poster 736 use of best practice alerts in electronic
TRANSCRIPT
Use of Best Practice Alerts in Electronic Medical Records to Promote
Anti-infective De-escalation** B Fox*, K Osterby1 ,M Lavitschke2, L Schulz1
*University of Wisconsin-Madison Medical School, Section of Infectious Diseases; 1University of Wisconsin Hospital Drug
Policy Program and Center for Clinical Knowledge Management; 2 University of Wisconsin Hospital and Clinics IT
Methods
2
Abstract Results
Contact Info and References
Introduction
Results
• Effective use of antimicrobials requires broad spectrum initial empiric therapy to optimize
the probability of a successful patient outcome. However, after 48-72 hours, the concept of
anti-infective de-escalation is crucial to minimizing anti-infective side effects, and
minimizing the potential for antibiotic resistance.1
Poster 736
Background: Continued exposure to antibiotics may lead to adverse effects and
antibiotic resistance. UWHC utilizes Epic (Verona , WI) for an electronic medical
record (EMR) and Safety Surveillor (SS) (Premier, Inc. Charlotte, NC) for antimicrobial
stewardship surveillance
Methods: Twelve unique Best Practice Alerts (BPA) tools were constructed in Epic for
AMS. We examined the utility of one BPA designed to notify providers when antibiotic
de-escalation could be accomplished. Antibiotic utilization trends including number of
antibiotics and antibiotic breadth were assessed using EMR records. The results
were analyzed using ANOVA, Chi-square and t-test statistics comparing response
date -1 to response date +1.
Results: 244 alerts were constructed in 393 days. Of the 244 BPAs, 169 (69%)
were accepted (A), 30 (12%) were accepted with modifications (M), and 45 (18%)
were rejected(R). Differences between A and M were not statistically different.
Comparisons between A and R BPAs are presented in the above table. Baseline
utilization was not statistically significantly different in any group. All Day +1
comparisons were statistically different except for oral antibiotics. Utilization at Day
+1 of piperacillin/tazobactam, vancomycin and ciprofloxacin was significantly less in
the A group than the R group
Conclusions: BPAs targeting antibiotic de-escalation was highly effective in
reducing several utilization parameters, including total number of
antibiotics,spectrum of antibiotic especially the use of broad spectrum agents,
including piperacillin/tazobactam, and targeting vancomycin. Since Epic is a
common EMR in the US, this BPA tool, and possibly others, may be utilized to
increase the efficiency of the antimicrobial stewardship program.
Baseline Antibiotic Use Day +1 and Change from Baseline
Antibiotic Use
Antibiotic use parameters
(n=135) Accept Reject p-value Accept Reject p-value
# of antibiotics on Day -1 2.44 2.19 0.28 -1.12 -0.33 <0.005
# of antibiotics on Day +1 1.32 1.85 0.006
# of broad-spectrum
antibiotics 2.11 2.00 0.81 0.806 (-1.31) 1.59 (-0.41) <0.0001
# of anti-MRSA antibiotics 0.69 0.63 0.60 0.18 (-0.51) 0.44 (-0.19) 0.01
# of anti-pseudomonal
antibiotics 1.44 1.41 0.93 0.61 (-0.82) 1.11 (-0.30) 0.004
# of IV antibiotics 1.60 1.56 0.68 0.6 (-1.00) 1.22 (-0.33) <0.0001
# of PO antibiotics 0.84 0.63 0.32 0.72 (-0.12) 0.63 (-0.00) 0.19
Barry C Fox MD - University of Wisconsin-Madison, Division of Infectious Diseases,
600 Highland Ave .MS#5158 .Madison, WI 53792
Phone: 608-263-1545
Email: [email protected]
** Supported in part by an independent investigator award from Merck
1. Masterton,RG. Antibiotic Deescalation. Crit Care Clin 2011:27:149-162
Materials • Twelve unique Best Practice Alerts (BPA) tools were constructed in EPIC for antimicrobial
stewardship. We examined the utility of one BPA designed to notify providers when
antibiotic de-escalation could be accomplished. Antibiotic utilization trends including
number of antibiotics and antibiotic breadth were assessed using EMR records. The
results were analyzed using ANOVA, Chi-square and t-test statistics comparing response
• date -1 to response date +1.
• 244 alerts were constructed in 393 days. Of the 244 BPAs,169(69%) were
accepted (A), 30 (12%) were accepted with modifications (M), and 45 (18%) were
rejected (R). Differences between A and M were not statistically different.
Comparisons between A, M and R BPAs are presented in the tables. Baseline
utilization was not statistically significantly different in any group. All Day +1
comparisons were statistically different except for oral antibiotics . Utilization at Day
+1 of piperacillin/tazobactam, vancomycin and ciprofloxacin was significantly less in
the A group than the R group ( data not shown)
• Antibiotic de-escalation is a crucial principle of antimicrobial stewardship. We
created many prescriber notification alerts, BPA, to facilitate the use of the EMR as a
communication and learning tool. We chose to examine a single BPA that we felt was
a crucial component of stewardship.
• The alerts created by the stewardship team were well received by prescribers with an
81% acceptance rate.
• The BPA tool successfully reduced the number of antibiotics, narrowed the bacterial
scope, and decreased utilization of target antibiotics, including vancomycin,
piperacillin/tazobactam, and ciprofloxacin.
• We are encouraged by the results of this single intervention and expect similar
outcomes with the 11 other alerts currently in use. The other alerts include: duration of
therapy, diagnostic recommendations, dose optimization, duplicate therapy, drug/bug
mismatch, and route interchange.
Discussion
Conclusion • BPAs targeting antibiotic de-escalation was highly effective in reducing several
antibiotic utilization parameters, including total number of antibiotics. The use of broad
spectrum agents, including piperacillin/tazobactam and vancomycin saw significant
reductions. Ciprofloxacin use was also reduced. Since EPIC is a common EMR in the
US, this BPA tool, and possibly others, may be utilized to increase the efficiency of
AMS programs. UW is currently working with EPIC to implement this and all BPA tools
into future EPIC versions to assist with AMS.
Accept Accept w/ Mod Reject
Oneway
ANOVA
Means
Comparison w
Student's t
N Rows 169 30 45
# Drugs BPA Response Date -1 2.38 2.33 2.27 p = 0.7117
# Drugs BPA Response Date +1 1.26 1.40 2.09 p <0.0001* A<R; M<R; A=M
Difference (Date -1) - (Date +1) 1.12 0.93 0.18 p <0.0001* A>R; M>R; A=M
BPA Response
0.0
1.0
2.0
# Drugs BPA Response Date -1 # Drugs BPA Response Date +1 Difference (Date -1) - (Date +1)
# A
I D
rugs
Accept Accept w/ Mod Reject
Total number of antibiotics before and after de-escalation BPA by Response
0.0
0.2
0.4
0.6
0.8
# Anti-MRSA Anti-Infectives,BPA Response Date -1
# Anti-MRSA Anti-Infectives,BPA Response Date +1
Difference (Date -1) - (Date+1)
# A
nti-M
RS
A D
rugs
Accept Accept w/ Mod Reject
# Anti-MRSA drugs before and after de-escalation BPA, by response
Accept
Accept
w/ Mod Reject
Oneway
ANOVA
Means
Comparison w
Student's t
N Rows 169 30 45
# Anti-MRSA antibiotics Date -1 0.645 0.600 0.756 p = 0.3296
# Anti-MRSA antibiotics Date +1 0.172 0.300 0.556 p < 0.0001* A<R; M<R; A=M
Difference (Date -1) - (Date +1) 0.473 0.300 0.200 p = 0.0053* A>R; A=M; R=M
BPA Response
0.0
0.5
1.0
1.5
# Anti-Pseudomonal Anti-Infectives, BPA Response Date -1
# Anti-Pseudomonal Anti-Infectives, BPA Response Date +1
Difference (Date -1) - (Date +1)
# A
nti-P
seudom
onal D
rugs
Accept Accept w/ Mod Reject
# Anti-pseudomonal drugs before and after de-escalation BPA, by response
Accept
Accept
w/ Mod Reject
Oneway
ANOVA
Means
Comparison w
Student's t
N Rows 169 30 45
# Anti-Pseudomonal antibiotic Date -1 1.391 1.533 1.267 p = 0.1871
# Anti-Pseudomonal antibiotic Date +1 0.538 0.867 1.067 p < 0.0001* A<R; A<M; M=R
Difference (Date -1) - (Date +1) 0.852 0.667 0.200 p < 0.0001* A>R; A=M; M>R
BPA Response
Accept
Accept
w/ Mod Reject
Oneway
ANOVA
Means
Comparison w
Student's t
N Rows 169 30 45
# Narrow Spectrum antibiotics Date -1 0.343 0.200 0.289 p = 0.3497
# Narrow Spectrum antibiotics Date +1 0.527 0.200 0.444 p = 0.0264* A=R; A>M; R=M
Difference (Date -1) - (Date +1) -0.183 0.000 -0.156 p = 0.2269
BPA Response
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
0.5
# Narrow Spectrum Anti-Infectives,BPA Response Date -1
# Narrow Spectrum Anti-Infectives,BPA Response Date +1
Difference (Date -1) - (Date +1)
# N
arr
ow
Spectr
um
Dru
gs
Accept Accept w/ Mod Reject
# Narrow spectrum antibiotics before and after BPA, by response
Accept
Accept
w/ Mod Reject
Oneway
ANOVA
Means
Comparison w
Student's t
N Rows 169 30 45
# Broad Spectrum antibiotics Date -1 2.041 2.133 1.978 p = 0.7413
# Broad Spectrum antibioticss Date +1 0.734 1.200 1.644 p < 0.0001* A<R; A<M; M<R
Difference (Date -1) - (Date +1) 1.308 0.933 0.333 p < 0.0001* A>R; A>M; M>R
BPA Response
0.0
0.5
1.0
1.5
2.0
# Broad Spectrum Anti-Infectives,BPA Response Date -1
# Broad Spectrum Anti-Infectives,BPA Response Date +1
Difference (Date -1) - (Date +1)
# B
road S
pectr
um
Dru
gs
Accept Accept w/ Mod Reject
# Broad spectrum antibiotics before and after BPA, by response