A QC Plan for A Quality Clinical Database

Before a clinical database can be locked, data management run a set of qualify control checklists to ensure that quality is built into the clinical database.

This process is also known as ″estimated error rate″ which is a calculation of an error occurred during the course of the clinical trial, from the data management point of view.

The formula is:

Percentage error rate = # of errors detected/total #’s of items checked x 100

Depending upon the tool you are using, i.e. OC or Clintrial, you will need to count and determine the number of items for a particular dataset. 80% of the time, SAS tool is used for this step. Other times, a manual count is required.

Not every issue or discrepancy found is considered an error. This is described in a section of the QC plan. Each Sponsor or CRO is different. For example, misspelling mistakes or changes in punctuation are not considered an error. Also, your statistical team member should be able to identified those items considered ‘critical’ for the statistical analysis.

Common panels/items are: Demography (date of birth, sex, race), lab data, randomization data, Study Conclusion/final status table.

When a clinical database is considered 80-90% clean, the data manager will randomly select a set of patients, as discussed with the statistical leader for the study, to be included in the final QC.

Sample size = √n + 1 where n=the number of patients in the study.

If a study has 34 patients and 1 study site then √34 = 5.83, 5.83 + 1= 6.83. We also determined that only 20% of the patient enrolled will be QC’d therefore, 7 case report forms will be included in the final review.

In order for a clinical data to be considered acceptable, the error rate must be below 0.1% or less.

In conclusion, a good QC plan identify what quality control tasks are needed and that all activities are completed in the most effective way through out the project.

Anayansi Gamboa has an extensive background in clinical data management as well as experience with different EDC systems including Oracle InForm, InForm Architect, Central Designer, CIS, Clintrial, Medidata Rave, Central Coding, OpenClinica Open Source and Oracle Clinical.

Disclaimer: The legal entity on this blog is registered as Doing Business As (DBA) – Trade Name – Fictitious Name – Assumed Name as “GAMBOA”.


3 thoughts on “A QC Plan for A Quality Clinical Database”

  1. As site staff enter data in the eCRF directly, a typical data management group in a CRO will not have a lot of control over how is data entered. Would this affect error rate calculation process for eCRFs compared to in-house clinical data management systems used by the DM groups in CROs?

Comments are closed.