MedDeviceGuideMedDeviceGuide
Back

Quality Investigation for Medical Devices: Complaint, NCR, and Audit Finding Investigation Complete Guide

How to conduct quality investigations for medical devices — complaint handling, nonconformance investigation, audit finding root cause analysis, investigation methods, CAPA integration, and regulatory requirements under ISO 13485:2016, FDA 21 CFR 820, and EU MDR.

Ran Chen
Ran Chen
Global MedTech Expert | 10× MedTech Global Access
2026-04-1716 min read

Quality Investigations Are Where Compliance Lives or Dies

A medical device company can have perfectly written SOPs, spotless production floors, and beautifully formatted quality manuals — but if it cannot investigate quality events effectively, its entire QMS is built on sand. Inadequate root cause investigation is consistently one of the top five FDA 483 observations and one of the most common reasons for Notified Body major nonconformities.

Under the FDA's Quality Management System Regulation (QMSR), effective February 2, 2026, which incorporates ISO 13485:2016 by reference, the expectations for investigation thoroughness, documentation, and corrective action effectiveness have only increased. This guide covers the complete quality investigation lifecycle — from complaint intake through root cause analysis to CAPA closure and effectiveness verification.

What Triggers a Quality Investigation?

Quality investigations in medical device manufacturing are triggered by three primary events:

1. Customer Complaints

A complaint is any written, electronic, or oral communication that alleges deficiencies in a medical device's identity, quality, durability, reliability, safety, effectiveness, or performance. Under ISO 13485:2016 Clause 8.2.2 and FDA 21 CFR 820.198, every complaint must be documented, evaluated, and investigated when warranted.

2. Nonconformances (NCRs)

A nonconformance is any deviation from a specification, standard, or requirement identified during incoming inspection, in-process manufacturing, final release, or post-market surveillance. ISO 13485:2016 Clause 8.3 requires organizations to identify, document, segregate, evaluate, and control nonconforming product.

3. Audit Findings

Internal audits (Clause 8.2.4), external audits by Notified Bodies, and FDA inspections all produce findings that require investigation. These findings may relate to product quality, process performance, documentation gaps, or systemic QMS deficiencies.

Comparison of Investigation Triggers

Aspect Customer Complaint Nonconformance (NCR) Audit Finding
Source External (customer, patient, distributor) Internal (manufacturing, QC, incoming) Internal or external (auditor)
Regulatory Basis ISO 13485 Clause 8.2.2; 21 CFR 820.198 ISO 13485 Clause 8.3; 21 CFR 820.90 ISO 13485 Clause 8.2.4; 21 CFR 820.22
Initial Priority Risk-based: patient safety first Risk-based: containment first Risk-based: major vs minor
Typical Investigation Depth Moderate to deep (depends on severity) Moderate (depends on criticality) Deep (systemic root cause focus)
Reporting Obligation May trigger MDR/vigilance reporting Usually internal Usually internal, may be externally tracked
CAPA Threshold Recurrence, trend, or high severity Systemic cause or recurrence Major findings or recurring minors

The Seven-Step Quality Investigation Process

Step 1: Event Detection and Intake

Every quality event begins with detection. The key requirements at this stage:

  • Complaints: Capture all relevant details — device identifier, lot/batch number, date of event, description of what happened, patient outcome, reporter contact information. Under FDA 21 CFR 820.198(e), even oral complaints must be documented upon receipt.
  • NCRs: Record the specific deviation — what was found, where, when, and by whom. Include the applicable specification and the actual measurement or observation.
  • Audit findings: Document the specific requirement, the observed gap, the evidence reviewed, and the finding classification (major, minor, observation).

Critical rule: Never delay intake. A complaint received by phone should be logged the same day. An NCR identified on the production line should be documented before the shift ends.

Step 2: Immediate Containment

Before any root cause analysis begins, you must prevent further harm or nonconformance:

  • Complaints: Determine if the affected device is isolated or if other devices in the same lot, batch, or product line are at risk. If product in the field is affected, evaluate the need for a field safety corrective action (FSCA).
  • NCRs: Segregate all affected product. Physically move it to a quarantine area. Lock the lot in your ERP system. If your NCR does not explicitly state that affected material is secured, an auditor will assume it might have been shipped.
  • Audit findings: Implement immediate corrective actions for any finding that poses an ongoing compliance or quality risk.

Step 3: Impact Assessment

This is the most frequently missed step in quality investigations. Finding a problem with one unit is not enough — you must answer:

  • Does this affect other units in the same lot?
  • Does this affect previous lots?
  • Does this affect product already in the field?
  • Does this affect other product lines using the same process, material, or supplier?
  • Does this affect product registered in other markets?

If your investigation record says "disposition: scrap" without answering "what else is at risk?" you have an audit finding waiting to happen.

Step 4: Root Cause Investigation

This is the core of the quality investigation. The goal is to identify the true root cause — not the symptom, not the most convenient explanation, but the fundamental reason the event occurred.

Regulatory expectation: ISO 13485:2016 Clause 8.5.2 requires organizations to identify root causes and implement corrective actions. FDA's QSIT inspection technique (now replaced by CP 7382.850 under QMSR) specifically directs investigators to evaluate whether failure investigations determine root cause and whether the degree of investigation is commensurate with the significance and risk of the nonconformity.

Root Cause Analysis Methods

Method Best For Complexity Key Strength
5 Whys Simple, single-cause events Low Quick, easy to facilitate
Fishbone (Ishikawa) Diagram Multi-factor investigations Medium Organized categorization of causes
Fault Tree Analysis Complex systems, safety-critical events High Quantitative, systematic
8D Problem Solving Customer-facing or supply chain events Medium-High Structured team approach
Pareto Analysis Trending data, prioritizing multiple issues Low-Medium Data-driven focus on vital few
FMEA Review Process or design-related nonconformances Medium Links to existing risk analysis
5W1H (What/Where/When/Who/Why/How) Initial problem definition and containment Low Ensures complete event description
Kepner-Tregoe Complex, multi-variable problems High Systematic, evidence-based

Choosing the right method: The complexity of the investigation method should match the complexity and risk of the event. Using 5 Whys for a Class III device recall is insufficient. Using fault tree analysis for a mislabeled carton is overkill. Match the tool to the problem.

Common Pitfalls in Root Cause Analysis

  1. Stopping at the first plausible cause: The most common error. The first answer to "why" is usually a symptom, not the root cause. Continue investigating until you reach a cause that is actionable and preventable.
  2. Assigning root cause to "human error": Human error is almost never the root cause — it is a symptom of a system that allowed or caused the error. Ask why the human made the error. Was training inadequate? Was the procedure unclear? Was the environment conducive to mistakes?
  3. Not involving cross-functional expertise: Root cause investigation requires representatives from all relevant functions — engineering, quality, manufacturing, regulatory, and sometimes sales or clinical. Quality teams investigating alone miss process and design context.
  4. Confusing correlation with causation: Two events occurring together does not mean one caused the other. Verify your root cause with evidence.
  5. Skipping verification: After identifying a root cause and implementing a corrective action, you must verify that the action actually eliminated the cause. Without verification, the investigation is incomplete.

Step 5: Corrective Action Development

Once the root cause is identified, develop corrective actions that:

  • Directly address the root cause, not the symptom
  • Are specific and actionable — not "improve training" but "revise SOP-X004 to include step-by-step instructions for torque verification, train all assembly operators by [date], and add torque verification to in-process inspection checklist"
  • Include a timeline and responsible party
  • Consider whether the corrective action could introduce new risks — update the risk management file if needed

Step 6: CAPA Integration

Not every quality event requires a CAPA. CAPA is the heavy artillery — it should be reserved for systemic issues, high-risk events, and recurring problems. Opening a CAPA for every NCR leads to "death by CAPA," where the quality team is investigating hundreds of root causes simultaneously and rushing closures.

CAPA trigger criteria (recommended):

  • Recurrence: The same or similar event has occurred 3+ times in a defined period
  • Severity: The event involved patient harm, risk of harm, or a significant regulatory noncompliance
  • Systemic cause: The root cause affects a process, system, or product family, not just a single unit
  • Audit requirement: A regulatory body or Notified Body explicitly required a CAPA
  • Management review decision: Leadership determines a CAPA is warranted based on trend data

Step 7: Effectiveness Verification

The investigation is not complete until you have verified that the corrective action actually worked. This means:

  • Define an effectiveness check metric before closing the CAPA
  • Wait an appropriate period (30, 60, or 90 days depending on the event)
  • Check that the metric has improved and the event has not recurred
  • Document the effectiveness check results

Common audit finding: CAPAs closed without effectiveness verification, or effectiveness checks that are superficial ("training was completed" rather than "operators demonstrated competency and no recurrence in 90 days").

Recommended Reading
Quality KPIs and Metrics for Medical Devices: Complete Framework + ISO 13485 & QMSR 2026 Alignment
Quality Systems ISO 134852026-04-17 · 14 min read

Investigation Documentation Requirements

What Must Be in Every Investigation Record

Element Requirement Common Gap
Event description Clear, factual, complete Vague or incomplete descriptions
Containment actions What was done to prevent further impact No evidence of segregation or quarantine
Impact assessment Scope of affected product/process Not extended to other lots or products
Root cause analysis Method used, data reviewed, team involved "Human error" without further analysis
Corrective actions Specific actions with owners and dates Generic actions ("retrain") without specifics
Effectiveness verification Metric, timeframe, result Missing or superficial
Link to risk management Risk file update if needed Not considered
Regulatory reporting MDR, vigilance, or other reporting assessed Not evaluated

Who Should Investigate?

Investigations should involve a cross-functional team:

  • Quality Assurance: Investigation lead, documentation, compliance assessment
  • Engineering/Design: Technical analysis, design input/output review
  • Manufacturing/Operations: Process data review, equipment assessment
  • Regulatory Affairs: Reporting obligations, regulatory impact assessment
  • Clinical/Medical (for complaint investigations): Clinical significance evaluation

The investigation lead should be trained in root cause analysis methods and should not be the person who caused or discovered the event (independence requirement).

Complaint Investigation Specifics

FDA 21 CFR 820.198 Requirements

FDA requires that medical device companies establish a formally designated unit to manage complaint investigations. Key requirements:

  • All complaints must be processed in a timely manner
  • Oral complaints must be documented upon receipt
  • When an investigation is deemed unnecessary, the reason must be documented along with the name of the person making that determination
  • Investigation records must include: device name, any reported similar complaints, investigation findings, and any corrective action taken
  • Complaint records must be accessible at the manufacturing establishment, even if processed off-site

ISO 13485:2016 Clause 8.2.2 Requirements

ISO 13485 goes beyond basic complaint logging:

  • Evaluate the need for regulatory reporting (vigilance, MDR)
  • Determine the need for CAPA
  • Assess the need for nonconforming product controls
  • Feed complaint data into risk management and post-market surveillance

EU MDR Vigilance Triggers

Under EU MDR Article 87, manufacturers must report any serious incident involving a device on the EU market. A "serious incident" is defined as any event that directly or indirectly led to, or could have led to, death, serious deterioration in health, or a serious public health threat. The investigation timeline is strict:

Event Type Reporting Deadline
Serious public health threat Immediately, no later than 2 calendar days
Death or unanticipated serious deterioration No later than 10 calendar days
Other serious incidents No later than 15 calendar days

NCR Investigation Specifics

The NCR Investigation Arc

When an auditor reviews your NCR log, they are reading a story — the narrative arc of the failure:

  1. What happened? (detection and documentation)
  2. What did you do immediately? (containment)
  3. How far could this go? (impact assessment)
  4. Why did it happen? (root cause analysis)
  5. What did you fix? (corrective action)
  6. Did the fix work? (effectiveness verification)

If any chapter of this story is missing, the auditor has a finding.

Disposition Decisions

For nonconforming product, the investigation must lead to a disposition decision:

Disposition When Appropriate Documentation Required
Scrap Product cannot meet specs, rework not feasible Scrap authorization, quantity, lot traceability
Rework Product can be brought into compliance Rework instructions, re-inspection results
Use As Is Deviation does not affect safety or performance Technical justification, risk assessment, authorized approval
Return to Supplier Nonconformance due to supplier issue Supplier NCR, supplier corrective action request
Recommended Reading
Cost of Quality (CoQ) in Medical Devices: Complete Framework — Prevention, Appraisal, Internal & External Failure Costs
Quality Systems ISO 134852026-04-17 · 14 min read

Audit Finding Investigation Specifics

Classification and Response

Finding Type Definition Investigation Timeline
Major Nonconformity Total breakdown of a QMS element; risk to product safety or compliance Immediate action, CAPA within 30 days typically
Minor Nonconformity Isolated lapse in a QMS element; not systemic Corrective action within 60–90 days typically
Observation / Opportunity for Improvement Area where improvement is recommended but no nonconformity exists Address at next management review

The Escalation Rule

An audit finding that recurs across multiple audits should be escalated in severity. A minor finding that appears in three consecutive audits is a systemic problem, not an isolated lapse. ISO 13485 Clause 8.5.2 explicitly requires analysis of recurrence patterns.

Investigation in the QMSR Era

The QMSR, effective February 2, 2026, replaced the Quality System Regulation (QSR) with ISO 13485:2016 incorporated by reference. Key changes affecting investigations:

  1. ISO 13485 Clause 8.2.2 (Feedback and Complaints): Explicitly requires evaluating the need for regulatory reporting and feeding complaint data into risk management.
  2. ISO 13485 Clause 8.3 (Control of Nonconforming Product): Requires documented procedures for handling nonconforming product, including evaluation of the need for action commensurate with risk.
  3. ISO 13485 Clause 8.5.2 (Corrective Action): Requires reviewing the effectiveness of corrective actions taken — this is not optional.
  4. CP 7382.850 (New FDA Inspection Program): Replaces the QSIT guide with a new inspection approach aligned to ISO 13485. Investigators will evaluate investigation processes against the standard's requirements.

Comparison: Manual vs. eQMS-Based Investigation Workflow

Aspect Paper/Email Investigation eQMS-Based Investigation
Intake Speed Days (email routing, manual logging) Hours (automated intake forms)
Containment Documentation Manual quarantine logs System-locked lots, automatic holds
Investigation Tracking Spreadsheets, shared folders Workflow-driven with role assignments
Root Cause Documentation Word documents, attachments Structured forms with built-in methods
CAPA Linking Manual cross-references Automatic parent-child linking
Effectiveness Verification Calendar reminders, manual follow-up Automated tasks, overdue alerts
Audit Trail Partial (depends on email retention) Complete, Part 11 compliant
Trend Analysis Manual pivot tables Built-in analytics and dashboards
Recommended For Very small companies (< 10 complaints/year) Any company with > 10 quality events/year
Recommended Reading
Root Cause Analysis for Medical Devices: RCA Methods, Tools, and CAPA Integration
Quality Systems ISO 134852026-04-17 · 15 min read

FAQ

When should we open a CAPA versus just closing the investigation? Open a CAPA when the root cause is systemic (affects a process or product family, not just one unit), when the event has recurred, when severity warrants it (patient harm or regulatory risk), or when a regulator explicitly requires one. If the event is isolated, non-recurring, and low severity, close it at the investigation level with a documented rationale.

How long should an investigation take? There is no regulatory maximum, but best practice is: high-severity events (patient harm, regulatory impact) within 30 days; medium-severity within 60 days; low-severity within 90 days. CAPA effectiveness checks should follow 30–90 days after implementation.

Can we use "human error" as a root cause? Almost never. Human error is a symptom. Ask why the human made the error. Was the procedure unclear? Was training inadequate? Was the work environment contributing to fatigue? Was the equipment design conducive to mistakes? The root cause must be something you can act on systemically.

Do all complaints need formal investigation? All complaints must be evaluated. Not all need formal investigation. If a complaint is clearly not product-related (e.g., shipping damage that is the carrier's responsibility) or is so minor that it poses no quality or safety risk, you can close it without investigation — but you must document the rationale for not investigating and the name of the person who made that determination.

How do we handle anonymous complaints? Anonymous complaints must be treated with the same rigor as identified complaints. You may have limited ability to gather additional information, but you must still evaluate the allegation, check device history records, and determine if similar events have been reported.

What if we cannot determine the root cause? Sometimes a definitive root cause cannot be established despite thorough investigation. In these cases, document the extent of your investigation, the methods used, the data reviewed, and the potential contributing factors identified. Then implement corrective actions addressing the most probable causes and monitor for recurrence more closely.

How does the QMSR change investigation requirements? The QMSR replaces QSR with ISO 13485:2016, which is more explicit about requiring effectiveness verification of corrective actions (Clause 8.5.2.e), integrating complaint data into risk management (Clause 8.2.2), and ensuring investigation depth is commensurate with risk (Clause 8.3). The new inspection program (CP 7382.850) aligns FDA investigators with these ISO requirements.

What records must be retained? FDA 21 CFR 820.198(e) requires complaint investigation records to include the device name, complaint nature, investigation findings, and corrective actions. ISO 13485 Clause 8.2.2 requires records of complaint evaluation and investigation results. Under both frameworks, records must be retained for the device's expected lifetime or as required by applicable regulations (minimum 2 years under FDA, typically longer under EU MDR).