Transcript Data Analysis for Disease Management

Data Analysis for Disease
Management
presented by:
Michael Mina
Senior Statistical Analyst
Medical Mutual of Ohio
October 13, 1999
Presentation Overview
•
•
•
•
Brief Disease Management Overview
MBSD DM Responsibilities
Competitor DM Data Issues
Possible Enhancements to MBSD DM
Processes
• Disease Management Resources
Brief Disease Management
Overview
• Disease management is the process of
coordinating and managing members who have
chronic conditions and health management
services along the full spectrum of health care
delivery while striving to improve both clinical
and economic outcomes through altering patient
and provider behavior.
Brief Disease Management
Overview (cont’d)
• Population based - different levels of
severity
• Education, Assistance
• Help patients control their own disease
• Concerned with quality of life, not cost
alone
Brief Disease Management
Overview (cont’d)
• Insurance companies usually understaffed
for this purpose
• Disease Management programs often
vendor-based, with vendors focused on a
specific disease
Brief Disease Management
Overview (cont’d)
• MMO DM programs include:
–
–
–
–
–
Breathe Easy (Asthma, COPD)
Heart Sense (Congestive heart failure)
Transplanting Health (Organ transplant)
BabyLink (Perinatal education)
Outpatient Diabetic Education
MBSD DM Responsibilities
•
•
•
•
Identify diseases to manage
Process information for vendor usage
Invite providers to participate
Determine effectiveness of DM efforts
– Outcomes reporting
– Over/underutilization measures
Tools Used by MBSD
•
•
•
•
DB2 / QMF / SQL
SAS
Access
Oracle
DB2 / QMF / SQL
• Above terms often used interchangeably
• DB2 - Relational Database Management
System (RDBMS)
• QMF - Query Management Facility
• SQL - Structured Query Language
• Easy access to data
DB2
• Relational Database
DB2 (cont’d)
DB2 (cont’d)
DB2
• Production systems - PReview, CMS, CPIMS
• Data storage - claims, premium, dependents
• Standard RDBMS for MMO data
warehousing / data mart efforts
– Note: In the industry as a whole, outcomes
reporting is a force behind data warehousing
• DB2 used widely inside and outside MMO
- BP pay-at-pump
QMF
• Manages queries
• Some reporting and formatting capabilities
• Can create procedures (procs) containing
instructions for running multiple queries
• More effective programming
SQL
• Queries - mini computer programs
• “The most difficult area of data
warehousing is the translation of simple
business analyses into SQL”
- Ralph Kimball Ph.D.,
CEO Red Brick Systems
SQL (cont’d)
• SELECT DISTINCT SUB_ID, DEP_NBR
FROM CORP.INPAT
WHERE
YR_MO=199909
AND
PAT_AGE>=65
AND
SPEC_CD=‘C1’
ORDER
BY 1, 2
SAS
•
•
•
•
SAS is a company with many products
Very powerful statistical analysis software
MMO has about 10 SAS programmers
Goodyear has over 700 SAS programmers
and a SAS data warehouse
• SAS is used by GE Card Services with an
Oracle data warehouse
Access
• Microsoft database
• Very versatile
– Easy to create reports with downloaded
mainframe data
– Very widely used
– Small business database of choice
– But also used by NCB ($80 billion bank) for
Sales & Incentive program
Oracle
•
•
•
•
Very powerful, very widely used RDBMS
Uses PL/SQL version of SQL
MMO uses Oracle for HEDIS reporting
Used by Yahoo! and other database-driven
web sites
Identify Diseases to Manage
• Based on Population Analysis (SAS)
• Look for:
– High patient volume
– High cost
Process Information for Vendor
Usage
• Initial and subsequent targetings
• Standardized methodology based on DB2
– Select claims related to disease state
• BabyLink: by Rx; Others: medical claims by dx
– Identify distinct members from claims
– Select all claims for those members
(comorbidities)
– Get member, provider, PCP information
• Methodology documented and flowcharted
Invite Providers to Participate
• Problems with provider address quality
• DB2/QMF/SQL and Access programming
used to improve information quality
– DB2: standard in-network provider addressing
– Access: “smart” formatting of mailing labels
– “Clean” and “proper” addresses still an issue
Determine Effectiveness
- Outcomes Reporting
• Vendor outcomes reporting
- based upon enrolled population
• MMO outcomes reporting
– Days/1000, Cases/1000
– ER visits/1000, Office visits/1000
– coming soon: Readmissions/1000
Determine Effectiveness
- Outcomes Reporting (cont’d)
• MMO outcomes reporting
– disease state outcomes for patients
– overall outcomes for patients
– overall outcomes for all MMO members
Determine Effectiveness
- Outcomes Reporting (cont’d)
• Breathe Easy:
Started with SAS, ran out of space
• Met deadline with a workaround
• Change in specifications
• Tried DB2 this time, still ran out of space
Determine Effectiveness
- Outcomes Reporting (cont’d)
• Vertical partitioning
– Another way to organize data
– 85% space savings over previous process
– Process enabler
• Data mart
• DB2 + Vertical Partitioning + Data Mart +
Access = Success!
Determine Effectiveness
- Outcomes Reporting (cont’d)
• Process still under development
• New specifications easier to implement
– Disease assignment module (5 SQL queries)
– Separate BH outcomes from Medical
Determine Effectiveness
- Over/Underutilization Measures
• What is a readmission? Ask me in a month!
• Developed in DB2 using claims, not PReview
• Access reports used to test methodology
– Process: review results > modify SQL query > rerun
query > import results into Access > review results
• Can ultimately be run from data mart
Competitor DM Data Issues
- Aetna
• Managed care data warehouse since 1996
• Combines some functions of PReview,
HEDIS
• Standard reports, member mailings
• Chronic disease registry
• Data warehouse soon to have web browser
front end
Competitor DM Data Issues
- Anthem BCBS
•
•
•
•
Data warehouse since 1991
Excellence in Business Information Award
Outcomes reporting, HEDIS reporting, fraud
“We did this not just to get into a leadership
position but as a matter of survival”
- Joe Bruscato, Chief Data Warehousing
Architect, Anthem BCBS
Possible Enhancements to MBSD
DM Processes
• Currently use FFS inpatient, outpatient,
professional claims for targeting, outcomes
– exception: BabyLink
• Likely inclusions
– Rx claims (major medical, freestanding)
– encounters
• Redundancy to ensure validity
– e.g., asthma dx and one Rx and one refill
Possible Enhancements to MBSD
DM Processes (cont’d)
• Data mart schema for other DM programs
• Evaluation of Data Mining/ Business
Intelligence products
Disease Management Resources
Recap
•
•
•
•
Brief Disease Management Overview
MBSD DM Responsibilities
Competitor DM Data Issues
Possible Enhancements to MBSD DM
Processes
• Disease Management Resources
Thank you for coming!