How does Partial Source Data Verification (PSDV) work in RDC?

There are two parts to PSDV in RDC:

• Critical CRF Plan: Marking certain CRFs as requiring verification every time they are collected in a study.

• Patient Plan: Defining a percentage of patients randomly assigned for verification of 100% of their records.

Only users with special privileges can see which data must be source-verified and record the source data verification progress.

Can I upgrade to RDC 5.1 and start using PSDV for existing live studies?

Yes.

To start a critical CRF plan:

1. Select the SDV? check box for a Study DCI to make the corresponding CRF require SDV for all patients.

2. Create and publish the SDV plan.

To start a patient SDV plan:

1. Make some or all patients eligible for PSDV. Enter a date for SDV eligibility for patient positions manually or using batch data load. (For new studies you can also use a procedure that is triggered by data entry in a particular CRF.)

2. Create SDV default values for the initial number of patients and percentage of patients requiring SDV. Either publish patient SDV plans for all sites from Oracle Clinical or for specific sites from RDC. Adjust the values for each site as needed In RDC.

3. Schedule the Pending Patient Updates job, which runs across all studies.

You can do these steps in any order, but the algorithm that randomly assigns some patients to SDV does not run until all are done and the Pending Patient Updates job runs. The job will take some time the first time it runs in an ongoing study because it is processing many patients. See Will enabling PSDV impact performance?.

What if I don't want to enable PSDV for ongoing studies?

Do nothing and your studies continue to work the way they did before the upgrade.

Can I have separate PSDV plans for different sites?

Yes. You can create patient and/or critical CRF SDV plan versions for individual sites after initial publication of the plans based on study-level default values.

Will enabling PSDV impact performance?

The first time the Pending Patient Updates job runs in an ongoing study may take a long time because it is processing many patients. To minimize this:

• The first time you run Pending Patient Updates in the database, schedule a one-time execution during the least busy time.

• After the initial run, schedule Pending Patient Updates to run frequently.

• For subsequent studies, publish the patient SDV plan for each site separately in RDC. The next time Pending Patient Updates runs it will process all patients from that site. Testing showed the time to publish plans for up to 2000 patients was under 20 seconds.

After the initial processing and in new studies, testing has shown an insignificant impact on RDC performance.

Why is the Pending Patient Updates job needed?

Patient eligibility updates may be deferred because another process has locked the Patient Positions table. These updates are maintained in a queue and committed by the first Pending Patient Update job to run after the lock is released.

Should I always publish both a patient plan and a CRF plan?

If you need one type of plan but not the other, you may need to publish an "empty" plan for the unneeded type:

• If you are publishing a patient SDV plan—some patients require 100% SDV—always publish a critical forms SDV plan too. If there are no critical forms that require SDV for all patients, publish an "empty" forms plan in which no forms are selected. This way, an RDC search for CRFs requiring SDV retrieves only the CRFs entered for patients requiring 100% SDV. Without any critical forms plan, the system retrieves all CRFs for all patients.

• If you are publishing a critical forms SDV plan—some CRFs require 100% SDV—but no patients require 100% verification, you may or may not want to publish an "empty" patient plan.

  • If you publish an empty patient plan—in which both the initial patient count and the auto-selection rate are zero and no patients are made eligible—RDC Search panes include the option to search for Patients requiring 100% SDV, although there are none. This maintains consistency in the RDC Search pane from one site to the next.

  • If you do not publish any patient plan, RDC users will not have the option to search for patients requiring 100% SDV at any study site, but such a search would never retrieve any records anyway.

• For sites requiring 100% SDV (all CRFs for all patients), publish a patient SDV plan that specifies an auto-selection rate of 100%. There is no need to publish an accompanying forms SDV plan in this case.

Can I feed revisions to the plan from an external statistical analysis system?

This is possible through the PSDV web services. You can get payloads with information about the new initial patient count and automatic selection rates coming from an external statistical analysis system. The web service creates a new plan version with the values and can publish the new plan too if it's set up that way.

See the Oracle Clinical API Guide for information on PSDV web services at http://docs.oracle.com/cd/E52785_01/doc.51/e53565/web_services.htm#CHDJAGHB.

Where do I find all the documentation for Partial Source Data Verification support in Release 5.1?

Start with the introductory section in Oracle Clinical Creating a Study at https://docs.oracle.com/cd/E52785_01/doc.51/e52787/plan.htm#CACBEJAC. It lists all related documentation.

Download all Oracle Clinical and Remote Data Capture user manuals here: http://www.oracle.com/technetwork/documentation/hsgbu-clinical-407519.html