Driving Knowledge Generation

What Does It Take?

James E. Tcheng, MD Professor of Medicine, Professor of

Community and Family Medicine (Informatics) IOM Digital Learning Collaborative

April 16th, 2013

2

Medical Therapies Proven to Reduce Death Reduction in deaths: Therapy # pts Relative Absolute C/E MI: Aspirin 18,773 23% 2.4% +++++ Fibrinolytics 58,000 18% 1.8% ++++ Beta blocker 28,970 13% 1.3% ++++ ACE inhibitor 101,000 6.5% .6% + 2nd prev: Aspirin 54,360 15% 1.2% +++++ Beta blocker 20,312 21% 2.1% ++++ Statins 17,617 23% 2.7% ++++ ACE inhibitor 9,297 17% 1.9% ++++ CHF: ACE inhibitor 7,105 23% 6.1% +++++ Beta blocker 12,385 26% 4% +++++ Spironolactone 1,663 30% 11% +++++

3

The Cycle of Quality: Generating Evidence to Inform Policy

Califf RM et al, Health Affairs, 2007

Measurement and

Education

Early Translational

Steps

Clinical Trials

Clinical Practice

Guidelines Performance

Measures

Outcomes

Discovery Science

Data Standards Network

Information

Empirical Ethics

Priorities and Processes

Inclusiveness

Use for Feedback

on Priorities

Conflict-of-interest Management

Evaluation of Speed and Fluency

Pay for Performance

Transparency to Consumers

FDA Critical Path

NIH Roadmap 1 2 3

4

5

6

7

8

9 10

11

12

4

Goals for CRUSADE Registry Improve Adherence to ACC/AHA Guidelines for Patients with Unstable Angina/Non-STEMI

Acute Therapies Aspirin

Clopidogrel

Beta Blocker

Heparin (UFH or LMWH)

Early Cath

GP IIb-IIIa Inhibitor All receiving cath/PCI

Discharge Therapies Aspirin

Clopidogrel

Beta Blocker

ACE Inhibitor

Statin/Lipid Lowering

Smoking Cessation

Cardiac Rehabilitation

Circulation, JACC 2002 — ACC/AHA Guidelines update

5

Goals for CRUSADE Registry Improve Adherence to ACC/AHA Guidelines for Patients with Unstable Angina/Non-STEMI

Acute Therapies Aspirin

Clopidogrel

Beta Blocker

Heparin (UFH or LMWH)

Early Cath

GP IIb-IIIa Inhibitor All receiving cath/PCI

Discharge Therapies Aspirin

Clopidogrel

Beta Blocker

ACE Inhibitor

Statin/Lipid Lowering

Smoking Cessation

Cardiac Rehabilitation

Circulation, JACC 2002 — ACC/AHA Guidelines update

Evaluating the Process of Care

An adherence score is applied to each patient, incorporating the components of process of care.

The score from each patient then combined for all patients at each hospital. Typical scores ranged from 50 to 95%.

All 400 hospital adherence scores then ranked in quartiles — best to worst.

6 Peterson et al, ACC 2004

Link Between Overall ACC/AHA Guidelines Adherence and Mortality

Every 10% ↑ in guidelines adherence → 11% ↓ in mortality

Electronic Health Records are the Obvious Transforming Factor

• Record every health transaction and use the information to improve quality, service and decision making

• Every major business in the US already does this using sophisticated decision support, and increasingly RCTs

• Seems simple

Duke’s Vision of the Data Mart

A partnership between Duke Medicine and the Durham community that seeks to improve the health status of Durham County residents.

In 2009, DHI funded 10 planning teams to

find ways to reduce death or disability from specific diseases or disorders prevalent in the community.

Sample Health Application: Preventing Lead Exposure

DM patients, no HbA1C (3434, 24%) DM patients, w/ HbA1C (10,811, 76%)

HbA1C < 7 (5817, 54%) 7 < HbA1C < 9 (3279, 30%)

HbA1C > 9 (1715, 16%)

Missing data

monthly data extract + risk algorithms

Lower intensity Higher intensity Intervention Spectrum

Medical-Social Risk Phenotypes

Spatially-enabled data architecture and analytics (who, what, where)

COMMUNITY PARTNERSHIP ZONE CLINICAL CARE

Accountability: real-time monitoring and evaluation (e.g., weight, HbA1c, vision, CVD, cancer, nutrition, nephropathy, neuropathy, physical

activity, self-care/management, health system trust)

Decision support systems

Feed

back

loop

Feed

back

loop

14

Driving Knowledge Generation: Where Are the Gaps?

Indications and

Problems

Application and

Prescription

Performance and

Outcomes Learning and

Education

Clinician Behavior

Integrated at “enterprise level”

Disease Registries—Granular, Detailed Primary

Care Cancer Mental Health

Cardio vascular Etc…

Health System A

Health System B

Etc…

ElectronicHealth Records

Adaptable to all!

Fundamental Informatics Infrastructure-Matrix Organizational Structure

What Is a Registry?

Data elements and definitions

Quality Assessment

Clinical Research

Quality Improvement

Clinical Data Observational

Database

Bufalino VJ et al. Circulation. 2011;123:2167-2179

That was then...

Launched 1998 1 Registry Focused on quality measurement / support local QI

This is now... 7 National Programs More than 2,500 hospitals and 1000 practices

Health plans, state and government adoptions >12 societal partners

>150 publications

FDA uses NCDR data for post market surveillance

This is our future…

Integrated programs

Point of care tool(s) to support needs of clinicians/hospitals (e.g. quality reporting)

EHR Integration / Structured reporting/ Meaningful use

Support MOC / MOL / LLL

International collaborations

Platform for clinical trials and CER; and FDA / CMS : post market, coverage with evidence

Maintenance of

Certification

Performance Improvement -

CME Value Based Purchasing

Physician Quality

Reporting System

Benchmarking • Appropriate Use

Criteria • Guideline

Adherence • Hospital and

Physician Level Performance Measures

What if…?

19

THE ESSENTIAL ISSUE The “Tower of Babel” of data from databases,

literature, and clinical trials: without controlled vocabulary and data standards we are lost!

Technical • EHR consistency

– Interfaces, standardized data, data access • Controlled terminologies

– >350 biomedical ontologies (NCBO) – Still lacking depth, specificity

• Workflow integration, usability – Lack of best practice

• Health Information Technology – Does not improve (physician) efficiency

The Impediments …

Policy / Culture • High value targets only?

– Data requires resources – What about needles in the haystack?

• Linking patient data – Registry contracts, HIPAA

• Clinician culture – Speed over data (pen, dictation) – Richness of language – E&M coding / remuneration / incentives

• Patient engagement?

The Impediments …