Islands of Information: Linking Clinical Data

University of New MexicoUNM Digital Repository

Historical and Administrative Collection Administration

10-18-2004

Islands of Information: Linking Clinical DataJ. Marc Overhage

Follow this and additional works at: https://digitalrepository.unm.edu/hslic-historical-administrative

This Presentation is brought to you for free and open access by the Administration at UNM Digital Repository. It has been accepted for inclusion inHistorical and Administrative Collection by an authorized administrator of UNM Digital Repository. For more information, please [email protected].

Recommended CitationOverhage, J. Marc. "Islands of Information: Linking Clinical Data." (2004). https://digitalrepository.unm.edu/hslic-historical-administrative/28

https://digitalrepository.unm.edu?utm_source=digitalrepository.unm.edu%2Fhslic-historical-administrative%2F28&utm_medium=PDF&utm_campaign=PDFCoverPages

https://digitalrepository.unm.edu/hslic-historical-administrative?utm_source=digitalrepository.unm.edu%2Fhslic-historical-administrative%2F28&utm_medium=PDF&utm_campaign=PDFCoverPages

https://digitalrepository.unm.edu/hslic-admin?utm_source=digitalrepository.unm.edu%2Fhslic-historical-administrative%2F28&utm_medium=PDF&utm_campaign=PDFCoverPages



https://digitalrepository.unm.edu/hslic-historical-administrative/28?utm_source=digitalrepository.unm.edu%2Fhslic-historical-administrative%2F28&utm_medium=PDF&utm_campaign=PDFCoverPages

https://digitalrepository.unm.edu/hslic-historical-administrative/28?utm_source=digitalrepository.unm.edu%2Fhslic-historical-administrative%2F28&utm_medium=PDF&utm_campaign=PDFCoverPages

mailto:[email protected]

Islands of Information: Linking Clinical Data

October 2004

2004 © Regenstrief Institute, Inc.D. 1

2004 © Regenstrief Institute, Inc.

Islands of Information:Islands of Information:Linking Clinical DataLinking Clinical Data

J. Marc Overhage, MD, PhDJ. Marc Overhage, MD, PhDIndiana Health Information ExchangeIndiana Health Information Exchange

Regenstrief Institute, Inc.Regenstrief Institute, Inc.Indiana University School of MedicineIndiana University School of Medicine


A Simple Country DoctorA Simple Country Doctor


Current Status of Medical Records


Inform Clinical PracticeStrategy 1. Incentivize EHR adoption

Strategy 2. Reduce risk of EHR investment

Strategy 3. Promote EHR diffusion in rural and underserved areas

Personalize CareStrategy 1. Encourage use of Personal

Health Records

Strategy 2. Enhance informed consumer choice

Strategy 3. Promote use of telehealth systems

Improve Population HealthStrategy 1. Unify public health surveillance

architectures.

Strategy 2. Streamline quality and health status monitoring

Strategy 3. Accelerate research and dissemination of evidence

Interconnect CliniciansStrategy 1. Foster regional collaborations

Strategy 2. Develop a national health information network

Strategy 3. Coordinate federal health information systems

Decade of Health Information Technology

Goals from the HHS Framework for Strategic Action – July 21, 2004


Registration Cross-over

<1%5

3%4

18%3

39%2

40%1

PercentHospital systems


Emergency Department Cross-over

0

5000

10000

15000

20000

25000

Hosp A Hosp B Hosp C Hosp D Hosp E

Hospital System

Num

ber

of P

atie

nts

Crossover Patients Unique Patients

*More than one visit


October 2004



The Indiana Network for Patient

Care (INPC)An operational community wide

electronic medical record2004 © Regenstrief Institute, Inc.

Indianapolis, Indiana

• 1.5 million population base• 12th largest city in U.S.A.• Home to Indiana’s only medical school• State Department of Health• Referral center for entire state (7 million)

2004 © Regenstrief Institute, Inc. 2004 © Regenstrief Institute, Inc.

Initial RMRS Aims

• Eliminate the logistic problems associated with the paper record

• Standardize the care process. Deliver information in a more organized and useful way. Actively process this record and provide decision support to clinicians.

• Analyze and understand the data to improve the health of populations


INPC Project Goal

Demonstrate the feasibility and benefit of a community wide electronic medical record system in acute care situations.


Resistance is futile!

You will be assimilated

INPC Project Motto


October 2004



Pilot Project• Participants

– Methodist– Community East– Wishard

• Data - Wishard only• Outcomes

– Charges– ED visits– Admissions


Pilot Emergency Departments

Institution ED Visits Beds

Wishard 100,000 350

Methodist 90,000 800

Community 48,000 350


METHODIST HOSPITALED

RMRS

COMMUNITY HOSPITALEAST ER

SMSMalvern, PA

ERRegistrationMessage

ERRegistrationMessage

Direct Accessto Patient Recordx 24 H

Direct Accessto Patient Record

x 24 H

Medical RecordAbstract

Medical RecordAbstract


Emergency Dept. Data Flows

EmergencyDepartment INPC

Registration Message

Merged Clinical AbstractMerged Clinical Abstract

Merged Result RetrievalMerged Result Retrieval

ED Charge DataED Charge Data


Study ED Methodist Hospital Community Hospital

Group Control Intervention Control Intervention

Patients (N) 10,526 10,694 5,568 5,680

Age years 32.7 ± 21 32.7 ± 21 34.2 ± 22 33.6 ± 22

Female % 56.9% 56.1% 56.8% 57.4%

Black % 56.4% 56.1% 39.6% 38.4%

Pilot patient demographics


Pilot ED visit charges

0

100

200

300

400

500

Community Methodist Wishard

Control Intervention

**

**


October 2004



Clinical INPC functions

• Results retrieval• Clinical messaging/document delivery• Data entry• Reporting• Clinical decision support• Public health surveillance• Medical reference access


Consolidating the Silos

Public HealthPublic Health

Hospital Electronic Hospital Electronic Medical RecordMedical Record

Patient ID: 123LMNOPPatient ID: 123LMNOPName: Jane Doe Name: Jane Doe DOB: 01/01/04DOB: 01/01/04SSN: N/A SSN: N/A Address: 555 Johnson RoadAddress: 555 Johnson RoadCity: IndianapolisCity: IndianapolisState: IndianaState: IndianaZIP: 46202ZIP: 46202

Patient ID: 6789XYZPatient ID: 6789XYZName: Jane Ellen DoeName: Jane Ellen DoeDOB: 01/01/04DOB: 01/01/04SSN:123SSN:123--4545--67896789Address: 555 Johnson RoadAddress: 555 Johnson RoadCity: IndianapolisCity: IndianapolisState: IndianaState: IndianaZIP: 46202ZIP: 46202

Global Patient Global Patient IndexIndex

Concept Concept DictionaryDictionary

Global ID:Global ID: 4567845678Name: Name: Jane Ellen Doe Jane Ellen Doe Lots of Demographics..Lots of Demographics..MRF1 ID: MRF1 ID: OU81247OU81247MRF2 ID: MRF2 ID: 45643564564356PH MRF ID: PH MRF ID: 123LMNOP123LMNOPMRF3 ID:MRF3 ID: 6789XYZ6789XYZ

DTaPDTaP Dose Count:Dose Count: 3093630936--99HIB Dose Count:HIB Dose Count: 3093830938--55IPV Dose Count:IPV Dose Count: 3355533555--44VZV Dose Count:VZV Dose Count: 3094330943--55MMR Dose Count:MMR Dose Count: 3094030940--11HepBHepB Dose Count:Dose Count: 3093730937--77

Jane DoeJane Doe’’s Immunizations:s Immunizations:

3/1/043/1/04 DipTetaPurDipTetaPur3/1/043/1/04 HemInfBHemInfB3/1/043/1/04 PolioVirPolioVir3/1/043/1/04 HepaBHepaB

Jane Ellen DoeJane Ellen Doe’’s Shots:s Shots:

5/1/045/1/04 DTaPDTaP ImmImm5/1/045/1/04 HIB HIB ImmImm5/1/045/1/04 IPV IPV ImmImm7/9/047/9/04 DTaPDTaP ImmImm7/9/047/9/04 IPV IPV ImmImm

3093630936--9 9 3093830938--5 5 3355533555--4 4 3093730937--77

3093630936--9 9 3093830938--5 5 3355533555--4 4 3093630936--9 9 3355533555--44


Consolidating the Silos

Public HealthPublic Health

Electronic Medical Electronic Medical Record SystemRecord System



St. Vincent St. Vincent MRFMRF

Clarian Clarian MRFMRF

Wishard Wishard MRFMRF

Community MRFCommunity MRF

Public Health Public Health MRFMRF


IUMG MRFIUMG MRF


The Indiana Network for Patient Care


INPC Storage Strategy

• Separate medical record vault per institution

• Each vault in separate physical files

• Standardized data structure – All use same

software and observation codes .

• Combine on the fly when needed

• Patient linking


Indiana Network for Patient Care (INPC)

• A local health information infrastructure (LHII)• We serve as the “Data Switzerland” for the city

of Indianapolis and the state• Up and running for more than 9 years• Started with a modest goal (Emergency Care)• Extended by adding projects one step at a time• Focus on clinical and public health issues,

particularly electronic laboratory reporting


INPC – Participants • Includes 11 hospitals from the 5 major

Indianapolis hospital systems (95% of non-office care)

• Includes all four homeless care systems• Includes county and state public health

departments• Many outpatient practices• Both major cardiology

referral centers


October 2004



Hospital Geographic Coverage


Clinical INPC Users

• Almost all med/surg hospital EDs• Hospital based providers (expanding)• Ambulatory physicians (approximately

35%)• Homeless care network• Public school clinics• Marion County Health Department• Indiana State Department of Health


INPC Contents• In the system:

– 1.3 million patients, 5 million registration “events”

– 24 million orders– 489 million coded results– 12 million dictated reports– 8.8 million radiology reports– 25 million prescriptions– 480,000 EKG tracings– 45 million radiology images

• Added Per Year:– 600,000 ambulatory encounters– 50,000 inpatient encounters



Patient: John DoeMRN: 123-0Diagnosis: 410.0WBC: 14,000/cm3

• Current– HL7 messages for most as the envelope– DICOM messages for images as the envelope– LOINC for laboratory results content– CPT-4 for procedures content– ICD-9 for diagnoses content– NDC and RxNorm for medications content

• Evolving– Organisms for microbiology content

Clinical Data Standards


October 2004



Patient Linkage

• When data is returned from multiple sites, it will need to be combined and linked.

• For example, if “John Doe” is seen for various aspects of his colon cancer at different institutions, the data must be aggregated:

Diagnostic Studies

Treatment Data

Outcome Data


Linkage Methodologies -Deterministic

• “All-or-None”• Rules based on exact agreement or

disagreement• Match first on a reliable and discriminating

identifier (Such as SSN)• Verify link using additional parameters

(Such as LN, FN)


Linkage Methodologies -Probabilistic

• Use statistical methods to generate frequency ratios, similar to likelihood ratios, for each variable

Agreement Ratio = frequency of agreement among linked recordsfrequency of agreement among non-linked records

Disagreement Ratio = frequency of disagreement among linked recordsfrequency of disagreement among non-linked records


Linkage Methodologies -Probabilistic

Example: Assume 94.1% of last names agree among true links, and last names randomly agree among non-links with a frequency of 0.1%

Agreement Ratio = .941 = 941, log2(941) = 9.88.001

Disagreement Ratio = 1-.941 = .059, log2(.059) = - 4.081-.001


Linkage Methodologies

Deterministic• Rapid Implementation• Simple calculations

• Relies on accurate data

• May not function well with other data sets

Probabilistic• Complex implementation• Computationally

intensive • More forgiving of data

errors• Algorithm is customized

to data being linked


Global Patient Registry

AssigningAuthority Global # Local Pat # Patient Name Birthdate Sex

Hospital A 99-1 231456 Sinkwell, Ralph J 12-2-59 M

Hospital B 123-0 A47239 Sinkwell, RJ 2-12-59 M

Hospital A 99-1 1032115 Sinkwell, Ralph 12-2-59

Hospital C 101-0 A3276 Fredrick, Alice 4-14-78 F

Hospital A 101-0 2314590 Fredrick, Alyce 4-14-78 F


October 2004



INPC - Confidentiality• While numerous measures in in place to

protect confidentiality of patient’s data, the provider has to know who the patient is.– Secure physical network– Encryption– Authentication– Agreements– Device controls


Participants’ Agreement•How can participants share health data to treat patients?•Who may have access to PHI for treatment purposes?•What information is to be stored on the network?•How may the PHI be used for research purposes?•What are other considerations?

–Equipment.–Consistency of data.–Other uses of information.–Indemnification.. –Governance–Disposition of information upon termination–Security



=LHII=LHII

RHIIRHII


NHII – A National Database


October 2004



INPC Data Access




October 2004




Patient Name

Details

Report

Image



October 2004






October 2004





Video Records•Implement video-based teleconferencing between health providers and patients in a nursing facility•Measure impact on health outcomes

Secure transmission


October 2004



Links to knowledge

sources

PraxisMD

Links to knowledge

sources


InfoButtons

Links to knowledge

sources


Specify Cohort


Specify statistical breakdowns


One way to show output as breakdowns


October 2004



Clincial Abstract


Specialty Abstracts with Merged Data


Winston Niles Rumfoord had run his private space ship right into the heart of an uncharted chrono-syn-clastic infundibulumtwo days out of Mars. Only his dog had been along. Now Winston Niles Rumfoord and his dog Kazak existed as wave phenomena -- apparently pulsing in a distorted spiral with its origin in the Sun and its terminal in Betelgeuse.

Sirens of Titan Kurt Vonnegut 19592004 © Regenstrief Institute, Inc.

INFUNDIBULUM

• A funnel-shaped cavity. It’s the Latin word for a funnel, derived from infundere, “to pour”, plus the ending –bulum which formed the names of instruments. In English, it turns up in various anatomical contexts for something funnel-shaped. For example, in the human body it describes the outermost section of the fallopian tubes, a structure in the cochlea of the ear, and a formation in the brain close to the pituitary, among others.

• Science-fiction fans may have come across the splendid phrase chronosynclastic infundibulum that was invented by Kurt Vonnegut in The Sirens of Titan, which he explained, perhaps less than helpfully, as being “those places ... where all the different kinds of truths fit together”


Vital statisticsNewborn screeningImmunizationsLeadReportable conditionsSentinel hospitalsISDH laboratory

HospitalizationsEmergency room visitsHospital laboratory resultsHospital pathology resultsHospital radiology resultsHospital medications

ImmunizationsPrimary care visitsMCHD laboratory

Statewide Regional County

Organization

Technology

Management

Repository

Clinical care Research

“On-line” surveillance

ELR BT

CDSS

Result delivery Clinical Trials

Data StandardizationPatient

ProviderLabels

Answers


Data reuse• Clinical care

– Emergency room– Primary care– Inpatient

• Public health (state and local HD)– Immunization registry– Reportable conditions– Surveillance

• Health services research• Clinical research• Accreditation reports


October 2004



The “Killer App” – Reusable Data

Increasingly structured clinical data

Patient data

Direct support of

patient care

Research

Decision-supportDisease management &

community health improvementPublic health lab reporting

Public health surveillance

Research

Clinical trials screening & management

Adminfunctions

Credentialing

Claims clearing

Eligibility/referrals/pre-certification

8

9

10

Applications & tools

2

3

4

5

6

7

Results delivery1

Major urban hospitals Non-hospital labs Pharmacies

Physician offices Biological samples Payors

E-prescribing1


To InfectionControl

Reportable condition processor

InboundInboundHL7HL7

PotentiallyReportable

ReportableCondition

To PublicHealth

ReportableConditionsDatabases

Abnormal flag,Organism name in Dwyer II, Value above threshold

Compare to Dwyer I

Record Countas denominator

E-mailSummary

Realtime Daily Batch

PrintReports


0

5

10

15

20

25

30

35

Date

1/22/20

00

2/1/2000

2/10/2

000

2/17/2

000

2/28/2

000

3/6/20

00

3/16/20

00

3/23/20

00

3/30/2

000

4/8/20

00

4/15/2

000

4/22/2

000

4/29/2

0006-M

ay

13-M

ay

20-M

ay

IU/Wishard St. Vincents Community Methodist MCHD

Shigella Outbreak: Timeline

Dr. MD



Public Health Outcomes• Reliable

– Real time delivery– 100%received (for participants)

• Reporting completeness (capture/recapture)– Greatly increased case reporting

• Reporting timeliness (versus result date)– 8.4±15.4 days faster than HD– 1.4±2.0 days faster than hospital


Integrating Non-traditional DataIndianapolis Electrolytes, Rotavirus, and RSV

-1.5

-1

-0.5

0

0.5

1

1.5

2

2.5

7/4/

99

8/4/

99

9/4/

99

10/4

/99

11/4

/99

12/4

/99

1/4/

00

2/4/

003/

4/00

4/4/

00

5/4/

00

6/4/

00

7/4/

00

8/4/

00

Date

Std

Dev

from

Mea

n

ElectrolytesalesRotavirus

RSV

The sales of OTC Electrolytes are a result of two outbreaks: RSV (in

gold) and Rotavirus (in pink)


October 2004



Healthcare Collaborative Network

Data Source Organizations

LegacySystems

•Diagnosis•Lab•Pharmacy•Other?


LegacySystems



LegacySystems



LegacySystems


HCN HUB


LegacySystems


Data Review Organizations

AgenciesPayers

andothers

whoanalyzeclinical

data

IntegrationBroker

InternetPortal

HCNGateway

FirewallsFirewalls

Internet

Internet

CMSFDACDC

HCNGateway



0

20

40

60

80

100

120

140

Credentialing Eligibilitychecking

Public health Results delivery Research Decisionsupport

Disease mgmt

Benefits Accrue to Stakeholders Across SystemAnnual

economicvalue ($M)

•Patients•Physicians

•Hospitals•Labs

•Physicians

•Hospitals


•Hospitals•Payors


•Hospitals•Payors•Employers


•Hospitals•Public health

•Payors•Employers

Beneficiaries1 •Patients•Hospitals•Labs•Public health

•Patients•Researchers•Investigators

•Pharmacompanies

(1) Includes quality of care/other non-economic beneficiaries as wellSource: BCG Analysis & Estimates

LONG-TERM ECONOMIC BENEFITTO CENTRAL INDIANA

WILL EXCEED $120M ANNUALLY


C!TL HIE economic model

Provider

Public Health

Pharmacy Laboratory

PayerOther Provider

$13.1B

$0.09B

$1.3B

$30.4B

$8.2B

N/A

-$1.0B

-$0.04B

$12.2B

$8.8B

$13.9B

$10.4B

Total value: $87 billion

Radiology

Provider system maintenance cost of $10.5B not reflected in diagram

Provider Net: $34B per year


HIEI Taxonomy

Secure e-mail of free text or incompatible/proprietary file formats, HL-7 message

Machine-organizable data3

PC-based and manual fax, secure e-mail of scanned documents

Machine-transportable data2

Mail, phoneNon-electronic data1

Automated entry of LOINC results from an external lab into a primary care provider’s electronic health record

Machine-interpretable data4

ExamplesDescriptionLevel

No PC/information technology

Fax/Email

Structured messages, non-standard content/data

Structured messages, standardized content/data


Achieving full value requires structured data

Percent

Capture electronically

Connect & interface

Standardize and store

data

0

10

20

30

40

50

60

70

80

90

100

76%

5%

19%

Source: Center for Information Technology Leadership, IHIE calculations


October 2004



CHCF Model


CFHC Model


1 year RCT done!

• 80% intervention/20% control• Intervention

– Printed abstract– On-line access for 24 hours

• Additional covariates


Tentative Results

Mean Charge Charge Savingb±SE 95% CI p-value

Control Intervention

All visits w/o covariates 572.93 563.23 -9.70±4.74 -19.01, -0.40 0.041

All visits w/ covariate 571.10 561.55 -9.55±4.56 -18.49, -0.62 0.036


INPCICareConnect

I3

BioCrossroads

Indiana Health Information Exchange

RegenstriefInstitute

Central IndianaLife Science Initiative

IUSM


–Founded in 1969 by Sam Regenstrief

–Affiliated with IU School of Medicine

–~$12M annual budget largely from federal grants

–Pioneers in medical informatics

• Standards: HL7, LOINC

• EMRs: RMRS, INPC

–Founded in 1999 by 50 community physicians – Central Indiana Coalition to Reinvent Health Care

–Early seed funding from Health & Hospital Corporation of Marion County

–Initiated community clinical messaging concept

–Founded in 2002 by the Central Indiana Corporate Partnership

–Economic development organization promoting academic/industry collaboration in life sciences


October 2004



Founding Members• Hospital systems

– Clarian Health Partners– St. Francis Hospitals and Health System– St. Vincent Health Care– Wishard Health Services– Community Hospitals of Indiana

• Government– City of Indianapolis

• Public Health– State Department of Health– Marion County Health Department

• Research– IU School of Medicine– Regenstrief Institute

• Medical societies– Indianapolis Medical Society– Indiana State Medical Association

• Economic development– BioCrossroads / Central Indiana Corporate Partnership


Reality Bites


Acknowledgement

• National Library of Medicine• Agency for Healthcare Quality and

Research• National Cancer Institute• Regenstrief Foundation• Eli Lilly and Company INGEN grant • BioCrossroads


J. Marc Overhage, MD, PhDRegenstrief Institute, IncIndiana University School of Medicine1050 Wishard BlvdIndianapolis, IN 46202

Voice: 317-630-8685Facsimile: 317-630-6962E-mail: [email protected] address: www.regenstrief.org


Three Hypothetical Communities Were Modeled

participation and 15% physician usage adoptionn participation and 35% physician usage adoption penetration percentages for institution participation not applicable

• Constituent type • Low*

• High**

• Total number in community

• Medium – Major hospital– Diagnostic imaging center– Independent laboratory– PBMs – Major physician groups– Physicians

621521

,000

211111

50

421323

50

• Large – Major hospital– Diagnostic imaging center– Independent laboratory– PBMs – Major physician groups– Physicians

10

53555

,000

321117

50

742331

,750

• Penetration

• Small***

– Major hospital– Diagnostic imaging center– Independent laboratory– PBMs – Major physician groups– Physicians

110013

0

111307

0

111502

00


Value Increased w/Community Size& Penetration

* Includes annual support costs and amortized implementation costs over 5 years

• Medium

• Large

• Low • High

•• PenetrationPenetration

•• Community sizeCommunity size

• $U.S. annual

•• $800,000$800,000

•• $900,000$900,000

•• Costs*Costs*

•• BenefitsBenefits

•• $490,000$490,000

•• $180,000$180,000

•• Costs*Costs*


•• $780,000$780,000

•• $600,000$600,000

•• Costs*Costs*


•• NetNet •• ($310,000)($310,000) •• NetNet •• ($180,000)($180,000)

•• NNetet

•• $100,000$100,000

•• $1,400,000$1,400,000

•• $2,600,000$2,600,000

•• Costs*Costs*


•• NeNett

•• $1,200,000$1,200,000

•• $1,000,000$1,000,000

•• $1,300,000$1,300,000

•• Costs*Costs*


•• NNetet

•• $300,000$300,000

•• $2,200,000$2,200,000

•• $7,900,000$7,900,000

•• Costs*Costs*


•• NetNet •• $5,700,000$5,700,000

•• ValueValue

• Small


October 2004



Modest Value For Each Constituent; First Mover Disadvantage Existed For All Constituents• $U.S. annual

1

• Per constituent• Total for all

constituents

• Intrinsic benefits of providing data

• Network benefits

• Total individual benefits

• Costs1,2

• Total benefits

• Number of constituents

• Total costs• $180,

000• $110,00

0• $290,00

0• $120,00

0• $2,000,0

00• 7• Hospital • $840,000

• Imaging center

• $44,000

• $(15,000)

• $29,000• $110,000

• $120,000• 4 • $440,000

• Laboratory • $70,000

• $170,000

• $240,000

• $110,000

• $480,000• 2 • $220,000

• Physician group

• $90,000

• $280,000

• $370,000

• $120,000

• $1,100,000

• 3 • $360,000

• Other physicians

• $0 • $2400 • $2400• $40 • $3,500,000

• 1,750

• $70,000

• PBM • $0 • $0 • $0• $110,000

• $0• 3 • $330,000

• ~$7,300,000

• ~$2,200,000

• LARGE COMMUNITY, HIGH PENETRATION

Documents

Islands of Information: Linking Clinical Data