Connecting Drugs and Lab Results to Prevent Inpatient Medication Errors

1
Connecting Drugs and Lab Results to Prevent
Inpatient Medication Errors
Linking Pharmacy and Data for Better Care Part
3 Concepts and Prospects

• January 20, 2011 130pm 230pm CST
• Voice conferencing 513-241-1028
• Conference ID 40278
• Participant, Please mute your phone by pushing 6
• Gordon Schiff MD
• Associate DirectorCenter Patient Safety Research
  and Practice Brigham Womens Hospital
  -Boston
• Associate Professor of Medicine Harvard Medical
  School
• Clinical Director UIC TOP-MED CERTCo-investigator
  BWH CERT

This project was supported by grant number
U18HS016973 from the Agency for Healthcare
Research and Quality. The content is solely the
responsibility of the authors and does not
necessarily represent the official views of the
Agency for Healthcare Research and Quality.
2
Schiff Arch Intern Med 2003
3
Hug JGIM 2010
4
(No Transcript)
5
(No Transcript)
6
Feldstein Arch Int Med 2006
7
CDS Lessons, Caveats

• - Speed is everything CDS needs to be easy to
  use and not time consuming must actually save
  time
• - Anticipate needs and delivery in real time.
• - CDS needs to fit into the users workflow.
• - Offer alternatives rather than trying to stop
  an action.
• -Simple interventions work best (single screen of
  info).
• -The more data elements requested, the less
  likely the guideline will be implemented
• -Signal to noise is bedeviling, important,
  improvable.

Adapted from Bates, Ten Commandments for CDS.
JAMIA 2003
8
More Lab-Pharmacy Link Relevant CDS Lessons

• -Monitor impact of interventions and act on
  findings and lessons. Including frequencies,
  satisfaction, glitches Continuous improvement,
  incorporate users feedback
• -Think about who will be receiving messages, when
  
• -CDS often turned off, before, during after
  implementation Signal of paradigm failure
• -Its about leadership, policy, goals,
  communication change management, not technology
  per se Mandatory, supportive adoption (VA)

9
Leapfrog Tests of CPOE Systems Most Fail
Lab-Pharm Checks
Metzger Health Affairs 2010
10
Metzger Health Affairs 2010
11
ItAint Just Alerts
National Quality Forum (NQF), Driving QualityA
Health IT Assessment Framework for Measurement
A Consensus Report, 2010.
12
Categories of interactions w/ IT system Each can be performed by a human role or system itself Categories of interactions w/ IT system Each can be performed by a human role or system itself
1. Access The act of retrieving data or a computer file.
2. Acknowledge To officially recognize, admit or accept receipt of an object or information.
3. Alert. To make someone aware of a possible danger or difficulty
4. Calculate To compute mathematically
5. Create To produce something as in a printed report or electronic copy.
6. Discontinue To stop or end an activity that is planned or is happening regularly also to remove an element from existing patient information such as an allergy from an allergy list.
National Quality Forum (NQF), Driving QualityA
Health IT Assessment Framework for Measurement A
Consensus Report, 2010.
13
Categories of interactions w/ IT system Each can be performed by a human role or system itself Categories of interactions w/ IT system Each can be performed by a human role or system itself
7. Document To create a record of facts, events, symptoms or findings.
8. Implement To put into effect or action.
9. Notify. To inform or warn officially to make something known
10. Order. An instruction or request to bring, supply, perform or activate something
11. Perform To carry out an action or accomplish a task, especially one requiring care or skill.
12. Receive To receive or take something provided.
13. Recommend To suggest something as worthy of being accepted use or done.
National Quality Forum (NQF), Driving QualityA
Health IT Assessment Framework for Measurement A
Consensus Report, 2010.
14
Piecing Together the Global Picture
15
Critical LabF/up
POC Testing
Med Legal Liability
Teamwork Roles Handoffs Efficiency
CDS Rules Alerts
System vs. MD vs. Patient Responsibilities
Anticoag Monitoring
Test Timing
Test Appropriateness
Diagnosis Error
SPC Rx Dosing
Tight Control
Data Mining
Costs Tests Drugs
Drug Marketing Safety
PharmacoGenomics
16
Linking Pharmacy and Data for Better Care
Part 3 Lab?Med Linkage Decision
SupportInteractive Discussion on Use of Project
Tools
William Galanter MD/PhD Medical Director,
Clinical Information SystemsChair Pharmacy
Therapeutics Committee Co-investigator UIC
TOP-MED CERTDepartment of Medicine/Department of
Pharmacy Practice University of Illinois at
Chicago (UIC)
This project was supported by grant number
U18HS016973 from the Agency for Healthcare
Research and Quality. The content is solely the
responsibility of the authors and does not
necessarily represent the official views of the
Agency for Healthcare Research and Quality.
17
Outline
-Background of Lab?Med Linkage decision support
-The development of high priority Lab?Med
pairs -Results of the first phase of
pairs -Issues Cost/Benefit Regulators/Leap
Frog/Meaningful use Relationship with CPOE
Timing of report Dissemination
Customization Bureaucracy Lack of lab
testing -How can we help each other? -Where
to go from here?
18
Synchronous Alerts (synchronous to CPOE)
Registration
EMR
Rules Engine
19
Alerts for Contraindication
Proportion of patients with renal
dysfunction receiving Metformin when order
started by clinician 4-months pre-alert vs..
4-months post-alert
Pre-Alert (63)
100
80
60
40
20
0
Galanter et al. J Am Med Inform Assoc. 2005
May-Jun12(3)269-74
20
Synchronous Alerts
Patient Days on Elevated K vs. time
Synch Alerts
21
Asynchronous (to CPOE) Alerts
EMR
Rules Engine
22
Asynchronous Alerts
Compliance with alert recommendations Low Mg
when on Digoxin
EMR

Galanter et al. J Am Med Inform Assoc. 2004
Jul-Aug11(4)270-7.
23
Asynchronous Alerts
Exposure to Metformin after a new eGFR lt60
Patients receiving Metformin
Post Pre
Minutes after new lab
24
Asynchronous Alerts
Discontinuation rate of K Supplementation with
?K
100 50 0
hours
25
Asynchronous Alerts
Discontinuation rate of K Supplementation with
?K
100 50 0
26
The timing of Lab?Med CDS?
Synchronous -Requires CPOE -Compliance is a
problem -Immediate feedback Real-time
asynchronous -Requires a 24o communication
channel -Immediate feedback Non real-time
asynchronous reports -Does not require CPOE
(lt20 of US Hospitals in 2009) -Can use a
dedicated team for a short period of time, thus
adds reliability -delayed feedback
Health Affairs, 28, no. 2 (2009) 404-414 Clin
Pediatr (Phila). 200948389-396.
27
Non real-time asynchronous Alerts
28
Development of an Asynchronous Daily Lab?Med
Report
Collaborators University of Illinois at Chicago
Bruce L. Lambert, Ph.D., Rob Didomenico, PharmD,
Mike Koronkowski, PharmD, Shengsheng Yu, MS,
Fang-Ju Lin, BPharm, MS, Jessie Moja, MD
Brigham and Womens Hospital Gordon D. Schiff,
MD Stroger Hospital Shane Borkowsky, MD, Mary
Wisniewski RN, MSN University of
Washington Beth Devine PharmD, MBA, PhD, Tom
Payne, MD Cerner Corporation David McCallie MD,
Margaret Kolm MD
29
Results of the Delphi Exercise Top 24 pairs
Yu, S. Galanter WL, Didomenico, RJ, Borkowsky S,
Schiff G, Lambert B. Consensus list of priority
drug-lab linkages for an inpatient asynchronous
alert program Results of a Delphi survey. Am J
Health-Syst Pharm. 2011. Mar 168. In press.
30
First Phase of Lab?Med pairs implemented at UIMCC
MEDICATIONS LAB RANK
ACE ?K 2
ARB ?K 2
potassium ?K 3
K Sparing Diuretic ?K 18
Warfarin Pregnancy 8
ergotamines Pregnancy 15
clozapine WBC or ANC 14
31
The Report
32
Performance of Alerts First 100 days of First
Phase
Type "Lab" Med False () Action No Action
Pregnancy Pregnancy Documentation Warfarin 100 0 0
Pregnancy Testing BHCG Warfarin 100 0 0
           
K Serum K Supplement 54 28 18
Serum K ACE or ARB 51 24 24
Serum K K Sparing Diuretic 88 5 7
    TOTAL K 67 17 15
           
WBC/ANC WBC ANC Clozapine 0 100 0
           
TOTAL 69 17 14
33
Example
Pt on ACE Aware of lab, forgot to act on med
Report
Asked to stop ACE
34
Example
Not aware of lab, cant act on med
35
Example
Not aware of prior result, not aware of
result,cant act on med
K
Creatinine
36
Example
No good indication, poor f/u labs, aware of
result,acted on one med, forgot the other
Stop Aldactone
Stop ACE
Admitted, placed on ACE/Aldactone for HTN?
37
Issues
Cost/Benefit -How to calculate ROI? -Problem
of small numbers -Very few abnormal links
turn into clinically relevant events -Severe
ADEs are rare -How many FTE to implement?
38
Issues
Regulators and measures of Quality CMS Quality
Indicators? Leap Frog -Only interested in
synchronous alerts -No credit for asynchronous
CDS work Meaningful use for eligible
hospitals and CAHs at 495.6(g)(10)(ii) to
Implement one clinical decision support
rule. In the proposed rule, we said that
clinical decision support at the point of care
is a critical aspect of improving quality,
safety, and efficiency. -Does
Asynchronous CDS count? Is it a rule?
39
Issues
Relationship with CPOE Can supplement CPOE with
Synch CDS Can supplement CPOE without Synch
CDS Can add safety without CPOE -A quick
win Can work in a hybrid paper/CPOE
system Is independent of CPOE -Only need lab
and pharmacy IS systems and a server to link
the two.
40
Issues
Timing of report Frequency -Running it very
frequently will decrease the potential lag from
problem to resolution, but manpower may become
unmanageable, assurance of resolution may be
lost -Running it daily would allow strong
assurance of resolution, but long lag
times. Time of day -running immediately after
labs are reported will increase
yield -Giving clinicians some time to act will
decrease annoyance and work in managing the
report.
41
Issues
Dissemination Customization -Technical
-Cultural -Clinical Bureaucracy

42
Issues
Customization -Technical -EMR/
(Lab/Pharm systems) vendor's -database/data
model -Variable names/tests
(POCT/Normal's/panels) -Types of tests
Critical care panel, ABG, VBG -Formulary
issues -Drug databases (multum, FDB, etc...)
-Cultural -Practice Variation
-Control and ownership -Clinical
-Patients/Dzs/specialties -Lab
differences -Changes in sensitivity/specifici
ty based on clinical differences
43
Issues
Dissemination Bureaucracy -Committees
-Vetting (To vet was originally a
horse-racing term, referring to the requirement
that a horse be checked for health and
soundness by a veterinarian before being allowed
to race. Thus, it has taken the general
meaning "to check".) -Med/legal
-Institutional priorities -TJC -Meaning
ful Use -Patient Safety Goals
44
Issues
Variation or Lack of lab testing -Variations
in lab testing will change the sensitivity,
annoyance, workload -If there is a problem
and no one is measuring it, the report will not
work -In the long run ancillary order decision
support can help with this
45
Issues
Where to go from here? -TOPMED
Cert -Validation of what we have
implemented -More pairs to be added -Try to
measure value -How can we help any interested
sites? -Potential dissemination for other
venues NH, Ambulatory
46
Issues
How can we help each other? -Suggestions for
us? -What help can we give you?
47
Some Topics for Discussion
Issues Cost/Benefit Regulators/Leap
Frog/Meaningful use Relationship with CPOE
Timing of report Dissemination
Customization Bureaucracy
Variation/Lack of lab testing How can we
help each other? Where to go from here?
48
Resources
Clinical logic for the first set of alerts can be
found at http//www.uic.edu/com/dom/gim/TOPMED/L
ogic-Phase1.pdf A list of all the pairs can be
found at http//www.uic.edu/com/dom/gim/TOPMED/P
airs.pdf An excellent source of references and
an explanation of the Delphi process can be found
in the reference below when available Yu et al.
Consensus list of priority drug-lab linkages for
an inpatient asynchronous alert program Results
of a Delphi survey. AJHP. 2010 In press.