total on the API response.
Filters
No query run yet
What the classes mean
Definitions: FDA 21 CFR 7.3(m)
Class I
Reasonable probability that use of, or exposure to, the product will cause serious adverse health consequences or death.
Class II
Use may cause temporary or medically reversible adverse health consequences, or the probability of serious consequences is remote.
Class III
Use is not likely to cause adverse health consequences — typically a labelling or manufacturing-standard issue.
Label lookup
The manufacturer's own FDA-submitted labelling (SPL). Boxed warnings — the FDA's strongest warning — are shown first. This is the label text as filed, not a summary and not advice.
No query run yet
Safety signals
Aggregate counts built with openFDA's count= parameter. These are report counts, not rates — there is no denominator of how many people took the drug, and reporting is voluntary and biased. A tall bar means "reported often", never "caused often".
No query run yet
No profile built yet
Enter a drug above to pull its most-reported reactions, seriousness split and recall history straight from openFDA.
Compare two drugs
Side-by-side record counts and top reported reactions. Both columns run identical queries, so the shapes are comparable — but the counts still reflect reporting behaviour, not risk.
No query run yet
Nothing to compare yet
Enter two drug names above. Pharma Watch will run the same four openFDA queries against each and line the answers up.
Watchlist
Drugs and firms you follow. Pharma Watch remembers the newest recall report date it has shown you for each one, then re-queries openFDA on load and tells you exactly how many recall records are newer than that.
Free and Student Boost
Workbench
Query building, cohorts, cross-tabs, firm
profiles, reports and the workspace as a file. Everything on
the other six pages stays free — this is the bench you use
when one search is not the whole question.
Build a query
Add a term to see the openFDA expression it builds.
Write it yourself
Nested groups, ranges, _exists_ and
_missing_ — the questions the filter row cannot ask.
The exact URL is written to the query log either way.
Saved searches
Cohorts
Compare a cohort
The timeline, your window
Firm profile
Reactions against seriousness
NDC directory lookup
Two labels, section by section
The query log
The field dictionary
What was not published
Build a report
The watchlist, in depth
The workspace as a file
About this data
What Pharma Watch shows you, where each number comes from, and — just as important — what it cannot tell you.
Not medical advice. Pharma Watch is an informational viewer over public U.S. FDA data. It does not diagnose, treat or recommend. Only your prescriber or pharmacist knows your history. If you think you are having a reaction to a medicine, seek care — in the U.S., report it to MedWatch.
The four datasets
How Pharma Watch handles numbers
- Every figure shown is read from a live openFDA response and stamped with the query that produced it.
- Result totals come from meta.results.total; charts come from count= aggregations.
- When openFDA returns no match (its 404 NOT_FOUND), you get an empty state — never a filled-in guess.
- When openFDA is unreachable you get an error naming the source, the URL and a Retry button.
- The demo dataset is synthetic and is labelled SAMPLE on every card, chart and export while it is loaded.
- Records served through the Feerebu proxy are marked as such, because the proxy substitutes placeholder text for a few blank fields.
Connection
Known limits
- Adverse-event reports are voluntary, duplicated and unverified. Counts are not incidence rates.
- Recall records are the firm's and FDA's description at the time of filing; status can change later.
- Labels are versioned — you may be seeing an older SPL version than the carton in your hand.
- openFDA rate-limits anonymous callers (240 requests/minute). Heavy paging can hit it.
- openFDA data covers the United States. Other regulators are not included.