An outline of data management in clinical research

Abstract

<p class="abstract"><span lang="EN-IN">Clinical data management (CDM) is an indispensable part of clinical research. CDM activities lead to the collection of reliable, high-quality and statistically sound data generating from the clinical trials. Several studies suggest that such data helps in extreme reduction in time from drug development processes to the marketing stage. Several practices in CDM including CRF annotation, case report form (CRF) designing, data extraction, data entry, data validation, database designing, database locking, discrepancy management and medical coding are evaluated for quality checks at regular intervals during clinical trial. In recent times, the huge gap about improvements of the CDM standards for meeting the regulatory requirements remains to be filled. Fulfilling these requirements will help the clinical trial sector to stay ahead of the game. The current article accentuates the practices followed and activities involved in CDM. Therefore, it enables the reader an outline of management of data in the clinical research.</span></p>

Overview of clinical data management and statistical analysis of bioequivalence study

“…29 In e-CRF based clinical trials, data validation processes will be running on frequent mode in order to identify the discrepancies in the entered data. 30 The discrepancies will be highlighted in the system and Data Clarification Form (DCF) is generated.…”

“…In eCRFs the site investigator has the authority to check the discrepancies notified/sent to him/her and provide the suitable clarifications online where the same will be updated in the database. 29,30 CDM team evaluates each one of the discrepancies regularly and the resolved data discrepancies are recorded as "closed". Not usually but in some cases the investigator will not be able to provide suitable answers for the discrepancy caused.…”

“…The statistical model for this study involves normal data which is not what we get in real life. 30,38,39 The raw data is highly skewed and needs some power transformation to fit the model. Log-transformation is one option as it decreases such skewness between data points resulting in a better fit data for our model.…”