Risk-Based Quality Management: Focusing on the Data that Matters Most

Over the past several decades, clinical data management has evolved from paper-based workflows to Electronic Data Capture (EDC) systems and an expanding ecosystem of digital data sources. Along with that shift, the volume and complexity of clinical trial data have grown significantly.

However, teams are still often expected to review more data with the same level of intensity. And that model is becoming increasingly difficult to sustain.

At this year’s SCDM Annual Conference, a panel on risk-based data management highlighted a shift that is becoming increasingly important across sponsors, CROs, and technology providers:

“Not all data should be treated equally.”

Risk-Based Quality Management (RBQM) has become an increasingly important focus across clinical research, with data management playing a critical role in how risk-based principles are put into practice. The industry is focused on maintaining the same high standard across a growing volume of data by focusing time, expertise, and review intensity where they have the greatest impact on patient safety, study integrity, and submission readiness.

Why Traditional Data Review Models Are Breaking Down

Modern clinical trials generate more data from more sources than ever before.

Sponsors may be managing data across:

• EDC
• eCOA/ePRO
• labs
• imaging
• IRT
• wearable devices
• site systems
• vendor platforms
• operational data sources

As data volumes grow exponentially, treating every field, every record, and every discrepancy with the same level of scrutiny creates several problems. It not only increases query volume, but also creates unnecessary burden for sites and consumes sponsor and CRO resources.

And most importantly, it can distract teams from the data that actually matters most. If every data element is considered critical, then in practice, nothing is truly prioritized.

Proportionality is the Core Principle

Risk-based data management depends on proportionality, meaning data review effort should reflect the actual importance and risk associated with the data. Primary and secondary endpoint data may require a much higher level of scrutiny than information that has little or no impact on the study’s conclusions.

Of course, that does not mean non-critical data should be ignored, but it means teams should clearly define:

• which data are critical
• what level of review each category requires
• when escalation is necessary
• what conditions should trigger deeper investigation

This creates a more intentional review model instead of one driven by habit.

Site Burden Matters, Too

Another important point from the discussion was the impact of excessive querying on sites.

Sites may receive hundreds or even thousands of queries, often without clear prioritization. It’s important to consider this from the site’s perspective. When a site receives a low-value query next to a clinically important one, both compete for attention, often creating unnecessary operational burden and making it harder for sites to focus on the issues that matter most.

A better model is to prioritize queries based on risk, severity, and relevance before they reach the site. AI may eventually play a role here by helping assess discrepancies, classify query importance, and surface the highest-value issues first, but the goal should not be to generate more queries faster.

The goal should be to generate fewer, higher quality queries.

Technology Alone Will Not Fix the Process

As AI is getting all the spotlight across the industry, it’s important to keep in mind that technology cannot compensate for a poorly designed workflow.

If a sponsor takes an inefficient process and simply adds AI on top of it, the likely result is a faster version of the same inefficient process. The better approach is to ask:

• What outcome are we trying to achieve?
• Which data are truly critical?
• Which activities add value?
• Which activities exist only because “this is how we have always done it”?
• Where can technology meaningfully reduce manual effort?

Only after those questions are answered should teams decide which tools to deploy.

Integrated Data Review Requires Collaboration Across Teams

Risk-based data management also challenges traditional functional silos. Data management, clinical operations, medical monitoring, and site management often review different parts of the same study through different systems and workflows.

There has been a growing shift toward more integrated delivery models, where teams:

• work from shared data
• review issues earlier
• coordinate actions across functions
• focus on study-level outcomes instead of function-specific metrics

The ultimate goal is to support better study execution, and when teams can identify emerging issues earlier, they can intervene before those issues become larger operational or data quality problems.

AI Has a Role, but It Should Not Define the Strategy

More and more biopharma teams this year are piloting AI as leadership wants an AI strategy, but technology often gets introduced before the underlying process has been redesigned.

That is backwards.

AI can certainly help with query prioritization, discrepancy detection, pattern recognition, risk identification, cross-system review, and automated execution of predefined data checks.

However, AI should support a clear operating model, not replace one.

The most valuable AI systems will likely be those that help teams apply human judgment more efficiently, rather than simply automate everything indiscriminately.

What Risk-Based Data Management Could Look Like by 2030

If the industry is still treating all data with equal intensity by 2030, risk-based data management will have failed.

The future will likely involve more automated data review, fewer low-value queries, stronger prioritization of critical data, more integrated roles across data management and clinical operations, broader use of AI as an execution and decision-support layer, and an earlier intervention based on operational and data signals.

At the same time, the role of the data manager may transcend what it is today and become broader. In addition to data cleaning, future data management teams may increasingly coordinate risk, quality, operations, and automation across the study lifecycle.

Focus is the Real Opportunity

The industry has spent decades building systems capable of collecting enormous amounts of clinical data.

The next challenge is learning how to focus attention where it creates the most value. That means defining what is critical, reducing unnecessary site burden, redesigning workflows, connecting data across systems, and using technology to support good judgment. No matter how good technology becomes, it can never replace good judgment.

As trial complexity continues to increase, successful risk-based data management will not be about doing less, but about being more intentional to know which data matters the most so that teams can act on it quickly.