The review method
Audit Trail Review as a Service (ATRaaS) follows a question-led, protocol-aware method for clinical studies. QA4Tech Experts establish what the study expects and prepare the evidence without losing its meaning, then review, investigate, and decide, assisted by technology and AI.
Perspective
Each audit trail records what changed in one system, when, and under which account. A review has to establish what that means for the study: which participant and assessment were involved, what the protocol expected at that point, and whether the evidence can answer the review question. A combined dataset helps, but integration alone does not establish that records mean the same thing, that an apparent discrepancy is real, or that an export is complete. The method is designed so that each question is asked explicitly, answered from evidence, and recorded — including when the honest answer is that the evidence cannot say.
Practical obstacles
These obstacles are not peculiar to one system or vendor. Naming them at the start is what separates a review that concludes something from one that produces a very large spreadsheet.
Principles
Source records keep their original fields, values, codes, and definitions. A common format is useful only when it preserves the differences that matter, so events from different systems are treated as equivalent only where a documented mapping supports it.
Review begins with study understanding and continues through direct examination, investigation, and reporting. It includes questions Experts initiate themselves as well as signals generated by automated checks.
The output explains what was available, what was reviewed, and what could not be established. Missing evidence and uncertain chronology remain visible rather than being resolved by assumption.
Engagement steps
The sequence is scaled to the agreed scope. An evidence readiness assessment concentrates on the first three steps and ends with a decision to proceed, revise the scope, or stop.
QA4Tech and the client agree the study, sources, period, priorities, responsibilities, and the scope of the review report. Together they establish what the study team, CRO, or vendors already review and where a cross-system perspective could add to it, and agree the routes for questions, urgent escalation, and follow-up.
Experts examine the applicable protocol versions, amendments, and supporting documents. They identify critical data, milestones, and permitted pathways, and establish the actual system configuration and assessment context. Each expectation the review will test is recorded against the document version it comes from. Specialist input is agreed where needed.
Experts assess the exports against their stated scope and establish early what is missing. Original records are preserved and connected into a consistent study view through documented, versioned rules. Identity and time are reconciled only where the evidence supports it; gaps and unresolved mappings are recorded, never treated as passed checks.
Experts prioritize significant changes and deletions, and protocol-critical data such as primary efficacy and safety data. They follow the participant journey as recorded in the evidence received and pursue their own questions. Automated checks raise signals for Expert assessment. Experts test alternative explanations, request further information, and decide whether the evidence supports a finding, supports no finding, or leaves the question unresolved.
Experts prepare the review report, separating observations, interpretation, and conclusions. It states coverage, findings with their evidence-linked case records, legitimate explanations, unresolved questions, recommended follow-up, and what the review does not establish. It also records the evidence baseline and the mapping and method versions, so each conclusion can be traced to the records and versions it rests on.
Evidence flow
Experts establish context, assess the evidence, and direct analysis at every stage. Automated checks — and targeted AI, where it is used — return their results to an Expert, never directly to a conclusion.
Synthetic example
The audit trail establishes that a participant’s answer changed, which account changed it, and when the system recorded it. It does not establish whether the change matters, whether it was appropriate, or how it relates to the rest of the study.
Synthetic illustration, not drawn from a real study. A legitimate explanation or an inconclusive assessment can be the correct outcome. Clinical, safety, or instrument questions go to an appropriate specialist through the agreed escalation route; ATRaaS does not assume the investigator’s clinical decision-making or the sponsor’s safety responsibilities.
Terms
These distinctions hold from the first automated check to the report, so an apparent anomaly does not become an unsupported claim.
A potential irregularity requiring evaluation; not automatically a confirmed finding. No signal becomes a finding without an Expert’s assessment.
An issue supported by investigated evidence and a documented Expert assessment. A finding can start from a signal or from evidence that no check flagged.
An Expert assessment that the evidence does not support an issue. Like every assessment, it requires a written rationale.
An Expert assessment that the evidence received neither supports nor rules out an issue. It is reported as unresolved, with the limitation that prevented a conclusion.
The outcome when a check’s prerequisite evidence is missing — for example, a milestone absent from the export. It is reported as a question that could not be assessed.
What systems, records, periods, event types, and evidence were actually available and reviewed.
Review informed by the applicable protocol version, amendments, study context, and critical data/process definitions.
The Expert’s role
Experts set priorities, examine evidence directly, initiate investigations, weigh alternative explanations, and decide what belongs in the report. They can challenge an automated result or pursue a concern that produced no signal. Nothing is cleared because a check did not flag it.
Technology and AI
Explicit expectations, such as ordering and timing rules, suit deterministic checks; some contextual comparisons can benefit from targeted AI. Neither establishes the significance of a record or the completeness of the evidence. That remains Expert work.
Evidence limits
Where one of these limits applies, the report says so, rather than letting a conclusion reach further than the evidence.
Evidence principles
Each conclusion is traceable to the source records it rests on, through a method another reviewer can follow.
Exports are kept as received, separately from anything derived from them, and every record read from an export is given a recorded disposition, including those set aside as out of scope.
Events in the combined study view link back to the records and fields they were built from, so the source behind a displayed value or time can be shown.
Links between participants, visits, forms, and items across systems need a documented basis. Ambiguous matches stay visible, and different kinds of timestamp are not sorted together without qualification.
Where a time, identity, or code meaning cannot be established, it is recorded as unresolved rather than guessed. A bare date is not shown as midnight; an event with no defensible time is listed, not dropped.
Transformation rules are defined for each study’s own system configuration and versioned. A corrected mapping creates a new version rather than overwriting the old one, and the review report records the versions it relied on.
The same product can be configured differently in two studies, so each study’s systems are interpreted from that study’s configuration and documents. Familiarity with a product is not knowledge of a study’s set-up.
Start a conversation
Describe the question that matters and where the evidence for it is held. A short conversation establishes whether an evidence readiness assessment, the usual first step, or a pilot review is the right place to begin.