How are subjects randomly selected for targeted source data verification?

If your study only requires partial source data verification, then you might opt for creating a partial source data verification strategy, which would guide Clinical Research Associates (CRAs) to verify the data of certain subjects and specifically-marked questions at a site, by a study designer.

When you choose to create a strategy for partial source data verification, you have to specify the exact number of initial subjects whose data a CRA must verify. You also have to specify the percentage of remaining subjects that a CRA can verify after they've performed source data verification for the initial subjects.

After the initial subjects are verified, the percentage of the remaining subjects is randomly selected and they have to be verified according to the study's design and the source data verification strategy.

Note:

Once a subject has been selected for SDV and the strategy changes, thus reducing the number of subjects requiring SDV, the subject can no longer be deselected. The new strategy will only apply to newly entered subjects.

Let's consider the following example:

A CRA must verify 3 initial subjects at a site and an additional 25% of the remaining subjects at a site. The percentage (25%) is a whole number, greater than or equal zero.

The random selection will proceed as follows, using a 25% example selection to illustrate.

The system strives to match and maintain the percentage of subjects to be randomly selected. When a subject is added, if that subject would cause the number of randomly selected subjects to fall below the specified 25% threshold, then the system randomly selects a subject from the set of subjects that aren't already selected. In doing so, the set of subjects considered for SDV only includes subjects counted since the last selected subject.

For example, consider a study having reached the 25% limit of remaining subjects with subjects 1 through 4, and subject 1 number being selected. Then, as subject 5 is added, another subject is randomly selected from subjects 2 through 5. Because subject 1 was the last selected subject, the system begins selecting subjects added after that subject 1.