EMA is changing how SmPC section 5.1 is assessed. What should RA teams review before 2027?

What did EMA publish, and when does it apply?

On 19 October 2026 EMA published the Methodological Principles for the Assessment of Section 5.1 of the Summary of Product Characteristics (EMA/283093/2016). Section 5.1 is the 'Pharmacodynamic properties' section, where the SmPC describes the clinical efficacy behind the indication. The document comes into effect on 1 January 2027.

It has been a long time coming. A draft went to public consultation from 15 December 2023 to 4 March 2024, the Methodology Working Party agreed the revised text in October 2025, and the CHMP adopted it on 26 June 2026. EMA lists it on its SmPC page under guidance for assessors, as the basis for a consistent assessment of section 5.1 across centrally authorised medicines. The 2023 draft (EMA/CHMP/566497/2023) is now marked as superseded.

The starting point is unchanged. The applicant or marketing authorisation holder proposes section 5.1, the rapporteurs assess it, and the final content remains a case-by-case CHMP decision. The 2009 SmPC guideline still asks for clinical trial information that is concise, clear, relevant and balanced. What is new is a written methodology for deciding what that means in practice.

This is guidance for assessors, not new legislation. But the principles assessors apply are the principles an applicant's proposed section 5.1 will be judged against, so they are the ones to draft to.

What belongs in section 5.1?

Results that are both clinically relevant and methodologically reliable, from the studies that provide confirmatory evidence of efficacy, limited to the indication, population and posology that are authorised. EMA defines methodological reliability as results that support an interpretation with an estimable and acceptable level of uncertainty. A clinically relevant endpoint with unreliable results stays out, because it is likely to be misread.

• Confirmatory studies first. Usually one or more randomised trials, occasionally single-arm trials, and in exceptional cases other sources.
• Supportive studies only if they add something the prescriber needs, such as a population excluded from the pivotal trials or a low-incidence outcome the pivotal trials could not capture.
• No non-informative studies 'for the sake of transparency or comprehensiveness' when positive pivotal studies already establish efficacy. The full assessment of every study belongs in the EPAR.
• Nothing on indications or populations that were not applied for or not granted, and nothing on off-label use. Consequential benefits, such as cardiovascular outcomes from weight management, may be added if relevant to the prescriber.
• Post-authorisation studies the CHMP requested to confirm efficacy are described once available, if relevant and reliable.
• The same principles apply before and after authorisation.

Which data cut should section 5.1 show?

Usually the confirmatory one: the data cut at which the null hypothesis was rejected. EMA's reasoning is that later updates are normally descriptive, their timing is often not pre-specified, and they can be biased by unblinding and crossover. Updated results can still be included after a case-by-case assessment of clinical relevance and reliability.

If they are, they replace rather than accumulate. Only one set of results per endpoint should appear, because two sets targeting the same estimand add no information. Where descriptive overall survival was shown before its hypothesis was tested and more mature, reliable results follow, the new results usually replace the old. The exception is when an update targets a different estimand, such as vaccine efficacy over several seasons, which should then have been defined as a separate endpoint in the protocol.

• Confirmatory endpoints whose pre-planned results come only at a later data cut are not included by default. EMA's example is overall survival in an oncology trial approved on progression-free survival.
• Confirmatory endpoints that were not statistically significant are generally excluded, with possible exceptions for rare diseases or long-term outcomes. If included, the text must make the intended message clear.
• In single-arm trials, only endpoints that can be interpreted without a control arm belong. Overall survival is usually left out because the size of the effect cannot be assessed.
• In the exceptional case that the main evidence comes from a formally failed study, reliable and relevant endpoints are shown, but the justification for the positive opinion stays in the EPAR.

Section 5.1 is not a results archive. Every new data cut is a decision about what to replace, not an addition to the pile.

How much statistical detail helps a prescriber?

Less than most drafts contain. The principles are explicit that methodological detail confirmed during assessment belongs in the EPAR, not the SmPC.

• No p-values. EMA says the numerical p-value does not support clinical interpretation.
• Confidence intervals, usually nominal 95%, as the measure of uncertainty. Take care where alpha was relaxed or designs were adaptive, because a nominal 95% interval can suggest a conclusion the hypothesis testing did not reach.
• Statistical significance only for multiplicity-controlled tests that rejected the null hypothesis under the pre-defined method, shown in text or with an asterisk. Asterisks are reserved for that purpose.
• Relative effects such as hazard ratios or odds ratios always alongside absolute values, including baseline values where applicable.
• Model-based results without an absolute effect, such as a negative binomial model or a log-rank test, supported by descriptive statistics such as annualised rates or Kaplan-Meier medians.
• No description of Type 1 error control or of how confidence intervals were calculated, no use of the word 'estimand', and no names of intercurrent-event strategies such as 'treatment policy' or 'hypothetical'.
• Report the estimand of highest relevance to patients and to the benefit-risk assessment, which is often, but not necessarily, the primary one.

What if the authorised indication is narrower than the studied population?

Then section 5.1 describes the whole trial population briefly, before focusing on the authorised subgroup. The main findings in the overall trial population come first, followed by the findings in the subgroup the indication covers. No off-label results are given.

• A subgroup excluded from section 4.1 because efficacy or safety could not be established can be mentioned only as part of the population description, for example its share of the trial population. Its efficacy results go in the EPAR.
• Where an indication is restricted on regulatory grounds, section 5.1 may still show results for the entire studied population, plus a short description of the results for the restricted indication.
• Consistency of effect across subgroups is highlighted only when it is essential to understanding efficacy, justified in the EPAR and in line with EMA's guideline on subgroups in confirmatory trials. A forest plot may then be appropriate.

Conditional marketing authorisations and paediatric data

For a conditional marketing authorisation, section 5.1 presents the relevant and reliable information that supports the positive benefit-risk, and does not state that efficacy has not been comprehensively assessed. When comprehensive data arrive, section 5.1 is re-assessed. If the specific obligation was more mature data from the same trial, the earlier results are usually replaced. If it was a study in a different disease setting or regimen, keeping both sets is likely to be informative.

Paediatric data are treated differently from adult data. All clinically relevant paediatric pharmacodynamic and efficacy studies are recommended for section 5.1, including negative or inconclusive results where no paediatric indication was applied for or granted, presented with enough text to prevent misreading. Confirmatory paediatric results usually supersede exploratory ones, including those from a paediatric investigation plan. EMA also reminds applicants that access to the paediatric reward may depend on paediatric data appearing in section 5.1.

When do tables and graphs earn their place?

When they make the main results easier to interpret, not when they make section 5.1 longer.

• Tables should be self-explanatory and readable without the text. Two or more similar studies can sit side by side in one table.
• Statistical models generally stay out of table footnotes. If a footnote is needed for correct interpretation, it should not be technical.
• Graphs only when they add relevant, reliable information for the prescriber. Immature Kaplan-Meier curves can mislead, for example by reflecting only a subgroup with early events.
• If a graph is included, numbers at risk and confidence intervals are encouraged; significance annotations and statistics already shown in text or tables are not.

What should RA teams review before 1 January 2027?

The principles take effect on 1 January 2027, and the procedures that will meet them are being planned now.

1. List the products where section 5.1 will open in 2027: new indications, extensions, updated survival analyses, conversion of a conditional marketing authorisation, and paediatric data submissions.
2. For each planned update, decide in advance whether the new results replace earlier ones or justify a second set because they target a different estimand.
3. Check draft section 5.1 texts for p-values, relative effects without absolute values, statistical-method descriptions and estimand terminology.
4. Where the indication will be narrower than the trial population, draft the overall-population summary and the subgroup results in the order EMA expects.
5. Agree with biostatistics, in the protocol and statistical analysis plan, how results will be presented in the SmPC, including which confidence intervals are appropriate for endpoints tested at a different alpha.
6. Prepare the justification for the EPAR, especially for subgroup presentations, rather than defending it inside section 5.1.
7. For generic and hybrid products, watch the reference product: generic product information normally follows the reference product's, so its section 5.1 changes will need to be tracked and reflected.

The document does not ask marketing authorisation holders to revise existing texts on 1 January 2027. Our reading is that existing sections will be measured against it the next time they are opened.

Our view: more data is not better product information

Section 5.1 has drifted towards completeness. Every new analysis is a reason to update it, and the text grows with each variation: another data cut, another secondary endpoint, another table. The result is product information that is accurate in every line and harder to use as a whole.

EMA's principles push the other way, and they are right to. A prescriber reading section 5.1 needs the evidence that supports the authorised use, with its uncertainty, in a form they can interpret. That is a judgement about relevance and reliability, not an extraction exercise. Deciding which data cut to show, what to replace and what to leave to the EPAR is regulatory work that needs clinical and statistical input. It cannot be automated by copying the clinical study report into the SmPC.

Where RafiHive fits

RafiHive answers EU regulatory questions from a reviewed knowledge base of official sources, with each answer citing the passage it relies on and each source's status and version recorded. A question about section 5.1 should be answered against a source like this one with its date of effect, not as if it had always applied.

RafiHive's packaging and labelling check compares the structure of the SmPC, labelling and package leaflet with the required particulars and the current QRD template, for qualified review. It does not decide which efficacy results belong in section 5.1. That decision stays with the RA, clinical and statistical reviewers who stand behind the product information.