Connecting all the dots…

David Lansky, PhDSenior Director, Health ProgramMarkle Foundation

What’s the goal of HIT?

• Reduce errors, improve quality, increaseefficiency?

• Empower consumers with information?

• Create platform for transformation?

Three layers of HIT planning

• Information to care for a patient

• Information to support better decision-making

• Information to help consumers improveown health and their care

Connecting for Health

• Broad-based, public-private collaborative of morethan 100 diverse stakeholders

• Founded and supported by Markle Foundation,with additional support from Robert WoodJohnson Foundation

• Purpose of Connecting for Health:

To catalyze changes on a national basis tocreate an interconnected, electronic healthinformation infrastructure to support betterhealth and healthcare

1. Information to care for a patient

Connecting for Health Approach

Architecture is Policy(“code is law”)

The Design Principles

• Designed to safeguard privacy—imposed therequirement first and then designed the functionalarchitecture

• This approach is harder and requires resisting “if only”thinking.

• It does not produce the easiest or simplest technicalsolutions

• You can’t build first and worry about the policieslater…

GET

SEND

Source: © 2004 The Markle Foundation Graphic adapted from Tom Benthin original.

De-IdentifiedData

ReportingRouter

MessageTransfer

FIND

Patient Index

Provider asks if thereare records forhis/her patient Index sends

location ofany records

Sourcesends indexinformation

LOCATOR

Provider asksfor and receives

records

Recordsare sent toProvider

TING

Source maypush data for

reporting

REPOR

DataSources

PatientsProviders

HospitalsDiagnosticServices

Payors

Public HealthProviders

DATA

IndividualCare

Providers

PatientsProviders

HospitalsDiagnostic

ServicesPayors

Public HealthProviders

PublicHealth

Over the last three years…

• The most challenging aspects of the agenda are “policy”related: What rules, business and organizationalrequirements govern the flow of information?– How is access to your information controlled?– Who has access to what, under what circumstances?– What protections are in place for information that is shared?– Who shares what and who bears the liability for it?– Which policies need to be uniform in a health information

exchange and which do not?

A Common Framework Is Needed

• The Common Framework is the minimumnecessary set of rules or protocols for everyonewho shares health information to follow.

• Helps organizations overcome the barrierswithout “reinventing the wheel”

• Enables nationwide interoperability…avoidingisolated islands of information

• Builds trust

Sharing Health Information = LinkingExisting Sources

• Health information can stay where it is—with thedoctors and others who created it

• Specific information is shared only when andwhere it is needed.

• Sharing does not require an all new “network” orinfrastructure

• Sharing does not require a central database or anational ID

• Sharing does require a Common Framework

Common Framework is based on a set of PolicyPrinciples

1. Openness and Transparency2. Purpose Specification and Minimization3. Collection Limitation4. Use Limitation5. Individual Participation and Control6. Data Integrity and Quality7. Security Safeguards and Controls8. Accountability and Oversight9. Remedies

Common Framework, p.4

Common Framework is based on a set ofTechnology Principles

1. Make it “Thin”2. Avoid “Rip and Replace”3. Separate Applications from the

Network4. Decentralization5. Federation6. Flexibility7. Privacy and Security8. Accuracy

Common Framework, p.5

The Connecting for Health Modelfor Health Information Sharing

• Sharing occurs via a network of networks—not acompletely new architecture

• The nationwide “network” is made up of smallercommunities or SNOs (Sub Network Organizations)

• The model relies on an RLS (Record LocatorService) to locate patient records

• Implementation and Experimentation areEssential

• Develop a policy and technical framework thatenables information sharing to happen for highquality patient care while protecting the privacy andsecurity of personal health information.

• Identify what needs to be common for interoperabilityand what does not.

• Design and develop the documentation and thematerials for communities on issues such as access,control, privacy and security.

• Share and disseminate broadly in order to continueto learn !!!

Connecting for Health Prototype Goals

Common Framework architecture

SNO SNO

SNO

ISB ISB

ISB

RLS

RLS

RLS

Implications

• Need architecture that anticipates needfor interoperability, protects privacy

• Need policies that create trust, areenforceable

• Need incentives to share data with others

2. Information to support better decisionmaking

GET

SEND

Source: © 2004 The Markle Foundation Graphic adapted from Tom Benthin original.

De-IdentifiedData

ReportingRouter

MessageTransfer

FIND

Patient Index

Provider asks if thereare records forhis/her patient Index sends

location ofany records

Sourcesends indexinformation

LOCATOR

Provider asksfor and receives

records

Recordsare sent toProvider

TING

Source maypush data for

reporting

REPOR

DataSources

PatientsProviders

HospitalsDiagnosticServices

Payors

Public HealthProviders

DATA

IndividualCare

Providers

PatientsProviders

HospitalsDiagnostic

ServicesPayors

Public HealthProviders

PublicHealth

Why Population Health?

• Population health objectives are realized in atleast three distinct domains:– bolstering research capabilities and enabling

clinical practice to fully participate in and makeuse of scientific evidence,

– increasing the effectiveness of our public healthsystem, and

– empowering consumers with information aboutcost and quality.

• Each of these requires authorized users toaccess information housed in manyfragmented data sources.

Hypothesis 1: “Gimme thedata….please…”

There is an increasing demand fromvarious constituencies for large sets of(frequently the same) population-levelhealth information, created byaggregating patient-level informationfrom clinical sources.

Aggregate Data Requests to a SingleCDO

• National level:– CMS – ORYX indicators– UHC– NACRI– Premier Benchmarking Project– Biosense– DAWN– AHRQ quality data– NCQA– Leapfrog Group

• State level:– State UB92 reporting

(quarterly)– Public Health – a variety of

reports where requirementsmandate reporting

– Cancer tumor registry –patient-specific at times

– State peri-natal system – someclinical reporting

– Genetic screening – someselected reporting through thestates

– Quality reporting and AQApilots

• Local Level:– IRB Approved Research

Projects– Internal Management reporting– Clinical Quality Reporting –

infection control, etc.– JCAHO Reporting– Teaching requests– Commercial plans – on-line

hospital assessment

Hypothesis 2: “All of these thingsare just like the others”

• The problem is large -- many requestors, manytargets of request, much siloization, littleshared infrastructure, cost, or concern foreconomies of scale or re-use.

• It's going to get larger.

• There's no one 800-lb gorilla, it's 800 gnats.As a result, any action we propose has tointroduce systemic change.

Syndromic Surveillance

-CDC-National Bioterrorism Syndromic Surveillance Demonstration

Program (NBSS).

Quality and Performance Measurement

AHA – Quality Reporting Architecture

Clinical Research

Hypothesis 3: “So much duplication…so little strategy…

deconstructing the data flows”We must understand the flow of healthinformation for population health, so as toidentify where common solutions may befeasible and desirable. We need to identifywhich areas lend themselves to commonsolutions across application areas thatincrease efficiency, and improve privacyprotections and public trust.

Hypothesis 4: “Imitation is thesincerest form of flattery”

Because the problem of compositeanalysis of distributed data is not unique tothe healthcare sector, we should examineinformation technology and policy solutionsdeveloped in other sectors.

A Better Solution Would…

Hypothesis 5: “…leverage currentinvestments.”

The business case for clinical IT adoption andinformation exchange has been weak, except inlarge delivery systems where local informationtransfer adds value. However, significantresources – both technical and financial – nowbeing deployed for the capture, aggregation,analysis, and reporting of clinical data could bemore efficiently used if a comprehensive clinicalinformation infrastructure existed and weredesigned to permit appropriate aggregateduses.

Hypothesis 6: “…anticipatechanging needs.”

Design of a nationwide health informationnetwork should allow for capture of all datathat may be pertinent to a high-prioritypopulation-level analysis, even though we cannot identify all such requirements today. Itshould permit collection of unanticipated dataelements, aggregation across currentlyuntapped information sources, and applicationof new analysis and distribution techniques.

• Design for decisions

• Design for many users

• Design for distributed environment

• Design for public values

• Design layers of protection

• … a virtuous circle of information flow

Some General Principles

Implications

• Need architecture that anticipates needfor efficient data acquisition &manipulation while protecting privacy

• Need to “close the loop” by providing valueback to contributors

3. Information to help consumersimprove own health and their care

National trends - 2007

• President’s 10-year commitment (2014?)

• HHS: American Health Information Community – ConsumerEmpowerment “Breakthrough”

• Congress – Carper, Porter, Kennedy bills for federalemployees, incentives

• AHIP and BCBS – 100 million Americans?

• Major employers – Dossia, IBM, PepsiCo

• Major providers – VA, Kaiser, Partners

• Commercial: Google, Microsoft, Intuit, WebMD

• Health Record Bank (Trust): Brownback, Wash. State …

The simple case

40

Doctor(s)

Health Care System

PHR

The reality…

41

Primary CareDoctor

SpecialistDoctor

Hospital X Pharmacy Q Pharmacy R

School Nurse

Hospital Y

Payer Data Center(health plan,Medicare)

Laboratory

HomeMonitoring

Device

The Networked PHR

42

PersonalHealthRecord

Primary CareDoctor

SpecialistDoctor

Hospital X Pharmacy Q

PharmacyData Hub

Pharmacy R

School Nurse

Hospital YHospitalSystem

Data Hub

Payer Data Center(health plan,Medicare)

Laboratory

HomeMonitoring

Device

Are we headed for integration or justmore silos?

Features of a “networked” approach

• ‘Networked’ consumers drive transformative changein other sectors.

– Content– E-commerce– Personal finance– Etc.

• A common ingredient is a fresh openness towardconsumer access to, and contribution of, information.

• Consumers and health professionals gainopportunities to transform care delivery and roles.

• A network needs common rules, particularly forprivacy.

Common Framework architecture

SNO SNO

SNO

ISB ISB

ISB

RLS

RLS

RLS

How does a consumeraccess informationacross the network?

Individual Consumers Will NeedMediating Bodies to Facilitate Their

Access to the NetworkFunctions:• Distribute services to populations of consumers.• Issue individuals’ identity credentials and “vouch”

for them as network users.• Help consumers access and aggregate their

personal health data and connect with variousservices.

• Assure that network-wide policies (e.g., privacyand information practices) are followed.

SNO SNO

SNO

ISB ISB

RLSRLS

NHIN

Authenticate/Aggregate

ISB

ConsumerAccessService

Consumers withNetworked PHRs

Consumer Access Services

• Affinity groups (e.g., AARP, labor unions)• “Retail” PHR providers (e.g., WebMD, Intuit,

Medem)• Consumer portals (e.g., Google, Yahoo)• Data clearinghouses (e.g., SureScripts)• Retail pharmacies (e.g., Walgreens, Wal-Mart)• Health plans (e.g., AHIP, BCBS)• Provider organizations (e.g., VA, Kaiser

Permanente)

Potential Sponsors of ConsumerAccess Services

Keys to Success?

• Defining a Consumer Access Service that istrusted by consumers.

• Defining a Consumer Access Service that istrusted by other participants on thenetwork.

• Determining minimum necessary privacyand security policies and practices.

Needed policy framework for CAS

• Does HIPAA address privacy and securityconcerns?

• Authentication• Authorization• Consent and notification• Consumer control of information sharing, including

audit• Rules for secondary use, data mining• Consumer annotations and edits to their data• Data management systems• Governance, transparency, remedies

Implications

• Need rules that create trust – by dataholder and by consumer

• Need interfaces

• Need reason to share data withconsumers

Three layers of HIT planning

• Information to care for a patient

• Information to support better decision-making

• Information to help consumers improveown health and their care

Implications

• The “network” has to support clinical care,population health, and consumerparticipation

• Implies many new players with a legitimaterole

• Adds new business interests, legalquestions, ethical questions

Implications for the Infrastructure

• Standards:

• Measures:

• Certification:

• Payment:

• Privacy policy:

• Enforcement of policy:

Implications for where we start

• Can only get there with collaboration

• Can only get there with shared vision as acommunity

• Can only get there with shared values

• … and leadership