Laboratory Informatics: Becoming a Key Hub in the Hospital Information Network

1
Laboratory Informatics Becoming a Key Hub in
the Hospital Information Network
Anand Dighe MD, PhD
Director, Core Laboratory Director of Information
Management Assistant Professor, Harvard Medical
School Massachusetts General Hospital Boston, MA
2
Outline

• Trends in information management
• Fusing information management strategies with
  operations
• Auto-identification (bar codes/RFID)
• Improving test ordering and utilization
• Improving test interpretation

3
The Science of Networks (Lazlo Barabasi)

• Most stable and robust real-world networks adopt
  a particular configuration (a scale free network
  with multiple hubs)
• Animal relationships in an ecosystem
• World trade
• The Internet
• Social networks
• Molecules involved in cellular metabolism

Trade Network
Social Network (science co-authorship network)
Internet E-commerce Sites
4
Hospital Information Networks As a Scale Free
Network
Lab testing
Decision support
Interpretations
RFID
Patient ID
Medicine
Ordering guidelines
Pathology
INFO SYSTEMS
Pharmacy
Clinician

• Information hubs
• provide
• Stability of expertise
• Efficiency
• Robustness

Radiology
5
Information Management Concepts

• The information based society has arrived
• Organizations that will succeed in the global
  information environment are those that can
  identify, value, create, and evolve their
  information assets
• Most of what goes on in a hospital is actually
  decision making driven by information processing
• The critical limiting factor is the individuals
  limited ability to process information and make
  decisions

6
IVD Companies are Selling Information Services
Rules-based auto-dilution
Reflex algorithms
Specimen management
Automatedadd-ons
Simplified instrument interfaces
Auto-verification
Middleware
Lab InformationSystem
Analyzers
Alert laboratory staff via page or email
Rules-based sample flagging
Communicate with pharmacy and HIS for complex
rules-based decisions
Interferencechecking
Middleware is

• The coming of the information age to the
  laboratory?
• A reaction to the dismal capabilities of most Lab
  Information Systems?
• A ploy to lock labs into the vendors technology
  platform?
• All of the above.

7
Healthcare Trends

• Goals
• Improve access to testing
• Improve efficiency and quality
• Improve (and document) patient outcomes
• Reduce costs
• Trends
• Improve pre-analytic steps with systems
  approaches
• Electronic order communication from collection
  areas
• Improve ordering and utilization
• Pre-analytic automation
• Auto-identification (RFID/bar codes)
• Centralize and automate core laboratory testing
• Eliminate batch mode in all its forms (Lean/Six
  Sigma)
• Decentralize high impact critical care testing to
  critical care areas and operating rooms under the
  auspices of the central lab

8
Vendor Perspective (Why vendors are starting to
pay attention to information management)

• Focus on analytic performance (quality) and ease
  of use has been successful in selling IVD
• This has worked as a strategy since vendors were
  selling into a fragmented micro-markets in each
  hospital with little oversight
• This is unlikely to be the case in the future
• Individual hospital lab programs are increasingly
  less fragmented. Formation of core labs and POCT
  increasingly under central (typically laboratory)
  control to handle all aspects of test management
  from ordering to reporting.
• Potential future customers will be the connected
  enterprises that demand POC connectivity (a major
  determining factor in purchase)
• Current non-connected customers will slowly cease
  to exist (will be the losers in marketplace of
  healthcare)
• Outcomes data will be required for pay for
  performance. POC results will need to be part of
  the EMR to enable PFP reporting.
• Survey What feature would cause you to switch to
  another device? gtgt Information management 1
  answer.

9
Customer Loyalty and Additional Purchases
Banking

• Connectivity costs are sunk costs for an
  institution. Cost of interfaces also includes
  cost of time and IT resources to interface,
  install and test.
• Connectivity may benefit vendor as well as
  purchaser
• Customer loyalty and additional analyte purchase?
• Customers that use Online BillPay are 55 more
  likely to remain with the bank in 12 months than
  customers that do not use Online BillPay
• Online BillPay customers are 45 more likely to
  purchase other services from the bank than
  customers that do not use Online BillPay

10
POC Information Management

• POC testing accounts for 10 of all testing
• As little as 10-20 of POC testing is managed by
  the central laboratory computer
• Critically important results are not in
  electronic record
• Management of POC data is often done manually
• Billing handled manually or not at all
• This is at odds with the increased reliance on
  mobile computing, wireless devices, dynamic data
  displays, and electronic flowsheets throughout
  healthcare
• ? All hospitals/laboratories need to develop a
  plan for POC data management

11
Moores Law and Medicine (A certain
inevitability to this ride)

• The number of transistors on an integrated
  circuit therefore performance will double every
  18 months. (Gordon Moore, 1965)

• What this means for medicine
• High dimensional, integrated data driven by
  Moores Law
• - Bioinformatics helping make sense of
  multi-marker testing (sepsis, ACS)
• Rich tool sets to analyze the data in real time
  and for outcomes analysis
• IT as the driver of workflows
• Central to preventive, efficiently delivered care

12
Clinical Information Systems A shift has
occurred Goal to create environments supporting
decision making

• Prior emphasis
• Electronic medical record, information retrieval
  and reporting, scheduling, billing
• New emphasis
• Cost effectiveness, error prevention, safety and
  quality
• Decision support at the point of care
• Increased attention to sharing and
  standardization of knowledge, data warehousing
• Mobile computing, wireless connectivity

13
Expectations For Data Access Are Higher Than Ever
However..
14
Recommended Care is Frequently Not Given

• In U.S. overall only 55 of patients get
  recommended level of care
• VA patients (that have access to decision support
  tools at point of care, mature EMR) get 67 of
  recommended care

NEJM 35411 (2006)
15
Reports on Medical Error
2001
1999
16
Increasing Medical Error as the Groans of a
Collapsing System?

• Against the background of an explosively growing
  knowledge base current models of clinical
  decision making by autonomous practioners relying
  on their memory and personal experience are
  inadequate
• Embedded in our culture is the notion of finding
  a good doctor when what we should be looking for
  is a good health care system
• Greater than the sum of its parts
• Acts on a knowledge base of accumulated best
  evidence that can change quickly if necessary
• Utilizes decision support to assist clinicians in
  diagnosis and treatment
• Focuses on outcomes

17
Types Of Outcomes

• Medical outcomes Live longer, better
• Very difficult to document
• Showing that statins improve medical outcomes
  took decades
• Financial outcomes Save money, more cost
  effective
• Complex and difficult to document
• Operations outcomes Improve length of stay,
  improve efficiency, streamline processes
• Easier to document

18
Why Outcome Focus is Important in a Complex
Organization
21 steps and handoffs required to discharge a
patient and turnover a bed
19
JCAHO National Patient Safety Goal 1 Reduce
patient identification errors
Proposed JCAHO 2008 NPSG The organization
investigates and initiates planning for the use
of technology to assist with patient
identification.
20
Comparison of Auto-ID Implementations
Bar codes have been around since the 1970s. Why
so poorly adopted in healthcare?
Retail
Healthcare

• No industry giant no standards
• Bar codes on many things
• Point of care
• New work flow
• Life impacting errors are possible
• Success better patient care, error avoidance

• Industry giant sets standards
• Bar codes on products
• Checkout
• Same work flow
• Minor monetary consequences of error
• Success reduced labor costs, increased customer
  and revenue growth

Bar codes generally work best in situations where
the bar code scanner is stationary and the
products are moving (checkout, specimen
accessioning). Problematic when the scanner is
mobile AND the products are mobile (POCT)
21
But.they work! Bar Coded Patient Wristbands
Reduce Errors
MGH Patient Wristband

• MGH patient wristband has 1D and 2D bar codes
• At MGH scanning the 1D bar code instead of
  typing it in dropped the glucometry patient ID
  error rate from 1-3 to zero

22
Strategy Maximize Use of Auto identification
Technologies

• Auto Identification Technology technology by
  which a physical object can be automatically
  identified.
• Bar coding
• Magnetic stripe cards (credit cards)
• Biometric (fingerprint and retinal scans)
• Voice recognition
• Optical character recognition
• Radiofrequency identification

23
Healthcare RFID Market

• 2006 global RFID market is 4 billion
• Healthcare market is small but growing fast
• 45 of hospitals have RFID spending planned for
  2007 (up from 10 in 2005)
• Hospital applications
• Patient tracking
• Equipment tracking
• Documentation enhancement
• Process/capacity management

Healthcare RFID Market in U.S.
1.2 B
90 M
24
Radiofrequency Identification (RFID) is
Everywhere

• Hospital (IDs for door access)
• Train (Subway pass)
• Car (FastLane Mass Pike)

Mass Pike Toll Booth Transponder
Hospital ID
MBTA Subway Pass
25
Types of Radiofrequency Identification

• 2 major types of RFID systems
• Passive RFID
• Active RFID
• Each has different strengths and weaknesses
• Essential to match the right technology to the job

26
Passive RFID
Integrated circuit with memory NMJim
SmithMRN301-93-9322
Reader antenna
Electromagnetic field of reader energizes the tag
Tag sends data in its memory to reader
Tag antenna
1
Passive RFID tag
2
301-93-9322 Jim Smith

• No battery (passively powered)
• Tag may hold 100-1000s of characters

RFID Reader
3

• Provides power to tag

• Server/Computer
• Connects tag data with user

27
Blood Transfusion Safety

• Transfusion of ABO incompatible blood occurs in 1
  of every 38,000 transfusions
• Mistransfusions are typically caused by
  distractions or failure to pay attention at a
  critical moment (slip errors)
• ? This category of error is often amenable to a
  technologic solution

2005 www.shotuk.org
28
Blood Transfusion Safety (Dr. Dzik, MGH)
MRN 187-96-45

• MRN 187-98-45

Blood product with RFID tag
Passive RFID Reader in OR
Patient wristband with RFID tag
29
Active RFID TagsWHAT is it and WHERE is it
right now?

• Active tags operate at a longer range than
  passive tags
• Active tags have a battery
• Active tags check in at regular intervals

Antenna
RFID Reader Network
1
Jim Smith, 301-93-9322
2
Radianse Tag
Server/Computer
Battery
301-93-9322 Jim Smith Location OR37
Integrated circuit with memory NMJim
SmithMRN301-93-9322
3
Reader 21 (located in OR37)
Jim Smith is close to Reader 21 ? Jim Smith is in
operating room 37
30
Active RFID (Radianse Indoor Positioning
System in MGH ORs )
Reader
After tags of patient and surgeon have left OR38
gtgt alert cleaning staff to clean OR38 via text
pager If ultrasound machine is not in OR53 and
needed for the next case gtgt Alert RN via text
pager to find ultrasound machine and bring it to
OR53 ? Results reduced time between cases,
increased number of cases in same amount of time
Tag (10)

• Patients
• Surgical staff
• Cleaning crew
• Equipment (e.g. ultrasound machines)

31
Bringing Technology to the Lab Testing Loop
Providing external memory to improve decisions
Error in result interpretation
Action
Help clinicians interpret complex tests Automate
the interpretive process
Interpretation
Ordering
Help clinicians pick the right tests Smart
Online test manual Computerized order entry
1
Reporting
Collection
3
2
Identification
Analysis
Transportation
Preparation
32
MGH Online Laboratory Manual

• Essentials of an online laboratory handbook
• Easily accessible in normal clinician workflow
• MUST be locally updated and maintained and not
  include generic test information
• Smart searching to allow searching by test name,
  synonym or disease
• Provides information regarding appropriate test
  usage and basic interpretive information

33
MGH Online Laboratory Manual
Search results
34
MGH Online Laboratory Manual
Details

• Differs based on audience.
• More interpretive info for clinicians, more
  specimen info for PCAs and phlebotomists

35
Monitoring Usage (Lab Handbook)

• Determine whos eyes are on the site
• Assess marketing efforts
• Use monitoring data to improve content,
  navigation, and search algorithms

Connect to CIS, POE
JCAHO
New residents
Lab Week
36
Learning from Users Search Monitoring

• Monitoring user searches (what was typed into the
  lab search engine)
• Gives insight into how users are looking for lab
  tests
• Allows rapid cycle improvements

• Realize that past searches all have value.
• Lab handbook data now used for POE applications
• Data capture/review should be part of all search
  applications

AJCP (2006) 126(2)1-7
37
Pathology Inside
Big Database in the Sky Model is OUT.
Collaboration is IN. (Use web services/XML to
serve up our content to the rest of hospital)
Outpatient Order EntryEnd of Visit Module
Inpatient Provider Order EntryPOE-Misys Order
Communication
Clinical Information System (KnowledgeLink)
SQL Database
Laboratory Handbook
Misys LIS
38
Bringing Technology to the Lab Testing Loop
Providing external memory to improve decisions
Error in result interpretation
Action
Help clinicians interpret complex tests Automate
the interpretive process
Interpretation
Ordering
Help clinicians pick the right tests Smart
Online test manual Computerized order entry
1
Reporting
Collection
3
2
Identification
Analysis
Transportation
Preparation
39
Daily Rounds to Interpret Complex Test Results

• Physician orders non-routine tests within certain
  areas of the laboratory (examples coagulation,
  immunology, hematology, toxicology)
• Test results as numbers and a narrative
  interpretation by the pathologist are provided to
  the clinician
• The interpretation includes a patient specific
  differential diagnosis and recommendations for
  future testing

40
Payment Requirements for Interpretive Services

• CMS Carrier Manual
• The interpretation must be requested by the
  patient's attending physician
• - A hospital's standing order policy can be
  used as a substitute for the individual request
• The interpretation is a written narrative report
  included in the patient's medical record
• The interpretation requires the exercise of
  medical judgment by the consultant physician

http//cms.hhs.gov/manuals/
41
Improving Interpretive Efficiency (Pathologist
and Clinician)
Seamless integration of interpretive services
with the Laboratory Information System Automate
the process with the use of interpretive software
and a direct HL7 interface of tests and
interpretations to facilitate the efficient
production of a high quality, reproducible
interpretation
Interp Request
Finalized Interpretation
Clinical Information System
Laboratory Information System
Interpretive Software
HL7
HL7
Clinicians request and view interpretations
Pathologists create interpretations
42
(No Transcript)
43
(No Transcript)
44
(No Transcript)
45
Click to expand view of labs
46
(No Transcript)
47
Metcalfe's law the value of network is
proportional to the square of the number of users
of the system.
Click to forward
Nodes all bring Cases Ideas Algorithms Coded
comments Experts
Interpretation Central
48

• Software selects comments based on
• Labs
• Patient demographics
• Past history of comment use
• Interpreter

Click to interpret
49
Click to sign
50
(No Transcript)
51
(No Transcript)
52
Enhanced Result Reporting
1) Test results for tests performed in
hypercoagulation reflex algorithm
2) Interpretation of complex results
3) Link to MGH Lab Test Handbook
53
KnowledgeLink Just in Time Information
54
Information Management in the Laboratory Testing
Process
Its not just about doing the lab test. Labs
should aim to provide an array of information
intensive services and become the hub for the
creation, maintenance, and dissemination of a
wide variety of lab related knowledge

• Info Buttons
• Guidelines
• Literature
• Online resources

• Middleware
• Interference checking
• Rules-based auto-dilution
• Automated add-ons

Test Result Auto-verification
PathologyInterpretativeServices

• Computerized Provider
• Order Entry (CPOE)
• Test panels
• Redundancy alerts
• Clinical guidelines

Post-Analytic
Analytic
Pre-analytic
PROCESS
Inter-pretation
Reporting
Processing/Analysis
Collection
Ordering
Automated Specimen Collection Process RFID/bar
coding
Enhanced Electronic Medical Record systems
Institutional ReflexAlgorithms
ElectronicTechnicalSupport
55
Just In Time Knowledge Management

• The key to success is to bake knowledge into the
  jobs of highly skilled workers to make
  knowledge so readily accessible that it cant be
  avoided

Glaser and Davenport, HBR, July 2002
56
Thanks