Table of contents
At a glance
How do you validate Veeva CDMS for your clinical data process?
Identify the Veeva applications, study versions, roles and data handovers in use. Assess supplier evidence and test the study-specific configuration, including failures. Reconcile all data after partial failures, control activation of changes and preserve a traceable review. A successful deployment or API status demonstrates neither business completeness nor sufficient investigator control. Test the release, licence and enabled functions actually used.
- Veeva EDC, DQS, and eCOA have different responsibilities and activation thresholds.
- For certain API calls, some may succeed while others remain unchanged.
- Deployment, active site version, and amendment of existing Casebooks are separate validation checkpoints.
- Signature capability and technically sufficient data review are not equivalent.
1. Describe the CDMS as a concrete clinical data process
Veeva CDMS is a collective term for several applications within Veeva Clinical Data. Veeva EDC supports study design, data entry, and review. DQS aggregates data from multiple sources; for example, eCOA captures questionnaire responses. The common vendor does not make these functions into an automatically fully licensed or validated end-to-end process.
This guide looks at clinical trials of medicinal products for human use in the EU-GCP context. For this purpose, the EMA guideline on computerised systems of 2023 and ICH E6(R3), Principles and Annex 1, which have been effective in the EU since 23 July 2025. Annex 2 will only take effect on 15 January 2027. A GMP system guide or a flat-rate FDA statement does not replace this classification.
Describe what the trial site collects, which data arrive from external sources and what decisions use them. Veeva Safety is a separate application area. An interface for safety information does not establish complete safety case processing or medical assessment within EDC.
2. Release, data model, and activated functions to be recorded
The public documentation specifies release dates of 31 July and 7 August 2026 for 26R2. With this release, CDB is renamed to DQS; DQS requires a separate license. Therefore, verify the actual contractual and functional scope when older project documents still reference CDB or generally CDMS.
Some changes are activated in phases. This applies in particular to EDC Audit 2.0 and Live Capture. Therefore, a Vault with release name 26R2 does not need to utilize every described function at the time of release. In addition to tracking release and maintenance status, also document the data model, relevant feature activations, environments, and API versions.
Veeva releases both major updates and ongoing maintenance patches. The publication of a planned or newly released patch does not constitute proof that your Vault currently includes it. Maintain an accurate record of the actual customer version and associate the relevant changes with the affected workflows. A bug fix may necessitate a reevaluation, even if the name of the major release remains unchanged.
3. Plan the data flow with specific outcomes and responsibilities assigned
The following internal work matrix describes a limited scope of application: EDC captures study data, integration updates selected events, and a controlled review assesses completeness. DQS is included only in the scope of evidence when it is actually implemented. Each row requires a designated responsible party and accessible documentation.
A small team benefits from agreeing the response to a failure in advance. Who distinguishes an incorrect mapping from an unauthorised source data change? Who may correct it, who involves the trial site and who decides whether processing can continue? Resolve these questions before the first production transfer.
| Area | Clear demarcation | Expected evidence |
|---|---|---|
| EDC Study Setup | Forms, events, rules, and Casebook version | Approved study specification aligns with the activated version. |
| External data source | Fixed source and authority for event modifications | Origin, participant linkage, and authorized modifications are traceable. |
| EDC API | Identified endpoint, version, and live capture status | Internal results are evaluated and reconciled against the target balance. |
| DQS, if used | Own license, data sources and transformation rules | Source reconciliation and significance of derived data are verified |
| Site Activation | Released conditions and active version | New Casebooks use the intended system version. |
| Review and Closure | Roles, Blind spots, Confirmation and required export | Decision and data context remain traceable |
4. Deployment and amendment of existing Casebooks differ
Veeva documents that the active version of a site determines the subsequently newly created case books. This does not imply that each existing participant record has already been adjusted through this single step. Plan existing and new case books separately and describe the handling of previously entered data.
Test the initial configuration, then deployment and the study version in use. Evidence of technical transfer is distinct from approval of a new protocol rule. Under ICH E6(R3), activate study-specific changes at individual trial sites only after the relevant approvals are in place.
The deployment status also requires a precise interpretation. Veeva distinguishes between Completed and Completed with Documentation Errors. In the latter case, the technical deployment was successful, while the generation of SDS or comparison report was compromised by errors. Explicitly assess missing documentation and remaining gaps in evidence. A green technical step does not constitute a justified decision to proceed with usage.
Sources: [7] [6] [11] [9] [12]
5. Internal API results instead of evaluating only the overall status
The EDC-API reference distinguishes between external and internal response statuses. External success can only indicate that a request was processed, while individual actions may have failed. Therefore, integration must evaluate the results within the response and unambiguously assign each result to the requested changes.
During active Audit 2.0 Live Capture, Veeva specifies for certain Bulk endpoints, including Update Event, success, or rejection per participant. If an object fails for a participant, their changes remain unsaved; changes from other participants may persist. This behavior applies across API versions under the specified activation condition. It is not a general rule applicable to all endpoints.
Establish a standalone control measure specifying an expected change inventory comprising: participants, event, starting value, target value, and processing status. A subsequent call must not blindly reprocess the entire original delivery. Prior to doing so, verify what has been stored and identify the underlying root cause of any errors. The manufacturer's indications of "Data Not Saved" must not be interpreted as an automatic retry rule for technical or scientific validation failures.
6. Example: Failure scenario involving four stored and two rolled-back changes
The example was not executed against a Veeva system. Prerequisites include active Live Capture and a relevant update event invocation. Three synthetic participants each have two events. Initially, events E1 and E2 record date values of 01.09.2026 and 08.09.2026, respectively, across all participants. The authorized change instruction specifies dates of 02.09.2026 and 09.09.2026.
For S02, a deliberate error is introduced: E2 receives the invalid date 31.09.2026. The intended acceptance response is a corresponding error response from the specifically tested endpoint. Both changes for S02 must be rolled back. Whether the actual error response exactly matches this expectation must be demonstrated by the product test. The scheduled dates do not justify a medical visit plan.
| Participants / Step | Requested Change | Expected target inventory and assessment |
|---|---|---|
| S01, first request | E1: 01.→02.09.; E2: 08.→09.09. | Both changes saved: 2 of 2 |
| S02, first call | E1: 01.→02.09.; E2: 08.→31.09. | Errors for S02: both remain at the start date; 0 of 2 saved |
| S03, first call | E1: 01.→02.09.; E2: 08.→09.09. | Both changes saved: 2 of 2 |
| First call reconciliation | 6 requested changes | 2 + 0 + 2 = 4 stored; 2 unchanged; editing remains open |
| Correction S02 | Clear the cause; correct E2 to 09.09.; send the 2 failed changes again | On successful completion, a total of 4 + 2 = 6 expected changes in the target. |
| Final check | Compare all 6 events based on participant and event identification | No missing change; no unintended reexecution of the 4 successful actions |
7. Check roles and blinding in the technical evidence as well
User permissions should arise from the study tasks and data requirements. An integration account must have a clearly defined purpose and restricted data access. The ability to technically modify events does not establish whether a data change at the inspection site is professionally authorized. Clarify these authorizations with the relevant stakeholders.
In addition, assess potential information pathways outside the visible form. An export, an API response, or an audit trail entry may contain information that a blind user must not receive. This also includes references that could indirectly lead to associations. Test the specifically used roles and output pathways using appropriate synthetic data.
Veeva defines separate authorizations for restricted data. Use these specifications as the basis for actual configuration review. The internal negative test should demonstrate that an unauthorized role does not obtain the protected information either through the intended interface or via the process-utilized outputs.
8. Review, open queries, and signatures to be evaluated in a cohesive and integrated manner
Veeva EDC supports signatures at the form, event, or case book level. Appropriate definitions and permissions are required for this functionality. Certain changes may invalidate signatures. At Data Model 2, the activated option "Enable Sign with Open Queries" allows signatures to be issued even when open queries are present. Therefore, a general statement that any open query prevents all signatures would be incorrect.
Define the meaning in the process of your approval and verify the actual data included. A technically signable state does not indicate that open questions relevant to the specific decision have been adequately resolved. Furthermore, a collective signature alone does not substantiate a proper prior review.
The review must align with the study, the risk profile, and the timing. Identify relevant data modifications and audit trail events that require separate evaluation. Following a change, it must be assessed whether the previous approval still holds validity or whether a new evaluation is required. The specific signature configuration is only one component of the overall technical control framework.
9. Operation, closure, and specific operational costs for SMEs to be clearly defined and prepared
For small sponsor and CRO teams, significant effort often arises from unclear handovers: a data manager expects corrections from IT, while IT awaits subject-matter approval. Therefore, establish escalation paths, information requirements, and decision-making authority for interface errors. The open status must remain visible, even when a transfer task is technically completed.
Supplier documentation can provide appropriate standard evidence. Complementary, you will need verified study rules, mappings, roles, relevant outputs, and operational workflows. Extensive custom logic, multiple data sources, or frequent amendments increase the need for coordination. Early clarification of these points enables a more robust resource planning compared to a generic estimation based solely on user count.
Plan the study closure during setup: What data and metadata are required, who receives them, in what form do they remain interpretable, and how is access secured? A table export may suffice for one type of analysis but be inadequate for another. Retention, contract end dates, and future readability are determined on a data set basis, not by a uniform retention period applicable to all information.
10. Checklist for demonstrable Veeva data flow
This checklist links configuration, processing and the resulting decision. Use it before introducing a new data flow and after changes that could affect what its output means. For each open item, the owner should be able to explain what evidence is missing and how to proceed in the meantime.
In the example, a complete closure after the first call is not permissible: two changes remain open. The root cause is the intentionally invalid value for S02; the example does not describe a defect in Veeva CDMS and is not a completed product test. Only after the corrected processing followed by a full reconciliation can this work item be closed. The study release and subsequent professional review remain separate decisions.
- Are EDC, DQS and other applications together with licenses explicitly defined and scoped?
- Is release, maintenance status, data model, and feature activations confirmed?
- Do the protocol specifications, study configuration, and data sources used align?
- Are new and existing Casebooks considered separately during changes?
- Are necessary site authorizations and activation conditions documented?
- Are internal API results evaluated and do failed changes remain visible?
- Have the root cause, correction, and targeted re-run been evaluated?
- Do role-based access restrictions and blinding also apply to logs, metadata and exports?
- Are signature scope, open queries, and technical review aligned?
- Are finalization data retained with relevant context for the required period?
Project experience and collaboration
Trace study changes through to the active data version
A technically successful transfer of a study change does not establish that it took effect everywhere intended. With clinical owners, we trace one case from configuration to the affected data, including partial failures and reconciliation of missing or unchanged entries. A single handover may need only a focused investigation. A new study configuration also needs planning for roles, activation and investigator oversight. We assess supplier evidence for applicable coverage first.
Frequently asked questions about Veeva CDMS
Are Veeva CDMS, EDC, and DQS the same?
CDMS is frequently used as a general term. The current vendor overview distinguishes Veeva EDC for study design, data entry, and review, as well as DQS for the consolidation and processing of data from multiple sources. CDB has been designated as DQS since version 26R2 and requires a separate license. Therefore, in the justification, describe the actually deployed applications, functions, and data transfer pathways rather than referencing a general suite name.
Does 26R2 prove that Live Capture is active in our Vault?
No. Veeva describes Audit 2.0 and Live Capture as phased deployment. In addition to the release designation, you require the actual activation status. This can affect the behavior of affected bulk calls. Maintain a record of the test and production environment configurations and evaluate differences. An older API version does not automatically prevent the described change when Live Capture is active.
Why might a successful API response still be incomplete?
The reference distinguishes between external and internal status indicators. An external success does not necessarily imply that every action has been recorded. Therefore, evaluate the results of individual entries and compare them against the requested and actually existing target inventory. In the example provided, after the first access, four out of six changes are recorded; two remain explicitly open.
Sources: [4]
Can we simply repeat the entire call after a partial error?
Check before retrying. A retry could repeat actions that already succeeded. Identify which changes were saved and which failed, investigate the cause and choose a suitable restart point. If the underlying data fails a business validation rule, correct the erroneous input. The vendor’s recommendation for Data Not Saved does not apply to every type of error and does not guarantee idempotence.
Does a new active site version update all existing Casebooks?
The documentation states that newly created casebooks shall use the assigned active version. Existing casebooks require individual evaluation of their amendments. Specify which participants and prior data are affected. Review the transition between legacy and new data; do not equate a technical site setting with a study amendment that has already been fully implemented.
How do we handle Completed with Documentation Errors during Deployment?
Veeva thus demonstrates a successful technical implementation in cases of errors in SDS or comparison report generation. Investigate which documentation is missing and whether the intended evidence is thereby incomplete. The current status does not permit automatic professional release. Document the evaluation, correction, and, if applicable, restrictions before relying on the amended study status based on these documents.
Are open queries always a technical signature lock?
This depends on the level, data model, and activated configuration. The Veeva guide provides an option for signing with open queries in Data Model 2. Therefore, the professional justification for your approval must be separately defined. Review which data is included in the signature, which questions remain unresolved, and how relevant future changes may impact the signature or required review.
Does a limited user interface suffice for blinding?
It is only part of the control. Consider also used exports, API responses, error messages and relevant metadata. Otherwise, a role outside the form could detect an unauthorized assignment. Check the specific output paths with synthetic data and document the expected access restrictions. The appropriate design depends on the study design and the actual tasks.
Do we need to re-test all manufacturer functions ourselves?
No, not generically. Evaluate existing supplier documentation with respect to version, intended use, and scope of coverage. The internal validation effort should focus on remaining gaps and your specific study configuration, integrations, and workflows. This may involve targeted confirmation of existing evidence as well as additional testing. Document this decision; a general supplier package does not replace the evaluation of specific usage scenarios.
What preparation is most beneficial for a small team?
Prepare a defined data pathway including source, mapping, study version, roles, and expected outcomes. Assign responsibility holders for technical and subject-matter inquiries. A fully completed failure scenario with partial error and retry enables early visibility of missing decisions. Subsequently, expand with additional risks and data routes. From this, the effort can be clearly justified without fixed time or cost commitments.
Sources and review date
Source date: 6 October 2026. Vendor reference: Veeva Clinical Data 26R2, EDC API 26.2, and current General Release Support. Live Capture, data model, and licensing scope are to be confirmed on a customer-specific basis. The scheduled maintenance outages announced for 7 October are not to be considered as already installed. EU-GCP serves as the regulatory reference framework. Tables and figures are internal target guidelines and do not reflect results from actual product testing.
- Veeva: Clinical Data Applications Overview
- Veeva: About the 26R2 Release
- Veeva: What's New in 26R2
- Veeva EDC API Reference 26.2
- Veeva: Providing Signatures
- Veeva: Deploying a Study
- Veeva: Managing Sites
- Veeva: 26R2 Maintenance Releases
- EMA: Guideline on computerised systems and electronic data in clinical trials
- EMA: ICH E6 Good Clinical Practice – Status
- ICH E6(R3): Principles and Annex1 – EU Step5
- Veeva: Managing Casebook Versions & Amendments
Sources checked: .
Manage Veeva data flows with clear validation evidence
We help align manufacturer records, study configurations, and interface controls with actual usage.
- Cloud and SaaS validation →
Maintain the validated state of cloud and SaaS systems as suppliers issue new versions. We define the boundary between your responsibilities and the supplier’s and agree how each release will be assessed.

