Back to the journal
RegDataLab Team

Recall → Facility Trace

6 min read

When a recall is announced, one of the first questions is: which facility produced the affected product? FDA doesn't always publish this directly in the enforcement record. fda_recall_facility_trace bridges this gap by cross-referencing the recalling firm against facility registrations and returning candidate facilities with confidence scores.

MCP tool calls — your AI assistant calls these automatically. You can also invoke them via the MCP protocol.

Running a trace

fda_recall_facility_trace({ "recall_number": "D-0591-2025" })

Sample response:

{
  "recall": {
    "recall_number": "D-0591-2025",
    "classification": "Class II",
    "recalling_firm": "Pfizer Inc",
    "product_description": "Eliquis (apixaban) Tablets, 5 mg, ..."
  },
  "candidate_facilities": [
    {
      "fei": "3002807852",
      "firm_name": "Pfizer Inc",
      "city": "New York",
      "state": "NY",
      "confidence": "high",
      "match_method": "exact_firm_name"
    },
    {
      "fei": "3004336298",
      "firm_name": "Pfizer Manufacturing Belgium NV",
      "city": "Puurs",
      "country": "BE",
      "confidence": "medium",
      "match_method": "alias"
    }
  ],
  "caveats": [
    "Facility matching is based on firm name similarity — not confirmed by FDA linkage.",
    "Multiple facilities may be candidates; confidence indicates match strength, not certainty."
  ]
}

Interpreting confidence levels

The confidence field indicates how strong the name match is between the recalling firm and the facility record:

  • high — exact or near-exact name match. Likely the right facility, but not FDA-confirmed.
  • medium — matched via company alias or fuzzy similarity. Plausible but warrants verification.
  • low — weak match. Treat as a lead, not a conclusion.

Each candidate includes its match type, confidence and a matching explanation, so you can show end users the evidence behind every candidate.

What to do with the results

  1. Take the high-confidence FEIs and call fda_get_facility to get full facility details and risk summaries.
  2. For medium-confidence matches, check whether the facility's product list (via fda_facility_products) aligns with the recalled product.
  3. Store the recall number + FEI + confidence together for audit trails.
  4. Use the candidate FEIs and matching explanations to investigate the recalling firm’s facilities.

When the trace finds nothing

For recalls issued by distributors or importers, start from the recalling firm’s company family and use each candidate’s matching explanation to find the manufacturing site.

All entries