The Parts of Source Data Verification That Still Need a Human Eye

In digital and decentralised trials, remote monitoring has become more common. But while tools have improved and automation helps catch many data issues, there are still parts of source data verification that rely on human judgement.

Some aspects of quality control cannot be fully automated. A smart form can flag missing fields or unexpected values. But only a trained reviewer can interpret what those values mean in context or spot subtle inconsistencies.

‍

What Technology Does Well

Systems that support eSource or direct data capture bring clear benefits:

• They can highlight incomplete forms
• They track timestamps and user actions
• They allow for faster and more frequent review
• They standardise terminology and logic across sites

These functions help monitors identify gaps and trigger follow-up more quickly than in traditional workflows.

‍

Where Human Review Still Matters

Even the best technology cannot replace:

• Contextual interpretation. For example, when a blood pressure reading is technically within range but does not match the participant’s history.
• Narrative review. Free-text descriptions of adverse events or protocol deviations often contain important nuance that automation misses.
• Source linkage. Making sense of how data from different systems align, especially when integration is imperfect.
• Identifying behavioural patterns. A site that always completes forms just before a monitoring deadline might look fine in the system but signal workflow issues worth exploring.

‍

Grey Areas Are Where Risk Hides

Many issues in data quality live in the space between clearly right and clearly wrong. These are the places where judgement is needed:

• When is a missing value a data error and when is it a genuine skip?
• Should a form be returned for correction or accepted with comment?
• Is a cluster of similar adverse events worth escalation or just routine?

These decisions rely on people who understand the protocol, the data model, and the study’s operational realities.

‍

Making Space for the Human Layer

Remote systems should not aim to remove the human layer. They should aim to support it by:

• Highlighting where attention is needed
• Reducing time spent on routine checks
• Giving reviewers clearer access to relevant history and context

In doing so, they make space for more meaningful work and better oversight.

‍