iHT² Health IT Summit Denver 2013 - Pamela Peele, PhD, Chief Analytics Officer, UPMC Health Plan,...
-
Upload
health-it-conference-iht2 -
Category
Education
-
view
731 -
download
1
description
Transcript of iHT² Health IT Summit Denver 2013 - Pamela Peele, PhD, Chief Analytics Officer, UPMC Health Plan,...
Pamela Peele, Ph.D.
Chief Analytics Officer
UPMC Insurance Services
Strategies for Building a Learning Organization
1
July 24, 2013
• $11 billion integrated global health
enterprise
• 2nd largest Integrated Delivery System
• 21 hospitals operating over 4,200
licensed beds; 187,000 admissions per
year
• 4.6 million outpatient visits; 480,000
emergency visits per year
• >2 million Health Plan members
• 400 outpatient locations
• 55,000 employees
• 3,400 employed physicians
• 20,000+ contracted physicians
• 5th NIH funding
UPMC BACKGROUND
Strong Commitment to Infrastructure and Technology: UPMC’S Information Technology Investment
$1.6 Billion
over the past 5 years
Advantages
• Creates synergistic provider and payer business growth and development strategies
• Combines provider and payer expertise to drive improved outcomes
• Aligns clinical and financial incentives to create value
• Creates administrative efficiencies
Challenges
• Balancing owned vs. non-owned network
• Balancing FFS and capitated models
• Balancing insurers and Blue dominance in our market
• Managing adverse selection
Integrated Delivery and Financing System Innovation Lab
UPMC
Health
Plans
UPMC
Clinical
Enterprise
Innovation Lab
Levels of Analytics Framework
5
Standard ReportsWhat happened?
AlertsWhat actions are needed?
Query DrilldownWhat exactly is the problem?
Ad hoc ReportsHow many, how often, where?
Statistical AnalysisWhy is this happening?
OptimizationWhat’s the best that can happen?
Predictive ModelingWhat will happen next?
ForecastingWhat if these trends continue?
Degree of Intelligence
Com
petit
ive
Adv
anta
ge
From Tom Farre, “The Analytical Competitor”, in Analytics: The Art and Science of Better, ComputerWorld Technology Briefing.
UPMC HP: 2009
UPMC HP: 2006
• Data that is “fit for consumption”
• Data Governance
• Tools
• Staff with strategic plans and skills
The Basics
6
• Many types of disparate data available
• Medical Claims
• Behavioral Health Claims
• Pharmacy Claims (allows medication possession ratio MPR)
• Worker’s Compensation Claims
• Short Term Disability
• Absenteeism Data from Time Cards
• On-Site Biometric Screening Results
• Health Risk Assessments – (self-reported)
• Care Management Assessments/ Phone interaction
• Enrollment & Demographic Data
• Lab Values
Integrated Data to Support Clinical ManagementPopulation Health Strategy and Clinical Support
Identifying Health Conditions by SEPARATE Data Source
Identifying Health Conditions by AGGREGATING Data Source
1,596 1,994 2,197 2,344
4,086 5,698 5,698 6,774
4,324 6,588 6,588 7,658
982 2,715 2,715 2,715
2,200 6,366 7,597 7,597
2,738 2,738 2,738 2,738
0 1,442 5,721 6,119
132 132 8,593 8,878
11,795 16,036 21,005 21,913
Stratification Data Flow
Health PlaNET
• Database: SQL, Toad
• Statistics: SAS, STATISTICA, STATA, R
• Data Mining: STATISTICA, R
• Text Mining: STATISTICA
• Modeling & Simulation: MATLAB, Mathematica, Vensim,
GEPHI
• GIS: ArcGIS
Tools
11
• Excel
• Access
• Crystal Reports
Staff - 2006
12
Business Analyst (30)
Accounting
Current Staff
13
Clinical Program
Evaluation (5)
Epidemiology
Biostatistics
Health Services Research
Strategic Business Analysis
(6)
Finance
Economics
Policy
Statistics
Database & Data
Quality (7)
Finance
Economics
Policy
Statistics
Modeling (3)
Physics
Mathematics
Biomedical Engineering
Statistics
Operations (3)
Economics
Industrial Engineering Operations
Communications
Statistics
• Industry Knowledge
• Data visualization skills
• Data ECTL (extraction, cleaning, transformation, loading) skills
• Statistics
• Health Services Research
• Data Mining
• Financial modeling & evaluation
• Presentation, writing, and communication skills
• Formally trained but NOT blinded by their training
– Challenge deeply held beliefs
Staff Skills and Backgrounds
14
• Predictive modeling
• Clinical program evaluation
• Financial modeling
• Practice variation
• Text mining
• Visualizations, Linkages
What you can do with your groomed data
15
0.990.880.770.660.550.440.330.22
500
400
300
200
100
0
Probability
Fre
qu
en
cy
0.70.5
Distr ibution of Probability for R eadmiss ion
FY09 Acute Inpatient Discharges (A ll LOB)
n = 38,840
Most impactable opportunity to
prevent readmission
Discharge Advocate: Risk Models Identify Readmission “Sweet Spot”
16
UPMC Project RED In Brief
• Before program, at discharge, patients lacked competency in their own conditions and care:
• 37% able to state the purpose of all their medications
• 14% knew their medication’s common side effects
• 42% able to state their diagnosis
• Readmission Model targets patients at admission most likely to be readmitted for avoidable reasons
• Not just for UPMC facilities: currently deployed at 10 sites –4 UPMC hospitals and 6 network facilities;
additional 4 UPMC and 6 network facilities launching in 2012
Lower Risk of
Readmission
Less impactable
despite high
readmission risk
Single Acute
Episodes
Early/Mid Stage
Chronic DiseaseEnd Stage
Chronic Disease
2.
Clinical Program Evaluation (Supportive Services Program)
18
• No significant change in 30 day readmit rates
• Time to readmission significantly longer by ~11 days
Clinical Program Evaluation (Supportive Services Program)
19
When they occur, readmissions cost significantly less by $4,000
0
2000
4000
6000
8000
10000
12000
9/28/2012 10/28/2012 11/28/2012 12/28/2012 1/28/2013 2/28/2013 3/31/2013
Influ
enza
Lik
e Ill
ness
Vis
its
Per
100,
000
Influenza Like Illness Epidemic Course With IBNR Adjusted Actual Costs Through January 2013 And Estimated Costs February-April 2013
SNP CHIP CMFI MC MA Pittsburgh ILI Visits
$7,908,217
$4,937,557
$5,193,713
$668,217$1,414,810
$6,690,009
$4,239,679
$3,591,971$708,044$635,277
Projected Influenza Like Illness course with IBNR-adjusted actual costs
through January 2013 and projected costs February-April 2013.
• New Medicare Enrollees
– No prior clinical or claims information
• Medicare Health Assessment Survey
– 24 questions
• What can you learn?
• Don’t return the enrollment questionnaire
– Non-returners have 22% higher annual expenditures
• 5 Questions produce 8 rules = high future expenditure
– 160% higher annual expenditures
Learning the Rules: Using Decision Tree Models
21
RuleQuestion 2
Response
Question 5
Response
Question 6
Response
Question 7
Response
Question 8
Response
Rule 1
(6.8%, N=633)
X X
Rule 2
(5.9%, N=549)
X
Rule 3
(5.7%, N=531)
X X
Rule 4
(6.5%, N=605)
X X
Rule 5
(5.9%, N=549)
X X
Rule 6
(8.6%, N=801)
X X
Rule 7
(10.5%, N=978)
X X
Rule 8
(9.3%, N=866)
X X
High Expenditure Rules
22
The percentage of members in the test set for which a given rule applies is stated below the rule.
290%
320%
290%
290%
325%
250%
275%
290%
23
Average # of Imaging Services Per Admit – CY 2008
DRG 470 – Major Joint Replacement without Major Complications & Comorbidities
Bubble size is proportional to the 30 day readmit rate
Confidence interval bars indicated by vertical extent
24
Average # of Consultation Services Per Admit – CY 2008
DRG 470 – Major Joint Replacement without Major Complications & Comorbidities
Bubble size is proportional to the 30 day readmit rate
Confidence interval bars indicated by vertical extent
Average # of Subsequent Attending Visits Following Hospital Discharge – CY 2008
DRG 470 – Major Joint Replacement without Major Complications & Comorbidities
Bubble size is proportional to the 30 day readmit rate
Confidence interval bars indicated by vertical extent
25
26
Average # of Laboratory Testing Services Per Admit – CY 2008
DRG 470 – Major Joint Replacement without Major Complications Comorbidities
Bubble size is proportional to the 30 day readmit rate
Confidence interval bars indicated by vertical extent
EMR Text Mining
27
Provider Network Plot
28
Provider Patient Sharing Patterns
29
30
• Executive team support
– Resources
• Analysis and knowledge creation
– Not an Information Technology (IT) function
– Reports outside of IT
• Institutional Wiki and Electronic Filing Cabinet
– Document, document, document
Lessons Learned
31
• Governance Structure
• IT governs data
• Analytics governs secondary data use
• Build capacity as needed, starting with the data
• Need a professionally trained analytics leader
• Centralized or decentralized?
• Hire for tomorrow
• Core analytics group needs diverse skillsets and
backgrounds
Lessons Learned
32
• Data Shopping
– Addictive
– Highly Infectious
– No known treatment once infected
– Attempts to help can make it worse
• All those one-off databases and marts
– Make something better and they go away
• Language fluency
– Matching words with meaning
Dangers
33
• Many Vendors
• Many Products
– Don’t interface easily
• Need a FLEXIBLE plan
• The wrong plan will costs the one thing you don’t have
TIME!
Dangers
34