A practical guide to MDR Article 88 trend reporting: what counts as a reportable trend, how to set statistical thresholds, and how to build it into your PMS Plan.
MDR Article 88 requires manufacturers to report a statistically significant increase in the frequency or severity of non-serious incidents or expected undesirable side-effects to their National Competent Authority (NCA), even though each individual event falls below the serious-incident threshold. To meet this, a manufacturer's PMS Plan must define which events are being monitored, the statistical method used to detect a significant increase, and the observation period, per Annex III, section 1.1(b).
Trend reporting isn't new. Article 88 turns an expectation that already existed under MEDDEV 2.12-1 Rev. 8 (2013) into a binding MDR requirement, a shift also anticipated by GHTF SG2/N54R8:2006. Manufacturers who already ran a MEDDEV-style trending process will find Article 88 a smaller lift than those building it from scratch.
What counts as a “non-serious incident” or “expected undesirable side-effect”?
Both are events a manufacturer should already have anticipated. Non-serious incidents are complaints that don't meet the legal definition of a serious incident; expected undesirable side-effects are the known, labeled complications of the device. Both should already appear in the risk management file, typically in a design, use, or application FMEA, with a pre-defined occurrence rate and severity rating drawn from literature or historical data.
The catch: because these events sit below the individual-reporting threshold, they're often under-reported as complaints. Unless a manufacturer runs a deeper review of historical data, literature, testing, or Post-Market Clinical Follow-up (PMCF) activity, the baseline it's trending against may understate what's actually happening in the field.
How do you determine a “statistically significant increase” in frequency?
Use of statistics must be justified, not automatic. With only 2–5 events of interest in an observation period, evaluate each one individually and document why a statistical method wasn't appropriate for that sample size. Once incidents are frequent enough for statistics to be meaningful, a method is expected, and the result still has to be paired with root-cause analysis and, where the trend affects the benefit-risk balance, a report to the NCA.
Two approaches see the most use. The one-sample t-test checks whether an observed event rate differs from a known or hypothesised baseline rate, but it's a parametric test, so it only holds up if the underlying data is confirmed to follow a normal distribution. Control charts, built on Shewhart rules, plot the occurrence rate across an observation period and are often more forgiving of the messy, non-normal data trend reporting usually produces.
How do you evaluate a “statistically significant increase” in severity?
Severity doesn't trend the way frequency does. An event scored as mild in the risk file (say, a 2 on a 1–5 scale) but that later causes more serious harm isn't a statistics problem, a shift on an ordinal scale doesn't carry statistical significance on its own. Each case has to be assessed individually against the benefit-risk analysis, asking directly whether it pushes the individual or overall risk from acceptable to unacceptable.
That assessment only works if the risk management file underneath it is built on quantitative occurrence and severity ratings in the first place, as ISO 14971 allows either qualitative or quantitative categorisation, Article 88 effectively pushes manufacturers toward the quantitative option, since a qualitative scale gives a severity shift nothing to be measured against.
The same logic extends forward: any new event type that shows up for the first time has to be evaluated on its own and folded into the risk management file, including its effect on the overall benefit-risk conclusion, not treated as a one-off that sits outside the system.
How does Article 88 trend reporting fit into the PMS Plan?
Article 88 doesn't just require reporting; it requires the PMS Plan itself to define how trending is done. Per Annex III, section 1.1(b), the plan must cover the methods and protocols for managing events subject to trend reporting, the methods for establishing a statistically significant increase, and the observation period. GHTF SG2/N54R8:2006, Appendix C is still the most useful practical reference for building that section.
At minimum, the PMS Plan should define or reference each of the following:
- Events subject to trend reporting. The specific non-serious incidents and expected undesirable side-effects being monitored, often maintained as a living “hazard list”
- Baseline. The reference rate to trend against, drawn from risk analysis, reliability testing, historical data, or published literature
- Observation period. Typically one month for higher-volume devices, kept short enough to allow timely corrective action, especially for higher-risk products
- Threshold value for reporting. A defined trigger (e.g. a specified percentage increase or a jump in severity rating) set to avoid both over- and under-reporting
- Trending statistical technique and rationale. The chosen method, justified, with a documented rationale wherever a statistical approach is judged inappropriate
Where should these details actually live?
Most of what feeds trend reporting already exists elsewhere: the device labelling, the risk management file, the outputs of the data-analysis procedure under ISO 13485:2016, Clause 8.4, the statistical-techniques procedure, and company policy. Since 2 February 2024, the FDA's Quality Management System Regulation (QMSR) has aligned US requirements with ISO 13485:2016 — including Clause 8.4 — which supersedes the older 21 CFR 820.250 statistical-techniques clause this guidance used to cite on its own.
Reference these sources in the PMS Plan rather than repeating their content in full. It keeps the plan shorter, and it means a change to a trending parameter or an event type doesn't force a PMS Plan revision every time.
Getting Article 88 audit-ready
Most gaps we see are in the paper trail: a threshold nobody wrote down, a hazard list nobody kept current, or a trending method with no documented justification for using (or not using) it. Qserve's vigilance and PMS specialists can audit your post-market procedures against Article 88, help set up defensible statistical thresholds, or update the relevant QMS SOPs. Get in touch through our regulatory affairs consultancy team to talk through where your PMS Plan stands today.