Quality Improvement in the Laboratory

1
Quality Improvementin the Laboratory

• Satellite Meeting 8
• Focus on Information Technology for Laboratory
  Medicine Bologna, Italy
• REED M. GARDNER, Ph.D.LDS Hospital University
  of Utah
• Salt Lake City, UT
• Reed.Gardner_at_hsc.utah.edu
• 5 June 1999 1230 to 1300

2
Most Important Technology to Improve Patient Care
1993 HIMSS Survey
Clinical decision support
Other technologies
Multimedia-images, voice, and data
Hand-held computers
Healthcare Information and Management Systems
Society
3
There are serious problems with the quality of
health care
AMA News 1997
4
Transcribed Dictation
ICU/SURGERY
Nurse Care Plans
Pulse Oximeters
Infectious Disease
Nursing Procedure Charting
Ventilators
KNOWLEDGE DATABASE
Pharmacy
Mixed Venous Saturations
IV Pumps
ECG Lab
DECISION MAKING PROCESSOR, DATA AND TIME DRIVER
MIB Data
Bedside Monitors
INTEGRATED CLINICAL DATA BASE
Surgery Anesthesia Charting
Physiologic Data
X-Ray
Laboratory
Blood Gas Lab
Data Review Alerts Computations Interpretations Pr
otocols
Pathology
Surgery Schedule
Blood Bank
Admitting
Catherization
Medical Records
Respiratory Therapy
OUTPUT
5
(No Transcript)
6
Monitor 13
Other 2
Observation 21
Laboratory 33
Drugs I/O IV 22
Blood Gas 9
7
(No Transcript)
8
Describe the situation that resulted in missing,
incorrect or misleading data.
9
Free Text Results(preliminary)
10
Data missing or Incorrect

• Laboratory results converted from Negative
  (true result) to Positive in transfer from lab
  system to hospital system.
• Unable to determine medication history so wrong
  orders given.
• Patient was on screen but then computer reverts
  to another pt house staff me got wrong lab
  (different pt) and reacted to it.?

11
Lab Results

• Labs were put in on the wrong patient with the
  same name.
• There is a problem with the lab data getting
  into the HELP system promptly.
• Usually missing laboratory data, sometimes from
  this admission sometimes from previous
  admissions.
• Computer report of labs didnt match phone
  report.

12
In your opinion, what is the greatest potential
for clinical computing systems to do Harm?
13
Free Text Results(preliminary)
14
Lack of Human Validation

• If the humans fail to confirm data when
  appropriate.
• Having inappropriate decision support
  suggestions followed without questioning it.
• To assume the data of (the) computer is
  infallible.
• Inability to recognize and correct mistakes.

15
Quality improvement is the science of process
management Health Care delivery is a system
made up of thousands of interlinked processes
16
Quality of Care Questions

• 1. Are we doing the right thing?
• 2. Are we doing things right?
• 3. How can we be certain that it is done right
  the first time every time?

17
(No Transcript)
18
The Circle of Patient Treatment Optimization
What To Do?
What Effect Did It Have?
Get It Done!
What Was Done?
Do It Right!
19
Medicine is inherently an information science!!
As a clinicians, the better knowledgeand
information that is available, The better they
can diagnoseThe better counsel and advice they
can giveThe better treatments they can offer
The better outcomes they can achieve
20
You manage what you measure!
21
Data Collection Issues

• 1. What data are needed ?
• 2. How frequently are data needed?
• 3. Who supplies the data?
• 4. How do you collect the needed data?
• 5. How reliable are the data?

22
10 Potholes in the Road toInformation Quality
Adapted from Strong, Yang, and Wang IEEE
Computer August 1997, 30(8)38-46.
23
1Multiple sourcesof the sameinformationproduc
e differentvalues.

• Example When a hospital used two separate
  procedures to assess severity of illness.

24
2 Information is produced using subjective judgme
nts, leading to bias.
Example When medical coders use their
independent judgment to select disease codes.
25
3Systemic errorsin informationproductionlead
to lostinformation.
Example Information was missing because an edit
check would not accept, or incorrect information
was entered so edit check would accept.
26
5Distributedheterogeneoussystems lead
toinconsistentdefinitions,formats, andvalues.
Example Different systems in each division,
each using a different format for diagnostic
codes.
27
6Non-numericinformation isdifficult toindex.
Example Medical image information is relatively
easy to store but difficult to access (large
files) doctors notes are difficult to enter
(mostly free-text) but relatively easy to access.
28
7Automatedcontent analysisacross
informationcollections is notyet available.
Example Difficulties in analyzing trends in
image and text information Is pneumonia
increasingly common in ICU patients?
29
8As informationconsumers tasks andthe
organizationalenvironment change,the
information thatis relevant anduseful changes.
Example The basis for insurance reimbursement
to hospitals changes (capitation of care),
requiring changes in information processes and
systems.
30
10Lack ofsufficientcomputingresourceslimits
access.
Example Unreliable communications lead to
incomplete information. Shortage of terminals
problematic.Complaints at LDS Hospital Always
down, not enough terminals, and system too slow
31
(No Transcript)
32
Data sources for Antibiotic Assistant
Antibiotic Assistant
Medical Records Diagnosis, Costs, Allergies
Findings
Lab Kidney Function, Liver Function, Micro
Results, Sensitivities, Susceptibilities
Dictation X-ray and Pathology transcription
to text natural language processing
33
Continuous Quality ImprovementInvolves
continuous search for small opportunities to
improve patient careComputers can be used to
optimize real-time quality improvement
34
(No Transcript)
35
Adverse Drug Events

• Most common adverse event (hospitals)
• Generic screen for quality
• Low voluntary reporting (U.S.)
• Cost
• Regulatory Agencies - WHO, FDA, JCAHO
• Preventable

36
Attributes of an Effective Protocol

• 1. Scientifically valid
• 2. Clinically relevant
• 3. Reliable sophisticated enough
• 4. Broad development team
• 5. Explicit, clear, unambiguous user friendly
• 6. Easy to review modify
• 7. Improve care make practice more
  efficient effective
• Clemmer Spuhler 1998

37
Protocol Implementation Strategies

• 1. Select area of focus carefully
• 2. Seek feedback from protocol users
• 3. Face-to-face dialog
• 4. Small group consensus conferences
• 5. Power of opinion setter
• 6. Computer-assisted systems
• Clemmer Spuhler 1998

38
Expected Response from Protocol Implementation

• 1. Innovators - a few creative thinkers
• 2. Early adopters - pioneer group
• 3. Early majority - willing to act on faith
  alone
• after this group you have a critical mass
• 4. Late majority - keep change honest
• 5. Late adopters - only reluctantly accept
• Clemmer Spuhler 1998

39
Identification of ADEs
Period ADEs Pats May 88 - April 89
9 (0.04) 25,142 May 89 - April 90 373
(1.6) 23,297 May 90 - April 91 560
(2.3) 22,247 May 91 - April 92 509
(2.3) 21,963
JAMA 1997 277301-306
40
Drug Administration Errors
Ordering Errors
ADE
Psychic
Neural
Compliance
Ann Pharmacotherapy 1994 28497-501
41
Adverse Drug Events
Mortality Length of Cost of
() stay(days)
Hospitalization()

• Case patients 3.50 8.19
  10,584
• Matched cohort 1.05 4.36
  5,350
• patients
• Attributable ------- 1.94
  1,939
• difference

plt.001 by chi square, plt.062 by paired t test,
p.147 by paired t test, plt.05 by t test
Classen DC. MS Thesis in Medical Informatics,
University of Utah 1993.
42
Continuous Quality Improvement Applying
Computers to Medicine

• IFCC WorldLab 99
• Florence, Italy
• REED M. GARDNER, Ph.D.LDS Hospital University
  of Utah
• Salt Lake City, UT
• Reed.Gardner_at_hsc.utah.edu
• 9 June 1999 1630 to 1700