ePRO in clinical trials: getting between-visit symptoms into the safety dataset

ePRO in clinical trials is the electronic capture of patient-reported outcomes: participants report symptoms, function and other outcomes on a phone, tablet or web form, and the answer is recorded as given, without interpretation by a clinician. In oncology trials, ePRO between visits records symptomatic adverse events when they happen rather than at the next appointment, often with items from NCI's PRO-CTCAE item library. FDA's 2024 guidance on core patient-reported outcomes in cancer trials says these reports complement standard CTCAE safety data and do not replace it.

The participant's answer stays their own report: a clinician on the study team grades the event, and an investigator signs it. This guide covers ePRO for safety reporting in oncology trials; the EMA data practices in the checklist below apply in any therapeutic area.

FDA notes that intermittently given intravenous cytotoxic chemotherapy often produces its most intense symptomatic adverse events earlier in each cycle. When the visit falls at the start of the cycle, that peak falls between visits. For a sponsor, what reaches the safety dataset can reach the label. FDA's guidance cites XALKORI (crizotinib): exploratory patient-reported results on the timing, frequency and impact of visual disturbances were included in the ADVERSE REACTIONS section of its labeling (FDA 2024, section VI).

What is ePRO in a clinical trial?

FDA defines a PRO measure as a report that comes directly from the patient, without amendment or interpretation of the patient's response by a clinician or anyone else. ePRO collects that measure on a phone, tablet or web form, which FDA notes may allow assessments outside the clinic (FDA 2024, section V.B).

EMA's guideline on computerised systems and electronic data in clinical trials (2023, Annex 5) adds that ePRO records when each entry was made. NCI notes that electronic capture of PRO-CTCAE can use conditional branching logic, which reduces respondent burden (NCI PRO-CTCAE overview).

For cancer trials, FDA's 2024 guidance recommends collecting and separately analyzing five core patient-reported outcomes:

Core PRO (FDA, 2024)	What it captures	Examples FDA gives
Disease-related symptoms	Symptoms of the cancer itself	NSCLC-SAQ; 0 to 10 ratings of pain, anorexia or fatigue
Symptomatic adverse events	Side effects the regimens are expected to cause	The PRO-CTCAE item library
Overall side effect impact	One global rating of overall side effect severity	FACIT item GP5; EORTC item Q168
Physical function	Activities that require physical effort	PROMIS physical function item bank; EORTC QLQ-C30 physical function scale
Role function	Work and daily activities	EORTC QLQ-C30 role function scale

The symptomatic adverse event row is the one that touches the safety dataset. For it, FDA says to:

1. Choose a concise set of the most important expected symptomatic adverse events from an item library.
2. In an active-controlled trial, ask about the side effects expected from either regimen in both arms.
3. Consider a free-text question, to mitigate the risk of missing an important symptom.

Why do symptoms between visits go missing from the safety record?

Three things lose them: a phone report is graded later by someone reconstructing the call, recall fades before the next visit, and clinicians record fewer symptoms than patients report.

The phone call. A participant with diarrhea calls the research line on a Thursday afternoon. A research nurse assesses, advises and writes a telephone note. The study work is still ahead: a CTCAE term, a grade against baseline, an onset date and attribution. It can wait until the next visit, done by someone reconstructing a call they were not on. The reconstructed dates are then what a monitor checks against source.

Recall. Severity and onset are easier to report on the day than three weeks later, and a symptom that resolved before the visit is the easiest to leave out. EMA notes that direct entry minimizes recall bias because data are entered at the time (EMA 2023, A5.1.1.4).

The reporter. Di Maio and colleagues compared patient and physician reporting of six symptoms across three randomized trials with 1,090 patients: anorexia, nausea, vomiting, constipation, diarrhea and hair loss. Agreement was low for all six, and physician-reported rates were always lower than patients' (Di Maio et al., J Clin Oncol 2015).

What does the evidence say about between-visit symptom monitoring?

Three randomized trials of symptom monitoring in cancer treatment found fewer emergency room visits, longer overall survival and small improvements in physical function and symptom control:

Study	Setting	What participants did	Result
Basch et al., J Clin Oncol 2016	766 patients on outpatient chemotherapy for advanced solid tumors	Reported 12 common symptoms, with weekly email prompts between visits for those with home computers; nurses alerted on severe or worsening symptoms	ER visits 34% vs 41% (P = .02); 75% vs 69% alive at one year (P = .05)
Denis et al., JAMA 2019	121 patients with advanced lung cancer without progression after initial treatment	Weekly symptoms online, with oncologist alerts on predefined criteria, against routine scheduled imaging	Median overall survival 22.5 vs 14.9 months (P = .03)
Basch et al., JAMA 2022	1,191 adults with metastatic cancer in 52 US community oncology practices	Weekly surveys online or by automated phone for up to a year, with care team alerts	Prespecified secondary outcomes at three months (EORTC QLQ-C30, 0 to 100): physical function 2.47 points and symptom control 2.56 points better than usual care

These trials tested monitoring as part of treatment, not as a trial data source. Read beside the under-reporting Di Maio found, they suggest a study that captures the interval holds a fuller safety record than one that samples it at each visit.

How does PRO-CTCAE fit with CTCAE grading in clinical trials?

NCI designed PRO-CTCAE as a companion to CTCAE. A PRO-CTCAE response is the participant's rating; a CTCAE grade is a clinician's determination against a written criterion, often relative to the participant's baseline. The item library holds 124 items representing 78 symptomatic toxicities drawn from CTCAE. Items rate frequency, severity, interference, amount or presence over the past seven days, and NCI reports linguistic validation in more than 60 languages (NCI PRO-CTCAE overview).

NCI notes there are as yet no standardized rules for combining attributes into one score, and recommends presenting CTCAE grades for the same period alongside PRO-CTCAE scores. A 2021 composite grading algorithm maps up to three items for a symptom to one score from 0 to 3 (Basch et al., Clinical Trials 2021), a scale separate from CTCAE's Grades 1 to 5.

Diarrhea shows why both records are needed:

• PRO-CTCAE asks how often the participant had loose or watery stools in the past seven days, on a five-point scale from never to almost constantly.
• CTCAE grades Diarrhea largely by the increase in stools per day over baseline. An increase of 4 to 6 is Grade 2 in both v5.0 and v6.0, and v6.0 rewrote Grade 1 as a change in consistency or frequency (Diarrhea in both versions).
• A participant whose baseline is one stool a day and who now has five may answer "frequently". The clinician, with the count and the baseline, assigns Grade 2.

Both records are correct, and they are reconciled, not merged. Keep the participant's answer as given. Link it by date to the adverse event it informed. Record why a reported symptom did or did not become an adverse event entry.

What should an ePRO system do with a between-visit report?

Between-visit data arrive in two modes. A scheduled questionnaire, such as PRO-CTCAE on a fixed schedule with a seven-day recall, produces PRO data for analysis. An event-driven report is sent when a symptom starts, and it feeds adverse event capture. Eight practices follow from FDA and EMA guidance; practices 1, 2 and 8 govern the schedule, and 3 to 7 apply to both modes.

1. Record a baseline before the first dose. FDA asks for a pre-treatment assessment as the reference point for change (FDA 2024, section V.A).
2. Ask more often early. FDA's example schedule for symptomatic adverse events is baseline, weekly from week 2 to week 8, then months 3, 4, 5, 6, 9 and 12, adjusted to how the drug is given (FDA 2024, section V.A).
3. Timestamp the entry, not the upload. EMA asks for the time of data entry, not only the time of transmission (EMA 2023, A5.1.1.1). FDA lists the participant as the data originator for ePRO and asks that each data element record who entered it and when (FDA 2013, Electronic Source Data in Clinical Investigations, sections III.A.1 and III.A.3).
4. Let participants see what they entered. EMA notes that without it, a participant may forget whether an event-driven report was already made (EMA 2023, A5.1.1.1).
5. Get the data to the investigator in time. ePRO data are often hosted by a service provider, and EMA expects them to reach the investigator in a timely manner to minimize the risk of missed adverse events (EMA 2023, A5.1.1.3). EMA also states that the sponsor should not have exclusive control of the data at any point (EMA 2023, section 6.6).
6. Audit every correction. EMA asks that ePRO corrections be documented in an audit trail, and that a correction made much later carry a good reason and justification (EMA 2023, A5.1.1.2 and A5.1.1.4). FDA asks that a change never obscure the original entry and record who made it, when and why (FDA 2013, section III.A.3).
7. Offer an alternative to the participant's own phone. EMA states that a participant unable or unwilling to use their own device should not be excluded from the trial (EMA 2023, A5.1.3).
8. Plan for missing data, and document how, where and why not. FDA asks for protocol procedures to mitigate missing data, such as reminders. Sponsors should record how and where each assessment was completed, and put the reasons for missing data in the analysis dataset (FDA 2024, section V.B).

How does Healthnotes carry a between-visit report to clinician grading?

On Burna AI, between-visit capture runs through Healthnotes, the participant app sponsors deploy through their sites.

• Entry. A site issues the participant a PRO token from app.burna.ai, tied to the study and the site. The token is the login, and every submission is filed against the participant's enrollment.
• Event-driven reports. The participant speaks or types what happened in their own words, or picks from a short list of symptoms common for their treatment. They review the entry before anything is sent.
• Scheduled instruments. A sponsor composes the study's instrument battery, schedules it and publishes it to participants without an app release, and PRO-CTCAE is supported. Answers are stored as given, against the instrument version the participant was shown. They stay the participant's report rather than a CTCAE grade.
• Grading. When the site's encounter is graded, the engine reads what the participant reported, proposes a CTCAE grade for each symptom reported and shows the criterion and reasoning behind it. The grade stays a suggestion until a clinician confirms it, and the audit trail records what was suggested and what was decided.
• The record. A confirmed event follows the same path as one recorded in clinic. It is graded against the CTCAE version the protocol names and scored for attribution against every drug in the regimen with WHO-UMC and Kramer. An investigator signs it with a 21 CFR Part 11 electronic signature, and it is available to the FDA MedWatch, IRB and serious adverse event reports the platform produces.

A participant with a symptom that cannot wait still calls the study team; Healthnotes keeps what they report, dated and in their words, for the study record. The capabilities behind each step are listed in what Burna AI does today. AI suggests, clinicians decide. Human-in-the-loop, always.

For a research site, the picture a nurse would otherwise reconstruct is assembled before the participant arrives, and the reconstruction work returns as capacity: more trials with the same team. For a sponsor, the participant's experience of the interval reaches the safety data a label is written from.

Common questions about ePRO in clinical trials

What are patient-reported outcomes in clinical trials?

Patient-reported outcomes (PROs) in clinical trials are reports about the participant's health that come directly from the participant, without interpretation by a clinician. For cancer trials, FDA's 2024 guidance names five core PROs: disease-related symptoms, symptomatic adverse events, overall side effect impact, physical function and role function.

Is ePRO the same as eCOA?

No. eCOA, electronic clinical outcome assessment, is the umbrella term. EMA's guideline describes it as covering outcomes reported by investigators, trial participants, caregivers and observers, and FDA also counts performance outcomes as clinical outcome assessments. ePRO is the participant-reported part, and PRO-CTCAE is one item library an ePRO system can collect.

Can trial participants use their own phones for ePRO?

Yes, with an alternative. EMA's guideline accepts bring-your-own-device ePRO when a sponsor-provided device is available, access controls make each entry attributable, and only the participant knows the password. FDA's guidance on digital health technologies for remote data acquisition (December 2023, section IV.A.4) likewise asks for a sponsor-provided option so no one is excluded for lacking their own.

---

Coordinators reconstructing between-visit calls, or a protocol that names PRO-CTCAE items? Book a 15-minute conversation.