The Purpose of Data Cleaning and Preparation

Cleaning focuses on missing values, invalid codes, impossible dates, duplicates, inconsistencies, and other findings that may require review or correction.

Preparation may include selecting variables, standardizing names, converting dates, reshaping repeated measures, joining datasets, creating derived variables, and producing summaries.

Recoding 1 and 0 into Yes and No is a preparation step. Changing an incorrect consent date after source review is a correction. The workflow should document which is which.

When clinical data need correction, the correction should normally occur in the validated source database through the query and audit trail process, then be reflected in a new export.