DA2 Update for CI
1 November 2013
Outline
• Situation Analysis
• Priorities for 2013
• New Technologies
• Results
2
Situation Analysis
Current State:
Future State:
Impact:
Reporting Performance Metrics
Analyzing Performance Metrics
Efficiency, Quality, Costs
Minimal Predictive Analytics
Focus on Predictive Analytics
Quality, Costs, Population
Health
Manual Data Abstraction
Automated Data Abstraction
Efficiency, Costs
Multiple Data Silos
Enterprise Data Warehouse
(EDW)
Efficiency, Quality
No Data Governance
Robust Data Governance
Safety, Efficiency, Quality
Minimal Self-Service Capacity
User Driven Data Access
Efficiency, Quality, Costs
Limited Awareness of Analytics
Data is a System Asset
Data Quality
Poor Data Quality
High Quality Data/One Source
of Truth
Data Quality, Efficiency
Unmet Analytic Needs
High Client Satisfaction
Efficiency, Quality, Costs
3
DA2 Priorities for 2013
1. Performance Metrics: Develop reliable and actionable
leading indicators for clinical and non-clinical operational
performance metrics
2. Quality and Value: Demonstrate CHS clinical quality,
value and improved patient outcomes to payers and
employers
3. Population Health: Provide predictive analytics using full
range of data available to positively impact patient health
by proactive delivery of preventative services.
4
Technology to Support Priorities
Accelerators for Rapid Value Creation:
Software
Description
Verisk
Allows capability to conduct population analysis,
employer group analysis, practice pattern
variation, episode analysis on claims data
Humedica
Allows capability to conduct population analysis
and provide medical home reporting using clinical
(EMR) data. Can compare, analyze and identify
clinical best practices near real time.
Predixion
Removes reliance on technical analysts to
accelerate predictive analytics capability. Point of
care predictive analytics through workflow.
5
Building an Analytic Competency
Service Lines
and
Functional Units
Requests
Client Services
Requests
Estimates
Models
Reports
Analytics
Opportunity Analysis
Demand Management
Requests
SLA
Abstraction Services
Shared Resource pool
SLA
Abstraction
CoE
Analytic Services
Shared Resource Pool
Requests
Think Tank
Leading metric development
Key capabilities
Data Services
Requests
Research
Analytics
CoE
Color Key:
Blue = Business
Maroon = DA²
Green = IT
Pink = External
Requests
Data, SLA
Services, SLA
Data Marts,
Extracts,
Automation,
SLA
Shared Resource pool
Information Architecture
Data CoE
Requests
Services, SLA
IT Infrastructure Services
6
DA² New Organization Structure
Allen Naidoo
VP
Michael Dulin
CCO
Client Services
Mike Trumbore
AVP
Melanie Spencer
Research Director
Analytic Services
John Carew
AVP
Abstraction Svcs
Angela Humphrey
AVP
Data Services
Tim Reagan
AVP
Financial Services
Director
Rodney White
Director
Marcy Neale
Director - Quality
IT EDW Team
Avery Ashby
Director
7
Organization Impact
INNOVATIVE LEADER
DA² will apply innovative techniques and tools to the analysis of health information and will challenge the status
quo uses of information and move the CHS culture forward to raise the collective CHS analytics IQ.
ADVANCED ANALYTICS AND BUSINESS INTELLIGENCE
DA² will apply a balanced approach to understanding business needs, promoting data governance and integrity,
analyzing routine and complex data, and providing relevant and targeted intelligence to solve business problems.
ELEVATE PATIENT OUTCOMES
Quality of care is the cornerstone of the CHS business model and DA² is fully aligned to support existing and new
initiatives to improve quality.
PREDICT HEALTH NEEDS
The true advancements will come from the modeling of data to predict the relationship of individuals with
disease processes so proactive interventions can be employed.
TRANSFORMATIVE SOLUTIONS
DA² will focus on projects apt to move CHS in a material and enduring direction toward a successful future state.
PROMOTE THE HEALTH OF OUR COMMUNITIES
Population health management is an advanced state comprising all of the activities of CHS and DA² to improve
the well‐being of the citizens in our services areas.
8
Results
• DA2 Acute Care Predictive Analytics
(Readmissions)
• DA2 Ambulatory Analytics
• Population Health – Population Segmentation and
Stratification
9
DA2 Predictive Analytics
Impacting Readmissions
Readmission Risk Modeling and Patient Segmentation
Readmission Risk Model
40 Key Predictors
200,000 Hospitalizations
600 Predictor Variables
Patient Segmentation
Segments
Low Risk
Medium
Risk
High Risk
Very High
Risk
Total
Volume
9 x 9 , Emergency and Medicare
3.90%
1.13%
3.42%
2.41%
10.86%
Lengthy Stays, Large Number of Orders
2.84%
2.10%
3.80%
5.77%
14.51%
New Patients, Longer Visits
3.44%
2.18%
7.32%
1.44%
14.37%
New Patients, Shorter Visits
7.94%
11.98%
3.79%
0.15%
23.85%
Older Frequent Flyers
Simpler than 9 x 9, Medicine, Discharged
Home
0.29%
0.72%
0.79%
8.57%
10.37%
1.02%
0.20%
0.50%
1.98%
3.70%
Younger and Healthy
2.71%
4.70%
2.37%
0.54%
10.32%
Younger Frequent Flyers
2.88%
1.98%
3.02%
4.15%
12.02%
25.00%
25.00%
25.00%
25.00%
100.00%
Total Volume
Care Management Analytics Application
Care Management Analytics Application
DA2 Ambulatory Analytics
Using Analytics to Improve Outcomes for
Patients with Diabetes
Capabilities
15
Capabilities Outline
Gaps in Care
• Identify patients who are out of compliance for recommended tests or visits
• Identify patients with heavy disease burden who are no being actively managed
Predictive Modeling
• Identify and stratify patients at high risk of hospitalization or disease complication
Cost & Utilization Management
• Track patients and providers with high utilization scores (ER or IP Visits within time
period)
• Monitor readmissions and post-discharge follow-up
High Risk Patient Management
• Profile poly-chronic patients
• Discover actionable clinical insights by patient cohort
Prescribing Patterns
• Identify costly drug prescribing patterns in relation to clinical outcomes
• Track compliance against medical protocols
16
Benchmarking Opportunities
Anonymous
Transparent
17
• Anonymous
• Built into Humedica MinedShare® Ambulatory tool
• CHSMG Ambulatory data compared to blinded
comparator group
• Comparator group includes all of Humedica’ s
ambulatory clients, but can be filtered based on
criteria such as region and size
• Transparent
• CHSMG is included in American Medical Group
Association (AMGA) and the Anceta Collaborative
through Humedica
• Anceta uses data to identify opportunities for
improvement and best performance
• Collaborative members share and learn from each
other
Sample Output
18
Sample Output
19
DA2 Population Health Efforts
Storyline
Current Status
Foundational Efforts Underway
Potential Care Management Activities
21
Vital Statistics
1,524,282 Unique Patients*
accessed CHS IP/OP Hospital Services and/or Medical
Group Practice Services during August 2012 – July 2013
Medical Group Practice Services
Charges
$2.641B
IP/OP Hospital Services
Charges
$10.171B
* Unique patients are determined using IBM® Initiate Master Data Service® products looking across IDX, STAR, and Cerner
22
Patient & Charged Amount Distribution
1% Patients (15,243)
$3,674,713,631 or
28.7% Total Charged Amount
Mean = $242,299
4% Patients (60,971)
$4,013,166,277 or
31.3% Total Charged Amount
Mean = $65,820
25% Patients (381,071)
$4,123,756,151
32.2% Total Charged Amount
Mean = $10,821
70% Patients (1,066,997)
$999,518,908
7.8% Total Charged Amount
Mean = $985*
23
Core Market* Vital Statistics
1,226,372 Core Market Unique Patients
accessed IP/OP Hospital Services and/or Medical Group
Practice Services during August 2012 – July 2013
80% of Unique Patients
84% of Charged Amount
1% Patients (12,264)
$3,011,776,739 or
27.9% Total Charged
Amount
Mean = $245,599
4% Patients (49,054)
$3,348,738,254 or
31.1% Total Charged
Amount
Mean = $68,265
25% Patients (306,593)
$3,576,082,446
33.2% Total Charged
Amount
Mean = $11,664
70% Patients (858,461)
$847,618,912
7.8% Total Charged
Amount (Mean = $1,038** )
** Core Market defined by patient’s zip code (Management Company); CHS standard for all planning & reporting purposes
** Does not include patients with zero total charges in the calculation
24
8 County View – Unique Patient Counts*
Anson
Cabarrus
Cleveland
Gaston
Lincoln
Mecklenburg
Stanly
Union
10,531
142,825
68,272
74,285
41,208
530,244
23,625
120,567
* Unique patients utilizing CHS IP/OP Hospital Services and/or Medical Group Practice Services during Aug 2012 – Jul 2013
25
Aug 2012 – July 2013
26
Aug 2012 – July 2013
27
FOUNDATIONAL FRAMEWORK
PCP Attribution
Indigent Identification & Modeling
DA2 Patient 360º Information
28
PCP Attribution
Provider
Level
CPT
Codes
Visit
Recency
DIVISION
Time
Period
Specialty
Volume
PROVIDER
Note: Attribution is first determined in the most recent 12 months, and then if needed, looks within the previous 12 months. Future
29
state may include variation in the time frame used, based on patient demographics/conditions.
Indigent Identification & Modeling
• Common system-wide definition
• Tagging of patients known to be indigent
• Predicting patients of becoming charity or
over due using social determinants/financial
and other information
• Micro-segmenting for outreach opportunities
to enable primary care and community
resource utilization
30
DA2 Patient 360º Information
31
POTENTIAL
INITIATIVES FOR IMPACT
DA2 Ambulatory Analytics Use Cases
DA2 Predictive Analytics Use Cases
DA2 Segmentation
32
DA2 Ambulatory Analytics Use Cases –
Diabetes
Focused Diabetic
Outreach based
on Lab Data
http://www.diabetes24-7.com
33
DA2 Predictive Analytics Use Cases –
Top 5% Readmission Risk at CMC
Statistically significant differences
between the top 5% and the
remaining 95% for:
• Insurance Type
• Race
• Age
• Admission Type
* Gender was not statistically different
Patient segments will be identified, mapped across the CHS
service area, and proper outreach developed
34
Patient Mapping
There were 183
patients with 100 in
Mecklenburg County
35
DA2 Research
Asthma CER
Table 1. Asthmatic Patients Seen in the CHS System (2008-2009)
Number of
Number of Clinic
Patients
Visits
Total Number of Unique Patients with
38,634
77,582
an Asthma Diagnosis
African American Race
14,168
34,551
Hispanic Ethnicity
2,043
4,596
37
Age < 18
11,058
21,357
Number within Mecklenburg County
16,458
41,961
Uninsured, Medicaid, or Medicare
Number of Hospitalizations for Asthma
Number Emergency Room Visits
CHS School Children with Asthma
13,564
10,321
30,121
8,500
31,022
NA
NA
NA
95 Primary Care Practices
(28,200 Asthmatic Patients and 49,692 Asthma Visits Per Year)
AHRQ Asthma
Group A
Control - Usual Care
n = 20 Practices
(5,820 Patients)
School System
(8,500 Asthmatics)
Group E
Intervention - SBC
171 Schools
(8,500 Patients)
Group B
Control - EAP
n = 65 Practices
(20,100 Patients)
Time Delay*
Group C
Intervention - EAP + IAC
n = 10 Practices at Start and 75 Practices after 18 months
(3,050 Patients at Start and 23,150 at 18 months)
Time Delay*
Group D
Intervention - EAP + IAC + SDM
n = 4 Practices
(2,580 Patients)
Evaluate Outcomes for Each Group On: Healthcare Utilization, Medication Compliance, Costs, School Attendance,
40
School Performance, Clinical Outcomes, Quality of Life, & Perceptions of Care
*Groups will be analyzed using a time delay study design
EAP - Electronic Medical Record with Decision Support, Asthma Action Plan, & Population Management Reports
IAC - Integrated Approach to Care
SBC - School Based Care for Asthma
SDM - Shared Decision Making Approach to Care
Asthma Comparative Effectiveness (ACE)
Research Database
Outcome measures for 12,581 eligible asthma patients




Service Utilization
Medication Compliance
Appropriate Care
Quality of Life (QoL)
Billing
Data
Patient
Surveys
Medicaid
Claims
ACER
Clinical
Data
Spatial
Data
School
Data
ACE Database: Data Collection
Patient-Level Measures
–
–
–
–
–
–
Insurance
QoL Score
Visit History
Demographics
Medication Orders
Comorbid Conditions
Facility-Level Measures
– Treatment Group
– Intervention Dates
Community-Level Measures
•
•
•
•
•
Crime
Poverty
Education
Housing values
Access to services
42
Shared Decision Making
Results
Time
N
Pre
Post
Chang
e
P-value
3 months (n=171)
171
18.70
%
9.40%
-9.40%
0.011
6 months (n=121)
121
31.40
%
23.10
%
-8.30%
0.114
43
AHRQ Asthma Population
*Patients with first SDM visit in 2012Q1 or earlier
Appropriate Care Measures
45