Transcript masica_1b

Targeted Injury Detection System
for Adverse Drug Events: An AHRQFunded Patient Safety Initiative
TIDS-ADE
The Quality Colloquium
August 20, 2008
Andrew Masica, MD, MSCI
Baylor Health Care System-Dallas, TX
TIDS-ADE Background
• Trigger tool methodology
focused mechanism for risk reduction
event precipitates a response
Example: IHI
• Adverse drug events
common/costly
usually actionable
clinical/IT interface
Project Goals
• Develop a functional trigger tool for ADEs in
hospitalized patients that can be disseminated
broadly
• Detection at multiple time points related to event
occurrence (before, during, or after)
• Potential benefits in clinical care setting:
1. Prevention of ADEs
2. Mitigation of ongoing ADEs
3. Capture of “true” ADE rate
• Toolkit for real-world implementation
Definitions
• Trigger = alert: any event prompting further investigation
by clinician.
• ADE criteria = if event attributed to drug and:
reaches a level of harm that is durable or
requires a change in the treatment plan due to
unacceptable level of risk for harm or patient discomfort
• Example of “unacceptable risk” for patient harm:
-INR ≥ 6.0 and active warfarin order
-event prompted discontinuation of drug=ADE
• Broader concept of ADEs
Organizational Structure
Coordinating Center
RTI
AHRQ
Site leads
Conference Calls
In-person meetings
Site System Leadership
Local Test Site
• Patient Safety
• Health Care Improvement
• Project champion
-oversight
-data management
• Pharmacy champion
• Pharmacy IT
• Pharmacy Staff
Implementation: Site Environments
Task
Timing
• Site leader meetings
• 3-9 months prior to start
• Activation of IT/programming
resources
• 2-6 months prior to start
• Project introduction to site staff
• 1-4 months prior to start
• Validation of triggers
• 4-6 weeks prior to start
• Launch
• Begin pilot
Implementation: Triggers
• Choice/set-up of triggers:
higher yield alerts (Classen, Evans JAMA 1991)
core set of 1520 consensus, tiered TIDS alerts
tailoring to local site capabilities/priorities
• Trigger validation steps (3-phases):
programmer’s bank of “dummy data”
real-time pre-launch tests by site IT pharmacist
post-launch troubleshooting for obvious “misses”
• Uniform process for evaluating trigger utility
TIDS-ADE Workflow
Central Pharmacy
Floor
Virtual
Alert Work List
Alert Review
Trigger Evaluation
 Patient ID
 Chart
 Respond to ?’s
 Date/location
 Patient
 Data Warehouse
 Intervention
 Biweekly meetings
 Trigger details
 Triage
15 minutes
1-2 minutes (review)
1 minute (response)
Per alert
1-2 minutes
Results†
Alert
# Times Fired
% Patients (N=4171)*
K+<3.4 and K+ reducer
85
K+>5.5 and K+ raiser
60
INR >3 and active warfarin
51
PTT > 100 and heparin
28
Platelets < 100 and platelet reducer
10
Platelets <50 and platelet inhibitor
9
K+<3.2 and digoxin
6
Creatinine >133% of previous
5
Digoxin level >2.5
4
Mg2+ < 0.75 and digoxin
0
Total
258
*8 week study period. Data from Providence Health Care System.
2.0
1.4
1.2
0.7
0.2
0.2
0.1
0.1
0.1
0
6.2%
Test site average: 5-10 alerts per 100 patient days
†Preliminary
data from alpha-site testing
Trigger Evaluation
Was the alert useful?
3 Site Total
Alert useful
% Useful
Baylor GRV
# Alerts
# Alerts
% Useful
Bleed-1
123
14
7
29
Bleed-1b
53
17
53
17
C diff-1
49
8
49
8
Creatinine-1
6
0
0
.
Digoxin related-1
20
20
13
23
Hyperkalemia-1
15
0
15
0
Hyperkalemia-2
18
11
12
0
Hypoglycemia-1
2
0
0
.
Hypoglycemia-2
41
12
23
0
Hypokalemia-1
496
7
111
15
Hypokalemia-2
48
10
11
18
Hypotension-1
16
31
0
.
Hypotension-2
2
0
0
.
Platelets-1
1
0
0
.
Platelets-2
Rash-2
Renal clearance-1
13
30
76
23
0
22
6
30
75
17
0
23
Renal clearance-2
82
20
21
29
Trigger Evaluation
Did the alert detect an adverse event or trend?
3 Site Total
Baylor GRV
Adverse event
# Alerts
% AE
# Alerts
%AE
Bleed-1
123
15
7
29
Bleed-1b
53
17
53
17
C diff-1
49
8
49
8
Creatinine-1
6
0
0
.
Digoxin related-1
20
20
13
23
Hyperkalemia-1
15
0
15
0
Hyperkalemia-2
25
48
12
0
Hypoglycemia-1
Hypoglycemia-2
Hypokalemia-1
2
41
496
0
12
7
0
23
111
.
0
15
Hypokalemia-2
Hypotension-1
Hypotension-2
Platelets-1
Platelets-2
Rash-2
Renal clearance-1
Renal clearance-2
56
16
2
1
13
30
76
82
23
31
0
0
23
0
22
20
11
0
0
0
6
30
75
21
18
.
.
.
17
0
23
29
Trigger Evaluation
Did the alert change patient care?
3 Site Total
Prompt change
% Changed
Baylor GRV
# Alerts
# Alerts
% Changed
Bleed-1
123
12
7
29
Bleed-1b
53
13
53
13
C diff-1
49
0
49
0
Creatinine-1
6
0
0
.
Digoxin related-1
20
10
13
8
Hyperkalemia-1
15
0
15
0
Hyperkalemia-2
18
11
12
0
Hypoglycemia-1
2
0
0
.
Hypoglycemia-2
41
10
23
0
Hypokalemia-1
496
5
111
5
Hypokalemia-2
48
8
11
9
Hypotension-1
16
19
0
.
Hypotension-2
Platelets-1
Platelets-2
Rash-2
2
1
13
30
0
0
23
0
0
0
6
30
.
.
17
0
Renal clearance-1
76
18
75
19
Renal clearance-2
82
10
21
24
Trigger evaluation
Did a drug cause the adverse event or trend?
3 Site Total
Drug cause
Bleed-1
Bleed-1b
C diff-1
Creatinine-1
Digoxin related-1
Hyperkalemia-1
Baylor GRV
% drug cause
# Alerts
% drug cause
15
7
29
13
53
13
0
49
0
0
0
.
10
13
8
0
15
0
# Alerts
123
53
49
6
20
15
Hyperkalemia-2
25
8
12
0
Hypoglycemia-1
2
0
0
.
Hypoglycemia-2
41
10
23
0
Hypokalemia-1
Hypokalemia-2
Hypotension-1
Hypotension-2
Platelets-1
Platelets-2
Rash-2
Renal clearance-1
Renal clearance-2
496
56
16
2
1
13
30
76
82
5
5
31
0
0
23
0
18
16
111
11
0
0
0
6
30
75
21
5
9
.
.
.
17
0
19
24
TIDS-ADE: Trigger Summary
Results can guide refinement of alerts.
3 Site Totals
Alerts
% Useful
% Changed
% Adverse
% Drug
Bleed-1
123
14
12
15
15
Bleed-1b
53
17
13
17
13
C diff-1
49
8
0
8
0
Creatinine-1
6
0
0
0
0
Digoxin related-1
20
20
10
20
10
Hyperkalemia-1
15
0
0
0
0
Hyperkalemia-2
18
11
11
48
8
Hypoglycemia-1
2
0
0
0
0
Hypoglycemia-2
41
12
10
12
10
Hypokalemia-1
496
7
5
7
5
Hypokalemia-2
48
10
8
23
5
Hypotension-1
16
31
19
31
31
Hypotension-2
2
0
0
0
0
Platelets-1
1
0
0
0
0
Platelets-2
13
23
23
23
23
Rash-2
30
0
0
0
0
Renal clearance-1
76
22
18
22
18
Renal clearance-2
82
20
10
20
16
Impact on ADE Detection rates
• Expanded definition of ADEs for project:
patient harm or unacceptable risk for patient harm
• TIDS Alerts considered to have detected an ADE if:
alert detected an adverse event or trend
adverse event or trend was caused by a drug
Baylor Grapevine
• >40 cases meeting both conditions over 10 weeks
• Approximately 4-5 ADEs detected per week with TIDS
2.3 ADEs per 100 admissions
•
Voluntary reporting: <0.5 ADEs per 100 admissions
Lower ADE rate at Baylor?
• Sites in published literature:
3-6 ADEs per 100 admissions
academic centers/training programs
mature EHRs/CPOE
experience with trigger tool methodology
vs.
• Community setting
paper based with varying degrees of IT support
staffing limitations
acceptance of trigger approach to ADEs
verification process can be difficult
Additional Outcomes
• Qualitative Feedback
level of detail in alert felt to be beneficial
favorable view of alerts with trending
evaluation piece undermined perceived usefulness
sharp learning curve
fits well into existing practice patterns
• Quantitative
80 hours of programming time for study triggers
45 minutes pharmacist time daily
Lessons learned from TIDS-ADE
• High risk situations can be captured prospectively
with use of a trigger tool
• Need to resource multi-site collaboration
general framework for implementation
• Outcomes are influenced by site characteristics
performance of specific triggers
ADE detection rate
• Dynamic evaluation process for alerts is critical
optimizes performance of the triggering system
reduction in alert fatigue
TIDS-ADE: Future Directions
• Full analysis/toolkit development in progress
• Incorporation of broader ADE definition into
daily patient care
• Clarify endpoints for “successful” triggers
• Cross-cutting projectrealistic planning for
resource allocations
TIDS-ADE Leaders/Sites…thanks to all!
Michael Harrison, PhD
AHRQ
Jim Battles, PhD
AHRQ
Amy Helwig, MD, MS
AHRQ
Shula Bernard, RN, PhD*
Jonathan Nebeker, MD, MS†
RTI
University of Utah/VA
Scott Evans, MS, PhD
Intermountain
Brent James, MD, MS
Intermountain
Bruce Bayley, PhD
Providence Health Care
Steve Pickette, PharmD, BCPS
Providence Health Care
Howard Peckman, PharmD
UNC-Chapel Hill
Baylor Grapevine Pharmacy Staff
Baylor Health Care
Andrew Masica, MD, MSCI
Baylor Health Care
*Principal Investigator; †Senior Scientist