ePRO cheat sheet

ePRO = any patient-reported outcome collected electronically (handheld device, smartphone, tablet, web, IVR); the mode of collection does not change the underlying construct, but it must be validated as equivalent to the paper version.

FDA 2009 PRO Guidance and 2013/2016 ePRO clinical/technical guidance, plus ICH E6(R2) and E9(R1), form the core regulatory frame; EU adds EMA/CHMP reflection papers on health-related quality of life and the 2022 ICH E6(R3) draft principles.

PRO instruments must be fit for purpose: documented content validity in the target population, psychometric evidence (reliability, construct validity, responsiveness), and a prespecified endpoint definition in the protocol before first patient in.

Endpoint hierarchy: primary, key secondary, and exploratory PROs must be prespecified with a multiplicity strategy; a PRO that is not in the statistical analysis plan is hypothesis-generating only.

Minimally important difference (MID) must be stated a priori and tied to the anchor or distribution-based method used; a statistically significant change without clinical relevance does not support labeling.

Blinding integrity: ePRO devices must not reveal treatment assignment; diaries and reminders must be identical across arms, and device settings (font size, alarms, language) must be locked and documented.

Reviewer-facing evidence: FDA and EMA expect a completed PRO evidence dossier (e.g., PRO labels database, patient-focused drug development history) and a mapping of instrument items to the endpoint concept.

Patient input: FDA PFDD and EMA patient engagement guidance mean instrument selection should be justified against the patient-important concept, not just prior use in the literature.

21 CFR Part 11 and EU Annex 11 govern electronic records: audit trails, secure time-stamped records, access control, and electronic signatures must be enabled and validated.

Usability validation per IEC 62366 / FDA human factors guidance is required; summative testing must include the intended patient population, including low-literacy, elderly, and visually impaired users.

Screen equivalence: each item must render fully on one screen without scrolling; visual analog scales and complex grids need specific justification or redesign for small screens.

Data provenance: the device must capture a timestamp for each entry, prevent back-filling, and store an immutable audit trail; sponsor edits must be logged with reason and user ID.

BYOD (bring your own device) requires a validated app, device compatibility matrix, and a plan for OS updates; mixed-device studies need documented equivalence testing.

Source data: ePRO entries are source data when entered directly by the patient; site transcription into the EDC creates a second source and must be reconciled with an audit trail.

Instrument licensing: use of copyrighted PRO instruments (e.g., EORTC, FACIT, SF-36) requires a license; electronic migration often requires a separate e-license and translation validation.

Disaster recovery: offline capture, backup, and a documented plan for device loss, theft, or network outage are expected in the data management plan.

Compliance is measured as entry compliance (proportion of expected entries completed) and timeliness; both should be prespecified and monitored by arm to detect differential compliance.

Diary windows: a recall period must match the instrument (e.g., 24-hour recall for daily diary); entries outside the allowed window are typically excluded per a prespecified rule, not ad hoc.

Missing data: ICH E9(R1) estimand framework requires specifying how intercurrent events (e.g., treatment discontinuation, device failure) are handled; complete-case analysis alone is rarely acceptable.

Compliance monitoring should be blinded where possible; differential reminder strategies across arms can bias PRO results and must be avoided or documented.

Data cleaning rules for ePRO (e.g., duplicate entries, out-of-range values, contradictory responses) must be in the data management plan and applied consistently.

Analysis of PRO endpoints often uses mixed models for repeated measures (MMRM) or generalized estimating equations; the choice must match the estimand and missing-data assumption.

Site and patient training records, device accountability logs, and help-desk tickets are inspection-ready evidence and should be retained with the trial master file.

Security and privacy: GDPR/HIPAA-compliant data flows, encryption in transit and at rest, and a documented data protection impact assessment for patient-facing apps.

Site activation requires device provisioning, user account creation with role-based access, and documented training before first patient enrollment.

Patient training should include a teach-back or demonstration; the training record and device serial number must be linked to the patient in the trial master file.

Help desk and escalation: a documented tiered support model with response times, and a process for reporting device malfunctions as protocol deviations when they affect data.

Protocol deviations for ePRO (missed entries, lost devices, late entries) must be classified and reported per the protocol; systematic issues trigger a CAPA.

Monitoring: centralized review of compliance dashboards, site-level outlier detection, and risk-based monitoring per ICH E6(R2) addendum.

Inspection readiness: be able to produce the audit trail, validation documentation (IQ/OQ/PQ), usability reports, and the mapping from device data to analysis datasets.

Change control: any device, app, or instrument version change mid-study requires impact assessment, validation, and documentation; version must be traceable per patient.

Vendor oversight: sponsor retains responsibility for ePRO vendor compliance; quality agreements, audits, and service-level metrics should be in place.

Paper-to-electronic migration requires cognitive debriefing and, where feasible, an equivalence study comparing paper and ePRO administration in the target population.

Measurement equivalence is supported by similar means, variances, and item-total correlations across modes; formal tests include differential item functioning (DIF) analysis.

Recall period and context effects can shift with mode (e.g., scrolling vs. paging); the migration report should document layout, instructions, and response option presentation.

Translation and cultural adaptation follow ISPOR good practices: forward and back translation, reconciliation, cognitive debriefing, and harmonization across languages.

Electronic migration reports (often called ePRO migration or equivalence reports) are expected in regulatory submissions when the electronic version differs materially from the validated paper version.

Response scale presentation (e.g., radio buttons, sliders, numeric entry) must preserve the original scale semantics; changing scale format requires revalidation.

Accessibility: WCAG 2.1 AA and section 508 considerations apply to patient-facing apps in many jurisdictions; font size, contrast, and screen reader support must be validated.

Instrument version control: item wording, response options, and recall periods must be locked and versioned; any change after validation triggers re-evaluation.

Decentralized trials rely on ePRO for remote data capture; the protocol must specify device logistics, connectivity requirements, and a fallback for patients without smartphones.

Direct-to-patient shipping of devices requires chain-of-custody, temperature/condition checks where relevant, and a documented return and decommissioning process.

Telehealth and remote consent: eConsent platforms must meet Part 11 and GDPR requirements and be validated; ePRO training can be delivered remotely with documented competency.

Equity and access: providing devices and data plans, offering multiple languages, and testing with diverse populations reduces selection bias in remote trials.

eSource and eConsent integration: ePRO data should flow into the EDC or eSource with an auditable transfer; avoid duplicate transcription.

Regulatory acceptance of decentralized elements varies by region; EU CTR and FDA guidance on decentralized trials should be checked for each country before implementation.

Site oversight in remote models: centralized monitors review compliance and safety signals; escalation pathways to sites must be defined and tested.

Patient burden: instrument length, diary frequency, and app complexity should be minimized and tested in usability studies to protect compliance and data quality.

PRO instruments can capture safety-relevant symptoms (e.g., suicidal ideation, severe pain); the protocol must define alert thresholds and a same-day escalation pathway to the site or medical monitor.

Serious adverse events identified from ePRO entries must be reported per ICH E2A and local timelines; the ePRO system should support flagging and triage.

Automated alerts must be validated and tested, with a documented false-positive/false-negative review; alert fatigue is a known failure mode.

Patient-facing apps should provide clear guidance on what to do in an emergency, distinct from study-related alerts.

Signal detection: aggregate PRO symptom data can inform DSMB reviews; the DSMB charter should specify what ePRO summaries are reviewed and how often.

Privacy of safety data: alert content and recipient lists must be minimized and documented in the data protection impact assessment.

Audit trail for safety alerts: who was notified, when, and what action was taken must be recorded and retrievable for inspection.

Training: sites must be trained on alert handling and documentation; unaddressed alerts are a common inspection finding.

Vendor qualification includes a quality management system review (ISO 9001/27001), validation documentation, and reference checks; the sponsor remains accountable for data integrity.

Computer system validation (CSV) follows GAMP 5: risk-based, with IQ/OQ/PQ and a validation summary report; the ePRO system is typically a GAMP category 4 or 5 system.

Part 11 assessment must cover audit trails, access controls, e-signatures, and record retention; a documented gap analysis is expected.

Service level agreements should define uptime, help-desk response, data delivery timelines, and breach notification.

Data transfer specifications: define file formats, frequency, encryption, reconciliation process, and error handling between vendor and sponsor.

Business continuity and disaster recovery testing should be performed and documented before go-live and periodically thereafter.

Vendor audits and quality agreements should be in place; the sponsor should have the right to audit and to receive validation evidence.

Exit strategy: data return in a usable format, destruction certification, and record retention obligations must be contractually defined.

PRO results intended for labeling must be prespecified, clinically meaningful, and supported by a validated instrument with a documented MID.

The FDA PRO labeling database and EMA EPAR summaries show how PRO endpoints have been accepted; reviewers expect consistency between the protocol, SAP, and label claim.

Patient-reported outcomes in medical devices: FDA and EU MDR require clinical evidence that may include PROs; the intended use and claim must match the instrument.

Real-world evidence: ePRO collected in registries and post-authorization studies must meet data quality standards; the FDA RWE framework and EU guidance apply.

Digital endpoints: novel digital measures (e.g., actigraphy, ecological momentary assessment) require analytical and clinical validation before use as endpoints.

Publication and transparency: PRO results should be reported per CONSORT PRO extension and posted on registries; selective reporting is a documented concern.

Health economic use: PRO data (e.g., EQ-5D) feed cost-utility analyses; the instrument and mapping method must be justified for HTA bodies.

Post-market surveillance: ePRO can support signal detection and patient experience reporting; GDPR and pharmacovigilance rules apply to any safety-relevant data.

A cross-functional ePRO governance team (clinical, data management, regulatory, IT, quality) should own standards, vendor strategy, and change control.

Standard operating procedures should cover device provisioning, training, data review, deviation handling, and vendor oversight; SOPs must be version-controlled and trained.

Training records must be contemporaneous and retrievable; role-based training matrices define who needs what before access is granted.

Quality metrics: compliance rate, timeliness, help-desk ticket volume, deviation rate, and data query rate should be tracked and trended.

CAPA: recurring ePRO issues (e.g., low compliance at a site) should trigger root-cause analysis and corrective action with effectiveness checks.

Risk assessment: a documented risk register for ePRO (device failure, privacy breach, compliance loss) with mitigations and owners.

Inspection findings in ePRO commonly involve inadequate validation, missing audit trails, and undocumented training; preventive controls should target these.

Documentation retention: ePRO records, validation packages, and training records must be retained per local requirements (often 2 years after last patient, longer for some regions).