september 27, 2007 health information exchange: linking practitioners and public health for decision...

September 27, 2007

Health Information Exchange:Linking Practitioners and Public

Health for Decision SupportStephen M. DownsIndiana University

Children’s Health Services ResearchRegenstrief Institute

September 27, 2007

Overview

• The Indiana Network for Patient Care (INPC) and the Indiana Health Information Exchange (IHIE)

• Real-time Decision Support for Practitioners – Adaptive Turnaround Documents

• Practitioner Data for Public Health – Disease Surveillance

September 27, 2007

Concept

• HIE connects health systems & providers• Creates opportunities for decision support• Patterns of care identify opportunities for

decision support & quality improvement– Patients with diabetes need annual urinary

microalbumin testing– Infants with positive newborn screens must be located

and treated immediately– Health departments need to respond to disease

outbreaks

September 27, 2007

What’s Needed

• A mechanism for 2-way communications with physicians in the INPC

• We have used Adaptive Turnaround Document Technology (ATD) for decision support in our clinics

• Adapting to the INPC

September 27, 2007

What is an Adaptive Turnaround Document (ATD)?

• Computer Generated

• Paper

• Delivers tailored information

• Scannable

• Captures structured data– Computer, scanner, or fax

September 27, 2007

Applying ATD Technology on a Regional Basis

• Indiana Network for Patient Care (INPC)

• 21 hospitals

• County and State Health Departments

• RxHub

• Medicaid administrative data

• Over one billion coded results

September 27, 2007

September 27, 2007

How it was accomplished (technical)

• Federated repository

• Data from difference sources are stored in separate physical files

• Global patient index

• Same data structure and data dictionary

• Data arrive by HL7 (also DICOM, NCPDP)– Parsed– Translated

September 27, 2007

Docs4Docs - Links to clinical providers

• Results delivery service

• All hospitals send reports through same mechanism (HL7)

• HL7 results messages flow through INPC and are routed to MD in-boxes or faxes

• Eliminates mail costs

September 27, 2007

INPCINPC

HL7 QUEUE

Repository

DSS

HL7

Docs4Docs ©

September 27, 2007

Improving Diabetes Management

• ADA Guidelines– Microalbumin measured annually

• Monitor HL7 messages for Hgb A1C– Proxy for diabetes

• Check INPC for microalbumin in the last 12 months– If none, alert physician

September 27, 2007

Methodist Hospital Enhanced Report Name: Mkhemtest, mhtest Department of Pathology MRN: 007222030 And Laboratory Medicine DOB: 09/10/67 Sex: Female 1701 North Senate Blvd Home: (457) 348-9794 Indianapolis, IN 46206 Copy for: Rougraf, Bruce Date: Mon 02/23/2004 01:23 PM Accession: #04-054-000014 Account: 010001000234 Ordered by: Rougraf, Bruce T Location: HRTF

** Glycosolated Hemaglobin **

Procedure Units Ref Range Hgb A1C 8.5 H percent [<7] All tests performed at Clarian/Methodist Lab, 1701 N Senate Blvd, Indpls, IN 46202 Goal 15 Jun

2005 22 Aug 2005

2 Mach 2006

29 Oct 2006

Due

Physician visit every 6 months or less

X [_}Done [_}Scheduled [_|_]/[_|_]/[_|_] (Date)

HgbA1C every 6 months 8.5 Microalbumin annually [_}Done

[_}Scheduled [_|_]/[_|_]/[_|_] (Date)

Creatinine clearance & serum protein yearly if positive microalbumin

[_}Done [_}Scheduled [_|_]/[_|_]/[_|_] (Date)

Lipid profile annually X Ophtalmologic exam annually [_}Done

[_}Scheduled [_|_]/[_|_]/[_|_] (Date)

Dental visit every 6 months X [_}Done [_}Scheduled [_|_]/[_|_]/[_|_] (Date)

Influenza vaccine annually X Pneumococcal vaccine once X

[_] This is not my patient. The additional clinical context above was useful in caring for this patient. Stongly Somewhat Neutral Somewhat Strongly Agree Agree Disagree Disagree

Clinical data provided should always be considered incomplete and you should

exercise appropriate clinical judgment.

Procedure Units Ref Range Hgb A1C 8.5 H percent [<7]

Goal 15 Jun 2005

22 Aug 2005

2 Mach 2006

29 Oct 2006

Due

Physician visit every 6 months or less

X [_}Done [_}Scheduled [_|_]/[_|_]/[_|_] (Date)

HgbA1C every 6 months 8.5 Microalbumin annually [_}Done

[_}Scheduled [_|_]/[_|_]/[_|_] (Date)

Creatinine clearance & serum protein yearly if positive microalbumin

[_}Done [_}Scheduled [_|_]/[_|_]/[_|_] (Date)

Lipid profile annually X Ophtalmologic exam annually [_}Done

[_}Scheduled [_|_]/[_|_]/[_|_] (Date)

Dental visit every 6 months X [_}Done [_}Scheduled [_|_]/[_|_]/[_|_] (Date)

Influenza vaccine annually X Pneumococcal vaccine once X

[_] This is not my patient.

September 27, 2007

Simulating the RuleMicroalbumins & A1Cs in 2005

Hosp Pats Albumin % Albumin %

1 8430 2898 34% 2931 35%

2 4055 168 4% 181 4%

3 7673 1270 17% 1345 18%

4 10927 2647 24% 2683 25%

5 16239 5113 31% 5213 32%

TOTAL 47324 A1Cs 73482 201 alerts per day

September 27, 2007

A Test by Any Other Name…

• Hemaglobin A1C– Glycated Hgb– Glycated Hb– Glycos Hgb A-1– Hgb Clycosylated– HGBA1C:EIA– HGBA1C:HPLC– HgbA1C % Ser EIA– Hgb A1c Bld Qn

• Microalbumin– Microalbomin 24h Ur– Microalbumin R-Ur Qn– Microalbomin Timed

Urine– Microalbumin 24h Ur

Cnc– Microalbomin Ur Qn

mg/dl– Albumin Ur Qn Elp

September 27, 2007

Another Example:Newborn Screening

• Virtually all newborns in the US

• Blood spot obtained by heelstick

• Tested for inborn errors of metabolism and other conditions

• Early treatment may be life saving

September 27, 2007

Expanded Newborn Screening

• Tandem Mass Spectrometry

• Over 50 Conditions

• Prevent Morbidity and Mortality

• Save Costs

September 27, 2007

Challenges

• Most Conditions Rare– MDs Unfamiliar

• Families Need Guidance

• Diagnosis & Treatment Must Be Timely– Who has the baby?

• No Mechanisms for LTFU

September 27, 2007

Solution

• Regional health information network

• Two way communication with providers

September 27, 2007

Project Objectives

• Adaptive turnaround document technology to facilitate communication: NBS program, subspecialists, medical home– Provide “just in time” information to the

primary care clinician & family– Reduce the risk of “missed opportunities” to

screen– Facilitate tracking of children with detected

conditions

September 27, 2007

HL7

INPC

ISDH

ATDATD

September 27, 2007

Three Ways ATDs Can Enhance the NBS Program

• Just-in-time information to the medical home (physicians and families)

• Prevent missed opportunities to screen

• Long term tracking of children with identified conditions

September 27, 2007

Jennifer D. Patient DOB: 14 May 2007 Newborn Screening Alert: Elevated C8 with Lesser Elevations of C6 and C10 Acylcarnitine Suggestive of Medium-chain Acyl-CoA Dehydrogenase (MCAD) Deficiency Condition Description: MCAD deficiency is a fatty acid oxidation (FAO) disorder. FAO occurs during prolonged fasting and/or periods of increased energy demands (fever, stress) when energy production relies increasingly on fat metabolism. In an FAO disorder, fatty acids and potentially toxic derivatives accumulate because of a deficiency in one of the mitochondrial FAO enzymes. MEDICAL EMERGENCY - TAKE THE FOLLOWING IMMEDIATE ACTIONS: Contact family to inform them of the newborn screening result and ascertain clinical status

(poor feeding, vomiting, lethargy). Consult with pediatric metabolic specialist. Evaluate the newborn (poor feeding, lethargy, hypotonia, hepatomegaly). If signs are present

or infant is ill, initiate emergency treatment with IV glucose. Transport to hospital for further treatment in consultation with metabolic specialist. If infant is normal initiate timely confirmatory/diagnostic testing, as recommended by specialist.

Educate family about need for infant to avoid fasting. Even if mildly ill, immediate treatment with IV glucose is needed.

Report findings to newborn screening program. Diagnostic Evaluation: Plasma acylcarnitine analysis will show elevated octanoylcarnitine (C8). Urine acylglycine will show elevated hexanoylglycine. Diagnosis is confirmed by mutation analysis of the MCAD gene. Please check ALL of the following that apply: [ ] Family contacted [ ] Newborn clinical status assessed

[ ] Problems (poor feeding, vomiting, lethargy, hypotonia, hepatomegaly) [ ] Treated with IV glucose [ ] Infant stable

[ ] Family provided attached educational materials Diagnostic Evaluation [ ] Plasma acylcarnitine sent [ ] Referral made to metabolic center [ ] Family could not be contacted [ ] This is not my patient

Just-in-Time Information for the Primary Care Physician

September 27, 2007








MEDICAL EMERGENCY - TAKE THE FOLLOWING IMMEDIATE ACTIONS: Contact family to inform them of the newborn screening result and ascertain clinical status




Report findings to newborn screening program.

Please check ALL of the following that apply: [ ] Family contacted [ ] Newborn clinical status assessed



X

XX

XX

September 27, 2007








RULES:IF lurelicubwliucTHEN isjfafkjdf;lksdfIF lurelicubwliucTHEN isjfafkjdf;lksdfIF lurelicubwliucTHEN isjfafkjdf;lksdfIF lurelicubwliucTHEN isjfafkjdf;lksdfIF lurelicubwliucTHEN isjfafkjdf;lksdfIF lurelicubwliucTHEN isjfafkjdf;lksdf

Database

Knowledge Base

Generating the ATD

September 27, 2007

Avoiding Missed Opportunities

• Capture HL7 message for neonates

INPCINPC

• Check against newborn screening reports

• Alert physician if missing or positive

September 27, 2007

MSH|^~\&|RLY_NB|RILEY|NEWBORN|REGEN|20051202070251||ORU^R01|||2.3PID|||72360601||Doe^John||20050908155100|M|Jones|White||||||||||||Non HispanicNK1||Doe^Jane|Mother|123 MAIN STREET^^GREENWOOD^IN^46143^^^^JOHNSON|3175350809|||||||||||19791116PV1|||||||^SMITH^LYNETTE

Probabilistic Matching withHL7 Messages

Infant LName = DoeInfant FName = JohnInfant DOB = 9/8/2005Infant Sex = MaleInfant Race = WhiteInfant Ethnicity = Non HispanicMother LName = DoeMother FName = JaneMother Street = 123 Main StreetMother City = GreenwoodMother State = INMother Zip = 46143Mother County = JohnsonMother Phone = 317-535-0809

Infant LName = DoeInfant FName = BabyInfant DOB = 9/8/2005Infant Sex = MaleInfant Race = WhiteInfant Ethnicity = Mother LName = DoeMother FName = JaneMother Street = 123 Main StreetMother City = GreenwoodMother State = INMother Zip = 46143Mother County =Mother Phone = 317-555-1828

September 27, 2007

318 un-linked HL7 records

Riley (100,785 unique records)

INPC (2,561 total unique records)

2,243 linked HL7 records

97,960 un-linked Riley records2,825 linked Riley records 1,107 un-linked Riley records with

Indianapolis address

September 27, 2007

Explaining Unmatched Cases

• Manual Review of 300 “children” whose HL7 messages did not match a newborn screen– 200 “junk” messages

• E.g., did not contain PID and NK1 segments

– 70 messages could be manually linked to an infant with a newborn screen

• Matching algorithm needs refining

– 30 appear not to have had a newborn screen

September 27, 2007

Long Term Follow-up Data Collection

Dr. Emmet Cal Holmes Kids Klinic Primary Care Ln Lebanon, IN RE: Jennifer Johnson, DOB: 9/30/06 Dear Dr. Holmes: We are following up children with conditions identified by newborn screening. Jennifer Johnson was identified as having VLCAD. Our records show that she is in your care. We need your help in following her growth, development and general health. Please answer the following few questions at your earliest convenience and fax to 1-800-555-2800. Is Jennifer still in your care? [_]YES [_] NO If NO, please provide any contact information you may have Street address [_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_] [_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_] City, state, zip [_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_], [_|_] [_|_|_|_|_|] Phone ([_|_|_]) [_|_|_]-[_|_|_|_] Data last seen (mm/dd/yy): [_|_]/[_|_]/[_|_] Development: Normal Concerning Gross motor [_] [_] Fine motor [_] [_] Personal-Social [_] [_] Language [_] [_] Growth: Height [_|_|_].[_] [_] inches [_] centimeters Weight [_|_|_].[_] [_] pounds [_] kilograms Has Jennifer been hospitalized in the last year? [_]YES [_] NO If yes, how many times? [_|_|_]

Is Jennifer still in your care? [_]YES [_] NO If NO, please provide any contact information you may have Street address [_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_] [_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_] City, state, zip [_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_|_], [_|_] [_|_|_|_|_|] Phone ([_|_|_]) [_|_|_]-[_|_|_|_] Data last seen (mm/dd/yy): [_|_]/[_|_]/[_|_] Development: Normal Concerning Gross motor [_] [_] Fine motor [_] [_] Personal-Social [_] [_] Language [_] [_] Growth: Height [_|_|_].[_] [_] inches [_] centimeters Weight [_|_|_].[_] [_] pounds [_] kilograms Has Jennifer been hospitalized in the last year? [_]YES [_] NO If yes, how many times? [_|_|_]

September 27, 2007

Electronic Lab Reporting

• Reportable disease HL7 messages are sent to ISDH and Local Public Health daily.

September 27, 2007

Reportable Condition Processor

InboundInboundHL7HL7

PotentiallyPotentiallyReportableReportable

ReportableReportableConditionCondition

ReportableReportableConditionsConditionsDatabasesDatabases

ReportableReportableConditionsConditionsDatabasesDatabases

Abnormal flag,Organism name, Value above threshold

Record Countas denominator

E-mailE-mailSummaSumma

ryryRealtime Daily Batch

PrintPrintReportsReports

To PublicTo PublicHealthHealth

To InfectionTo InfectionControlControl

September 27, 2007

Syndromic Surveillance Basics

Signal: Alarm threshold

Weekly ED Gastrointestinal Syndromes

27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46

Week

Alarm Threshold

Event:

Time

GI Outbreak

September 27, 2007

89.5, -36.788.6, -35.589.1, -36.188.9, -36.488.7, -35.889.2, -36.088.8, -36.289.3, -35.9

89.5, -36.788.6, -35.589.1, -36.188.9, -36.488.7, -35.889.2, -36.088.8, -36.289.3, -35.9

GIRESPRESPCONSTRASHRASHGIRESP

89.5, -36.788.8, -36.2

GIGI

88.6, -35.589.1, -36.189.3, -35.9

RESPRESPRESP

88.9, -36.4CONST88.7, -35.889.2, -36.0

RASHRASH

Geoco

de

Syn

dro

me

Cla

ssify

Seg

regat

e

ED Data

September 27, 2007

September 27, 2007

Current PHESS

Hospitals (n=72)

Goal of 78 hospitals by August 2007

> 6,500 visits per day~ 15 MB data per day

September 27, 2007

[email protected]

www.ichsr.org

mailto:[email protected]

September 27, 2007

Maintenance: Title: MCAD Reminder;; Filename: mcad;; Version: 0.2;; Institution: Indiana University School of Medicine;; Author: Steve Downs;; Specialist: Pediatrics;; Date: 05-22-2007;; Validation: ;; Library: Purpose: Provides a specific reminder, tailored to the patient who identified one or more fatty acid disorders;; Explanation: Based on AAP screening recommendations;; Keywords: fatty, acid, fatty acid disorder;; Citations: Screening for fatty acid disorder AAP;; Links: ;; Knowledge: Type: data_driven;; /* Priority: 232;; Evoke: ;; Urgency: ;; */ Data: fa := read last {FATTY ACID DISORDERS} ; hal := read last {TYPE OF PROTEIN FEED} ; ;; Priority: 232;; Evoke: ;; Logic: If hal = 'HYPERAL(TPN)' then conclude False; If fa = 'C8-C-01-003' then conclude True; ;; Action:write ("The above results are suggestive of Medium Chain ACYL-CoA Dehydrogenase Deficiency. An immediate recollection is necessary to further evaluate this infant"); ;; end:

Arden SyntaxMedical Logic Module

Data: fa := read last {FATTY ACID DISORDERS} ; hal := read last {TYPE OF PROTEIN FEED} ; ;; Priority: 232;; Evoke: ;; Logic: If hal = 'HYPERAL(TPN)' then conclude False; If fa = 'C8-C-01-003' then conclude True; ;; Action:write ("The above results are suggestive of Medium Chain ACYL-CoA Dehydrogenase Deficiency. An immediate recollection is necessary to further evaluate this infant");

September 27, 2007

Probabilistic Scoring ExampleAGREEMENT RATES

• 2 files, 10 Records each

• Form all possible pairs: 10 x 10 = 100 pairs

• Human review of all 100 record pairs shows that 10 are true-links, 90 are non-links.

September 27, 2007

Probabilistic Scoring Example AGREEMENT RATES

• Among the 10 true-links, the last names agreed in 9/10 pairs (e.g. one of the last names was misspelled)

• This represents a 90% AGREEMENT RATE for last name among TRUE LINKS.

• Similarly, among the 90 non-links, last names agreed (by random chance) in 2/90 pairs

• This represents a 2% AGREEMENT RATE for last name among NON-LINKS.

September 27, 2007

Probabilistic Scoring Example AGREEMENT RATES

90%90%

2%2%

= = 4545 Records that agree on last Records that agree on last name are name are 45 45 times more likely times more likely to be a true-link than a non-to be a true-link than a non-linklink Weights for each field are combined to Weights for each field are combined to

form a composite record pair score.form a composite record pair score. Field disagreement contributes a Field disagreement contributes a

negative weight, and reduces the overall negative weight, and reduces the overall record pair score.record pair score.

September 27, 2007

No Change in Workflow

• Since PHESS requires information that is already being collected by the registration system, there is – no change in work flow, (unless chief complaint is not

currently captured)

• The only upfront work involved is setting up connectivity and an HL7 interface. – On average it takes approximately 11 hours of a network

engineer and interface programmer’s time for this project– Record go live time set on 3/6/2006 – under 3 hrs

september 27, 2007 health information exchange: linking practitioners and public health for decision...

Documents

inpc slide

day slide

translated slide

fax slide

alert physician slide

data dictionary data

patient care inpc

data structure