MDR Article 88 Trend Reporting: Statistical Thresholds and Workflow
How to set MDR Article 88 trend reporting thresholds, choose denominators, test significance, escalate severity, link CAPA, and align PMSR/PSUR outputs.
What This Article Covers / Does Not Cover
This article covers one specific regulatory obligation: how to design and implement the statistical methodology for trend reporting under MDR Article 88 (and IVDR Article 83). It addresses denominator selection, expected frequency baseline calculation, statistical significance testing methods, severity escalation rules, complaint-to-trend coding, CAPA/PSUR/PMSR linkage, the new MDCG 2025 draft Q&A requirements, and the Manufacturer Trend Report (MTR) form.
This article does not cover the general post-market surveillance framework, PSUR writing, clinical evaluation, or FSCA reporting. For the PMS overview, see Post-Market Surveillance Plan. For PSUR requirements, see PSUR Periodic Safety Update Report Guide. For risk management fundamentals, see Benefit-Risk Analysis.
Article 88 Requirements: The Regulatory Text
MDR Article 88 requires manufacturers to report "any statistically significant increase in the frequency or severity of incidents that are not serious incidents or that are expected undesirable side-effects that could have a significant impact on the benefit-risk analysis and which have led or may lead to risks to the health or safety of patients, users or other persons that are unacceptable when weighed against the intended benefits."
Three conditions must all be met for a trend to be reportable:
Statistically significant increase in frequency or severity of non-serious incidents or expected side effects
Impact on benefit-risk analysis — the trend could significantly affect the benefit-risk determination
Unacceptable risk — the trend has led or could lead to risks that are unacceptable relative to intended benefits
The manufacturer must specify the methodology, statistical techniques, threshold values, and observation periods in the PMS plan (Article 84).
MDCG 2025 Draft Q&A on Trend Reporting
In September 2025, the MDCG released a draft Q&A document (MDCG 2025-X) clarifying trend reporting obligations. Key clarifications:
| Topic | MDCG Clarification |
|---|---|
| Scope of devices | Applies to MDR devices, IVDR devices, legacy devices, and Annex XVI products |
| Geographic scope | Events worldwide must be considered; trend reports should be submitted to CAs where related events occurred |
| Reporting method | Until EUDAMED vigilance module is functional, submit MTR via national vigilance systems |
| Event types | Statistically significant increases in: non-serious incidents, expected undesirable side effects (MDR), expected erroneous results (IVDR) |
| Integration | Trend detection must be part of PMS plan, integrated with QMS and risk management |
| Statistical approach | Manufacturers must define methods, thresholds, and observation periods in PMS plan; justify appropriateness |
| Unexpected events | If an unexpected incident type is detected, it may trigger a serious incident report (MIR) rather than a trend report |
| Custom-made devices | Trend reporting obligations apply to custom-made devices as well |
Step 1: Define the Data Collection Framework
Incident Categories for Trend Monitoring
| Category | MDR Reference | Examples | Data Source |
|---|---|---|---|
| Non-serious incidents | Art. 88 | Malfunctions not causing serious injury, minor use errors, cosmetic defects | Complaint system, service reports |
| Expected undesirable side effects | Art. 88 | Known local reactions, expected transient symptoms, known minor adverse events | Clinical feedback, literature, PMCF |
| Expected erroneous results (IVDR) | Art. 83 IVDR | Known false-positive/negative rates within expected range but trending up | QC data, proficiency testing, complaint data |
Complaint Coding for Trend Detection
Use a structured coding system that maps complaints to trend-monitorable categories:
| Code Level 1 | Code Level 2 | Code Level 3 | Trend-Relevant? |
|---|---|---|---|
| Device performance | Output degradation | [Specific failure mode] | Yes — frequency trending |
| Device performance | Intermittent malfunction | [Specific failure mode] | Yes — frequency trending |
| Use error | Patient-initiated | [Specific error type] | Yes — frequency trending |
| Biocompatibility | Local reaction | [Specific reaction type] | Yes — severity trending |
| Software | Display error | [Specific error type] | Yes — frequency trending |
| Serious incident | [Any] | [Any] | No — report via MIR, not trend report |
Step 2: Select the Denominator
The denominator defines the population at risk. Choosing the right denominator is critical — it determines whether an increase in complaints reflects a genuine device issue or simply increased usage.
Denominator Selection Table
| Denominator Type | When to Use | Data Source | Limitations |
|---|---|---|---|
| Units sold/distributed | Devices with one-time use or defined replacement cycle | Sales records, distribution logs | Does not account for actual usage intensity |
| Units in active use | Reusable devices, implanted devices | Install base tracking, registration data | Harder to obtain; requires active tracking |
| Procedure count | Devices used per procedure (e.g., surgical instruments) | Hospital usage reports, procedure logs | Requires customer cooperation |
| Patient-days / patient-exposures | Devices with continuous use (e.g., infusion pumps, monitors) | Usage logs, prescription data | Data may be incomplete |
| Test runs / analyses performed | IVD analyzers, laboratory equipment | Instrument telemetry, QC logs | Requires instrument connectivity |
Rule: The denominator must be the same one used in the risk management file to establish expected incident rates. If your risk file uses "per 10,000 units sold" as the rate basis, your trend denominator must match.
Step 3: Establish Expected Frequency and Baseline
Baseline Calculation Methods
| Method | Description | When to Use |
|---|---|---|
| Historical data | Calculate average incident rate over a defined lookback period (e.g., previous 12-24 months) | Established devices with sufficient history |
| Pre-launch estimates | Use rates from clinical investigation, bench testing, or predicate device data | New devices without field history |
| Literature-based | Use published rates for similar device types | Novel devices with no direct predicate |
| Risk file estimates | Use the expected frequency from the risk management file (ISO 14971) | Always — this is the regulatory anchor |
Expected Frequency Table (Example)
| Incident Type | Risk File Expected Rate | Historical Rate (12 mo.) | Observation Period | Denominator |
|---|---|---|---|---|
| Sensor drift outside spec | 1 per 5,000 units | 0.8 per 5,000 units | Quarterly | Units sold |
| Skin irritation at adhesive site | 2 per 1,000 patients | 2.3 per 1,000 patients | Quarterly | Patient-exposures |
| False-positive alert | 3% of tests | 2.8% of tests | Monthly | Test runs |
| Battery early depletion | 0.5 per 10,000 units | 0.4 per 10,000 units | Quarterly | Units in active use |
Step 4: Choose Statistical Methods and Set Thresholds
MDR Article 88 requires manufacturers to specify the methodology for determining statistically significant increases. The MDCG draft Q&A reinforces that methods and thresholds must be pre-defined in the PMS plan, not determined ad hoc when a trend is suspected.
Statistical Method Selection Table
| Method | Best For | Assumptions | Pros | Cons |
|---|---|---|---|---|
| Poisson rate test | Low-frequency events (count data) | Events independent, constant rate | Simple, appropriate for rare events | Assumes constant rate (may not hold) |
| Chi-squared test | Comparing observed vs. expected proportions | Sufficient sample size (>5 expected per cell) | Familiar, widely accepted | Requires adequate counts |
| Control chart (p-chart or u-chart) | Ongoing monitoring of rates over time | Sequential data collection | Visual, detects shifts quickly | Requires baseline period; sensitive to denominator changes |
| CUSUM | Detecting small, sustained shifts early | Sequential data, known baseline | Very sensitive to small persistent shifts | More complex to set up and explain |
| Moving average comparison | Smoothed trend detection | Sufficient data window | Reduces noise from random variation | Lag in detection; requires window size justification |
| Proportion test (binomial) | Percentage-based rates (e.g., false-positive rate) | Independent Bernoulli trials | Directly applicable to rate-based metrics | Requires binary outcome data |
Threshold Setting Decision Framework
| Threshold Component | What to Define | Example |
|---|---|---|
| Statistical significance level | Alpha value (probability of false positive) | α = 0.05 (standard) or α = 0.01 (conservative) |
| Minimum observation window | Time period for rate calculation | 3 months for high-volume devices; 6 months for low-volume |
| Minimum event count | Minimum complaints before testing | ≥5 events before statistical testing is applied |
| Magnitude threshold | Minimum fold-increase to trigger review | ≥2× expected rate before formal testing |
| Severity escalation trigger | When severity of events increases even if frequency does not | Any event at Severity Level ≥3 (on a 1-5 scale) where risk file estimated ≤2 |
Threshold Definition Template
For [Device Group / Product Family]:
- Expected incident rate for [Incident Type]: [X] per [denominator unit]
- Statistical method: [Method name]
- Significance level: α = [0.05 / 0.01]
- Observation period: [N months]
- Minimum event count for testing: [N events]
- Trend trigger: observed rate exceeds expected rate with p < α AND magnitude ≥ [threshold]
- Severity trigger: any incident at severity level ≥ [N] where risk file estimated ≤ [N-1]
- Review authority: [Role/Function] reviews within [N business days] of triggerStep 5: Trend Detection and Escalation Workflow
START: PMS data review cycle (monthly/quarterly)
│
├── Collect incident data for observation period
│ ├── Code incidents per complaint coding scheme
│ ├── Calculate observed rate per denominator
│ └── Compare to expected rate from risk file / baseline
│
├── Is observed rate ≥ magnitude threshold?
│ ├── NO → Document review; no trend detected; file in PMS records
│ └── YES → Apply statistical significance test
│ ├── Is result statistically significant (p < α)?
│ │ ├── NO → Document review; elevated monitoring; increase review frequency
│ │ └── YES → Is there impact on benefit-risk analysis?
│ │ ├── NO → Document rationale; continue monitoring
│ │ └── YES → Does the trend present unacceptable risk?
│ │ ├── NO → Update risk file; consider labeling update; document rationale
│ │ └── YES → INITIATE TREND REPORT
│ │ ├── Complete Manufacturer Trend Report (MTR) form
│ │ ├── Attach supporting Trend Report document
│ │ ├── Submit to Competent Authorities where events occurred
│ │ ├── Evaluate need for CAPA / FSCA
│ │ └── Update PSUR / PMSR with trend findings
│ │
│ └── ALSO CHECK: Severity escalation
│ ├── Any incident severity exceeds risk file estimate?
│ │ ├── NO → Continue frequency-based analysis
│ │ └── YES → Evaluate severity trend separately
│ │ ├── Use severity distribution comparison (non-statistical)
│ │ ├── Escalate to risk management review
│ │ └── If risk is unacceptable → INITIATE TREND REPORTStep 6: Link Trend Reporting to Other Regulatory Outputs
Trend-to-Output Traceability Table
| Trend Finding | Linked Output | What to Update |
|---|---|---|
| Statistically significant frequency increase | PSUR (periodic safety update report) | Include trend analysis in PSUR section on incident trends; describe actions taken |
| Severity escalation detected | Risk management file (ISO 14971) | Update severity estimate; reassess risk acceptability; update RMP |
| Trend triggers CAPA | CAPA record | Open CAPA with trend as input; link to MTR reference number |
| Trend indicates field risk | FSCA evaluation | Evaluate whether trend warrants Field Safety Corrective Action; document decision |
| Trend affects CER claims | Clinical Evaluation Report | Update CER with new safety data; reassess clinical benefit claims |
| Trend related to PMCF findings | PMCF report | Integrate trend data into PMCF evaluation; adjust PMCF plan if needed |
| Trend changes risk-benefit balance | SSCP (if applicable) | Update SSCP residual risk section and benefit-risk summary |
| Trend report submitted | PMS plan | Review and update PMS plan thresholds and methods if needed |
PMSR/PSUR Integration
| PSUR Section | Trend Content to Include |
|---|---|
| Summary of serious incidents | Reference trend reports submitted during reporting period |
| Analysis of non-serious incidents | Include statistical trend analysis tables and charts |
| Benefit-risk analysis update | Describe how trend findings affected benefit-risk determination |
| PMCF findings | Cross-reference any PMCF data contributing to trend detection |
| Corrective actions | List CAPAs opened in response to trend findings |
| Conclusions | State whether trend reporting led to changes in risk management, labeling, or device design |
Step 7: Complete the Manufacturer Trend Report (MTR)
The MDCG released the draft Manufacturer Trend Report (MTR) form v1.0 in September 2025. Until EUDAMED's vigilance module is fully functional, manufacturers must submit the MTR via national vigilance systems per MDCG 2021-1 (MDR) and MDCG 2022-12 (IVDR).
MTR Content Checklist
| MTR Section | Content Required | Evidence Source |
|---|---|---|
| Device identification | Basic UDI-DI, UDI-DI, device name, risk class, manufacturer SRN | EUDAMED data |
| Trend description | Nature of the trend, incident types affected, observation period | Complaint data, PMS records |
| Statistical methodology | Method used, expected rate, observed rate, significance level, p-value | PMS plan, statistical analysis |
| Baseline justification | How expected frequency was established | Risk management file, historical data |
| Benefit-risk impact | How the trend affects the benefit-risk analysis | Risk management file, clinical evaluation |
| Actions taken/planned | Corrective actions, additional monitoring, labeling updates, FSCA consideration | CAPA system, change control |
| Geographical scope | EU member states where events occurred | Complaint data, distribution records |
| Supporting documentation | Statistical analysis output, complaint data summary, risk assessment | Attachments |
Common Failure Modes and Remediation
| Failure Mode | Root Cause | Remediation |
|---|---|---|
| No statistical methodology defined in PMS plan | PMS plan is generic; no device-specific thresholds or methods | Update PMS plan with device-specific trend methodology, thresholds, and observation periods before audit |
| Using inappropriate statistical test | Chi-squared test used with very low expected counts (<5) | Use Fisher's exact test or Poisson rate test for low-frequency events |
| Denominator mismatch | Denominator for trend calculation differs from risk file denominator | Align trend monitoring denominator with risk management file; document any deviation with justification |
| Severity escalation missed | Only frequency monitored; severity increases not detected | Add severity distribution monitoring alongside frequency monitoring; define severity escalation triggers |
| Over-reporting | Thresholds set too low; every minor fluctuation triggers a report | Calibrate thresholds using historical data; use magnitude thresholds in addition to statistical significance |
| Under-reporting | Thresholds set too high or observation periods too long | Review thresholds against actual complaint data; consider shorter observation periods for high-risk devices |
| No linkage to CAPA | Trend detected and reported but no corrective action evaluated | Build trend-to-CAPA trigger into PMS procedure; include CAPA evaluation as mandatory step in trend workflow |
| MTR submitted late | Trend detected but report preparation delayed | Pre-prepare MTR template; train personnel on form completion; set SLA for MTR submission after trend confirmation |
| Legacy devices excluded | Manufacturer assumed trend reporting only applies to MDR-certified devices | MDCG draft Q&A clarifies that legacy devices and Annex XVI products are in scope |
| Global data not considered | Only EU complaints analyzed | MDCG draft Q&A requires global data consideration for trend detection |
RACI: Trend Reporting Process
| Task | Regulatory Affairs | Quality/Complaint Handling | Clinical/PMCF | Risk Management | Management |
|---|---|---|---|---|---|
| Define trend methodology in PMS plan | R | C | C | C | A |
| Code complaints for trend monitoring | I | R | I | I | I |
| Calculate observed rates and run statistical tests | R | C | I | C | I |
| Evaluate severity escalation | C | R | C | R | A |
| Determine benefit-risk impact | R | I | C | R | A |
| Complete MTR form | R | C | C | C | I |
| Submit MTR to Competent Authorities | R | I | I | I | A |
| Evaluate need for CAPA/FSCA | R | R | C | R | A |
| Update PSUR/PMSR with trend findings | R | C | C | C | I |
| Review and update PMS plan | R | C | C | C | A |
Sources
EU Regulation 2017/745 (MDR), Article 88: Trend Reporting.
EU Regulation 2017/746 (IVDR), Article 83: Trend Reporting.
MDCG 2025-X (Draft). "Questions and Answers on Trend Reporting under MDR Article 88 and IVDR Article 83." September 2025.
MDCG 2025-10. "Guidance on Post-Market Surveillance of Medical Devices and In Vitro Diagnostic Medical Devices." December 2025.
Manufacturer Trend Report (MTR) Form v1.0. European Commission, September 2025.
MDCG 2021-1 (MDR) and MDCG 2022-12 (IVDR): Guidance on vigilance reporting.
ISO 14971:2019. "Medical devices — Application of risk management to medical devices."
ISO 24971:2020. "Medical devices — Guidance on the application of ISO 14971."
Qserve Group. "How to Incorporate Article 88 into Post Market Surveillance." qservegroup.com
Johner Institute. "PMS Trend Analysis — A Complex Issue Where You Cannot Fail." blog.johner-institute.com
Casus Consulting. "MDCG 2025-10: New MDR/IVDR PMS Guidance." casusconsulting.com
Regulatory Rapporteur (TOPRA). "Lean Six Sigma in Post-Market Surveillance: Insights for EU MDR Article 88 and GB SI 2024/1368." April 2026.