Programmed range and logic checks are designed to assess the plausibility of data within the database, not its accuracy against original source documents. The most probable reason for a discrepancy discovered by an auditor during Source Data Verification (SDV) is a transcription error made during data entry. An incorrect value may have been abstracted from the source but still fall within the pre-defined acceptable ranges and logical rules, thereby passing all automated system checks. The primary purpose of manual SDV is to detect exactly these types of plausible but inaccurate data entries by comparing the database directly against the source.

A. While human error is possible, assuming auditor error is less likely than a data entry error at the site, which SDV is specifically designed to detect.

B. This describes the condition that allowed the error to go undetected by automated checks, but it is not the root cause of the discrepancy itself.

C. Post-check data changes are possible but less common than initial transcription errors, and a compliant system's audit trail should capture such modifications.

1. Society for Clinical Data Management (SCDM). (2018). Good Clinical Data Management Practices (GCDMP).

Section 5.6.2 (Edit Check Programming): This section details how edit checks are used to identify "missing data, and inconsistent or illogical data values," focusing on plausibility within the dataset.

Section 7.3 (Source Data Verification): This section defines SDV as the process to "ensure that the data recorded in the CRF/eCRF are accurate, complete, and verifiable from source documents." This distinction highlights that edit checks and SDV find different types of errors.

2. Krishnankutty, B., Bellary, S., Kumar, N. B. R., & Moodahadu, L. S. (2012). Data management in clinical research: An overview. Indian journal of pharmacology, 44(2), 168–172.

Page 170, "Data validation": The article explains that validation plans include checks for data type, valid values, range checks, and consistency. It notes these are automated processes.

Page 170, "Source data verification": It is explicitly stated that SDV is performed to "check the accuracy of the data." This confirms that automated validation cannot guarantee accuracy, which is the scenario presented in the question. (https://doi.org/10.4103/0253-7613.93842)

3. Lopienski, K., & Kress, W. (2014). Clinical Data Management: A Practical Guide. Johns Hopkins University.

Chapter 4, "Data Validation": Course materials and texts from such programs consistently differentiate between programmed validation (for plausibility) and source verification (for accuracy), reinforcing that a plausible value can still be an incorrect transcription from the source.