RTI International
Health IT Safety Webinar Series
CPOE, CDS and Health IT Safety
April 23, 2015
1:00-2:30pm EDT
RTI International is a trade name of Research Triangle Institute.
1
www.rti.org
RTI International
Housekeeping





2
For any technical questions, please type your question into the
Questions panel at lower right.
All telephone lines are muted. Due to the number of attendees,
please use the Questions panel to ask any questions during the
webinar.
Closed captioning for today’s session is available at
http://www.captionedtext.com/ Event #: 2595609
Q&A will take place at the end of each presentation. Slides and a
copy of the recording of this session will be posted at
www.healthitsafety.org
For general questions about the webinar series, please contact
healthitsafety@rti.org
RTI International
Health IT Safety Webinar Series
This series of 10 webinars focused on health IT and patient
safety issues will occur monthly through September 2015.
These webinars are funded by the Office of the National
Coordinator for Health Information Technology (ONC) and
are being conducted by RTI International, a non-profit
research organization, as part of a year-long project to
develop a road map for a Health IT Safety Center for ONC
(contract HHSP23320095651WC).
Additional information is available at:
www.healthitsafety.org
3
RTI International
Today’s Presentations
David Classen, M.D., M.S., Chief Medical Information Officer,
Pascal Metrics; Associate Professor of Medicine at the
University of Utah
David W. Bates, M.D., M.Sc., Chief Innovation Officer and
Senior Vice President, Brigham and Women’s Hospital; Chief,
Division of General Internal Medicine, Brigham and Women’s
Hospital; Medical Director of Clinical and Quality Analysis,
Partners HealthCare
(Moderator) Barry Blumenfeld M.D., M.S., is a Senior
Physician Informaticist in the Center for the Advancement of
Healthcare IT, RTI International.
4
RTI International
Today’s Speakers
Dr. Classen is an Associate Professor of Medicine at the University of
Utah and conducts Patient Safety Research at the University of Utah,
School of Medicine. Dr. Classen was a co-developer of the Institute for
Healthcare Improvement (IHI) Global Trigger Tool. He was a member
of the Institute of Medicine (IOM) Committee that developed the
National Healthcare Quality Report and was also a member of the
IOM Committee on Patient Safety Data Standards as well as the IOM
Committee on HIT and Patient Safety. Dr. Classen is an advisor to the
Leapfrog Group and has developed and implemented the CPOE/EMR
flight simulator for the Leapfrog Group and National Quality Forum
which has been funded and overseen by AHRQ. He is board certified
in Internal Medicine and Infectious Diseases.
Presentation:
“Optimizing the Patient Safety Impact of Operational EHRs”
5
Optimizing the Patient Safety
Impact of Operational EHRs
David C Classen MD, MS
Associate Professor of Medicine University of Utah
and
CMIO Pascal Metrics
April 2015
© 2006 HCC, Inc. CD000000-0000XX
6
A Patient Safety Case in The HIT Era
• 69 year old women admitted for elective colon resection
for diverticuli
• 2 days post op she develops pneumonia and is
transferred to the ICU
• On the second ICU day the patient suffers a prolonged
period of unrecognized hypotension and is ultimately
found to be septic and ultimately dies
• On review of the case a malfunction in the bedside
monitor/EHR Interface led to an inaccurate blood
pressure reading in the EHR blood pressure display
7
A Patient Safety Case in The HIT Era
• 27 year old women evaluated in the ER for severe lower
abdominal pain
• Taken to surgery for what was felt to be an acute
abdomen
• At surgery she was found to be pregnant and the fetus
did not survive
• On review of the case a problem with interoperability
lead to another patients lower abdominal ultrasound
report being inadvertently inserted into this patients EHR
record
8
Health IT and Patient Safety:
Building Safer Systems for Better Care
Safety Comparison Between Industries
JCAHO SE:
14 events
per 1,000,000
admissions*
IRS - Tax Advice
(phone-in)
(140,000 PPM)
PPM
1,000,000
100,000
IHI GTT – 45
events per
100
admissions
10,000
1,000
•
•
Airline Baggage Handling
Blood transfusion
Domestic
Airline Flight
Fatality Rate
(0.43 PPM)
100
10
Nuclear
Industry
1
DEFECTS
50%
31%
7%
1%
0.0003%
Ultra safe
Very unsafe
© 2013 Pascal Metrics
0.02%
•
REFERENCE: René Amalberti
*JCAHO sentinel events statistics 200610
AHA : hospital admissions, 2006 survey
US Government Study
11
Sample for National Incidence Study
October
2008
780
• Medicare beneficiaries
discharged from acute
care hospitals
• Sample Medicare
beneficiaries
661
• Hospitals represented
999,645
12
• Sample month
Incidence Rates – of all beneficiaries
13.5%
0.6%
• NQF Serious Reportable Events
1.0%
• Medicare Hospital-Acquired
Conditions
13.5%
13
• Adverse Events
(NQF, HAC, F– I Level)
• Temporary Harm Events (E
Level)
Health IT and Patient Safety:
Building Safer Systems for Better Care
Recommendation 1 (continued)
b. The Office of the National Coordinator for Health IT (ONC)
should expand its funding of processes that promote safety
that should be followed in the development of health IT
products, including standardized testing procedures to be
used by manufacturers and health care organizations to
assess the safety of health IT products.
c. ONC and AHRQ should work with health IT vendors and
health care organizations to promote post-deployment safety
testing of EHRs for high prevalence, high impact EHRrelated patient safety risks.
d. Health care accrediting organizations should adopt
criteria relating to EHR safety.
e. AHRQ should fund the development of new methods for
measuring the impact of health IT on safety using data
from EHRs.
17
Can CPOE Cause Errors?
© FCG 2007 | Slide 18
CDS
Unexpected Increased Mortality After
Implementation of a Commercially Sold
Computerized Physician Order Entry System
Scott Watson, Trung C. Nguyen, Hülya Bayir and
Richard A. Orr
Yong Y. Han, Joseph A. Carcillo, Shekhar T.
Venkataraman, Robert S.B. Clark,Richard A Orr.
Pediatrics 2005;116;1506-1512
© FCG 2007 | Slide 19
CDS
20
Recommendation 1 (continued)
b. The Office of the National Coordinator for Health IT (ONC)
should expand its funding of processes that promote safety
that should be followed in the development of health IT
products, including standardized testing procedures to be
used by manufacturers and health care organizations to
assess the safety of health IT products.
c. ONC and AHRQ should work with health IT vendors and
health care organizations to promote post-deployment safety
testing of EHRs for high prevalence, high impact EHRrelated patient safety risks.
d. Health care accrediting organizations should adopt
criteria relating to EHR safety.
e. AHRQ should fund the development of new methods for
measuring the impact of health IT on safety using data
from EHRs.
21
Features of safer health IT
22
SAFER Guides:
Safety Assurance Factors for EHR Resilience
Kathy Kenyon, JD MA, Office of the National Coordinator
Joan Ash, PhD MLS, MS, MBA, Oregon Health & Science University
Hardeep Singh, MD MPH, Houston VA and Baylor College of Medicine
Dean Sittig, PhD, University of Texas School of Biomedical Informatics
January 30, 2014
23
SAFER: Safety Assurance Factors
for EHR Resilience
• Foundational Guides
– High Priority Practices
– Organizational Responsibilities
• Infrastructure Guides
– System Configuration
– System Interfaces
– Contingency Planning
• Clinical Process Guides
–
–
–
–
Patient Identification
Computerized Provider Order Entry with CDS
Test Results Reporting and Follow-up
Clinician Communication
24
Information Transfer
and Clear Communication
25
-Excerpt-
CPOE may be adopted
with a stage approach
once integrated information
systems are in place to
support safety and effective
CPOE systems…
The CPOE system is tested
against The AHRQ/NQF
Inpatient CPOE Testing
Standards…developed to
provide organizations that are
implementing CPOE with
appropriate decision support
about…
26
AHRQ EHR Flight Simulator
“Anyone here know how to play
Microsoft’s Flight Simulator?”
27
Principles Behind the
Evaluation Methodology

Principle #1: Target the Harm
– Common sources of ADE’s (not errors)
– Sources of severe harm (existing literature and expert consensus)

Principle #2: Encourage Quality Improvement
– Categorize test set by type of error
– Provide feedback to the provider organization for each category
– Provide advice about nuisance alerting

Principle #3: Accentuate the positive
– Encourage quality, as well as harm reduction (ADE’s)
Address errors of commission and omission
Include corollary orders and duplicate interventions
© FCG 2007 | Slide 28
CDS
© FCG 2007 | Slide 29
CDS
The Assessment Methodology
Many Research Databases Used
Research background, combined with the practical experience of the EHR pioneers,
was first used to define the focus.
Preventable ADEs in 10.4/100 admissions to six community hospitals
Types of CPOE-preventable ADEs
Percentage*
Patient Diagnosis
1
Duplicate Med Check
1
Drug-drug
2
Drug Frequency
3
Drug Allergy
4
Drug-specific Guidelines+
7
Drug-age
9
Drug dose Suggestion (typical)
9
Renal Check
19
Drug-lab Check
27
* All sites
+ Ondansetron
Source: Bates et al. “Saving lives, Saving money: The Imperative for Computerized Physician Order Entry in
Massachusetts Hospitals.” The Clinical Baseline and Financial Impact Study. MTC and NEHI. February 2008.
Web-Based Evaluation Tool

Self-administered testing managed by a Web application

Separate tests for pediatric and adult, inpatient and outpatient

Test order set
– To be entered into the site’s CPOE system or EHR, against Leapfrogsupplied “test patients”
– System responses recorded and reported back to Leapfrog (Overall score)
and to the organization taking the test (detailed feedback)

Test orders representing nine categories of potentially dangerous errors
developed by FCG and ISMP

Three additional order categories developed based on literature and
advisor experience
– Corollary
– Cost of care
– Nuisance (important feedback)
© FCG 2007 | Slide 31
CDS
The Assessment Methodology
Simulations of EHR Use with CPOE
The assessment pairs medication orders that would cause a serious adverse drug event with
a fictitious patient.
A physician enters the order …
Patient
AB
Female
52 years old
Weighs 60 kg
Allergy to morphine
Normal creatinine
and observes and records the type of CDS-generated advice that is
given (if any).
Coumadin (Warfarin) 5 mg po three times a day.
32
Web-Based Evaluation Tool (cont.)

For ambulatory test: additional capability to test basic health
maintenance prompting

Outputs received immediately after submitting results
– Individual site performance feedback
 Indicating performance in each medication order category
 Indicating performance for health maintenance (ambulatory only)
– Sensitivity = the ones that you got right (percentage)
– Specificity = how many did you get that you should not have
(percentage)
– Aggregate score for public reporting - similar to the Leapfrog Hospital
Quality and Safety Survey
© FCG 2007 | Slide 33
CDS
Web-Based Evaluation Tool
Hospital
Hospital
Logs
Logs-On
-On
(Password
(Password
Access)
Access)
Hospital
Complete
Logs-On
Sample
(Password
Test
Access)
Obtain
Obtain
Patient
Patient
Criteria
Criteria
(Adult
(Adult or
or
Pediatric)
Pediatric)
Program
Program
Patient
Patient
Criteria
Criteria
Review Patient
Descriptions
Hospital
Hospital
Self
Self
Reports
Reports
Results
Results
on
on
Website
Website
Score
Score
Generated
Generated
Against
Against
Weighted
Weighted
Scheme
Scheme
Review Scoring
© FCG 2007 | Slide 34
Download
Download
and
andPrint
Print
30
30--40
40
Test
Test
Orders
Orders
(HM if AMB)
Enter
Orders
into
intoCPOE
CPOE
Application
Application
& Record
Results
Review Orders
and Categories
Report
Report
Generated
Generated
Aggregate
Score to
Leapfrog
Order
Category
Category
Scores
Scores
Viewed
Viewedby
by
Hospital
Hospital
CDS
Pick Adult Inpatient or Pediatric inpatient and Download the patients
Page 35
Print the patient descriptions
See pdf file for ALL patients
Page 36
Sign-out and set-up the patients (you have four hours to come back to the
application).
Page 37
Download the orders
Page 38
Print the orders and enter into your
CPOE system against the assigned
test patients.
Use this sheet as a worksheet /
notes page. Sign-out at the end of
the page.
Page 39
You have two hours to enter the orders into your CPOE system and come back to the site to enter in your
responses.
Page 40
Enter in your results.
Page 41
Submit your results.
Page 42
The Assessment Methodology
The team of advisors helped to define the order categories in the
assessment to reflect the sources of common, preventable ADEs
identified in research.
Order Category
Description
Example
Therapeutic duplication
Medication with therapeutic overlap with
another new or active order; may be same
drug, within drug class, or involve components
of combination products
Codeine AND Tylenol #3
Single and cumulative
dose limits
Medication with a specified dose that exceeds
recommended dose ranges or cumulative dose
Ten-fold excess dose of
methotrexate
Allergies and crossallergies
Medication (or medication class) for which
patient allergy has been documented
Penicillin prescribed for
patient with documented
penicillin allergy
Contraindicated route
of administration
Order specifying an inappropriate route of
administration (e.g., oral, intramuscular,
intravenous)
Tylenol to be administered
intravenously
Drug-drug interaction
Medication that results in known, dangerous
interaction when used in combination with a
different medication in a new or existing order
for the patient
Digoxin AND Quinidine
43
The Assessment Methodology
The team of advisors helped to define the order categories in the
assessment to reflect the sources of common, preventable ADEs
identified in research. cont.
Order Category
Description
Example
Contraindication/dose
limits based on patient
diagnosis
Medication either contraindicated based on
patient diagnosis or diagnosis affects
appropriate dosing
Nonspecific beta blocker in
patient with asthma
Contraindication dose
limits based on patient
age and weight
Medication either contraindicated for this
patient based on age and weight or for which
age and weight must be considered in
appropriate dosing
Adult dose of antibiotic in a
newborn
Contraindication/dose
limits based on
laboratory studies
Medication either contraindicated for this
patient based on laboratory studies or for which
relevant laboratory results must be considered
in appropriate dosing
Normal adult dose regimen
of renally eliminated
medication in patient with
elevated creatinine
Corollary
Intervention that requires an associated or
secondary order to meet the standard of care
Prompt to order drug levels
when ordering Dilantin
Cost of care
Test that duplicates a service within a
timeframe in which there is typically minimal
benefit from repeating the test
Repeat test for Digoxin
level within 2 hours
44
Print your results and sign-out.
Page 45
46
Growth in Participation and Performance
900
Hospitals Reporting
800
Total Hospitals
700
600
Fully
Implemented
500
Good Progress
400
Good Early
Stage
300
200
Completed
Evaluation
100
Pre-Implementation
0
2009
© 2013 TMIT
2010
2011
2012
2013
48
48
Handled Correctly by Checking Category - 1
100%
90%
% Handled Correctly
80%
70%
Duplication
60%
Allergy
Drug Drug
50%
Route
40%
Drug Dx
30%
Corollary
20%
Lab
10%
0%
2009
2010
2011
2012
2013
49
© 2013 TMIT
49
IOM – Improving Safety Requires a Learning
System Built from a Sociotechnical Approach


Safety is a characteristic of a sociotechnical system
System-level failures occur almost always because of unforeseen
combinations of component failures
50
50
51
Questions?
Comments
© 2006 HCC, Inc. CD000000-0000XX
52
RTI International
Today’s Speakers

Dr. Bates is an expert in patient safety, using information technology to
improve care, quality-of-care, cost-effectiveness, and outcomes assessment
in medical practice. He is a Professor of Medicine at Harvard Medical School
and a Professor of Health Policy and Management at the Harvard School of
Public Health, where he co-directs the Program in Clinical Effectiveness. He
directs the Center for Patient Safety Research and Practice at Brigham and
Women’s Hospital, and serves as external program lead for research in the
World Health Organization’s Global Alliance for Patient Safety. He is
president of the International Society for Quality in Healthcare (ISQua) and
the editor of the Journal of Patient Safety. He serves as the principal
investigator of the Health Information Technology Center for Education and
Research on Therapeutics. He has been elected to the Institute of Medicine,
the American Society for Clinical Investigation, the Association of American
Physicians and the American College of Medical Informatics, and was
chairman of the Board of the AMIA. He has over 600 peer-reviewed
publications.
Presentation:
“CPOE and CDS—Lessons from Research and Implementation.”
53
David W. Bates, MD, MSc
Chief Innovation Officer, Brigham and Women’s Hospital
Medical Director of Clinical and Quality Analysis, Partners
Healthcare
54

Evidence that CPOE and CDS work
◦ Key specifics

Implementation
◦ Case examples
◦ Common pitfalls
 Overcoming them
◦ Getting value!

Conclusions
55
Total
Site 1
Site 2
Site 3
Site 4
Site 5
Site 6
ADE Rate*
15
19.5
11
15.5
17
15
12.5
% Prev
75
72
82
71
85
73
68
*Per 100 admissions
Range: 11-19.5 for rate
68-85 for percent preventable
56

Is centrally important because most things
that occur in a hospital happen as the result
of a physician’s order
◦ Need to get physician to use the computer
◦ Key opportunity to change behavior
 Many opportunities to improve performance

CDS is the mechanism that can be used to get
providers to change what they do
57








Alerting--high lab value
Reminding--mammogram
Critiquing--rejecting an order
Interpreting--interpreting an ECG
Predicting--risk of mortality using
severity score
Assisting--tailoring antibiotic choices
Diagnosing--ddx in CP
Suggesting--for adjusting mechanical
ventilator
Randolph et al, JAMA 1999, from Pryor, 1990
58

Alerts, reminders, critiques often simple ifthen rules
◦ (sometimes other Boolean operators)
◦ Alerts use event monitors, evaluate streams of data
 Finding right person hard
◦ Reminders notify patients of tasks to be done before event
occurs
◦ Critiques--alternative suggestions, evaluate plan after
started

Interpreting/predicting/diagnosing/assisting
and suggesting are higher order
◦ Harder to program, require more data
59
1.
2.
3.
4.
5.
Speed is everything
Anticipate and deliver
in real time
Fit into the user’s
workflow
Little things can make
a big difference
Physicians resist
stopping
6.
7.
8.
9.
10.
Changing direction is
fine
Simple interventions
work best
Asking for information
is OK—but be sure you
really need it
Monitor impact, get
feedback, and respond.
Knowledge-based
systems must be
managed and
maintained.
Bates DW et al, JAMIA 2003
60
Effect of real-time decision support for
patients with renal insufficiency
 Of 17,828 patients, 42% had some
degree of renal insufficiency
Interv
Control
Dose
Frequency

67%
59%
LOS 0.5 days shorter
54%
35%
Chertow et al, JAMA 2001
61



In most systems most alerts get overridden
We identified a highly selected set of drug alerts
for the outpatient setting
Over 6 months, 18,115 alerts
◦ 12,933 (71%) non-interruptive
◦ 5,182 (29%) interruptive
 Of interruptive, 67% were accepted
Shah, JAMIA 2006
62



Two academic medical centers
Same knowledge base
◦ Site A used 3 tiers
◦ Site B had all of the alerts as interruptive (Level 2)
Results
◦ 100% of most severe vs. 34% at non-tiered
◦ Overall alert acceptance higher at tiered site (29%
vs 10%, p<.001)
Paterno, et al, JAMIA 2009
63

Interrupt with only most important warnings
and tier
◦ Jury still out regarding non-interruptive warnings



Have regular review
Track how providers are responding as
practices change
Sharing regarding this would help
◦ Would be a common good
◦ Could be international
Kesselheim et al, Health Affairs, 2011
64



Need uniform alerting mechanisms and
standardized alarm responses
Alarm philosophies should minimize false
alerts
Placement of alerts impacts the likelihood
that users will see these alerts
◦ Visibility is critical, and font size should be large
enough to be readily legible
◦ All visual alerts should be prioritized
Phansalkar, JAMIA 2011
65



Color should help cue the user about the level
of a specific alert, and the number of colors
used should be minimized
To make visual alerts more distinct, it is
important to minimize the number of visual
features that are shared between alerts
Text-based information should be succinct
66



50,788 DDI alerts analyzed, both inpatient
and outpatient
Providers accepted only 1.4% of the noninterruptive alerts
For interruptive alerts, user acceptance
positively correlated with:
◦ Alert frequency (1.30, 1.23-1.38)
◦ Quality of display 4.75, (3.87-5.84)
◦ Alert level (1.74, 1.63-1.86)
Seidling et al, JAMIA 2011
67

Alert acceptance was higher:
◦ In inpatients (2.63, 2.32-2.97)
◦ For drugs with dose-dependent toxicity (1.13,
1.07-1.21)

Textual information influenced reaction
◦ Providers were more likely to modify the
prescription if the message contained detailed
advice on how to manage the DDI
68


Content is more generalizable
Management may have an even bigger impact
◦ Can have great content and no impact
◦ Management has at least two dimensions
 Attention to human factors issues in delivery, display
 Implementation

Have to get good score on all of these right
to get benefit!
69

15 drug-class pairs endorsed as highly
clinically significant DDIs
◦ Should never be co-prescribed
◦ Candidates for “hard-stop” alerts
◦ Checking completeness would require further research, but
represents best available consensus

Less-significant DDIs are still significant

To improve sensitivity and specificity of DDI
warnings:
◦ Much more prevalent and probably cause much more harm
◦ Tend to depend on patient characteristics, drug dosages
and timing, concomitant conditions such as hypokalemia,
etc.
◦ Need much more investment in evidence review and
generation
◦ Methods to make DDI alerts conditional on other patient
data
Phansalkar et al, JAMIA 2012
70


Alert fatigue is a serious problem
Used consensus approach to identify lowyield DDIs
◦ Used data from several sources to identify potential
candidates

Created a list of 33 DDIs that do not warrant
interruptive status
◦ Account for many of the DDIs displayed in some
systems

A consortium to maintain this list would be
helpful
Phansalkar, JAMIA, 2013
71
11
10
/1
3
/1
6
/1
0
/1
0
/1
0
/1
0
/1
0
/1
0
12000
9/
18
8/
21
7/
24
6/
26
10
10
10
10
/1
0
5/
1/
4/
3/
3/
6/
2/
6/
10
/0
9
/0
9
1/
9/
/1
2
/1
4
5/
29
12
11
Cumulative Number of Problems
14000
Intervention
10000
5,491
8000
6000
Control
4000
2000
0
Date
72
Differences in Achievement of Composite Standards for Diabetes Care and
Outcomes at Practices with Electronic Health Records (EHRs) and Those with Paper
Records.
Cebul RD et al. N Engl J Med 2011;365:825-833.
73
LCD monitor
Full floor overview
at a glance
Nurse’s phone Central Nurse’s Station Bed side monitor
Real time alerts to
nurses &
supervisors +
reports on team
performance
Nurse / physician
communication
support
Facilitation of
critical thinking
by nurse
74
Study Outcomes Comparing Study Units Before and After Implementation of
Monitor
Control Unit
LOS in Med.
Surg./ Units
(mean)
LOS in ICU for
patients
coming from
Med/Surg.
units (mean)
Code Blue
Events/ 1000
Pt.
3
Arms
p
value*
Intervention (Study) Unit
Baseline
(Pre)
Intervention
(Post)
P Value
%
Reduction
0.07
4.00
3.63
0.02
9%
<
0.01
0.01
4.53
(2.33)
2.45
(1.85)
0.1
45%
0.04
0.36
9 (6.3)
2 (0.9)
0.05
86%
0.01
Baseline
(Pre)
Control
(Post)
P
Value
3.80
(1.264.25)
3.61
(1.194.12)
1.73
(1.062.28)
4.48
(0.944.09)
3.9
2.1
* P – value comparing 3 arms: intervention unit
post, intervention unit pre and control unit post
75


BWH saved $28.5 million over 10 years, net
operating savings $9.5 million
Elements creating greatest savings:
◦
◦
◦
◦
Renal dosing guidance
Nursing time utilization
Specific drug guidance
Adverse drug event prevention
Kaushal et al, JAMIA 2006
76

Koppel et al evaluated on a commercial
CPOE application at U Penn and asked users
about their impressions about the system
◦
◦
◦

Found many situations in which “a leading
CPOE system facilitated medication error risks”
Often took many screens to do things
Needed views not available
Others including Ash have also reported on
this
Koppel, JAMA, 2005
77




Didn’t actually count errors or adverse
events
Said that other studies focused only on
advantages—not accurate
CPOE application studied was an old one
Nonetheless, paper stimulated valuable
debate and identified key points
◦
◦
Need change systems after implementation
Software alone is insufficient
Bates DW, J Biomed Inform 2005
78



Studied children transported in for special
care
Mortality rate increased from 2.8% to 6.3%
(OR=3.3) after introduction of a commercial
CPOE application
Study design was before-after
◦ Other changes were made at same time as CPOE
was implemented
◦ Overall mortality wasn’t reported
Han, Pediatrics 2005
79






CPOE was introduced very rapidly—over 6 days!
After implementation, order entry wasn’t
allowed until the patient had actually entered
hospital and been logged into system
After CPOE implementation, all drugs including
vasoactive agents were moved to central
pharmacy
Pharmacy couldn’t process medication orders
until after they were activated
Many order sets weren’t available initially
Result was substantial delays in care delivery
80


Study was very weak methodologically
Nonetheless, increase in mortality rate was
very large and of obvious concern
◦ Introducing substantial delays in this group could
easily have caused


Organization broke many of the rules for
implementation
Essential for other organizations to handle
sociotechnical aspects better
Phibbs et al, Pediatrics 2005
81

Don’t recognize how big a change this truly
is
◦ Expensive
◦ Huge process change!


Failure to sufficiently engage both
administrative and clinical leadership
Failure to do necessary preparation with key
stakeholders
◦ Often takes 2 years to have all the key groups
meet
82


Going too fast early on—e.g. turning on
whole hospital at once
Trying to fix previously existing policy
problems at the time you implement
◦ Easy to get stuck

Turning on too much decision support early
on
◦ Much better to phase in
83



Failure to provide users an easy mechanism
for reporting on-going problems
Failure to make sufficient changes to
application
Failure to devote sufficient resources to
making changes to the application
◦ Won’t get value

Insufficient support for the underlying
system
◦ Keeping network up to speed
◦ Having enough terminals
84

Much is the inverse of common pitfalls
◦ But not all


Anyone will have issues that leadership
need to deal with
Keep in mind that it will be worth it
◦ Have to pay attention to details to achieve value—
doesn’t simply come with successful
implementation

Is a much bigger change than anything you
have previously attempted on the IT front
85





Strong leadership and long term commitment
Creating a culture of innovation
Excellent project management
Attention to clinical processes
A focus on quality
86

Strong teams at point of care
◦ Someone on site for first 2 weeks of
implementation, 24/7
◦ Good helpdesk afterward
◦ Tracking of feedback
 Personal follow-up about changes

Made needed changes to fit clinical issues
◦ Some groups needed special attention
◦ Worst in most highly stressed groups
87




“Solving the technological issues gets you 25%
there. You need leadership to provide the
vision to take you the rest of the way.”
“Commitment of key leadership is as important
as the quality of the technology”
“If leadership isn’t clear in its conviction, clear
in its communication, and clear in its
steadfastness, then I think your chances of
success start to drop rapidly.”
“When there was an issue, the CIO and CMO sat
down and addressed it quickly.” – CFO at a
hospital with a very successful implementation.
88

“When there were bumps and bruises along
the way and some people questioned whether
they should be doing this they would get a
friendly call from (the CEO) [saying] ...this is
the direction we are going in and everyone is
going to march in this direction.”
89





“The interesting thing is, once we get
physicians beyond the initial perception that
CPOE is time-consuming, they cannot think of
going back without it”
“Anticipate the needs of the physicians… Have
IS people make rounds with the physicians”
“Find a successful site, and have them sell it [to
your medical staff]”
“We kept redoing the system for 5 years after it
came up”
“Our CEO said that this was going to be a
clinician-driven process from the beginning.”
90

“We would go out and hover in the areas, we
made rounds frequently. Often times we
would be there before the physician, just as
they were getting ready to call. We would say
‘Here we are’.”
91

“What screws you up is realities that need
differences. It’s a chaotic system. Differences
that you would not imagine would matter do,
when you analyze it to the level of unique
work flows.”
92



“You just can’t buy anything that works out of
the box from the vendors. Smaller hospitals
will not be able to afford to customize the
products to suit their needs.”
“If there is a realistic, non-vendor-based
assessment of the [CPOE] technology and
where it will be in 2-3 years, then I as a leader
could leverage my political capital with some
reassurance that there’s gonna be some flesh
on the bones.”
“It would be helpful if hospitals interested in
CPOE can share the contract or RFP, so that
nobody has to re-invent the wheel when they
deal with the vendors.”
93

“I am not sure any of us know exactly how we
have succeeded. A lot of it has been so
evolutionary and we have been at it year after
year.”
94

“Staff, money, money, staff. I don’t think that
it is physician acceptance which it used to be.
I think that something terrible has happened.
We have been successful… I think that
physicians can just take more of this than we
ever have the capability of delivering.”
95


Have to have successful implementation
But also need to decide on a core of
decision support
◦ Implies having organizational structure enabling
group to reach consensus

Will have to make many changes
◦
◦
◦
◦
Need architecture enabling agility
Sufficient resources to keep up
Rule is to have a long queue
Want to start low, go slow—but need to end up
with enough
96


Leadership is key in all change—IT is no
exception—sociotechnical harder than
technical
Many decisions when implementing decision
support
◦ Degree of benefit is directly proportional to how
much decision support is included
◦ Need to test post-implementation


Will need to make many decisions in future
about autonomy and quality, safety
Have to track post-implementation,
systematically fix problems, especially those
newly created
97
RTI International
Questions and Wrap Up
Speaker Contact
Information
Next Webinar

David Bates
dwbates@bics.bwh.harvard.edu

David Classen
david.classen@pascalmetrics.com
98

Reducing Misdiagnosis: How
Can We Improve Diagnostic
Safety in the Health IT Era?
May 8, 2015
1-2:30p EDT
Please visit:
www.healthitsafety.org or
contact healthitsafety@rti.org for
more information on the entire
webinar series