1 Page summary
HCS 420:
RESEARCH METHODS IN HEALTHCARE
Week 12
Data Management and Analysis
Chapter 16
Video
https://www.coursera.org/learn/data-management/lecture/njyKm/why-is-data-management-important
Data Tables
Arrangement of data with the rows corresponding to individual records (i.e. subjects, events, or transactions) and the columns corresponding to fields (i.e. attributes of the records).
Single table vs multi-table relational database?
Single table
Easily accommodated in a spreadsheet or statistical package.
Database consisting of two-dimensional tables are known as “flat file”
Multi-table relational database
One field in the row must have the subject identification number to link the measurement back to subject-specific fields.
The relationship between the table of subjects and the tables of measurements is termed one-to-many.
Data Dictionaries?
They make column definitions explicit.
A table with rows representing fields and columns for field name, data type, and field description.
It’s also refereed to as a meta data since it’s a table of information about the database itself
Data Domain?
Range of allowed values
E.g. “Sex” would require “M” or “F”
Data Entry?
A mechanism for populating the data tables
That is, entering the data
What are the different data entry approaches?
1. Keyboard Transcription
Once data has been collected on paper forms, it can be transcribed via keyboard from the paper forms into the computer tables.
Case report form (CRF) are data collection form corresponding to a specific subject
What is double data entry, and what is it’s usefulness?
It ensures the fidelity/accuracy of the transcript.
The database program compares the two values entered for each variable and presents a list of values that do not match
2. Distributed Data Entry?
When data collection occurs at multiple sites, the sites can e-mail or fax paper forms to a central location for transcription into the computer database
Has become increasingly rare
3. Electronic Data Capture?
Collection of data primarily using online forms.
In clinical trials, electronic forms are called electronic case report forms (eCRFs)
Electronic Data Capture: Advantages
- Data are keyed directly into the data tables
- The computer form can include validation checks
- Can incorporate skip logic
- Data could be entered on portable, wireless devices
4. Coded Responses vs Free Text?
Free text responses can be used if the range of possible responses in unclear. These can be subsequently used to develop coded response options
Similar to “closed-ended” vs “open-ended” questions
“Exhaustiveness” vs “mutual exclusivity”?
“Exhaustive” means all possible options are provided
“Mutually exclusive” means no two options can both be correct
5. Importing measurements and laboratory results?
Many medical equipment (e.g. x-ray absorptiometry scanners and Holter monitors) are already in digital electronic format.
Data could be imported directly into the study database to avoid the labor and potential transcription errors involved in data re-entering.
6. Data management software?
Back end vs front end?
Back end: data tables
Front end or interface: online forms used for entering, viewing, and editing the data
Examples of statistical packages?
SAS, SPSS, STATA, and REDCap
SAS: Have data entry modules
REDCap (Research Electronic Data Capture): Web-based research data collection system
SPSS Video
https://www.youtube.com/watch?v=irwGg45ZX4c
Data Extraction
Also mean “queries”
It involves organizing, sorting, filtering, and viewing (“query”) of the data that has been entered in the database
Data Analysis
May require creating new, derived variables based on the raw field values in the data set.
Continuous variables may be made dichotomous, new categories created, etc.
Confidentiality and security
HIPAA?
HIPPA (10): Regardless of whether the subjects are also patients, the investigator is obligated both ethically and legally to protect their confidentiality
Database system should audit all data entry and editing
It must be backed up regularly and stored off-site
At the end, the original data, data dictionary, final database, and the study analyses should be archived