constructing a data management system
DESCRIPTION
Constructing a Data Management System. Pam Kennedy Analyst, McKing Consulting. Regional Training Workshop on Influenza Data Management Phnom Penh, Cambodia July 27 – August 2, 201 3. National Center for Immunization & Respiratory Diseases. Influenza Division. Course Objectives. - PowerPoint PPT PresentationTRANSCRIPT
Constructing a Data Management System
National Center for Immunization & Respiratory Diseases
Influenza Division
Regional Training Workshop on Influenza Data Management
Phnom Penh, Cambodia
July 27 – August 2, 2013
Pam KennedyAnalyst, McKing Consulting
Course Objectives
• Database Management System (DBMS)
What is it?
Essential functions
• Data collection forms
• Considerations in building a DBMS
Structure design
Data quality and control
Database Management System
• Definition
“…set of programs that enables you to store, modify, and extract information from a database,…
… provides users with tools to add, delete, access, modify, and analyze data stored in one location …
….provide the method for maintaining the integrity of stored data, running security … and recovering information if the system fails.”
• Basic database functionsDate Entry Storage
Modification Extraction
Searching Analysis
http://en.wikipedia.org/wiki/Database_management_system
Considerations in Building a DBMSEssential Functions
• How will the data be used?
Understand the study objective Types of data needed Data relationships Capture data collected from the questionnaire
and study forms
Understand the data flow
Understand what output is visualized
Ask questions – no assumptions
Data Collection Forms
• Use data collection forms as the basis of the electronic database
Identify all collection forms Understand the form sequence Understand each question and desired output
Yes/No Date field Data lists Eliminate redundant or unneeded information
Define interdependent information – Date of Birth vs. Date Of Hospital Admission
Gender vs. Pregnant
Date of Hospital Admission vs. Date of Hospital Discharge
Data Collection Forms
• Identify Data Rules
Identify variables that can be skipped – if any
If ‘Male’ then skip questions on pregnancy
Decide on variable options
Drop down lists
Yes/No fields
Option fields
Decide how to treat missing information
Not available vs. Unknown vs. Not applicable
Considerations in Building a DBMSStructure Design
• To increase effectiveness a good DBMS should have the following control functions enforced
Data access & relational functions Security
Control access rights Enforce data integrity
Relationship functions Data accuracy review process
Database salvage functions Backup and restore functions
Considerations in Building a DBMS Structure Design
• Questions to ask during design
How much data will be collected and stored?
How will data be analyzed?
Will year to year comparisons be conducted?
Will more than one person need access to data at
same time?
Where will backup data be stored?
What is Data Quality (DQ)?
• Aspects of data quality include: Accuracy
Date of birth expressed in day/months/years and not only years
Completeness Missing information
Update status Timeliness
Relevance Data relevant for the purpose of the activity
Consistency across data sources Data collection form to data management system
Reliability Recorded temperature or respiratory rates within
acceptable ranges
• Methods to ensure data quality include: Data validity checks Review procedures Limited access to enter and edit data once entered in
system• Documentation of changes/edits to system data
• Error log
• Standard operating procedures (SOPs) can aid in ensuring quality of data collected
• Data quality cannot be “fixed” one time and then left alone Will revert to poor quality if not controlled Issues will change over time
What is Data Quality (DQ)? (cont)
• Quality Control Strategy Steps Determine parameters (data) to be controlled Establish criticality and whether control is needed before (data
entry), during (data storage) or after results (reporting) are produced
Establish a specification which provides limits of acceptability – For example - range of acceptable temperatures (x to x)
Produce plans for control Specify how to achieve data quality, variation detection and removal Install a ‘validation check’ at an appropriate point in the process Collect and transmit data to location for analysis Verify the results and diagnose causes of variance Propose remedies and decide on the action needed
How to develop a Data Quality (DQ) Strategy?
http://www.transition-support.com/Quality_control.htm
• Identify possible sources of poor data quality Data capture and entry procedures Data collection tools Poor or lack of training Equipment calibration Data transfer from form to computer/site to site
• Identify the responsible person(s) Data source - surveillance and laboratory sites Data transfer/entry level
Data Quality (DQ) Actions
• Develop methods to address data quality issues Review of CIF/Lab results by a second reviewer to
check for missing information, etc. Identification of data “errors” at data entry level (missing
field, data inconsistency) Procedure to query source (sentinel site/laboratory)
to correct data “errors” identified (missing field, data inconsistency)
Random check of records Refer back to data sources (e.g. CIF/Lab report) to
correct errors originated at data entry level Double data entry
Data Quality (DQ) Actions
• Standard operating procedures are a systematic way of collecting, managing and storing data
• Standard operating procedures (SOPs) should include: Review and documentation of entire data collection
system Identification of people/team responsible for DQ Definition of roles and responsibilities for all data
collection personnel Methods to identify and address data quality
problems
Data Quality (DQ) Standard Operating Procedures
Validity Check (Example)
When you enter an invalid value, an error message prompts you to correct before allowing you to move to next item
• Validity checks help identify errors at data entry
Double Data Entry identify errors at data entry levelData Entry Check (Example)
Data Quality Verification• Suggested indicators that can be used for
surveillance systems Completeness
% of patients enrolled over total screened that meet the case definition in use (screening and enrollment logs) Example:
Total screened = 1000 # of patients enrolled = 1100 % enrolled = 110%
% of enrolled patients with CIF
% of enrolled patients with laboratory results
% of available CIF fully completed
% of completeness for key variables
Data Quality Verification (cont)
Timeliness % of CIF sent to central level within a defined time period % of Specimens sent to central laboratory within a defined
time period % of CIF entered in the database from reception within a
defined time period % of laboratory results available from reception within a
defined time period Example:
Total lab results = 1000 # lab results available within 7 days = 200 % available within 7 days = 20%
% of laboratory results sent to site from testing within a defined time period
Remember!!!• Understand the data and why you are collecting
• Collection forms should collect data you will use
• Define data rules and variable options• Document process and ensure everyone is
aware and understands process• Develop SOPs
• Data quality problems can occur at many points in the data collection process• To control data quality, you must control it at many
different points• If not controlled, data may become inaccurate and
begin to hinder its usefulness
• Questions???
For more information please contact Centers for Disease Control and Prevention1600 Clifton Road NE, Atlanta, GA 30333Telephone, 1-800-CDC-INFO (232-4636)/TTY: 1-888-232-6348E-mail: [email protected] Web: www.cdc.gov
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
THANK YOU
National Center for Immunization & Respiratory Diseases
Influenza Division