The FDA have published their inspection findings consistently over the years. The patterns don’t change much – late case entries, an outdated PSMF, a signal review that was never formally documented. These issues are raising because of compounding and preventable gaps in documentation across the workflows.
Here are the five common mistakes in pharmacovigilance that come up most often, and what each one actually looks like inside a PV operation.
5 mistakes to avoid during pharmacovigilance
1. ICSR [Individual Case Safety report] submission timelines slipping
The 7-day and 15-day clocks start the moment your company receives safety information. That’s a distinction that gets teams into trouble more often than people expect.
In practice, it usually looks like this: a serious case comes in through a non-standard channel like a medical affairs email, a market research call, a customer service log. It sits for a few days before anyone recognizes it as a reportable event. By the time it reaches case processing, the clock has already been running.
Why it matters to an inspector: A single late submission is a documentation gap. A pattern of late submissions tells them your intake process has no structured triage – and that’s a systemic finding, not an isolated one.
2. MedDRA coding that isn't consistent across cases
Ask three case processors to code the same adverse event and you’ll sometimes get three different preferred terms. That’s not a training failure, it’s a process design gap. Without a coding guideline that’s actively enforced, variation is the default.
The downstream consequence is signal detection. Your analytics are only as clean as the data underneath them. If the same reaction is split across multiple PT codes, disproportionality analysis misses what’s actually building in your database.
Why it matters to an inspector: Inconsistent coding signals that your quality oversight of case processing isn’t working. And if signal detection is built on inconsistent data, the entire signal management process comes into question.
3. Signal detection that exists in practice without documentation
A lot of PV teams do the work, they review cases, they notice trends, they escalate concerns internally. But under GVP Module IX, what matters is what’s documented. If there’s no formal record of a signal being identified, evaluated, and either escalated or closed with a rationale, the activity effectively didn’t happen from a regulatory standpoint.
Inspectors will ask for evidence of signal management over the past 12–24 months. Meeting minutes, signal tracking logs, written justifications for closure decisions. A busy inbox and a few Slack threads won’t satisfy that request.
Why it matters to an inspector: No documented signals in a large database looks like no oversight and not a clean safety profile. It raises questions about whether the team is analyzing data or just filing it.
4. Source documents that don't reconcile with the database
Table of Contents
ALCOA+ principles apply directly to pharmacovigilance data. Every case in your safety database should be traceable to an original source like a healthcare professional’s report, a clinical site notification, a literature abstract. If the details don’t match, or the source document is missing entirely, that’s a data integrity observation.
This one tends to emerge during case traceback exercises. An inspector pulls a case, asks for the source, and finds the original report says something different from what’s in Argus. The discrepancy might have an explanation, but if there’s no audit trail documenting why a change was made, the explanation doesn’t hold.
Why it matters to an inspector: Data integrity findings don’t stay contained. Once an inspector has reason to question the reliability of your database, the scope of the inspection expands.
5. A PSMF that describes the system as it was, not as it is
The PSMF is always the first document an inspector requests. It’s their map, they use it to understand how your PV system is structured, who is responsible for what, and which SOPs govern which processes. When the map is outdated, the inspection gets more intensive, not less.
Common gaps: an org chart that doesn’t reflect recent team changes, an SOP list referencing versions that have since been updated, or vendor agreements that have changed but aren’t documented.
Training records are reviewed alongside the PSMF that includes staff who haven’t completed current-version training on active SOPs is among the most cited findings in EMA inspections.
Why it matters to an inspector: An inaccurate PSMF doesn’t just fail on its own, it undermines confidence in everything the document is supposed to describe.
How can AI fix these mistakes in pharmacovigilance?
Intake and triage
NLP reads incoming reports on receipt, flags serious cases immediately, and starts the regulatory clock by removing the delay between information received and triage initiated.
Coding consistency
AI-assisted coding surfaces the correct MedDRA preferred term at the point of data entry, reducing variation across case processors without adding a manual QC layer.
Continuous signal monitoring
Rather than scheduled reviews, signal detection runs against your database in real time and the outputs are documented automatically, satisfying GVP Module IX expectations.
Audit-ready trails
Every case update, change, and decision is timestamped and linked to its source. The audit trail builds itself irrespective of when there is any inspection or not.
PSMF and training tracking
Automated alerts surface SOP version mismatches and training expiry dates before they become findings, so the PSMF reflects current reality, not last year’s.
Here is how a global pharma company improves pharmacovigilance scores with real-time analytics. Saxon built a Power BI solution covering employee performance, product quality, and compliance monitoring – leading to measurably higher PV scores.
Bottom line
Each of the five red flags above is detectable before an inspector walks in. The question is whether your system surfaces them early enough to act on them.
Saxon helps pharma companies build real-time PV analytics, automated safety monitoring, and AI-powered oversight that are built for drug safety teams.
Frequently asked questions
Pharmacovigilance is the science and process of monitoring the safety of medicines after they've been approved and are in use. It covers collecting adverse event reports, detecting safety signals, assessing risk, and communicating findings to regulators and healthcare professionals. It's a legal requirement for any company holding a marketing authorization.
Fatal or life-threatening unexpected serious adverse reactions require expedited submission within 7 calendar days. All other serious unexpected reactions must be submitted within 15 calendar days. Both timelines start from the date the company first receives information — not from when the case is fully assessed.
The Pharmacovigilance System Master File (PSMF) is typically the first document requested. Inspectors use it to understand how the safety system is structured before sampling cases, checking timelines, and reviewing signal management records.
A signal is information suggesting a new or changed risk associated with a medicine that warrants further investigation. Under GVP Module IX, signal detection must be a continuous, documented process — not a periodic review. Inspectors look for evidence that signals were identified, evaluated, and formally closed or escalated with documented rationale.