Wearables in Clinical Trials: Digital Endpoints, Adherence and Audit Trail

Wearables have moved from pilot curiosity to standard tooling in clinical trials over the past five years. Sponsors now use these wearable technologies to measure sleep quality, physical activity, heart rate and continuous physiological parameters that scheduled clinic visits miss.

--> More data, richer endpoints, less reliance on the patient diary and its recall bias.

This raises a practical question:
How do you run wearables in clinical trials so the data clears the regulatory bar, adherence holds across a long study, plus the audit trail survives a sponsor inspection?

This guide maps the qualification path for a digital endpoint and other digital biomarkers, the adherence reality across long studies, the audit-trail architecture that EMA and FDA rules expect, the operational layer that has to absorb the new technology, plus the five mistakes that most often stall wearables in clinical trials.

On the device side, our companion piece on the wearable sensor stack covers the hardware and its MDR classification.

Table of Content

Why Wearables in Clinical Trials Went Mainstream

The case for wearables in clinical trials rests on three shifts. First, the sensors inside modern wearables and medical devices became cheap and accurate enough to leave the lab. That same sensor technology powers remote patient monitoring in everyday clinical research.

Second, phones turned into reliable gateways that patients already own. Meanwhile, regulators issued guidance that makes the path from sensor data to an accepted endpoint walkable rather than experimental.

Three Shifts Drive the Trend

The demand side matched that. Sponsors want endpoints closer to how patients really live. For example, a six-minute walk test at a visit shows one snapshot, while a continuous step count over twelve weeks shows the trajectory.

That granular data reveals patterns a single visit hides. The payoff is better patient outcomes and clearer patient safety signals across clinical studies.

As a result, pharma, MedTech and biotech sponsors have moved from asking whether to add a wearable to asking which wearable, for which endpoint, with what evidence.

The Costs

Still, the shift carries hidden costs. Continuous data needs continuous adherence and continuous adherence needs hardware that does not annoy the patient.

Continuous capture and auditing also need infrastructure beyond what a consumer platform offers. Each of these problems is solvable and each is also where studies stumble when the wearable was bolted on after the protocol.

-> In short: Sensors got cheap, regulators issued guidance and sponsors want richer endpoints. The delivery problem is what you design for from day one.

What Counts as a Digital Endpoint

Trial wearables produce continuous sensor data and increasingly, digital biomarkers. Data becomes an efficacy endpoint only when regulators accept that it reliably measures something clinically meaningful in the specific context of use, whether for drug approval or wider clinical development.

So there are two routes: EMA qualification through CHMP and its Scientific Advice Working Party, plus FDA qualification through the digital health technology programme.

The EMA Qualification Route

The European route runs through the CHMP Scientific Advice Working Party. Sponsors apply for a qualification advice or a qualification opinion. The opinion is not binding, yet it carries real weight across EU processes.

For instance, the EMA has qualified stride velocity 95th centile for Duchenne muscular dystrophy, the first EMA-qualified digital endpoint measured by a wearable. The EMA also expects early involvement of academic and patient organisations as co-applicants. Qualified endpoints now span fields from cardiovascular medicine to mental health.

The FDA DHT Pathway

For trials that also target the US, FDA's 2023 DHT guidance sets the frame. It runs through use case, measurement definition, verification, validation and qualification.

Verification shows the sensor and algorithm measure what they claim. Validation shows the measurement is clinically meaningful.

Then qualification is the regulatory decision that the combination fits the context, after which the endpoint can be reused across studies. The evidence bar is similar to the EMA, though the procedural mechanics differ.

PRO Instruments vs Continuous Sensor Data

Not every wearable signal needs qualification. The wearable devices in play range from fitness trackers and continuous glucose monitors to patches that track vital signs such as oxygen saturation, with sleep tracking on top.

Many teams use wearables in clinical trials as supportive or exploratory endpoints, or as a richer data layer beside classic measures. Qualification is the bar for use as a primary endpoint in a pivotal study.

So the design question is which evidence tier and which measurable endpoints the protocol needs. An exploratory Phase 2 measure uses the wearable as a learning instrument. By contrast, a primary Phase 3 endpoint needs the full EMA or FDA qualification path, which most sponsors plan from Phase 2 onward.

-> In short: Wearable data becomes an endpoint when the EMA or FDA qualification path clears it. Plan the path from Phase 2 when a primary endpoint is the goal.

What To Do:

Define the use case and context of use before you choose the wearable, because the evidence plan follows from both.
For pivotal endpoints, plan EMA and FDA qualification in parallel, since the evidence packages overlap heavily.
Treat exploratory and primary endpoints as different evidence problems, without over-engineering the exploratory ones.

The Reality of Adherence and Dropout

Adherence, or patient compliance, is where trial wearables fail.

The protocol assumes patients wear the device daily.
However, the data shows a steady decline.
By week 12 a standard study often sees daily wear below 40 percent.
By week 48 it can fall under 20 percent.

What the Published Data Says

Published wearable studies of remote monitoring show a consistent pattern that clinical researchers see across research settings. Adherence is highest in the first two weeks and then it drops sharply over the following month.

By month three, daily wear under 50 percent is normal. Meanwhile, the drivers are mostly non-technical:

patients forget
devices are uncomfortable
the charging rhythm clashes with routine
the pairing breaks after a phone update.

None of these show up in a six-week pilot, which is why pilots over-promise on adherence.

Architecture Decisions That Move Adherence

A handful of architecture levers move the curve reliably. The charging rhythm matters most.

For example, weekly charging is easier to sustain than daily. Cradle-synced devices that transfer on contact remove the pairing failure entirely.

The form factor is second. A chest patch under clothing has different dropout patterns than a visible wristband, while a smart ring sits between them. Choosing the form factor against the patient group is an important study-design question.

Charging, Sync and UX

The third lever is engagement design, which lifts patient engagement and lowers participant burden. Patients last longer when the device returns something they value. For instance, a simple weekly summary the patient sees at a visit turns the device from a black box into a participation nudge.

Site-coupled adherence reminders also work. For example: a nurse who sees the summary at each visit can have a short conversation that resets the routine. Most adherence loss happens between visits, so the remedy is visibility.

-> In short: In standard deployments adherence falls. Charging rhythm, form factor and engagement design are the three levers that move the curve most.

The Audit Trail Architecture

Sensor data only has value in a trial when its origin is traceable. In the EU, the EMA guideline on computerised systems and EU GMP Annex 11 set the rules, alongside 21 CFR Part 11 in the US.

All of them require:

electronic signatures
audit trails
timestamps
access control
a validation status
ALCOA+ data integrity.

Most consumer platforms miss this.

Annex 11, the EMA Guideline and Part 11

The EU and US frameworks overlap heavily. Each requires every record to be attributable to:

a person
legible
captured at the time of the event
kept in original form and accurate.

The EMA's 2023 guideline extends this with ALCOA++, adding complete, consistent, enduring and available.

For wearable data, attributable means bound to a subject ID with a provable link. Contemporaneous means the timestamp on the sensor packet has to be trustworthy. Accurate means calibration and validation evidence exists, so a consumer wearable whose timestamp is only the phone clock at sync fails on contemporaneous.

Consumer Platforms vs Purpose-Built

Off-the-shelf consumer wearables usually struggle on contemporaneity and signed packet integrity.

Their data flows are designed for the consumer case. Instead, purpose-built study platforms add UTC-anchored hardware clocks, packet signatures, write-once storage and validated end-to-end pipelines.

Several vendors now offer study-grade wrappers around consumer hardware. The wrapper usually adds a study-only gateway app that captures timestamps, signs payloads and writes to an auditable backend.

--> So consumer hardware can pass in trials, but only with a non-consumer layer on top.

Timestamps, GDPR and ALCOA+

Timestamps deserve their own attention. A sensor packet reading 10:32:14 is meaningless without knowing whose clock produced it. Therefore the remedy is to anchor the device clock to a verifiable source and to record the sync events themselves as part of the audit trail.

Data protection runs in parallel. Trial wearable data is personal health data under the GDPR, so data privacy controls on this patient data are expected:

encryption in transit and at rest,
strict access control,
a lawful basis.

As a result, many sponsors evidence this through an ISO 27001 programme to ensure compliance with the regulatory requirements alongside the ALCOA+ controls.

-> In short: The EMA computerised-systems guideline, Annex 11 and Part 11 demand attributable, contemporaneous data with provable origin. Consumer wearables fail without a study-grade wrapper. GDPR adds the data-protection layer.

What To Do:

Test every consumer wearable candidate against ALCOA+ before you plan a study around it.
Treat timestamp anchoring as a study-design topic, not a backend detail.
Map the wearable data flow step by step against the EMA guideline, Annex 11 and Part 11, since every weak point compounds in the audit.

Trial Operations with Wearables in the Loop

Wearables change day-to-day trial operations. Site workflow, patient onboarding, data monitoring committee access and the option of decentralised models all shift. In fact, ICH E6(R3) Good Clinical Practice now addresses electronic data sources directly, which includes trial wearables.

Site Workflow and Patient Enrolment

Sites must issue the wearable, train the patient, pair it with the gateway app and confirm the first data transfer before the patient leaves the visit. A failed day-one setup is the most common reason a patient never delivers usable data. Therefore study teams need:

a short, scripted enrolment flow,
clear study protocols,
a named owner.

Shipping logistics matter too. Devices must arrive at the site charged and pre-configured.

Returns at study end need a documented process to extract data before the device is wiped. Both sit with the sponsor or CRO.

Data Monitoring Committee Access

The data monitoring committee sees data on a schedule. For wearable data, that schedule has to handle continuous streams rather than discrete visit snapshots. A committee reviewing adherence and signal data needs aggregated views.

So the analytics layer above the auditable storage must deliver committee-relevant cuts on demand. Machine learning can turn continuous streams into predictive insights for that review.

That layer is part of the trial infrastructure and should be treated that way. Sponsors who plan it in the protocol design avoid the late dashboard scramble under review pressure.

Decentralised and Hybrid Trial Models

Continuous data from trial wearables enables fewer clinic visits and real time monitoring between them. Decentralised and hybrid trials lean on this remote patient monitoring to reach more diverse patient populations. This way patients take part remotely, receive devices directly and replace some visits with telehealth checkpoints backed by wearable data, provided they have the digital literacy to run the app.

Especially in long studies with mobility-limited groups this saves so much. Still, the trade-off is that everything the site used to do now needs a remotely supported equivalent. Adherence is harder to win back when nobody sees the patient in person.

-> In short: Sites, the monitoring committee and decentralised models all need wearable-aware workflows. Plan them in the protocol design already.

Common Mistakes with Wearables in Clinical Trials

From real studies, the same five mistakes recur around trial wearables. Each is fixable when caught early and each is expensive when caught late.

1. Bolting the Wearable on After the Protocol

Sometimes sponsors pick the wearable late, once the protocol structure is set. The result is a device that:

does not match the endpoint
an adherence model the protocol assumed away
plus an audit trail the chosen platform cannot produce.

Instead, scope the wearable as part of protocol design.

2. Trusting Pilot Adherence Numbers

Six-week pilots show high adherence. That is like the honeymoon phase.

So naturally the cliff follows after. Therefore power the study against the month-six adherence reality, not pilot data, or plan adherence interventions from the start.

3. Relying on Consumer-Platform Timestamps

The consumer wearable shows perfect-looking timestamps. However, the audit shows they come from the phone clock at sync and not the sensor at capture. The data fails on contemporaneous, so the reviewer flags the whole dataset.

4. Skipping the Qualification Evidence Package

Sponsors use a wearable as a primary endpoint without working through verification and validation. But reviewers of a pivotal study expect the full qualification evidence stack. A submission without it triggers a major information request and costs quarters.

5. Treating the Companion App as Unregulated

The gateway app for enrolment, sync and patient-facing summaries is part of the regulated system. When it shows the patient clinical data or influences trial conduct, it can itself need qualification as software as a medical device under Rule 11. Treating it as a generic mobile build creates a compliance gap.

-> In short: Late wearable choice, inflated pilot adherence, consumer timestamps, no qualification evidence, plus a regulated app treated as generic. Five routes to the same trial delay.

FAQs on Wearables in Clinical Trials

Does a wearable need EMA or FDA qualification to be used in a trial?

A1: No, not always. Wearables can be used as supportive or exploratory endpoints without qualification. Qualification is the bar for use as a primary endpoint in a pivotal study. So a Phase 2 study can use unqualified wearable signals to learn what matters, while Phase 3 needs the full EMA or FDA evidence package.

Q2: Can we use a consumer smartwatch for trial wearables?

A2: Yes, but with a study-grade wrapper layer. The consumer hardware can pass in trials when a study-only gateway app adds UTC-anchored timestamps, packet signatures and a validated pipeline to an auditable backend. Several specialist vendors supply this layer.

Q3: What adherence rate should we plan for?

A3: Plan for 40 to 55 percent daily wear by month three in a standard deployment. Plan for 60 to 75 percent when you have designed for adherence with cradle charging, a comfortable form factor and engagement loops. Anything above 80 percent beyond the first month usually points to a measurement artefact rather than real wear.

Q4: How does the GDPR affect wearables in clinical trials?

A4: Wearable readings are personal health data, so the GDPR applies in full. You need a lawful basis, encryption in transit and at rest, strict access control and clear data-processing roles between sponsor, CRO and platform. In decentralised trials, where data flows straight from the patient, many teams evidence this through an ISO 27001 programme alongside the ALCOA+ audit controls.

Q5: Do decentralised trials need different wearables?

A5: No different hardware, but different supporting infrastructure. Direct-to-patient shipping, remote configuration, telehealth-coupled engagement and tighter time-zone handling matter more in decentralised models. The device is the same, while the platform around it has to be stronger.

Closing Thoughts

Wearables in clinical trials work when the digital endpoint, the adherence model and the audit-trail architecture are designed together. They fail when one of the three is bolted on late. The good news is that all three are now well-understood problems with practical patterns.

The catch is that those patterns need a commitment at protocol-design time, months before the first patient reaches a site. For sponsors and platforms scoping a wearable-backed study, structured support to digitise the clinical study is the fastest way to align endpoint, device, protocol and audit infrastructure before the study locks.