Chicago Learning Effectiveness Advancement Research Network

Report
Building a Learning Healthcare City:
The Chicago Learning and Effectiveness Advancement Research Network
(Chicago LEARN)
David Meltzer M.D., Ph.D.
May 10, 2013
Comparative Effectiveness and PatientCentered Outcomes Research Infrastructure
• Comparative effectiveness research (CER) seeks to
understand the relative benefits of medical interventions
– Patient-centered outcomes research (PCOR) focuses on
relative benefits for patient subgroups and individuals
• Chicago LEARN (Learning Effectiveness Advancement
Research Network) objective:
– To establish state of the art infrastructure for CER and PCOR
in the inpatient setting in Chicago with the potential for
national impact through partnership with UHC (University
HealthSystem Consortium)
– Embrace concepts of a learning healthcare system
Timeline for Chicago Inpatient CER
1997
2001
2004
2006
2007
2008
2009
2009
2010
2010
2010
2010
2010
2011
2011
2012
UC Hospitalist Project started
– Interview hospitalized patients @ admission & 30 days post D/C
– ~ 90,000 subjects enrolled to date
Multicenter Hospitalist R01
– Medicare linkage
Hospitalist Scholars Program started
UC CTSA Funded – focus on personalized medicine
Center for Education and Research in Therapeutics (CERT) (UC, UIC)
NU CTSA funded
UIC CTSA funded
ARRA $1.1 Billion for CER – CONCERT RC2 (UC, UIC), NIH KM1 (UC, UIC)
ACA Passed – PCORI established with focus on patient-centered outcomes
UC CTSA External Advisory Committee encourages development of CER
Chicago Effective Health Care Conference - CER Infrastructure, Informatics
Infrastructure conversations with Computation Institute
Chicago CTSAs Meeting – Julian Solway’s charge
Chicago Effective Health Care Conference – UHC/Indiana University
CTSA CER Supplement
UC CTSA Renewal – CER Core, CMMI Innovation Challenge Award
Key Ideas for Chicago LEARN
• Observational and experimental studies
– Patterns/context of use
– Causal inference
• Large sample sizes
– Faster trials that can include subgroups (PCORI)
– Cluster randomized trials to study health service organization (IOM CER, CMMI)
• Multiple inpatient data sources
– Administrative
– Chart/EHR (Learning Health Care System Model)
– Patient interviews (PCORI)
• Collect/use biospecimens
• Follow patients after discharge
– Integrated health system (Northshore)
– Medicare data
→ Consent essential
→ Multipurpose use
– Economies of scale in subject screening and recruitment (vs. single-study RAs)
– Economies of scale in informatics infrastructure (multi-disease, multi-mission)
Chicago LEARN
(Learning Effectiveness Advancement Research Network)
• Multifunctional infrastructure for hospital-based patient-centered
outcomes research for Chicago-area AMCs (UC, Rush, NS, UIC,
NU, Loyola) with potential for national impact
– Human infrastructure (RAs) with aim to spread hospitalist project model
– Informatics infrastructure with UHC
• Clinical Data Base/Resource Manager has chargemaster-level administrative
data for >250 hospitals from >100 AMCs
• All disease categories
• All inpatient and some outpatient data
• Supported by AMC CEOs for operations/quality benchmarking/ improvement
• Invest $30 million over 3 years to add EHR data
• Local partners can help develop and demonstrate value
• Situates us to be early adopters in use of resulting data
• Progress/Funding
– Monthly meetings, UHC data transfers starting, QA/QI studies
– CTSA CER Supplement, UC CER Core, CMMI grant, NIA Anemia study
– Medicare data
Early Studies
• Pharmacogenetic testing for warfarin dosing
– EHRs to screen, Shared RAs
• Transfusion for anemia
– UHC for study design & collaboration
Table 3: Means of Hemoglobin Values*. All Diagnoses and Dx's subgroups. Full Sample
Obs
293
Mean Std. Dev.
7.64
1.64
All Diagnoses
Dx's Subgroups
Digestive system disease
87
8.18
1.87
Blood related disease
51
6.45
1.18
Infectious disease
42
7.86
1.67
Injury and poisoning
26
7.78
1.40
Circulatory system disease
20
7.36
1.26
Neoplasm
14
7.51
1.34
All the rest
53
7.78
1.34
Specific Diagnoses
Upper GIB
46
8.35
2.16
Lower GIB
17
7.94
1.71
*Lowest hemoglobin value in 24 hrs before FIRST transfusion
Min
3.50
Max Median
13.40
7.60
4.10
3.50
5.00
5.30
5.20
5.50
4.00
13.40
8.80
13.10
10.40
9.90
9.50
11.50
8.10
6.30
7.75
7.50
7.30
7.40
7.80
4.00
4.10
13.40
9.80
8.50
8.40
Physician Location and Quality/Cost of Care
• Controlling healthcare costs requires focus on high cost patients
• Most spending for high cost patients is repeated hospitalization
• Repeated hospitalization strongly affected by poor care coordination
– Worse since traditional primary care physicians replaced by hospitalists
– 10-50% lower utilization possible with care coordination by own physician
• Growth of hospitalists partially a spatial problem – transport costs
• $6.1 Million CMMI Comprehensive Care Physician study to test if
having same doctor care for patients in inpatient &
outpatient setting can improve costs/outcomes
– Locate clinic in hospital; focus on high cost patients
– 2,000 patient randomized clinical trial - $50 mil/yr
– Innovative delivery mode l for Accountable Care
Organizations (ACOs)
Urban Contextual Data to Improve Health
• Neighborhood resources and context
• Environmental data to study healthcare productivity
– Hospital length of stay (LOS) a critical policy variable
– Unclear if faster discharge increases readmissions, costs
• LOS is a behavior
– LOS = f (observables, unobservables)
• Hard to understand, study effects of LOS
– Readmission = f (LOS, observables, unobservables)
• Wish one had experimental data
– Contextual data to infer effect of increased
LOS on costs and outcomes
• Hint: Chicago winter, summer
Instrumental Variable Approach to Identification
Admit
Date
Diagnosis-driven probability of ideal discharge
1/3
1/3
1/2
1/6
Extend LOS
1/2
Extend LOS
1/6
Shorten
LOS
Urban Contextual Data to Improve Health
• Neighborhood resources and context
• Use of environmental data to learn about productivity
– Hospital length of stay (LOS) a critical policy variable
– Unclear if faster discharge increases readmissions, costs
• LOS is a behavior
– LOS = f (observables, unobservables)
• Hard to understand, study effects of LOS
– Readmission = f (LOS, observables, unobservables)
• Wish one had experimental data
– Contextual data to infer effect of increased
LOS on costs and outcomes
• Hint: Chicago winter, summer
• Use to test productivity across cities, clinical areas
Conclusions
• Academic medicine and healthcare institutions in
cities provide important opportunities for research
to improve healthcare outcomes and costs
– Chicago LEARN as model for collaboration among
urban academic medical centers and across cities
• Chicago’s rich set of institutions in healthcare
create unique opportunities for researchers and for
the city

similar documents