Patient Experience Data: Why Linguistic Validation Matters

FDA’s July 2026 assessment examines how patient experience data appears in regulatory review. It combines interviews with FDA staff and a review of documentation for approved submissions. Within the assessment cohort, the report describes patient experience data mentions in 85% of NME reviews in 2025, and a 91.5% prevalence among original NME applications with orphan-drug status. These figures concern documented mentions, not the proportion of approvals caused by patient evidence.

The report also identifies challenges involving methodological rigor, representativeness and clarity about the intended use of submitted material. Its findings support attention to the quality and purpose of evidence. They do not evaluate NovimedGroup, demonstrate that one translation process is sufficient, or establish that every language version is comparable. The language-planning recommendations that follow are our operational interpretation of the broader evidence-quality problem.

Begin with the question the research needs to answer. Are interviews intended to discover relevant concepts, test understanding of an existing questionnaire, describe treatment burden or help interpret an observed change? Each purpose requires different questions and a different analysis plan. A fluent translated interview guide cannot repair uncertainty about what the study is trying to learn. Write the purpose in language that researchers, translators and interviewers can all use.

FDA’s PFDD Guidance 2 addresses methods for identifying what matters to patients, including approaches to gathering information about symptoms, impacts and treatment experience. For a multilingual project, turn the agreed research purpose into a concise language brief. Specify the population, country, language, setting and intended use of the findings. Record where an existing instrument must remain unchanged and where an exploratory interview permits follow-up questions.

A country list is not a complete participation plan. Within one country, patients may use different languages, dialects or levels of health-related vocabulary. Recruitment through a highly engaged online community may also reach a different group from recruitment through a clinic. Ask which people could struggle to understand an invitation, consent discussion, interview question or digital task. Decide how the research team will address those barriers before recruitment begins.

These considerations are especially important when an interview depends on describing everyday experiences in detail. Offer appropriate language support and accessible arrangements within the approved study design. Document exclusions and practical limitations rather than implying that the resulting sample represents everyone. Translating materials into additional languages can improve access, but it does not by itself establish that a sample is representative or that the analysis can support comparisons between groups.

Concept elicitation explores experiences that may matter to patients. Cognitive debriefing examines how intended respondents understand and use particular wording or response options. Treating these as interchangeable can produce an interview that does neither well. A project exploring treatment burden should allow patients to describe unexpected experiences; a project checking a translated item needs a clear way to investigate its intended meaning and response task.

Prepare interviewers for the purpose of each activity. In an exploratory discussion, avoid forcing an unfamiliar experience into a predefined category too early. In a comprehension interview, ask the participant to explain their interpretation rather than simply agree that a sentence is clear. Record the wording used, the relevant follow-up and the reason for the interpretation. The language specialist and qualitative researcher should be able to examine the same evidence.

A concept guide should explain what an item is intended to capture and what it excludes. It can clarify the recall period, intensity, frequency, activity, assistance and perspective involved. It should also identify the person authorized to resolve source questions. A list of preferred translated terms is helpful, but it cannot substitute for an explanation of how those terms function within a complete assessment.

Consider a hypothetical question about difficulty leaving home. One participant might describe physical effort, another anxiety, and another the absence of accessible transport. These are different experiences even when the same broad phrase describes them. The research purpose determines whether all are relevant or whether the item has a narrower intended concept. The translation team should raise that ambiguity for resolution rather than independently choose the most convenient interpretation.

When interviews take place in several languages, decide how recordings, transcripts, translations and analysis notes will be connected. Use stable identifiers and version labels, with access and retention governed by the approved privacy arrangements. If an analyst works from a translation, make it possible to raise a query about an expression and return to the relevant source passage through an authorized reviewer.

A polished summary can lose hesitation, self-correction or a distinction between a patient’s words and an interviewer’s suggestion. Preserve uncertainty where it affects interpretation. For example, a participant who initially agrees and then gives a conflicting explanation should not be summarized simply as having understood the item. Document how the team resolved the discrepancy, or why it remains unresolved. A clear audit trail makes that judgment easier to evaluate later.

An approved text can behave differently when implemented on a phone or tablet. A line break may separate a qualifier from the question; a response label may be truncated; a help message may introduce language that was never reviewed. Plan eCOA review around the actual respondent journey, including instructions, response options, branching, reminders and completion messages that fall within the project scope.

For a hypothetical diary, imagine that the question specifies the previous day but a reminder asks about how the patient feels now. Each sentence might be understandable in isolation, yet the combination creates a conflicting task. Link implementation issues to the approved wording, device version and corrective action. Confirm the correction in the final build, including the relevant language and screen size, before treating the issue as closed.

AI may help organize source queries, compare versions or flag potential omissions. Define the permitted task and the authorized input before using it. A generated summary of patient interviews is a draft requiring review against the underlying evidence; it is not evidence that participants expressed a particular view. Do not allow a confident model explanation to replace a missing observation or an unresolved linguistic question.

For each material recommendation, retain enough context for a qualified reviewer to accept, reject or revise it. Avoid pooling confidential materials across projects without authorization. If a model or prompt changes, consider whether earlier quality checks still address the task being performed. The useful outcome is a reviewable process with clear responsibility for meaning, rather than a larger volume of apparently consistent text.

A focused handover can include the approved source and concept definitions, language-country scope, permissions, workflow responsibilities, query history, interview evidence summaries, change decisions and final delivery versions. Agree which records the sponsor or instrument owner needs and who approves each stage. Connect every final file to its relevant review status so that an outdated export cannot be mistaken for the approved release.

Start the planning discussion with three questions: what decision must this evidence support, which people and language contexts must it reflect, and how will another reviewer follow the path from patient expression to final interpretation? NovimedGroup can help scope linguistic-validation and eCOA language-review activities around those questions. The appropriate methodology depends on the instrument, population and intended use; translation quality is one part of the wider evidence strategy.