Clinical Trial Innovation (3) – Risk based Monitoring and beyond

This topic is closely related to the Innovation area – Measurable Quality.   There is great concern in the industry that actions that involve the reduction or even removal of Source Data Verification will result in data quality issues that later are found to compromise the integrity of the study.

I would like to put forward 4 reasons for the slow take-up of reduced SDV today;

1). Source Data Verification takes around 30% of a typical Monitoring site visits time (on average).  Monitoring visits occur for reasons other than SDV.  Determining the portion of monitoring visits before the start of a study can be hard to predict as it is determined on site compliance.  Contracts between a CRO and a Sponsor will typically accommodate known facts. The costs will therefore be set up front based on a default monitoring approach (usually 100%). Exceptions exist.. but these are exceptions. CRO’s need to be more flexible, adjusting monitoring, and therefore costs, based on pre-defined rules.

2). Monitoring visits are often as much about ensuring site satisfaction than performing SDV.  The costs will often not reduce as the monitor will continue to visit the site – travel and durations may be the same, even if SDV is not done.

3). It is difficult to see the quality impact of full SDV versus partial SDV.   Sponsors will be penalized far more for placing a study at risk, than saving dollars.  If evidence that a study is not being impacted is not available, then many sponsor company managers will go with the safer approach. Better realtime tools (i.e. inside EDC) that show performance and compliance across sites will help give the ammunition Managers need to make brave (partial SDV) decisions.

4). Performing effective partial SDV without the capabilities in EDC can be impossible.  In Rave for example, it is torture without the Targeted SDV module. Unfortunately, the licensing, configuration and deployment of the partial SDV capabilites in EDC often occurs outwith the scope of CRO contracting – as a result it is not factored in, and the savings often not achieved.

For a minute, I am going to go with the assertion that risk based partial SDV is an entirely valid approach.

The question I would like to raise – is why stop at risk based SDV.  Can we not extend this to risk based data cleaning in general. What is so special about Source Data Verification?  Is it just the fact that real measurable costs are involved?

A good measure of quality is the number of times that a challenge (automatic or manual query) results in a change of data.   If this is high, then it indicates that the original quality was low.  In principle, this means that there is a greater likelyhood that data entry for fields that have not been reviewed or edit checked will also result in a change.   But what if a query is raised many times across a study, and never results in a change of data…. should the system take this into account in measuring cleanliness?

So – how can we make Partial SDV work?

Tools today target either the planning, or the execution. I am not aware of any EDC tool that allows Partial SDV planning as well as supporting the workflow of executing partial SDV.  A system must adapt the fields and forms that need to be SDV’d based on a). pre-defined rules and b). the ongoing performance of a site.  A system must also be able to plan out the need for Monitoring visits by being aware of the work inside and outside of EDC.

In reality – it probably just takes simplification.  For established site… only check a % of patients. For new sites… check more… and check quality is achieved.


Leave a Comment