Q J Med 2006; 99:397406 Advance Access publication 8 May 2006

doi:10.1093/qjmed/hcl050

Prospective drug utilization evaluation of three

broad-spectrum antimicrobials: cefepime,
piperacillin-tazobactam and meropenem
D. RAVEH1, E. MUALLEM-ZILCHA1,3, A. GREENBERG2, Y. WIENER-WELL1,
Y. SCHLESINGER1 and A.M. YINNON1,4
From the 1Infectious Disease Unit and 2Hospital Pharmacy, Shaare Zedek Medical Center, Jerusalem,
affiliated with the Faculty of Health Sciences, Ben Gurion University of the Negev, Beer Shevah, and
3
School of Pharmacy, 4Hebrew-University-Hadassah Medical School, Jerusalem, Israel

Received 1 November 2005 and in revised form 21 March 2006

Downloaded from http://qjmed.oxfordjournals.org/ by guest on January 12, 2012

Summary
Background: Cefepime, piperacillin-tazobactam
and meropenem are among the broadest-spectrum
and most expensive antimicrobials.
Aim: To evaluate guidelines for appropriate use of
these drugs.
Methods: We developed guidelines for use of these
antibiotics, and conducted a two-phase drug
utilization evaluation. We included all patients
who received one of the study drugs during two
3-month periods, with an educational intervention
in the intervening period. Appropriateness was
determined for initiation of treatment, and for
adaptation or continuation of established treatment.
Results: Overall, 205 patients received 271 courses
with one of these antibiotics, for a total of 709
defined daily doses (DDD) of cefepime, 543 of
piperacillin-tazobactam, and 680 of meropenem
(8.3, 6.3 and 7.9 DDD/1000 admission days,
respectively). Of these 271 courses, 234 were
appropriate (86%). Treatment was continued for
55 days in 60%, of which 88% were appropriate
(NS). Of the 271 courses, 210 (77%) were empirical
(83% appropriate), while 61 (23%) were based on a
relevant culture result (97% appropriate) (p < 0.001).
Appropriateness differed significantly between
departments (p < 0.001), and between the two
phases (p < 0.001). The major difference between
the two surveys was a decrease in meropenem
usage (p < 0.05).
Discussion: The vast majority of courses with
cefepime, piperacillin-tazobactam and meropenem
are empirically selected and continued, underlying
the importance of an optimal initial choice.
Antibiotic guidelines, in conjunction with formal
infectious disease consultation, can contribute to
more appropriate use of these drugs.

Introduction
The development of bacterial resistance to anti-
biotics has become a major problem throughout the
world.14 Resistant organisms may emerge as a
result of many factors, including widespread usage,
while their spread is mainly caused by factors
in the health care setting, including the health
care providers behaviour. The broadest-spectrum
antibiotics, such as fourth-generation cephalosporins,
piperacillin-tazobactam and carbapenems, play an
important role in the empiric therapy of serious
nosocomial infections. These antimicrobials are also
among the most expensive.5 Concern about escalat-
ing rates of multi-drug-resistant organisms and
spiralling expenditure on broad-spectrum anti-
microbials has induced most hospitals to implement
a range of measures.616 These include supervision
of their use by infectious disease consultants and/or
clinical pharmacists,6,7 provision of continuing

Address correspondence to Professor A.M. Yinnon, Infectious Disease Unit, Shaare Zedek Medical Center,
PO Box 3235, Jerusalem 91031, Israel. email: yinnon@szmc.org.il
! The Author 2006. Published by Oxford University Press on behalf of the Association of Physicians.
All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org
398 D. Raveh et al.
education regarding appropriate antimicrobial
drug use,10 and implementation of automatic stop
orders.14,15,17 However, there is evidence that, in
order to be effective, a multidisciplinary approach is
warranted, with application of a range of measures,
some of which should be individualized according
to the hospitals circumstances and means.
Another method increasingly used in this era
of cost constraints and quality assurance is drug
utilization evaluation (DUE).18 This tool was
adapted by pharmacists to assess appropriateness
of usage of various medications.8,9 The purpose of a
DUE is generally to detect possible problems with,
and improve, drug use. DUEs have traditionally
focused on drugs with frequent side-effects, high
price tags or complicated dosing regimens. Very few
DUEs have addressed broadest-spectrum antibiotics,
and none has included all three last-line agents.
We developed guidelines for the use of cefepime,
piperacillin-tazobactam and meropenem, and
conducted a two-phase DUE using these guidelines,
with an educational intervention in between, to
improve use of these expensive broad-spectrum
agents.
Methods
This study was done in a 550-bed university-
affiliated general hospital, Jerusalems second
largest. The hospital includes all major departments
and services, including three medical and two
geriatric wards, haematology and oncology;
paediatrics; two surgical departments, of which
one specializes in vascular surgery; gynaecology
and obstetrics, heart and chest surgery, urology,
orthopaedics, plastic surgery, ophthalmology,
otorhinolaryngology; and several intensive care
units (medical, surgical, paediatric and neonatol-
ogy). Transplantations are not performed. Many
patients are admitted through the emergency
department, where annually about 75 000 patients
are seen. The number of admissions increased
from 18 783 in 1990 to 44 458 in 2004, an increase
of 137%; number of admission days increased
correspondingly from 111 949 to 186 213 (66%).
These changes reflect the population growth in
the Jerusalem area, the increase in the hospitals
services as well as the near-universal shortening of
hospitalizations.
A drug utilization evaluation (DUE) program was
carried out over two 3-month periods: January
March and AugustOctober 2001. It included all
16 departments in which the study drugs (cefepime,
piperacillin-tazobactam and meropenem) were in
use, and all patients who were treated with at least
one of these three antibiotics. Patients receiving a
study drug were identified by daily review of the
patients drug records in each of the participating
departments. Patients were followed from initiation
until discontinuation of treatment.
Drug treatment was divided into two periods: the
initial period (45 days, i.e. the time interval during
which relevant culture results might become avail-
able and hence influence subsequent change or
continuation of drug therapy); and the often more
protracted period of definite treatment, influenced
by culture results or empirically continued. The
appropriateness of drug treatment was assessed for
the initial period in both phases of the study, and for
the definite period in phase 1 only.
Appropriateness was determined using a pre-
defined, two-page guideline (Appendix A), prepared
as part of the study. It was based upon the following
Downloaded from http://qjmed.oxfordjournals.org/ by guest on January 12, 2012
underlying principles: (i) accordance with the
hospitals protocols, summarized in the guideline;
(ii) treatment targeted according to susceptibility
data of an organism from a relevant clinical speci-
men; and (iii) drug therapy as recommended by
an infectious disease consultation. All files were
reviewed between two investigators, one of whom
was an infectious disease consultant: post factum
agreement with prescribed treatment was consid-
ered to indicate appropriate treatment, even if this
deviated from the guideline. However, this over-
rule was applied only in a few cases of uncertainty,
e.g. when relevant data were not available or had
not been recorded, as these sessions were not held
in real time, and with the patients charts at hand
only. The hospitals protocols for antibiotic usage for
common clinical situations consists of a 100-page
booklet, which has been distributed to all physi-
cians19 since the early 1990s, and is periodically
updated. These protocols, in turn, were based on the
literature,2024 local susceptibility patterns,2527 and
available antimicrobial agents.
Antibiotic use was expressed according to the
internationally accepted defined daily dose (DDD)
system. According to this method, the usual daily
dose of a drug, prescribed to an adult patient
without renal or liver impairment, is 1 DDD.
Adoption and implementation of this method
allows hospitals to compare their overall as well as
particular drug use with that reported and published
by other medical centres.28,29
Demographic and clinical data were retrieved
from the relevant patients files. Data gathered
included: demographic information; blood urea
nitrogen and creatinine levels; indication for initia-
tion of treatment; therapy with other broad spectrum
antibiotics during the present hospitalization; source
of infection (nosocomial or community-acquired);
 Use of broad-spectrum antimicrobials
involvement of an infectious disease consultant
(prescription of the three study drugs requires prior
authorization by an infectious disease consultant);
results from the microbiology laboratory regarding a
pertinent clinical specimen, including bacterial
identification and susceptibilities. Results of the
first survey were analysed, and subsequently an
educational campaign was mounted based on the
findings of that survey. The campaign consisted of a
written distribution to all the medical centres
physicians of the results of the initial survey,
accompanied by the practice guideline. In addition,
oral presentations were made during departmental
staff meetings, stressing the studys major findings
and discussing the hospitals guidelines for use of
the study antibiotics. Several months later, a second
survey, identical to the first, was conducted. During
the course of data collection, the investigators
refrained from influencing clinical decisions by
clinicians. No attempt was made to verify the
infectious diseases diagnosis for which the broad-
spectrum antimicrobial had been selected.
Infections were classified as community- or
hospital-acquired according to standard definitions:
basically, if an infection was acquired 472 h after
admission, it was considered nosocomial.
Collected data were entered into the Epi info
6.04d computer program (CDC), which was also
used for data analysis. Proportions were compared
using the 2 or Fishers exact test, where appro-
priate. Continuous variables were compared by the
Students t-test. All p values were two-tailed, and
p < 0.05 was considered statistically significant.
Table 1 Pharmaceutical data of drugs used in both phases of the survey
Variable Antimicrobial agent
Cefepime
Total coursesa 88
Phase 1 (n 143) 32 (22%)
Phase 2 (n 128) 56 (44%)
Antimicrobial duration (days)b
Phase 1 6.6
3.6
Phase 2 8.9
6.2
Daily dose (g
SD) 2.0
0.5
Duration of treatment (days
SD) 7.7
4.9
Total dose/course (g) 15.4
Total use in study periods
DDD (defined daily doses) 709
DDD/1000 admission days 8.3
DDD/1000 patients 38.8
a
The number of cefepime courses increased significantly (p < 0.05), while the number of meropenem courses decreased
significantly (p < 0.05) from phase 1 to phase 2. bTotal duration of antimicrobial treatment, in days (mean
SD).
399
Results
During the first survey, conducted in all departments
where cefepime, piperacillin-tazobactam and
meropenem were prescribed, 102 patients received
143 courses with at least one of these three
antibiotics. During the second survey, 103 patients
received 128 courses with at least one of the these
antibiotics. Table 1 shows drug and pharmaceutical
data of the study drugs during the two phases of the
survey, including number of courses given with
each antibiotic, duration of treatment and dosing.
Notably, meropenem use decreased from 44% of all
broad-spectrum therapy in the first period to 22% in
the second (p < 0.05). An opposite trend was seen
in cefepime use, which increased from 22% of all
broad-spectrum treatment in the first period to 44%
in the second (p < 0.05). Analysis of individual
Downloaded from http://qjmed.oxfordjournals.org/ by guest on January 12, 2012
antibiotic use for separate indications revealed that
for nosocomial pneumonia in phase 1 of the study,
44% of broad-spectrum courses were with mero-
penem; this percentage decreased to 18% in
phase 2 (p < 0.05). Not surprisingly, the opposite
happened with cefepime: in phase 1, 17% of
courses for nosocomial pneumonia were with
cefepime, increasing to 36% in phase 2 (p < 0.05).
There was a similar, although statistically insignifi-
cant, shift for the treatment of sepsis.
Table 2 shows data regarding appropriateness of
antibiotic use in both study phases. Overall, initia-
tion of treatment was justified in 234/271 (86%)
cases: 122/143 (85%) instances in phase 1 and
112/128 (87%) in phase 2 (NS). The rate of
Piperacillin-tazobactam Meropenem
92 91
48 (34%) 63 (44%)
44 (34%) 28 (22%)
6.8
3.7 9.6
6.6
7.5
4.2 10.8
10
11.6
2.8 1.5
0.7
7.1
3.9 10.2
8.3
82.4 15.3
543 680
6.3 7.9
29.7 37.3
400 D. Raveh et al.

Table 2 Overall appropriateness of treatment, with comparison of the two phases

Total Phase 1 (n 143) Phase 2 (n 128)

Antimicrobial
Cefepime 76/88 (86%) 29/32 (91%) 47/56 (84%)
Piperacillin-tazobactam 83/92 (90%) 43/48 (90%) 40/44 (91%)
Meropenem 75/91 (82%) 50/63 (79%) 25/28 (89%)
Indication for treatment
Sepsis (nosocomial) 86/100 (86%) 54/61 (89%) 32/39 (82%)
Pneumonia (nosocomial) 84/96 (87%) 48/57 (84%) 36/39 (92%)
Other 68/75 (91%) 20/25 (80%) 48/50 (96%)
Initiationa
Overall 234/271 (86%) 122/143 (85%) 112/128 (87%)
Empirical 175/210 (83%) 90/110 (82%) 85/100 (85%)
Based on a relevant culture result 59/61 (97%) 32/33 (97%) 27/28 (96%)
Initiation by departmentb
Intensive care 53/56 (95%) 25/26 (96%) 28/30 (93%)
Medical departments 109/118 (92%) 56/62 (90%) 53/56 (95%)

Downloaded from http://qjmed.oxfordjournals.org/ by guest on January 12, 2012

Heart surgery 12/26 (46%) 9/15 (60%) 3/11 (27%)
Other 60/71 (84%) 32/40 (80%) 28/31 (90%)

Data are numbers of appropriate treatments/total treatments (percentage). In phase 1, 44% of broad-spectrum courses for
nosocomial pneumonia were with meropenem; this percentage decreased to 18% in phase 2 (p < 0.05). The opposite
happened with cefepime: in phase 1, 17% of courses for nosocomial pneumonia were with cefepime, which increased to
36% in phase 2 (p < 0.05). A similar, although less pronounced shift, occurred for the treatment of sepsis (NS). In phase 2, the
appropriateness level was significantly lower for patients with sepsis than for other indications (p < 0.05). aEmpirical
treatment (175/210, 83%) was significantly less appropriate than treatment based on relevant culture results (59/61, 97%)
(p < 0.001). bThe difference in appropriateness of treatment differed significantly between the departments (p < 0.001),
both overall and in each of the two phases (p < 0.001).

appropriateness of treatment for sepsis in phase 2
was significantly lower than for other indications
(p < 0.05). In addition, the overall rate of appropri-
ateness of empirical treatment (175/210, 83%) was
significantly lower than that for treatment based on
relevant culture results (59/61, 97%) (p < 0.001).
Finally, the rate of appropriateness of treatment
differed significantly between the departments
(p < 0.001), and in each of the two phases
(p < 0.001).
Figure 1 shows the characteristics of drug therapy
in the two study phases. In phase 1, 77% of all
newly instituted treatment was empirical, compared
to 78% in the second phase (NS). Accordingly, only
23% and 22%, respectively, of treatment was
initiated based on the results of a relevant culture.
The appropriateness rate of the latter, i.e. treatment
started according to a relevant positive culture, was
97% in phase 1, and 96% in phase 2 (NS). On the
other hand, the rates of appropriateness of empiri-
cally initiated treatment were 82% and 85%,
respectively (NS). Not unexpectedly, the appropri-
ateness rate of empirically chosen treatment was
significantly lower than that of treatment selected
according to a positive relevant culture (p < 0.001).
Data collected from the first phase of the survey
showed that only 60% of initiated treatment was
continued beyond 5 days, of which 88% appeared
appropriate.
Sub-analyses revealed that of 90 appropriately
initiated, empirically-chosen antibiotic courses in
phase 1 of the study, 54 (60%) were continued:
7 (13%) according to culture results, of which
5 (71%) were deemed appropriate. The remaining
47 (87%) were continued empirically (of which
100% were deemed appropriate). On the other
hand, of 20 inappropriately initiated, empirically-
chosen antibiotic courses in phase 1 of the study,
12 (60%) were continued, all empirically: of these,
only one (8%) was deemed appropriate. Therefore,
the rate of appropriateness of definite therapy
was considerably higher for those courses whose
initiation was considered appropriate vs. those
whose initiation was considered inappropriate
(p < 0.001).
The patients records were also reviewed
for documentation of an infectious disease
consultation. The hospitals guidelines regarding
restricted antimicrobials, which include cefepime,
piperacillin-tazobactam and meropenem, require
an infectious disease consultation prior to initiation
of these drugs. Such a consultation may take place
 Use of broad-spectrum antimicrobials 401

Treatment Initiated
Phase 1: 143
Phase 2: 128

Positive relevant culture*

Empirical treatment*
Phase 1: 110 (77%) Phase 1: 33 (23%)
Phase 2: 100 (78%) Phase 2: 28 (22%)

Inappropriate* Appropriate** Appropriate ** Inappropriate *

Phase 1: 20 (18%) Phase 2: 90 (82%) Phase 1: 32 (97%) Phase 1: 1 (3%)

Phase 2: 15 (15%) Phase 2: 85 (85%) Phase 2: 27 (96%) Phase 2: 1 (4%)

Treatment continued
> 5 days

Downloaded from http://qjmed.oxfordjournals.org/ by guest on January 12, 2012

Phase 1: 86 (60%)
Phase 2***

Appropriate ** Inappropriate **

Phase 1: 76 (88%) Phase 1: 10 (12%)

Figure 1. Distribution of drug therapy between the two study phases. *The difference between the two phases was not
significant for all parameters. **The difference between empirical treatment and treatment based on culture results was
significant in both phases (p < 0.001). ***This aspect was not examined in phase 2.

at the bedside (preferable), and is therefore recorded
by the consultant in the patients record, or
alternatively, via telephone, in which case the
consulting physician is required to document the
results of the consultation in the patients record.
The rate of documentation was low: only 33% and
39% of antibiotic courses in phase 1 and phase 2,
respectively (NS). However, the rate of consultation
could conceivably have been much higher than
the rate of documentation. The education effort
in between the phases evidently did not affect
a substantial improvement.
Expenditure on cefepime, piperacillin-tazobactam
and meropenem constitutes
40% of the hospitals
outlay on antimicrobial agents. Of 709 defined daily
doses (DDDs) of cefepime, 99 (14%) were con-
sidered inappropriate; for piperacillin-tazobactam,
54/543 (10%) were deemed inappropriate; and for
meropenem, 122/680 (18%) were considered inap-
propriate. These amount to inappropriately spent
sums of 3498 on cefepime, 2832 on piperacillin-
tazobactam and 7049 on meropenem, totalling
13 379 over the 6 months of the study. Only part of
this sum could have been saved, as more appro-
priate therapy with alternative agents would have
been less costly but certainly not negligible.
However, as fewer courses of these broad-spectrum
agents are prescribed, selective pressure may be less
severe, possibly with slower emergence of drug-
resistant organisms.
Table 3 shows the use of the three study drugs, in
addition to intravenous ciprofloxacin and ceftazi-
dime, during the study periods, as well during
comparable periods in the year before and the year
following the present study. There has been a
decrease in use of ciprofloxacin over these years,
as a result of diminishing susceptibility of
Enterobacteriaceae and particularly Pseudomonas
to this agent.9,25,26 Ceftazidime has been removed
from the hospitals formulary during 2002, on
account of its potential for induction of extended
spectrum beta-lactamases in Enterobacteriaceae. As
a result, there has been a significant increase in
cefepime use. Nonetheless, there appears to have
been a modest reduction in overall use of broad-
spectrum agents, expressed in DDDs per 1000
hospitalization days and per 1000 admissions.
Table 4 shows the susceptibility rates of all strains
of Enterobacteriaceae and Pseudomonas isolated
during the study phases and comparable periods in
the years before and after the study. The number of
depicted isolates increased artificially between 2000
and 2002: in 2000 only blood culture results were
computerized, while during the subsequent
years results from other culture specimens were
increasingly computerized as well. There appears
402 D. Raveh et al.

Table 3 Annual use of broad-spectrum antimicrobials, in defined daily doses (DDD) per 1000 admissions and per 1000
admission days (20002004)

Year Months Ciprofloxacin IV Ceftazidime Cefepime Piperacillin-tazobactam Meropenem Total

DDD/1000 hospitalization days

2000 JanMar 20 16 9 8 8 61
AugOct 12 21 8 12 8 63
2001 JanMar 12 22 6 8 12 61
AugOct 11 16 9 6 6 49
2002 JanMar 12 17 9 7 9 54
AugOct 13 0 25 9 9 57
DDD/1000 admissions
2000 JanMar 40 32 18 15 16 121
AugOct 24 38 14 22 15 113
2001 JanMar 23 41 12 16 23 114
AugOct 19 28 16 11 10 85
2002 JanMar 23 30 17 13 16 99
AugOct 23 0 42 16 15 96

Downloaded from http://qjmed.oxfordjournals.org/ by guest on January 12, 2012

IV, intravenous.

Table 4 Antimicrobial susceptibility rates (%) of all strains of Enterobacteriaceae and Pseudomonas isolated during the
study phases, and comparable periods in the years before and after the study

Year Months (n) Ciprofloxacin IV Ceftazidime Cefepime Piperacillin-tazobactam Meropenem

Enterobacteriaceae
2000 JanMar (52) 60 74 74 74 100
AugOct (99) 67 84 89 86 100
2001 JanMar (312) 69 89 94 89 99
AugOct (750) 70 85 92 86 100
2002 JanMar (627) 73 80 85 87 100
AugOct (804) 73 81 82 89 100
Pseudomonas
2000 JanMar (7) 100 100 NA 100 83
AugOct (15) 93 93 100 100 78
2001 JanMar (45) 75 89 76 86 86
AugOct (89) 72 93 97 91 97
2000 JanMar (68) 76 93 93 96 100
AugOct (120) 76 93 100 90 100

IV, intravenous.

to be a slow decline in susceptibility rates of hospitals outlay on antimicrobial agents, although

Enterobacteriaceae to ceftazidime and cefepime,
and a similar decline in Pseudomonas susceptibility
to ciprofloxacin, although not to advanced genera-
tion cephalosporins.
Discussion
This study was conducted in order to evaluate and
improve the rate of appropriate use of cefepime,
piperacillin-tazobactam and meropenem, three of
the broadest-spectrum antibiotics. Together, these
agents consume a significant proportion of most
they constitute only a small percentage of drug use
in terms of defined daily doses. Our methods
consisted of the development and modification of
practice guidelines, based on relevant literature and
susceptibility data from organisms isolated from
various clinical settings to guide physicians regard-
ing optimal use of these antibiotics, and the conduct
of two 3-month drug utilization surveys, with an
educational effort in between.
The principal findings of our study were as
follows. First, the appropriateness rates for cefepime,
piperacillin-tazobactam and meropenem were
 Use of broad-spectrum antimicrobials
relatively high (86%). Treatment was continued for
55 days in 60% of cases, of which the majority was
deemed appropriate. Second, most antibiotic
courses were empirical and only a minority (22%)
were based on a relevant culture result; the rate of
appropriateness of empiric treatment was signifi-
cantly lower than that of treatment based on a
relevant culture results. Third, the rate of appropriate
treatment differed significantly among the depart-
ments and between the two phases. The major
difference between the two surveys was a decrease
in meropenem usage, while an opposite trend
occurred with cefepime use. Over the years
subsequent to the study, the use of meropenem
and piperacillin-tazobactam has decreased, while
that of cefepime has increased; part of this increase
can be explained by the concomitant decrease in
ciprofloxacin use as well as phasing out of
ceftazidime (Table 3). Finally, an infectious disease
consultation was recorded in about one third of the
cases only; this is remarkable, because use of these
drugs in our institution, like in most others, requires
prior approval by an infectious disease consultant.
Some of these prior authorizations may have not
been recorded in the patients files. However, it
appears that a sizeable percentage of use of these
agents bypasses the formal route. Only a system of
unit dosing (supplying every single dose of anti-
biotics directly from the pharmacy to the patients
floor) would solve this issue, although requiring
presence of pharmacists around the clock.
The problem of resistance to antibiotics is world-
wide, but should be addressed locally in every
medical centre.617 Resistance is foremost a function
of the extent of use: for most antimicrobials, the
more they are prescribed, the higher the resistance
rates.30 Generally, some 1540% of antimicrobial
drug use is inappropriate.8,9,31 Although the key to
appropriate use consists of education, most medical
centres have instituted certain administrative mea-
sures with the dual purpose of cost control and
reducing the rate of emergence of resistant organ-
isms.10,17 These measures range from the simple (the
requirement to obtain infectious disease approval
prior to initiation or continuation of certain drugs) to
the sophisticated (interactive computer programs for
ordering these drugs, offering advice at various
stages of the process).15,32 Appropriate selection of
antibiotic treatment requires a thorough knowledge
of various issues, including the most likely patho-
gens causing the patients infection (taking into
account individual host factors), local susceptible
patterns of these pathogens (which change over
time), pharmacokinetic and pharmacodynamic
properties of the relevant antimicrobials, possible
drug interactions, hypersensitivity and adverse
403
effects. Since this study, our hospital has introduced
such a computer system, but it is abundantly clear
that the input of infectious disease physicians, a
clinical pharmacist and/or clinical microbiologist
remains essential.
The majority (77%) of our broadest-spectrum
antibiotic treatment was initiated empirically. As
expected, the rate of appropriateness was signifi-
cantly lower for empirically selected treatment than
for that tailored according to relevant microbiology
results. Importantly, the rate of continuation of drug
treatment beyond 5 days, during which time
laboratory results may be expected that often lead
to adaptation of treatment, was similar for appro-
priately and inappropriately initiated courses (60%).
Therefore, the rate of appropriateness of definite
therapy (after 5 days) was considerably higher for
those courses whose initiation was considered
Downloaded from http://qjmed.oxfordjournals.org/ by guest on January 12, 2012
appropriate than for those whose initiation was
considered inappropriate. Our data, although lim-
ited in scope, support the crucial role of the
infectious disease consultant33,34 when antibiotic
therapy is selected or adapted empirically, which is
evidently the situation in the vast majority of the
cases. If the selected spectrum of treatment is too
narrow, complications may ensure, or worse. If
treatment is too broad-spectrum, improving patients
usually remain on the chosen regimen; if they do not
respond, this may well lead to change to even
broader spectrum agents. Both situations, with
heavy use of broad-spectrum agents, will lead to
emergence of multi-drug-resistant organisms, as well
as significant expenses.
Our study has several limitations. The first
concerns the guidelines for appropriate use of the
broad-spectrum antimicrobials. Other teams of
infectious disease physicians, pharmacists and/or
clinical microbiologists would quite likely define
appropriate use of these agents in similar, but not
necessarily identical ways. Nonetheless, we believe
that the basic approach to developing guidelines for
appropriate use of these broadest-spectrum agents
will be comparable with that elsewhere. Second,
appropriateness was evaluated as adherence with
these guidelines, rather than as an objective fact.
This, however, is the case with most drug utilization
evaluations. Third, in our drug utilization evalua-
tion, we compared pre-intervention with post-
intervention results, rather than with those of a
simultaneous control group. Consequently, uneval-
uated confounding factors may have influenced the
study results. However, long-term follow-up of these
agents has shown an overall stability or even modest
decrease in their use. Indeed, since publication of
another study on antimicrobial usage in our
hospital,29 we have recorded a modest but persistent
404 D. Raveh et al.
decrease in defined daily doses, both per 1000
admissions and per 1000 hospitalization days,
as well as a modest decrease in expenditure on
antimicrobials. This is particularly remarkable since
the patient population in most hospitals and in ours
as well, continues to increase in age and disease
complexity.35 These patients often suffer from
infections acquired in nursing homes or the hospital
itself, secondary to multi-drug-resistant organisms,
mandating use of broad-spectrum antimicrobials.
We believe that an actively involved infectious
disease team, ongoing educative efforts and
up-to-date microbiology and pharmacy data-
bases are at least partially responsible for these
improvements.
In conclusion, we developed guidelines for use of
three of the broadest-spectrum antimicrobial agents
and conducted a drug utilization evaluation before
and after an educational intervention. We believe
that the presence of such guidelines, their wide-
spread publication, ongoing education of hospital
staff regarding their use, and continued need to
obtain approval prior to prescribing these agents, are
all necessary components in the effort to control
spiralling expenditure on such agents, and the
emergence of multi-drug-resistant organisms. We
believe that infectious disease consultation should
be mandatory prior to prescription of these broad-
spectrum agents, as well as 35 days subsequently,
when laboratory results and clinical changes could
influence a change in antibiotic management.
Acknowledgements
This study was conducted in part as thesis for a
Master of Pharmacy degree at the Hebrew
University-Hadassah School of Pharmacy,
Jerusalem, Israel.
References
1. Kollef M, Niederman M. Antimicrobial resistance in the ICU:
The time for action is now. Crit Care Med 2001;
29(4 suppl.):N63.
2. Jarvis W. Preventing the emergence of multidrug resistant
microorganisms through antimicrobial use controls: the
complexity of the problem. Infect Contr Hosp Epidemiol
1996; 17:4905.
3. Kunin CM. Resistance to antimicrobial drugsa worldwide
calamity. Ann Intern Med 1993; 118:55761.
4. Marcus EL, Clarfield AM, Moses AE. Ethical issues relating
to the use of antimicrobial therapy in older adults.
Clin Infect Dis 2001; 33:1697703.
5. Roark MK, Reed WE Jr. Econotherapeutics. Diagn Microbiol
Infect Dis 1995; 22:20917.
6. Schentag JJ, Ballow CH, Fritz AL, et al. Changes in
Antimicrobial Agent Usage Resulting from Interactions
Among Clinical Pharmacy, the Infectious Disease Division,
and the Microbiology Laboratory. Diagn Microbiol Infect Dis
1993; 16:25564.
7. Ibrahim KH, Gunderson B, Rotschafer JC. Intensive care unit
antimicrobial resistance and the role of the phramacist.
Crit Care Med 2001; 29(4 suppl.):N10813.
8. Yinnon AM, Skorohod Y, Schlesinger Y, Greenberg A.
Cefuroxime utilization evaluation: impact of physician
education on prescribing patterns. Isr Med Assoc J 2000;
2:18791.
9. Hammerman A, Greenberg A, Yinnon AM. Drug use
evaluation of ciprofloxacin: impact of educational
efforts on appropriateness of use. J Clin Pharm Ther 1997;
22:41520.
10. Kane RL, Garrard J. Changing physician prescribing prac-
tices: Regulation vs education. JAMA 1994; 271:3934.
11. Masterton RG. Antibiotic policies and the role of
Downloaded from http://qjmed.oxfordjournals.org/ by guest on January 12, 2012
strategic hospital leadership. J Hosp Infect 1999;
43(Suppl.):S2614.
12. Goldmann DA, Weinstein RA, Wenzel RP, et al. Strategies
to prevent and control the emergence and spread of anti-
microbial resistant microorganisms in hospitals. A challenge
to hospital leadership. JAMA 1996; 275:23440.
13. Marr JJ, Moffet HL, Kunin CM. Guidelines for improving the
use of antimicrobial agents in hospitals: A statement by the
Infectious Disease Society of America. J Infect Dis 1988;
157:86976.
14. Rifenburg RP, Paladino JA, Hanson SC, Tuttle JA, Schentag JJ.
Benchmark analysis of strategies hospital use to control
antimicrobial expenditures. Am J Health Sys Pharm 1996;
53:205462.
15. Pestotnik SL, Classen DC, Evans RS, Burke JP. Implementing
antibiotic practice guidelines through computer-assisted
decision support: clinical and financial outcomes.
Ann Intern Med 1996; 124:88490.
16. English S, Scott E. Empiric therapy for treatment of infection
and the influence of antibiotic guidelines on outcomes.
Int J Pharm Practice 1995; 3:2315.
17. Belognia EA, Schwartz B. Strategies for promoting judicious
use of antibiotics by doctors and patients. Br Med J 1998;
317:668671.
18. ASHP guidelines on the pharmacists role in drug-use
evaluation. Am J Hosp Pharm 1988; 45:3856.
19. Yinnon AM, Schlesinger Y, Jersassy Z, Raveh D, eds.
Recommendations for antimicrobial usage in common
clinical conditions. Jerusalem, Pocketbook (Hebrew).
Almog Ltd, 2004.
20. Moellering RC Jr., Eliopoulos GM. Principles of anti-infective
therapy. In: Mandell GL, Bennett JE, Dolin R, eds. Principles,
& Practice of Infectious Diseases. 6th edn. Philadelphia,
USA, Churchill Livingstone, 2005:24253.
21. Chambers HF, Hadley KW, Jawetz E. Chemotherapeutic
Drugs. In: Katzung BG, ed. Basic and Clinical Pharmacology.
7th edn. Connecticut, Appleton & Lange, 1998:72487.
22. Marra F, Reynolds R, Stiver G, et al. Tazobactam versus
imipenem: a double-blind, randomized formulary feasibility
study at a major teaching hospital. Diagn Microbiol Infect
Dis 1998; 31:35568.
 Use of broad-spectrum antimicrobials
23. Hurst M, Lamb HM. Meropenem: a review of its use in
patients in intensive care. Drugs 2000; 59:65380.
24. Beaucaire G. Clinical activity of cefepime in severe
infections. Clin Microbiol Infect 1999; 5(Suppl. 1):S614.
25. Raveh D, Rudensky B, Schlesinger Y, Benenson S,
Yinnon AM. Susceptibility trends in bacteraemias: analyses
of 7544 patient-unique bacteraemic episodes spanning
11 years (19902000). J Hosp Infect 2003; 55:196203.
26. Raveh D, Rudensky B, Huerta M, Aviv Y, Yinnon AM. Use of
time-trend analysis in the design of empirical antimicrobial
treatment of urinary tract infection. Eur J Clin Microbiol Infect
Dis 2003; 22:15864.
27. Tallis E, Rudensky B, Attias D, Raveh D, Schlesinger Y,
Yinon AM. In vitro activity of cefepime and other broad
spectrum antimicrobials against several groups of Gram-
negative bacilli and Staphylococcus aureus. Diagn Microbiol
Infect Dis 1999; 35:1216.
28. Cars O, Molstad S, Melander A. Variation in antibioitc use
in the European Union. Lancet 2001; 357:18513.
29. Raveh D, Levy Y, Schlesinger Y, Greenberg A, Rudensky B,
Yinnon AM. Longitudinal surveillance of antibiotic use in the
hospital. Q J Med 2001; 94:14152.
30. Gaynes R. The impact of antimicrobial use on the emergence
of antimicrobial resistant bacteria in hospitals. Infect Dis Clin
North Am 1997; 11:75765.
31. Drori-Zeides T, Raveh D, Schlesinger Y, Yinnon AM.
Practical guidelines for vancomycin usage, with prospective
drug utilization evaluation. Infect Control Hosp Epidemiol
2000; 21:457.
32. Bailey TC, Troy McMullin S. Using information systems
technology to improve antibiotic prescribing. Crit Care Med
2001; 29(4 Suppl.):N8791.
33. Yinnon AM. Whither infectious diseases consultations.
Analysis of 14,005 consultations from a 5-year period.
Clin Infect Dis 2001; 33:16617.
34. Thuong M, Shorthen F, Zazempa V, Girou E, Soussy CJ,
Brun-Buisson C. Appropriate use of restricted antimicrobial
agents in hospitals: the importance of empirical therapy
and assisted evaluation. J Antimicrob Chemother 2000;
46:5018.
35. Raveh D, Gratch L, Yinnon AM, Sonnenblick M.
Epidemiological and clinical characteristics of patients
admitted to medical and geriatric departments. A prospec-
tive, comparative and observational study. J Eval Clin Pract
2005; 11:3344.
Appendix A: Guidelines for
appropriateness of therapy with
cefepime, piperacillin-tazobactam
and meropenem
I. Initiation of treatment, based upon
culture result
Treatment was initiated based upon
susceptibility results of an organism isolated from
405
a clinically relevant specimen. If so, check the
following:
1. The organism was susceptible to cefepime,
piperacillin-tazobactam or meropenem, but not
to a narrower-spectrum antimicrobial.
2. The organism was not susceptible to an amino-
glycoside, or, if susceptible to an aminoglyco-
side, either the creatinine clearance (CrCl) was
430 ml/min, or the CrCl was430 ml/min but the
associated infection was purulent, e.g. pneumo-
nia or intra-abdominal.
3. The organism was isolated from a clinically
relevant specimen, including:
a. Blood or cerebrospinal fluid.
b. If isolated from synovial, pleural or perito-
neal fluid, or from a removed foreign body
(e.g. heart valve, joint, goretex graft),
Downloaded from http://qjmed.oxfordjournals.org/ by guest on January 12, 2012
there has to be clinical evidence of infection
that cannot be easily ascribed to another
source.
c. If isolated from sputum, a Gram stain has to
show a paucity of epithelial cells (<10 per
HPF), an abundance of leukocytes (420 per
HPF) and there has to be clinical evidence
of infection, e.g. pneumonia by chest
radiogram.
d. If isolated from the tip of a removed
intravascular device, rolled on an agar
plate, there have to be more than 15 CFUs
and there has to be clinical evidence of
infection that cannot be ascribed easily to
another source.
e. If isolated from a urine culture, the urinalysis
shows 52 leukocytes and/or 5nitrates
and/or there is clinical evidence of a urinary
tract infection that cannot be ascribed easily
o another source.
4. An infectious disease consultant recommended
the particular agent, even if the patients condi-
tion deviated from the abovementioned
definitions.
Treatment is considered justified if the following
criteria were met: 1 AND 2 AND at least one
parameter of 3(ae); OR 4.
II. Initiation of empirically selected
treatment
1. Infection was acquired in the hospital.
2. The CrCl was 430 ml/min OR the CrCl was
430 ml/min (in which case treatment should
include an aminoglycoside), but the associated
infection was purulent, e.g. pneumonia or intra-
abdominal.
406
3. Patient was previously treated with ciprofloxacin
or ceftazidime without clinical response, or
there was substantial reason to believe that the
causative organism would be resistant to the
latter agents.
4. In case of empirical meropenem treatment:
considered appropriate only in intensive care
units or equivalent patients in regular wards,
i.e., immuno-compromised patients, patients on
ventilators, etc.
5. An infectious disease consultant recommended
the particular agent, even if the patients condition
deviated from the abovementioned definitions.
Treatment is considered justified if the following
criteria were met: 1 AND 2 AND 3 AND 4 (in case
of meropenem treatment); OR 5.
III. Continuation of treatment 55 days,
based upon a culture result
As section I.
IV. Continuation of empirically selected
treatment 55 days
14. As section II.
5. There was significant clinical improvement
according to the patients physician, defined
D. Raveh et al.
as improvement of at least two of the following
parameters:
a. Successful reduction in dosage of vaso-
pressor drugs.
b. Improvement in arterial blood gases, or
successful partial-complete weaning from a
respirator.
c. A decrease of at least 1 C if there had been
fever, or an increase of at least 1 C if there
had been hypothermia.
d. A decrease of 415% if there had been
leukocytosis or increase of 15% if there had
been leukopenia.
e. Improvement of blood acid-base
abnormalities.
f. Improvement of renal abnormalities, such as
uraemia or decreased urine output.
g. Improvement of a lung infiltrate on the chest
Downloaded from http://qjmed.oxfordjournals.org/ by guest on January 12, 2012
radiogram.
6. An infectious disease consultant recommended
the particular agent, even if the patients
condition deviated from the above-mentioned
definitions.
Treatment is considered justified if the following
criteria were met: 1 AND 2 AND 3, AND 4 (in case
of meropenem treatment), AND at least two para-
meters from 5(ag); OR 6.