SAS® Master Data Management Provides One Version of the Truth
9/13/2016
Thomas P. Donia, R.Ph.
VP, Corporate Medical Informatics
AmeriHealth Caritas
AmeriHealth Caritas Company Overview
Mission-driven organization with over 30 years of experience serving low-income and chronically ill populations
Serve more than 5.4 million Medicaid, Medicare and CHIP members in 19 States and the District of Columbia:
Full-risk Medicaid
Dual-eligible products (MMP’s/D-SNP’s)
Behavioral Health Services
Pharmacy Benefit Management
Backed by Independence Blue Cross and Blue Cross Blue Shield of Michigan
Originally founded by the Sisters of Mercy in the ER of Misericordia Hospital in 1980 as Mercy Health Plan
1992 Mercy became the first MA plan to gain full NCQA accreditation
ACFC holds 2 of the 8 NCQA Multicultural Health Care (MHC) Distinctions nationally
7 plans hold NCQA Commendable Accreditation
The Organizational Growth – Complexity Paradox
Growth since 2013: Full-risk Medicaid:
Louisiana
Florida
District of Columbia
Iowa
Michigan
MMP’s: Michigan
South Carolina
“Every successful company’s growth contains seeds of failure. At some point, organizational complexity could outweigh the business benefits. It may be almost impossible at that stage to unbundle complexity and simplify the system, due to the nature of the interdependence of all the components that that complexity has generated.”
Leandro Herrero
Architect of Organizations
Growth-induced Complexities
6 Different FACETS claims payment instances
Visual Cactus credentialing system with no interface to FACETS:
No profiling or data standardization of delegated provider rosters received from health systems
6 “Enterprise Data Warehouses” each using different reference data values
No time or budget during implementations for:
Master Data Management
Data Governance
Reference Data Management
Net Result of Rapid Growth
• Non-standardized critical demographic data elements:
• Practitioner types
• Specialties
• Board Certifications
• Mailing addresses
• Undeliverable / returned mail:
• Notifications
• Benefit and policy changes
• Payments
• Print and online Provider Directories wrought with duplication and data inconsistencies
• Provider Analytics, network adequacy reports problematic
• A plethora of short term fixes that created long-term maintenance challenges
Enterprise Challenge
“Transform AmeriHealth Caritas into a data-driven organization with world-class Advanced
Analytics and predictive modeling capabilities.”
Pathway to Advanced Analytics
Enterprise Information Management
Standard Business Intelligence
(what occurred?)
Data Exploration
(why did it occur?)
Forecasting
(what will happen in the future?)
Predictive Modeling / Advanced Analytics
(why will it happen in the future?)
Without Enterprise Information Management the accuracy and reliability of everything that follows the analytic pathway is in question
Data GovernanceMaster Data Management
Semantic Data Management
Global
Financials
Data
Warehouse
load
load publish
publish
Master Data
Registry Hub
Source A
Source B
Data
Stewards
govern
govern
govern
MASTER DATA MANAGEMENT (MDM) is the practice of cleansing, rationalizing and integrating data into an enterprise-wide “system of record” for core business activities. MDM is a critical component of an Enterprise Information Management strategy.
Business information problems MDM solves include:
Data Duplication
Lack of Standardization
Process Disharmony
Difficulty in Integrating Similar Datafrom Disparate Sources
Master Data Management
Conceptual MDM Model / Data Flow
MDM Goal:
Demonstrate that the SAS MDM product suite supports ACFC requirements for Data Quality and Enterprise Master Data Management.
Scope: Multi-Iteration Approach (2 Iterations focused on Provider Data attributes)
Iteration #1 – 60 Days (Start Date 9/15/2014)
Individual Practitioners
Iteration #2 – 90 days (start date 1/1/2015)
Groups and Facilities
177 Attributes covering Demographics, Location, Specialty and Identifiers
Data profiling and standardization
Standardize Reference Data
Removal of Duplicates
USPS Address Verification, Cleansing and Enrichment
NPI Remediation using NPPES
Trust Rules to address data value mismatches between sources
MDM Scope
MDM Implementation Timelines
11
Immediate Benefits Realized
Considerable reduction in undelivered/returned mail
Print and On-line Provider Directories are best-in-class
Repurposed 7 FTE’s from manual data manipulation for directories to data remediation
teams
Provider-based analytics turn-around time has been significantly reduced, while accuracy
is no longer in question
Advanced Analytics and predictive models are based on a single source of truth, thus are
more accurate and reliable
Estimated 3 year ROI on Provider MDM is approximately $15M
Profiling and Standardizing Data
Profile of Provider City & Zip Code Data in PA
CITY COLUMN PROFILE
INVALIDZIP CODEPATTERNS
ZIP CODE COLUMN PROFILE
INVALID ZIP CODEDATA SAMPLE
“PHILADELPHIA” DATA SAMPLE
NOSTANDARD
City Standardization in SAS MDM
REF_SRC_SYS REF_SRC_LOV_CODE REF_STD_LOV_CODE REF_STD_LOV_LONG_DESCFCTS DO DO DOCTOR OF OSTEOPATHY
VC D.O DO DOCTOR OF OSTEOPATHY
VC D.O. DO DOCTOR OF OSTEOPATHY
VC DO DO DOCTOR OF OSTEOPATHY
VC M. D. MD MEDICAL DOCTOR
VC M.D MD MEDICAL DOCTOR
VC M.D. MD MEDICAL DOCTOR
VC MD. MD MEDICAL DOCTOR
VC MD MD MEDICAL DOCTOR
VC Md MD MEDICAL DOCTOR
VC MD, MD MEDICAL DOCTOR
VC md MD MEDICAL DOCTOR
VC CNM CNM CERTIFIED NURSE MIDWIFE
VC CNMW CNM CERTIFIED NURSE MIDWIFE
VC NMW CNM CERTIFIED NURSE MIDWIFE
Profiling and Standardizing Reference Data
MDM standardized 163 Practitioner titles in FACETS /VC down to 92 standard values
Profiling and Standardizing Reference Data
Amount of Values
Decreased in MDM due to
Standardization
Parse Name Suffix in Sources & MDM
Facets Application does not provide a Name Suffix field. Suffix is included within the Last Name field.
Visual Cactus Application provides a Name Suffix but not consistently being utilized.
Finding Duplication Within Source Systems
NPI Duplicate or Practitioner Duplicate in Facets
34 NPI’s Assigned to Different Practitioners 67 Practitioners with Duplicate Set-up
*Clear Duplicates with Multiple Common Practitioner ID’s in Facets*NPI’s MUST be UNIQUE per Practitioner
Identifying Disconnects Across Systems
Mismatches in Sources – NPI & Name
NPI Mismatches – Trusted Source NPPES * 264 Instances of NPI Mismatches
*NPI Inconsistency Between Sources *Fixed in MDM using NPPES NPI
Trust rules establish NPPES as the only trusted source for NPI numbers. NPPES supersedes all sources.
Address Standardization and EnrichmentLocation Domain
Provider Location Analysis & MDM Results
Location Source Data Location MDM Data
Locations – Facets
Locations – Visual Cactus
MDM Unique Location
* All Address Types (Practice, Mail, Remit)
* All Address Types
MDM Verified Addresses
Location Comparison
* 20,691 Duplicates Merged/Removed in MDM
* Facets & Visual Cactus
FACETS SOURCE CountNumber of Locations 41,238
Number of Locations - PAR 15,729
Number of Locations - NON PAR 31,859
VC SOURCE CountNumber of locations 16,509
Location Duplicate Breakdown CountDuplicate Locations in FACETS 12,868
Duplicate Locations in VC 7,823
MDM CountNumber of Unique locations 31,127
Number of unique Locations FACETS(in MDM) 28,370
Number of unique locations VC (in MDM) 8,686
Unique
Location Count
USPS Verified
Count
USPS
Unverified
Count
GEO Code
Enriched
Count
Delivery Pt
Confirmed
31,127 27,907 3,220 31,001 27,078
ADDRESS VERIFICATION & CLEANSING RESULTS
Address Anomalies for Same Practitioner
Source - Address Anomalies
MDM – Address Cleansed, Merged & Enriched Address Enrichment
Source System Postal Code Missing Postal Code Added Postal Code Corrected in MDM
(Value provided was incorrect)
FACETS 197 126 19,558
VISUAL CACTUS 65 24 1,276
TOTAL 262 150 20,834
POSTAL CODE ENRICHMENT
Source System City Value Missing City Value Added City Corrected in MDM (Value
provided was incorrect)
FACETS 82 22 3,350
VISUAL CACTUS 37 6 532
TOTAL 119 28 3,882
CITY VALUE ENRICHMENT
SAS MDM Address Enrichment
AmeriHealth SAS MDM Screen Shots
Data Management Console Screen
28
Total Open Issues – 1,089
Identified Issues by ISSUE TYPE
Data Management Console Screen
Closed Issues
Open Issues
Using SAS Visual Analytics with SAS MDMfor Real-time Remediation Metrics
Visual Analytics Remediation Summary
Visual Analytics New Issues Summary
Visual Analytics Issues by Type
Benefits of SAS MDM
Best-in-class Provider Directories
Sample Discrepancy with Provider Directory (pre-MDM)
Snapshot of the
Online Provider Directory
Duplication
NPI First Name Last Name City State Address Line 1 Provider ID Specialty1316998412 FREIDA FISHER Reading PA 145 N 6TH ST 1ST FL 30184937 FP
1316998412 FREIDA FISHER Reading PA 145 N 6TH ST 2ND FL 20092634 CRNP
1316998412 FREIDA FISHER Reading PA 145 N 6TH ST 2ND FL 30148179 NP
1578678884 VICTORIA STELLA Reading PA 145 N 6TH ST 30017869 GP
1578678884 VICTORIA STELLA Reading PA 145 N 6TH ST 1ST FL 30184809 FP
1578678884 VICTORIA STELLA Reading PA 145 N 6TH ST 2ND FL 1006260 FP
1578678884 VICTORIA STELLA Reading PA 145 N 6TH ST 2ND FL 30148183 FP
Duplicates Removed Post-MDM
Online Provider Directory Search MDM Web Service Search
Online Provider Directory Search MDM Web Service Search
Duplicates Removed Post-MDM
Net Results
Mastered 177 provider data elements
5 entities = practitioners, groups, facilities, locations, organizations
Hierarchies established for health systems, to organizations, to groups, to providers
Standardized: Board certifications
Specialties
Titles, degrees
Name Prefix / Suffix
Addresses
Established 1 record for each entity with links to all contributing records in all source systems
Why SAS MDM?
1) Most flexible, configurable and affordable solution we reviewed
2) Phased approach enabled us to get critical pieces in production and then scale
3) Aggressive implementation was supported with staff augmentation by SAS team
4) “No party model” approach best fit our data and vision for MDM use
5) Role-based data remediation screens
6) Built-in Visual Analytics reporting engine
7) Knowledge transfer to FTE’s for in-house ownership was painless
Q & A