Trial machinery, without the vendor.
Investigator-initiated trials get industry-grade infrastructure at academic cost: eCRF and EDC, randomisation with concealment that holds up, e-consent, monitoring views your monitor can drive alone, a validation package sized to the study — and a CONSORT diagram that falls out of the database instead of being counted by hand.
An investigator-initiated trial fails for administrative reasons far more often than for scientific ones. The intervention was fine. The database opened nine months late.
Commercial trial infrastructure is priced and scoped for pharma. Academic alternatives are affordable and frequently under-built, which surfaces at the first monitoring visit. Those are not the only two options.
We build trials on REDCap or on a commercial EDC where the sponsor requires one, with the parts that actually get audited done properly: randomisation, blinding, audit trails, validation evidence, change control, and a documented reason for every decision.
Validation should be proportionate to the risk. A single-centre feasibility study should not carry the validation package of a phase III, and we will tell you where the line is.
Randomisation you cannot peek at.
Participants arrive, get stratified by site and disease severity, and are allocated inside permuted blocks so the arms stay balanced without the sequence becoming guessable. Switch between what the investigator sees and what the pharmacy sees — that difference is the whole point of concealment.
The machinery, sized to the study.
The database, built from the protocol
Case report forms derived from the schedule of assessments, with edit checks at entry, per-field audit trail, electronic signatures and a locking workflow at the end.
- REDCap, or a commercial EDC where the sponsor requires it
- Edit checks that catch errors at entry, before they become queries
- Form versioning with a documented effective date
Concealment that survives scrutiny
Stratified permuted blocks or minimisation, sequences generated and sealed before enrolment, and blinded roles enforced at the data layer.
- Allocation list generated, hashed and version-stamped in advance
- Blinded investigator views; unblinded pharmacy and safety views
- Emergency unblinding with an attributable, timestamped record
Version-controlled, and provable
Consent captured electronically with the exact text the participant saw, the version they signed, and a re-consent workflow when the protocol is amended.
- Witnessed, remote and proxy consent flows
- Consent status visible in the monitoring view
- Paper fallback, reconciled against the electronic record
A monitor who does not need you
Risk-based monitoring views, a query workflow with an audit trail, and source data verification lists that target the fields that matter to the endpoint.
- Central monitoring reports by site and by risk indicator
- Query raise, respond, close — with a resolution trail
- Protocol deviations logged as structured data
Adverse events, on the clock
Adverse event and serious adverse event capture with coding, causality and expectedness assessment, and the escalation timers the regulation actually specifies.
- Coded event terms reconciled against the source, in your chosen dictionary
- SUSAR workflow with reporting deadlines tracked
- Safety listings for the DSMB, generated from the database
Scaled to the study's risk
Computer system validation scaled to the study's risk: requirements traceability, installation and operational qualification, test evidence and a validation summary report.
- Every edit check and logic rule exercised once, with evidence
- Change control from go-live, with impact assessment
- Ready for an inspector without a three-month scramble
The file that keeps you inspectable
Trial master file structure, the data management plan, the data validation plan, user access records, training logs, and the version history that ties them together.
- eTMF structure aligned to the reference model
- Documents generated from the system where possible
- EU CTR submission annexes prepared with you
Pre-specified, then executed
Statistical analysis plan before unblinding, blinded data review, and the primary analysis documented and reproducible. The independent DSMB statistician is appointed by you — we prepare the data they work from.
- Sample size with the assumptions written down
- DSMB data packages and safety listings, delivered securely and on time
- CONSORT diagram and tables generated from the data
The CONSORT diagram, drawn by the data.
Every box in that diagram is a query against the trial database, so nobody transcribes a number from a spreadsheet at midnight. When a participant's status changes, the diagram changes with it.
This pays off twice. During the study, a discrepancy between the diagram and the data is how you discover a data problem. At submission, a reviewer who recomputes your numbers gets the same answer.
- Live during recruitment, showing the current flow at any moment
- Reproducible at submission from the frozen dataset version
- The same for STROBE, PRISMA and TRIPOD+AI where those apply
Four frameworks, one build.
Most academic trials are subject to fewer of these than the team fears, and more of them than the team has documented. We spend the first hour of any trial engagement establishing which of them actually apply.
| Framework | Applies when | What it means for the system |
|---|---|---|
| ICH-GCP E6(R3) | Any trial of a medicinal product, and increasingly expected of device and behavioural trials. | Audit trail per data point, role-based access with training records, risk-proportionate validation, and a documented quality-by-design rationale. |
| EU Clinical Trials Regulation | Interventional trials of medicinal products conducted in the EU. | CTIS submission structure, protocol and substantial-modification handling, transparency obligations and result reporting deadlines. |
| EU MDR / IVDR | Clinical investigations of devices or performance studies of diagnostics. | Different documentation set, device deficiency reporting alongside adverse events, and traceability to the device identifier. |
| GDPR | Always. | Lawful basis per category, pseudonymisation, DPIA, retention aligned to the trial retention obligation — see secure data handling. |
| 21 CFR Part 11 | Data intended to support a US submission. | Electronic signature controls, closed-system requirements and a validated audit trail. We have not delivered a Part 11 study — the controls are within reach on our stack, and we would scope that honestly rather than imply a track record. |
| National requirements | Everywhere, and they differ. | Danish scientific ethics committee practice (VEK and VMK), Medicines Agency expectations, and the local specifics of insurance, archiving and site agreements. We have been the technical partner on multiple international trials, from committee submission through to CTIS. |
From protocol to first randomisation.
Protocol & risk
The schedule of assessments becomes a variable list; the risk assessment decides how much validation the study actually needs.
- Endpoints and derivations
- Risk-proportionality agreed
- Regulatory scope confirmed
Plans, then build
Data management plan, data validation plan and randomisation specification written and approved before anything is configured.
- Edit checks specified
- Blinding matrix defined
- Roles and access mapped
Configure & qualify
eCRF, randomisation, consent and monitoring views built, then tested against the specification with evidence captured as output.
- Every check exercised once
- Allocation list sealed
- Validation summary written
Site training & UAT
Your coordinators and monitor use the system on synthetic participants and try to break it. Findings are logged and closed before go-live.
- Training records generated
- Site-facing guides written
- Go / no-go documented
Run, monitor, lock
Production support, change control, monitoring reports, DSMB data packages, and a clean database lock with a documented freeze.
- Named engineer on call
- Amendments handled formally
- Lock, then unblind
What an inspector will ask for.
- Audit trail
- Per-field, append-only: old value, new value, who, when, and the reason where one is required. Queryable by the monitor without our involvement, and exportable in full.
- Electronic signatures
- Bound to the record and the signer, with meaning stated (author, reviewer, approver), re-authentication at signing, and a signature manifestation visible on the form.
- Access & training
- Role-based access mapped to the delegation log; user accounts tied to named individuals with training records attached. Shared logins are never used, including by our own support staff.
- Randomisation integrity
- Sequence generated by a documented method with a recorded seed, sealed before first enrolment, and stored so that no study-team account can read it.
- Change control
- Post-go-live changes assessed for impact, tested, approved and dated. Mid-study form versions are recorded so the analysis knows which version produced each value.
- Database lock
- Documented freeze procedure, a hashed export, and a record of who approved the lock — before unblinding, in that order.
- Archiving
- Retention for the statutory period in a readable, migration-proof format, with the audit trail stored alongside the data so it survives the retirement of the original system.
- Business continuity
- For systems we operate: backups with tested restores, a documented disaster recovery plan, and an escrow or exit route so a trial in year four is not dependent on our continued existence. Where the trial runs in your institution's REDCap, continuity is theirs — we tell you what to confirm, and we confirm it with them in writing.
What sponsors and monitors ask.
Is REDCap acceptable for a regulated trial?
For a great many investigator-initiated trials, yes, and it is used that way across academic Europe. The real distinction is between the software and a qualified installation of it: a validated instance, a validated project, documented change control, and a sponsor who can produce that evidence. The instance is your institution's to qualify, because it is their licence and their server; the project-level validation is ours to produce, and a monitor will want both.
Where a commercial sponsor mandates a specific EDC, or where the trial is a pivotal registration study, we build in that system instead. We will tell you which situation you are in rather than argue REDCap into a role nobody asked it to play.
Can you be our data management provider of record?
Yes. We take named responsibility for data management in the protocol and the trial master file, provide the CVs and training records for the people involved, and are available to the monitor and the inspector directly.
We are not the sponsor. Sponsor obligations stay with your institution, and we will be explicit about which tasks are delegated to us and which are not. That boundary should be in writing before enrolment.
How do you handle emergency unblinding?
Through a defined path that is available at three in the morning without calling us, and that records itself. The unblinding account is separate, the action is logged with who, when and the stated reason, and the trial team plus the DSMB are notified automatically.
Every unblinding event appears in the study report. Trials that hide this behind a support ticket discover during their audit that they cannot prove how many times it happened.
What does validation actually cost us?
Less than the folklore suggests, if it is proportionate and done alongside the build. For a typical single-product academic trial it is a package of requirement traceability, test scripts with evidence, and a summary report, roughly a week of work.
It becomes expensive in only one situation, when it is attempted retrospectively on a system that has already been collecting data for a year. We would like to have that conversation with you before go-live.
Can you support the DSMB?
As the data management provider, yes — which is a real but bounded role: the data packages, the safety listings, the correctly blinded or unblinded extracts the committee's statistician needs, and the unglamorous part of getting all of it to members securely and on time, in a format they can read on a laptop in another country.
We do not provide the independent DSMB statistician. That person has to be independent of the team holding the trial data, and since that is us, it cannot also be us. You appoint them; we make sure they get clean, complete data on the agreed schedule.
Our trial is already running and we are worried. Now what?
A focused assessment, usually two weeks: what the system does and does not record, whether the randomisation was genuinely concealed, what the audit trail can prove, and which of the gaps are fixable forward versus which need to be documented as limitations.
You get a written report with a prioritised remediation list. It is candid, because a monitor will be equally candid later and it is cheaper to hear it from us. Most trials in this position are more recoverable than the team fears, provided the honest inventory happens now.
A trial is four of these at once.
Send the protocol before you send the application.
The cheapest hour in a trial's life is the one spent on infrastructure before the ethics submission goes in. Send us the protocol and we will come back with the variable list, the randomisation specification, the regulatory scope as we read it, and a fixed price.