Data integrity of GCP‑relevant computerized systems.
Systems and data carry two things at once: the protection of clinical trial participants and the reliability of trial results. ICH E6(R3) and the EMA guideline on computerised systems set the bar. We help sponsors, trial sites, CROs and software vendors put it into practice.
Process orientation over technology for its own sake.
Data integrity of GCP-relevant computerized systems is achieved when the data of a clinical trial is collected securely, kept accessible and maintained across its entire life cycle and meets the ten ALCOA++ attributes; the bar is set by ICH E6(R3) and the EMA guideline on computerised systems.
ICH E6(R3) makes data governance a shared task for sponsor and trial site. The EMA guideline spells out what inspectors expect from computerized systems and electronic data. We turn both, together with Regulation (EU) 536/2014, into processes and quality management systems that hold up.
The ISPE GAMP Good Practice Guide "Validation and Compliance of Computerized GCP Systems and Data - Good eClinical Practice" (2nd Edition, July 2024) took shape with QFINITY: Frank Henrichmann and Oliver Herrmann as Co-Leads, Jenny Gebhardt and Marcus Schwabedissen on the author team. Oliver Herrmann led the ISPE GAMP R&D and Clinical Systems Special Interest Group (SIG).
The system landscape of clinical research.
GCP system landscapes are heterogeneous, networked and used around the globe. ICH E6(R3) groups the capture tools as Data Acquisition Tools. Data that originates electronically at the source counts as eSource.
The EMA guideline expressly covers the use of AI in clinical trials, for example in recruitment, coding or query processes. AI-specific requirements may follow in a future annex. What an AI-specific framework can look like is shown on the GMP side by the Annex 22 draft.
Who controls the device the data is created on?
Trial data increasingly originates on devices owned by neither the sponsor nor the site: for instance participants' own smartphones, tablets or smartwatches. The EMA guideline gives Bring Your Own Device its own section within Annex 5, because this route moves control to a place where the sponsor cannot exercise it. The sponsor remains responsible for the integrity of the data all the same. That control gap belongs in the risk assessment and helps define the scope of validation. Since June 2026, ICH E6(R3) with its finalized Annex 2 has also covered decentralized elements, digital health technologies and real-world data. The annex takes effect in the EU on 15 January 2027.
The requirements for clinician reported outcomes match those for ePRO systems; here the data is captured by the investigator, site staff or an independent assessor. Access rights must preserve blinding.
Sponsor and trial site carry the data together.
The heart of the new GCP rules: data governance is shared responsibility across the entire data life cycle. The sponsor never controls the data alone, and whoever delegates remains accountable.
The site's own systems are part of this. At site selection, the sponsor assesses whether they suit this trial, down to the audit trail of the medical record; the guideline calls this site qualification. If the assessment is negative, the site selection has to be reconsidered, and a system used nonetheless requires effective mitigating actions or a documented residual-risk assessment.
Together with you, QFINITY defines the responsibilities, control points and contract content for how sponsor, trial site and service providers work together, and checks these agreements in vendor and CRO audits, from the eClinical platform to the SaaS provider.
Complex platforms, validated with focus.
Rather than documenting every module wholesale, we trace the data flow through the eClinical platform from capture through the interfaces to analysis. ICH E6(R3) calls for risk-based proportionality. The ISPE GAMP eClinical Good Practice Guide describes the methodology for it, as proven practice rather than a regulation. Effort follows intended use, data criticality and the impact on the protection of trial participants and the reliability of results.
Two case studies show how this works: the implementation of an investigator notification system and the replacement of a drug safety solution covering pharmacovigilance and E2B. The basis is set out in our book chapter on validating computer systems in clinical trials.
The path to a validated system landscape.
Risk-based and process-oriented: that is how even a complex eClinical environment is validated with proportionate effort and kept in the validated state through decommissioning.
- 1
Process and data flow analysis
Map the GCP-relevant processes and systems, chart the data flow from source data to analysis including interfaces and data criticality.
- 2
Service provider assessment & agreements
Assess and select service providers, conclude agreements before services start: data access, audit rights, subcontractors.
- 3
Risk assessment & validation strategy
Risks to participants and data integrity set the scope and depth of validation.
- 4
End-to-end verification
Standard functionality, trial-specific configuration and interfaces are verified along the data life cycle, traceable back to the requirement.
- 5
Access & user management
Access is granted only once the protocol and the agreements are in place. It is reviewed at regular intervals and withdrawn in good time. Privileged accounts stay independent of the conduct of the trial, and shared accounts are ruled out.
- 6
Operation, changes & reviews
Change control and periodic review maintain the validated state. The review assesses infrastructure, application, configuration, deviations, security incidents and provider agreements together. Changes to eCRF or eCOA forms reference the protocol amendment and are communicated to the investigator together with their effective date. Audit trail review runs alongside as a planned, risk-based activity.
- 7
Migration & decommissioning
Migration runs as a validated process with reconciliation, archives are read-only, and decommissioning ends with a certified copy and a defined recommissioning path. Dynamic data remains available for dynamic analysis.
What do the ten ALCOA++ criteria mean in the GCP context?
ALCOA++ is the GxP convention for data integrity: the EMA guideline uses exactly these attributes, and ICH E6(R3) demands them in substance. Security and reliability come in as aspects of data governance, not as an eleventh criterion.
| Criterion | Group | Examples in the GCP context |
|---|---|---|
| Attributable | ALCOA | Every entry is attributable to the person or system that generated it. For critical data, attribution reaches down to the device. |
| Legible | ALCOA | Trial data stays legible throughout the entire retention period. Compression and encryption must be fully reversible. |
| Contemporaneous | ALCOA | Data is captured at the time of the observation. Date and time come automatically from an external time source, not from the device. |
| Original | ALCOA | Source records are preserved, as the original record (eSource, if captured electronically) or as a certified copy. Dynamic data remains available for dynamic analysis. |
| Accurate | ALCOA | Clinical data is correct, safeguarded by edit checks and source data verification. |
| Complete | + | All visits, queries, metadata and the audit trail are present. |
| Consistent | + | Definition, capture and management stay consistent across the life cycle, migration included. |
| Enduring | + | Trial data remains intact in the GCP-compliant archive throughout the retention period. |
| Available when needed | + | Data is accessible for monitoring, audit and inspection. |
| Traceable | ++ | Every change is fully traceable through the audit trail. |
Clinical data whose integrity you can demonstrate.
The benchmark is clear: the EMA guideline, anchored in EudraLex Vol. 10 and complemented by ICH E6(R3). Inspectors expect direct read-only access to systems, data and audit trails, even after decommissioning. That is what we assess your systems against, using the methodology QFINITY co-defined in the eClinical Good Practice Guide.
Frequently asked questions.
ICH E6(R3) dedicates chapter 4 to data governance, addressed to investigator and sponsor alike. It describes the data life cycle from capture through audit trails, corrections, migration and finalization before analysis to retention and destruction. Section 4.3 requires documented responsibilities, procedures, training and access control for computerized systems.
Article 58 of Regulation (EU) 536/2014 requires sponsor and investigator to archive the content of the trial master file for at least 25 years after the end of the trial. Medical files follow national law. The medium must keep the content complete and legible throughout that period and every alteration traceable. ICH E6(R3) adds retrievability and access protection.
Every computerized system that creates clinical trial data for a medicinal product or controls processes affecting participant protection and data reliability. These include the sites' electronic medical records, eCRF/EDC, eCOA and BYOD applications, wearables, laboratory and imaging systems, eTMF, eConsent, IRT, portals, remote monitoring, plus the sponsor's CTMS, pharmacovigilance databases and statistics software. AI applications fall under the general expectations.
Through validated systems with clear roles and access controls, and through records that meet ALCOA++ and are directly readable by inspectors: audit trails for creating and changing data, certified copies where originals leave the system, a complete eTMF and a planned, risk-based audit trail review. After decommissioning, data and metadata must remain promptly accessible.
The responsible party in each case: the sponsor for systems it provides, the trial site for its own. The EMA guideline allows a vendor's evidence to be used once the responsible party has assessed it as adequate; additional checks remain that party's duty. The vendor supplies product evidence. Validation in the trial process against the intended use is not transferable.
Good Clinical Practice, supported in practice.
Last updated:
Data integrity has many contexts.
Your eClinical landscape with demonstrable validation.
Together with you, we develop a risk-based end-to-end validation strategy in line with the EMA guideline and ICH E6(R3), from data flow analysis to documented verification. It starts with a free intro call about your system landscape.
Book an intro call
