Data Collection & Analysis Flashcards

What must usually happen before a clinical trial database is locked?

Answer: All outstanding data queries must be resolved and data cleaning completed

Database lock happens only after cleaning is finished and queries are closed, so the data are final for analysis.

An EDC system flags an entered age of 250 years right away. What kind of feature is this?

Answer: Range (edit) check

Range checks are automated edit checks that flag values outside plausible limits when they are entered.

During analysis, an RA notices an extreme outlier weight value. What is the appropriate first action?

Answer: Verify the value against the source document

Outliers should be checked against the source first, because they may be real data or transcription errors, and any exclusion must follow the prespecified plan.

Which standardized dictionary is commonly used to code adverse events in clinical trials?

Answer: MedDRA

MedDRA is the standard medical terminology for coding adverse events for regulatory reporting.

Which document defines each variable's name, format, allowed values, and coding (for example, 1 = Male, 2 = Female)?

Answer: Data dictionary (codebook)

A data dictionary records variable definitions and coding so data are entered and interpreted consistently.

Under the HIPAA Safe Harbor method, how many types of identifiers must be removed for data to be considered de-identified?

Answer: 18

Safe Harbor requires removing 18 listed identifiers, such as names, most dates, and geographic units smaller than a state.

Which statistic is commonly used to measure agreement between two raters classifying the same categorical outcomes, corrected for chance?

Answer: Cohen's kappa

Cohen's kappa measures inter-rater agreement for categorical ratings and adjusts for agreement expected by chance.