FDA Recall Analytics helps you explore historical FDA drug recall reports, identify recurring product-quality problems and read the FDA description behind each result. The quality categories and GMP/PQS associations shown here are GMPNews analytical assessments, not official FDA classifications.
Start here: Choose a product area and quality issue, or click a bar in a chart. A second click on the same bar removes that filter. Select a product in the table to read the original FDA reason, supporting evidence and linked recall entries.
1. What data are included?
The dashboard covers FDA Drug Enforcement Reports with report dates from 1 January 2020 onward, limited to the information in the loaded official FDA archive. The last reporting year is partial. A recall may have started before its FDA report date. Each published update is a dated source snapshot, not a live status-tracking service. Automated refreshes depend on successful FDA ingestion and quality checks.
The number of FDA recall entries and the counts for each product group are calculated from the official source archive loaded into the builder. Only Finished Medicines, Compounded Medicines and APIs are displayed. Other records remain in the original source archive, not the public denominator.
2. What do the numbers mean?
| On-screen number | What it means |
|---|
| FDA Recall Entries | The number of distinct FDA recall entries matching your filters. One entry can cover several batches or package presentations; it is not a count of batches. |
| Recall Events | The number of separate FDA recall events among those entries. One event can include several recalled products or entries. |
| Companies Involved | Distinct recalling-company names reported by FDA. The recalling company is not necessarily the manufacturing company; names are not consolidated across affiliates. |
| Entries with Reported Findings | Entries for which FDA reports a specific quality finding in the relevant product or material. This does not imply that every listed batch or unit was separately tested. |
Example: If you see 584 FDA recall entries and 298 recall events, this means the selected group contains 584 individual FDA entries belonging to 298 events. These are two different ways of counting the same selected dataset; neither means 584 batches or 298 formal administrative decisions.
3. How do filters and charts work?
Product Area: Finished Medicines, Compounded Medicines and APIs are shown separately. All in scope combines these three groups. Technical review categories do not appear in public filters.
Quality Issue: The reported quality problem, such as a chemical impurity, potency problem, labelling issue or lack of sterility assurance. Some recalls have several independently reported concerns.
Quality Issue Evidence: Separates findings reported in the recalled product from possible or related risks, complaints and general GMP/regulatory issues. A risk-related recall remains searchable without being described as a positive test for every affected product.
GMP/PQS Area: An optional filter showing only areas linked to explicit source-described nonconformities in the current data. It does not list every possible GMP topic, and it does not suggest that every recall has a proven root cause in a particular quality-system process.
Other optional filters: Chemical subtypes (including nitrosamines), reported mechanisms such as cross-contamination, and the context in which a problem was identified (for example, stability testing).
Options show the number of matching FDA recall entries in brackets, recalculated using the other active filters. Quality topics are ordered from most frequent to least frequent; companies are ordered alphabetically. A selected option remains available so you can clear it. Year and month options are listed chronologically (2020 onward; January through December within each year), independent of their counts. Other issue filters show the most common categories first.
Within a filter, you can select more than one option. Different filters narrow the selection together. Clicking a chart element applies its filter; clicking that selected element again removes it. Clear filters returns to the default Finished Medicines view.
4. Why might a product appear in several issue groups?
FDA may report multiple findings in one recall. For example, a hand sanitizer could contain methanol and also have ethanol below its stated strength. GMPNews preserves both problem types, rather than reducing them to one broad label. In addition, a result obtained for one product does not automatically mean that all products recalled from the same site had the same test outcome.
5. FDA facts versus GMPNews interpretation
Classification methods: Most historical entries were classified automatically using source-text rules. A limited pilot used an AI model; those records are identified separately in their details. Automatic classifications are preliminary and may contain errors. The original FDA fields remain unaltered.
FDA supplies: recall identifier, product and company information, the original reason, FDA risk class and dated records.
GMPNews adds: normalized quality-issue labels, evidence scopes, related GMP/PQS topics, interpretation summaries and recorded limitations. These are analytical proposals, not FDA endorsements or proven manufacturing root causes.
Detection setting is not the defect. For example, low enzyme activity is the product-quality issue even if detected during a stability study. Cross-contamination is kept separate from a claim that cleaning validation failed unless the source specifically establishes that failure.
Evidence can be incomplete. In such cases, the dashboard should identify the known concern while explaining precisely what remains uncertain: the tested material, the exact substance, the affected batches or the underlying root cause.
6. Opening the original FDA source
Click a product in the table for its detail panel. The FDA source tab reproduces the archived FDA wording for the reason, description, batches, distribution and recall dates. The Search FDA Enforcement Reports link opens the official human-readable FDA search page; use Copy Recall Number to search for the exact entry.
The dashboard does not offer raw JSON downloads or JSON exports. You can export the filtered table as CSV. The complete original structured data are retained separately in the internal regulatory archive for reproducibility, source checks and future reclassification.
7. FDA class and historical recall status
FDA Class I, II and III are the FDA's own recall risk classes, not GMPNews severity assessments. Historical recall-status values in the details reflect the archived data and should not be interpreted as confirmation that an old recall is currently ongoing. The public dashboard therefore does not offer a filter for current recall status.
8. What the charts cannot establish
Counts describe reported FDA recall entries and events; they do not measure failure rates in medicines sold or manufactured. Differences among companies may reflect product portfolios, reporting structures and the number of entries in a single event. They are not quality rankings. No country-of-manufacture comparison is provided because the available source information is too incomplete for meaningful comparisons.
9. Data currency and limitations
The source snapshot is taken from the official FDA bulk archive selected by the operator. The available reporting period ends on the most recently loaded report date and does not imply live updating. Retained 2020 analyses are combined with conservative source-rule topic assignments for later years. These GMPNews assignments are preliminary analytical labels, not official FDA classifications or independently validated scientific conclusions. Check original FDA evidence for critical decisions.