Content uploaded by Heather L Evans
Author content
All content in this area was uploaded by Heather L Evans on Sep 09, 2014
Content may be subject to copyright.
O
VER THE PAST SEVERAL DECADES
, we have
witnessed an explosion of resistant mi-
croorganisms within the inpatient population
of hospitals worldwide [1]. Methicillin-resis-
tant Staphylococcus aureus (MRSA) now is re-
sponsible for more than 50% of all nosocomial
staphylococcal infections, and 25% of entero-
coccal isolates from hospitalized patients are
SURGICAL INFECTIONS
Volume 7, Number 5, 2006
© Mary Ann Liebert, Inc.
Can We Define the Ideal Duration
of Antibiotic Therapy?*
TRACI L. HEDRICK, HEATHER L. EVANS, ROBERT L. SMITH,
SHANNON T. MCELEARNEY, ALISON S. SCHULMAN, TAE W. CHONG,
TIMOTHY L. PRUETT, and ROBERT G. SAWYER
ABSTRACT
Background: Because of the increasing development of antimicrobial resistance, there is a
greater responsibility within the medical community to limit the exposure of patients to an-
tibiotics. We tested the hypothesis that shorter courses of antibiotics are associated with sim-
ilar or better results than longer durations. We also sought to investigate the difference be-
tween a fixed duration of therapy and one based on physiologic measures such as fever and
leukocytosis.
Methods: All infectious episodes on the general surgery units of the University of Virginia
Health System from December 15, 1996, to July 31, 2003, were analyzed retrospectively for the
relation between the duration of antibiotic therapy and infectious complications (recurrent
infection with the same organism or at the same site). All infections associated with either
fever or leukocytosis were categorized into quartiles on the basis of the absolute length of
antibiotic administration or the duration of treatment following resolution of fever or leuko-
cytosis. Multivariate logistic regression models were developed to estimate the independent
risk of recurrence associated with a longer duration of antibiotic use.
Results: Of the 5,561 treated infections, 4,470 were associated with fever (temperature 38°C)
or leukocytosis (white blood cell count 11,000/mm
3
). For all infections, whether analyzed
by absolute duration or time from resolution of leukocytosis or fever, the first or second quar-
tiles (0–12 days, 0–9 days, 0–9 days, respectively) were associated with the lowest recurrence
rates (14–18%, 17–23%, 18–19%, respectively). Individual analysis of intra-abdominal infec-
tions and pneumonia yielded similar results. The fixed-duration groups received fewer days
of antibiotics on average, with outcomes similar to those in the physiologic parameters group.
Conclusions: Shorter courses of antibiotics were associated with similar or fewer compli-
cations than prolonged therapy. In general, adopting a strategy of a fixed duration of ther-
apy, rather than basing duration on resolution of fever or leukocytosis, appeared to yield sim-
ilar outcomes with less antibiotic use.
University of Virginia Surgical Infectious Disease Laboratory, Charlottesville, Virginia.
*Presented at the 24th Annual Meeting of the Surgical Infection Society, Indianapolis, IN, April 27–29, 2004.
419
420 HEDRICK ET AL.
resistant to vancomycin (VRE) [2]. Infections
caused by resistant bacteria have been associ-
ated with increased length of stay, longer du-
ration of mechanical ventilation, and higher
cost [3,4].
Primary prevention of the development of
resistant organisms has thus become an im-
portant topic for clinical researchers. Several
methods of prevention have been proposed, in-
cluding better infection control methods, for-
mulary restriction for antibiotic prescription,
mandatory infectious disease consultation
prior to use of certain antibiotics, rotational an-
tibiotic regimens, and setting limits on the du-
ration of antimicrobial therapy [5]. In the sur-
gical patient, inappropriate duration of
antibiotic administration continues to plague
current practice [6,7]. Although guidelines gen-
erally suggest that short-duration therapy re-
sults in outcomes similar to those of longer
courses of treatment [8–10], the surgical com-
munity has been slow to embrace this princi-
ple. We recently reviewed the duration of an-
tibiotic administration at our institution among
the last 1,343 surgical patients treated for intra-
abdominal infections [11]. Despite Surgical In-
fection Society (SIS) guidelines from 1992 and
2002 that recommended 5–7 days of treatment
for intra-abdominal infections [12], the mean
duration of treatment at our institution was
14.5 days.
Two methods have been proposed for deter-
mining the length of treatment. The first ap-
proach utilizes the resolution of physiologic
measures of disease (e.g., fever and leukocyto-
sis) to guide antibiotic duration on the basis of
data demonstrating a low rate of treatment fail-
ure in patients who are afebrile with a normal
white blood cell count at the time of antibiotic
cessation [13–15]. The other approach relies on
a fixed duration of therapy determined at the
initiation of treatment. Using a large, prospec-
tively collected dataset, we analyzed the rela-
tion between time to resolution of fever and
leukocytosis and failure of treatment of infec-
tion, and tested the hypothesis that shorter
courses of antibiotics are not associated with
higher rates of recurrence. We theorized that
shorter courses of antibiotics would be associ-
ated with results similar to or better than those
of longer-duration antibiotic therapy, and that
no difference would exist between fixed-dura-
tion and physiologically based treatment regi-
mens.
PATIENTS AND METHODS
Patient selection
This cohort study, using a large, prospec-
tively collected surgical infections database,
was designed to evaluate retrospectively the
effect of antimicrobial duration on recurrence
of infection. The study was conducted at the
University of Virginia Health System from
December 15, 1996, to July 31, 2003, and was
approved by the Human Investigation Com-
mittee with waiver of informed consent. All pa-
tients admitted to the adult general, transplant,
and trauma surgery units (ward or intensive
care unit) were evaluated for the presence of
infection. All infectious episodes were entered
in the database, and patients were followed
prospectively until discharge or death. Patients
with infection were identified by the investi-
gators by alternate-day chart review, physician
interview, and review of daily antibiotic usage
and laboratory and microbiologic data. Infec-
tious episodes occurring more than 72 hours
apart were considered separate episodes. Be-
cause we were interested in evaluating only the
duration of therapy in this study, patients who
died while on antibiotics and thus never com-
pleted their course of therapy were excluded;
all patients who died after cessation of therapy
were included. To ensure consistency, all data
were reviewed and entered in the computer-
ized format by a single investigator (RS).
Study design
During the study period, an antibiotic
scheme designed to maintain heterogeneity
was followed for the implementation of broad-
spectrum initial empiric therapy at the diag-
nosis of infection. A de-escalation method for
empiric antibiotic therapy was utilized. Broad-
spectrum antibiotics were initiated on the
suspicion of infection. After speciation of the
offending organism, therapy was tailored ap-
propriately. The dose and duration of therapy
were left to the discretion of the individual
CAN WE DEFINE THE IDEAL DURATION OF ANTIBIOTIC THERAPY? 421
treating physician, as no guidelines were in
place with regard to these treatment parame-
ters. Generally, antibiotics were stopped on the
basis of clinical indicators such as fever, leuko-
cytosis, and clinical improvement.
Study definitions
U.S. Centers for Disease Control and Pre-
vention (CDC) definitions of infection [16] were
utilized, with the exception of catheter-related
infections (see below). The criteria for the di-
agnosis of pneumonia included systemic evi-
dence of infection, purulent sputum produc-
tion, isolation of a predominant organism from
an appropriately obtained culture, and a new
or changing infiltrate or effusion on chest ra-
diograph. Diagnosis of a urinary tract infection
(UTI) required the isolation of greater than 10
5
organisms/mL of urine or greater than 10
4
or-
ganisms/mL with symptoms. Blood stream in-
fections were diagnosed by isolation of organ-
isms from any blood sample obtained using
sterile technique, with the exception of Staphy-
lococcus epidermidis or other coagulase-negative
staphylococci, which required isolation from
two separate sites. Catheter-related infections
were identified by isolation of 15 or more
colony-forming units (cfu) from the catheter tip
by the semi-quantitative roll plate technique in
the setting of clinical infection. We have ob-
served that blood stream infection in the setting
of catheter-related infection is not an inde-
pendent predictor of outcome by multivariate
analysis [17]. As a result, we do not require pos-
itive blood cultures to establish the diagnosis
of catheter-related infections in patients with
evidence of systemic infection; patients were
treated similarly regardless of the presence of
blood stream infection. Antibiotic-impregnated
catheters were not utilized at any time during
the study. Surgical site infections and intra-ab-
dominal infections were diagnosed on the ba-
sis of clinical findings.
Study parameters
Parameters recorded for each infectious
episode included multiple demographic vari-
ables and complete microbiological and phar-
macological data. Antibiotic days, including
all discharge medications, were calculated
prospectively. Severity of illness was assessed
using the Acute Physiology and Chronic
Health Evaluation score (APACHE II) [18] cal-
culated at the time of diagnosis of infection. The
FIG. 1. Sample set of 20 infectious episodes and method of defining quartiles.
422 HEDRICK ET AL.
primary outcome variable was in-hospital re-
currence of infection, defined as either recur-
rence of clinical infection at the original site or
recurrence of the original organism at an alter-
native site after treatment was completed. Mor-
tality was included as a secondary outcome,
recognizing the limitations associated with at-
tributing death to an infectious etiology in sur-
gical and trauma patients.
Statistical analysis
The general strategy was to compare quar-
tiles of antibiotic duration determined either by
the actual duration of therapy or by days
following resolution of leukocytosis (WBC
11,000/mm
3
) or fever (temperature [T] 38°C
during any calendar day). Because the groups
were defined retrospectively, they are not uni-
form in number. For example, using a sample
dataset including 20 episodes (Fig. 1), each
quartile ideally would contain five episodes.
However, because six patients were treated for
three days, and only four patients were treated
for four days, the quartiles do not contain the
same numbers. This methodology was used to
define each quartile in this study, attempting to
assign as close to 25% of the patients to each
quartile as possible. Once the quartiles were de-
fined, the data were analyzed to determine the
quartile with the lowest rate of recurrence, and
univariate analysis was performed to compare
the quartile with the lowest recurrence rate with
all others. Categorical variables were analyzed
using a chi-square test, and continuous vari-
ables were analyzed using the Student t-test. All
p values are two-tailed, and values of 0.05
were considered statistically significant. Values
are expressed as mean standard error of the
mean or percentage of the group of origin.
A multivariate logistic regression model was
developed to estimate the independent risk of
recurrence and death associated with duration
of antibiotic use. Age, sex, APACHE II score,
ventilator dependence, and transfusions were
chosen a priori and entered in the logistic re-
gression model. Statistical analysis was con-
ducted using S-Plus 2000 Release 3 (MathSoft,
Inc., Seattle WA) with the Design Library func-
T
ABLE
1. D
EMOGRAPHIC
V
ARIABLES OF
A
LL
P
ATIENTS
WITH
E
ITHER
F
EVER OR
L
EUKOCYTOSIS
N 4,470 (80.9)
Sex, M/F (%) 2,500/1,970 (55.9/44.1)
Age, years 51.8 16.9
Race (%)
White 3,616 (80.9)
Black 758 (17.0)
Hispanic 40 ( 0.9)
Other 56 ( 1.3)
Underlying disease (%)
Diabetes mellitus 839 (18.8)
Hypertension 1,399 (31.3)
Coronary artery disease 703 (15.7)
Ventilator dependence 1,104 (24.7)
Steroids 959 (21.5)
Renal insufficiency 247 ( 5.5)
Dialysis 439 ( 9.8)
Obesity 281 ( 6.3)
APACHE II score 14.0 0.1
WBC, per microliter 16.1 0.1
Temperature, °C 38.4 0.0
WBC white blood cell count; APACHE Acute
Physiology and Chronic Health Evaluation.
T
ABLE
2. C
HARACTERISTICS OF
I
NFECTIONS
Sites of infection (%)
Abdomen 929 (20.8)
Lung 762 (17.0)
Blood 636 (14.2)
Wound 594 (13.3)
Skin/Soft Tissue 283 ( 6.3)
Vascular catheter 262 ( 5.9)
Colon 156 ( 3.5)
Community-acquired (%) 1,047 (23.4)
Nosocomial (%) 3,423 (76.6)
Organisms (%) 2,965
No culture 1,092 (24.4)
No growth 109 ( 2.4)
Gram-positive cocci 1,780 (60.0)
MSSA 186 ( 6.3)
MRSA 213 ( 7.2)
S. epidermidis 132 ( 4.5)
MRSE 297 (10.0)
E. faecalis 213 ( 7.2)
E. faecalis VRE 3 ( 0.1)
E. faecium 38 ( 1.3)
E. faecium VRE 73 ( 2.5)
Streptococcus spp. 191 ( 6.4)
Gram-negative bacilli 1,463 (49.3)
E. coli 261 ( 8.8)
P. aeruginosa 246 ( 8.3)
K. pneumoniae 153 ( 5.2)
E. cloacae 142 ( 4.8)
Acinetobacter spp. 99 ( 3.3)
Anaerobes 173 ( 5.8)
B. fragilis 52 ( 1.8)
MSSA methicillin-sensitive Staphylococcus aureus;
MRSA methicillin-resistant S. aureus; MRSE methi-
cillin-resistant S. epidermidis; VRE vancomycin-resis-
tant enterococci.
CAN WE DEFINE THE IDEAL DURATION OF ANTIBIOTIC THERAPY? 423
tions [19] and MedCalc software, version
6.10.001 (MedCalc, Mariakerke, Belgium).
RESULTS
There were 5,561 treated infections during the
study period, of which 2,934 were associated
with fever, 2,092 with leukocytosis, and 4,470 in-
fections were associated with either fever or
leukocytosis. Only patients with fever or leuko-
cytosis were included in this analysis because an
antibiotic scheme based on resolution of fever or
leukocytosis was one of the primary goals of the
study. Patient demographic data are character-
ized in Table 1, and a breakdown of the causative
infectious agents is presented in Table 2.
All infections
We initially analyzed all 4,470 infections as-
sociated with fever or leukocytosis and cate-
gorized them into quartiles on the basis of the
absolute duration of antibiotic administration
from the time of initiation to cessation (Table
3). The groups were comparable with regard to
age except for quartile 4, which had a lower
mean age (p 0.0001); also, quartile 1 had a
higher proportion of women (p 0.0001) and
a lower APACHE II score (p 0.0001). The
mean duration of treatment for Quartile 1 was
4.4 0.1 days. Quartile 1 had the lowest re-
currence rate (14.2%) as defined. The adjusted
odds ratios (OR) of quartiles 2–4 compared
with quartile 1 with regard to recurrence were
T
ABLE
3. A
LL
I
NFECTIONS
C
ATEGORIZED INTO
Q
UARTILES
A
CCORDING TO
T
OTAL
D
URATION OF
A
NTIBIOTIC
T
HERAPY
Quartile 1 Quartile 2 Quartile 3 Quartile 4
Range
All patients 0–7 days 8–12 days 13–17 days 17 days
N
All patients 1,231 1,146 1,112 981
Age (years)
All patients 52.9 0.5 52.0 0.5
a
52.2 0.5
a
49.4 0.5
a
Sex, M/F
All patients 608/623 671/475
a
651/461
a
570/411
a
APACHE II
All patients 12.0 0.2 13.5 0.2
a
15.1 0.2
a
15.8 0.3
a
Days abx
All patients 4.4 0.1 10.0 0.0
a
14.7 0.0
a
27.2 0.7
a
Recurrence (%)
All patients 175 (14.2) 201 (17.5) 269 (24.2)
b
377 (38.4)
a
abx antibiotics; APACHE Acute Physiology and Chronic Health Evaluation.
Adjusted odds ratios are reported using variables of age, APACHE II score, need for blood transfusion, ventilator
dependence, and sex to compare rate of recurrence for each quartile against Quartile 1.
a
P 0.001 and
b
P 0.05 compared with Quartile 1.
*P 0.05 and
†
P 0.001 compared with Quartile 1.
Risk of recurrence
Adjusted OR
0.40 0.60 0.80 1.20 1.60 2.20
Quartile 2 vs. Quartile 1
Quartile 3 vs. Quartile 1
✽
Quartile 4 vs. Quartile 1
†
424 HEDRICK ET AL.
calculated using the variables of age, APACHE
II score, sex, ventilator dependence, and need
for blood transfusion. There was no statistically
significant difference in outcome between
Quartiles 1 and 2. Quartiles 3 and 4 demon-
strated a higher rate of recurrence, with ORs of
1.40 (1.14–1.72) and 2.27 (1.85–2.79), respec-
tively (p 0.001). The mortality rates were
similar among all quartiles at 6%, 5%, 8%, and
6%, respectively, for quartiles 1–4.
Tables 4 and 5 represent all infections cate-
gorized into quartiles on the basis of the num-
ber of days of antibiotic treatment administered
after either normalization of the WBC count
(Table 4) or fever (Table 5). Of the 2,092 patients
with WBC 11,000/mm
3
, the lowest recur-
rence rates were observed in Quartiles 1 and 2
(OR 1.69 [1.26–2.25] and 2.58 [1.92–3.46] for
quartiles 3 and 4, respectively). The mortality
rates were similar among all quartiles at 6%,
5%, 7%, and 8% for Quartiles 1–4, respectively.
The 2,934 infections associated with fever
demonstrated similar results; all patients in
Quartiles 1 and 2 had a better outcome than
those in Quartiles 3 and 4. Again, mortality
rates were similar for each quartile, ranging
from 4% to 6%. All analyses also were per-
formed for general surgery, trauma, and trans-
plant patients separately, with similar results.
The first quartiles for each of the three treat-
ment strategies (fixed duration, duration after
resolution of fever or leukocytosis) were then
compared for all infections with regard to an-
tibiotic days and recurrence. Quartile 1 of the
T
ABLE
4. A
LL
I
NFECTIONS
D
IVIDED INTO
Q
UARTILES
A
CCORDING TO
D
URATION
OF
A
NTIBIOTICS
R
ECEIVED AFTER
R
ESOLUTION OF
L
EUKOCYTOSIS
Quartile 1 Quartile 2 Quartile 3 Quartile 4
Range
All patients 0–4 days 5–9 days 10–14 days 14 days
N
All patients 503 552 481 556
Age (years)
All patients 52.4 0.8 52.1 0.7
a
53.1 0.8
a
51.4 0.7b
Sex, M/F
All patients 252/251 302/250 277/204
a
319/237
a
APACHE II
All patients 13.7 0.3 14.6 0.3
a
14.6 0.3
a
15.8 0.4
b
Days abx
All patients 7.5 0.2 10.6 0.2
a
14.4 0.1
a
26.1 0.8b
Any comp. (%)
All patients 86 (17.0) 125 (22.6) 118 (24.5)
a
185 (33.3)
b
abx antibiotics; APACHE Acute Physiology and Chronic Health Evaluation.
Adjusted odds ratios are reported using variables of age, APACHE II score, need for blood transfusion, ventilator
dependence, and sex to compare rate of recurrence for each quartile against Quartile 1.
a
P 0.05 and
b
P 0.001 compared with Quartile 1.
*P 0.05 and
†
P 0.001 compared with Quartile 1.
Risk of recurrence
Adjusted OR
0.40 0.60 1.00 1.40 2.00 2.80
Quartile 2 vs. Quartile 1
Quartile 3 vs. Quartile 1
✽
Quartile 4 vs. Quartile 1
†
CAN WE DEFINE THE IDEAL DURATION OF ANTIBIOTIC THERAPY? 425
fixed-duration group represented the shortest
duration (4.4 0.1 days), followed by fever
(6.2 0.2 days; p 0.0001 versus Quartile 1 of
fixed-duration analysis) and leukocytosis
(7.5 0.2 days; p 0.0001 versus Quartile 1 of
fixed-duration analysis). The recurrence rate
was 14.2% in Quartile 1 of the fixed-duration
group, 17.0% (p 0.14) in Quartile 1 of the
leukocytosis group, and 18.6% (p 0.05) in
Quartile 1 of the fever group.
Intra-abdominal infections
Surgeons are particularly concerned with in-
tra-abdominal infections and pneumonia (the
most frequently lethal nosocomial infection);
therefore, we examined these individual infec-
tions separately using the same approach as for
all infections (Tables 6–11). All intra-abdominal
infections associated with either a fever or
leukocytosis (N 929) were analyzed by fixed
duration of therapy (Table 6), duration after res-
olution of leukocytosis (Table 7), and duration
after resolution of fever (Table 8). The results
were similar to those reported for all infections,
indicating that Quartiles 1 and 2 were associ-
ated with the lowest recurrence rate in the fixed-
duration (OR 1.81 [1.12–2.92] and 2.79
[1.75–4.47] for Quartiles 3 and 4, respectively)
and leukocytosis (OR 2.13 [1.10–4.13] and 2.42
[1.25–4.67] groups comparing Quartiles 3 and 4,
respectively, with Quartile 1). In the analysis of
duration of treatment following resolution of
fever in intra-abdominal infections, similar out-
T
ABLE
5. A
LL
I
NFECTIONS
D
IVIDED INTO
Q
UARTILES
A
CCORDING TO
D
URATION
OF
A
NTIBIOTICS
R
ECEIVED
A
FTER
R
ESOLUTION OF
F
EVER
Quartile 1 Quartile 2 Quartile 3 Quartile 4
Range
All patients 0–6 days 7–9 days 10–14 days 14 days
N
All patients 673 753 783 725
Age (years)
All patients 50.5 0.7 51.3 0.6 51.3 0.6
b
48.9 0.6
b
Sex, M/F
All patients 371/302 421/332 490/293
a
426/299
APACHE II
All patients 13.6 0.3 14.2 0.3 15.3 0.2
b
15.9 0.3
b
Days abx
All patients 6.2 0.2 10.2 0.1 14.4 0.1
b
26.1 0.7
b
Any comp. (%)
All patients 125 (18.6) 139 (18.5) 185 (23.6)
a
248 (34.2)
b
Risk of recurrence
Adjusted OR
0.50 0.75 1.00 1.40 2.00 2.60
Quartile 2 vs. Quartile 1
Quartile 3 vs. Quartile 1
✽
Quartile 4 vs. Quartile 1
†
abx antibiotics; APACHE Acute Physiology and Chronic Health Evaluation.
Adjusted odds ratios are reported using variables of age, APACHE II score, need for blood transfusion, ventilator
dependence, and gender to compare the rate of recurrence for each quartile against Quartile 1.
a
P 0.05 and
b
P 0.001 compared with Quartile 1.
*P 0.05 and
†
P 0.001 compared with Quartile 1.
426 HEDRICK ET AL.
comes were demonstrated in the first three
quartiles (no statistically significant differences
among quartiles 1–3; OR 2.27 [1.29–3.99] for
Quartile 4 vs. Quartile 1). Mortality rates were
similar for each group in each quartile, ranging
from 2% to 6%. All analyses were performed for
general surgery, trauma, and transplant pa-
tients separately with similar results.
The total lengths of treatment were com-
pared for Quartiles 1 of fixed duration of ther-
apy versus time to resolution of fever or leuko-
cytosis. Quartile 1 of the fixed-duration group
represented the shortest duration (4.5 0.1
days), followed by fever resolution (7.0 0.5
days; p 0.0001 vs. Quartile 1 of the fixed-du-
ration group) and leukocytosis resolution
(7.2 0.4 days; p 0.0001 vs. Quartile 1 of the
fixed-duration group), with a recurrence rate
of 10.6% in Quartile 1 of the fixed-duration
group, 14.6% (p 0.31) in Quartile 1 of the
leukocytosis resolution group, and 21.1% (p
0.05) in Quartile 1 of the fever resolution
group.
Pneumonia
Tables 9–11 represent the data for patients
with pneumonia associated with either fever
or leukocytosis (N 762). Quartiles were
again based on a fixed duration of therapy
(Table 9) and duration from resolution of
leukocytosis (Table 10) or fever (Table 11). For
T
ABLE
6. I
NTRA
-A
BDOMINAL
I
NFECTIONS
C
ATEGORIZED INTO
Q
UARTILES
A
CCORDING TO
T
OTAL
D
URATION OF
A
NTIBIOTIC
T
HERAPY
Quartile 1 Quartile 2 Quartile 3 Quartile 4
Range
All patients 0–7 days 8–12 days 13–17 days 17 days
N
All patients 218 217 246 248
Age (years)
All patients 53.0 1.2 53.1 1.1
a
53.2 1.1b 50.8 1.0
b
Sex, M/F
All patients 106/112 125/92 143/103
a
138/110
APACHE II
All patients 10.1 0.4 11.9 0.4
a
13.3 0.4
b
14.4 0.5
b
Days abx
All patients 4.5 0.1 10.0 0.1
a
14.8 0.1b 29.5 1.8
b
Any comp. (%)
All patients 23 (10.6) 35 (16.1) 53 (21.5)
a
86 (34.7)
b
Risk of recurrence
Adjusted OR
0.50 0.75 1.50 2.50 4.00
Quartile 2 vs. Quartile 1
Quartile 3 vs. Quartile 1
✽
Quartile 4 vs. Quartile 1
†
abx antibiotics; APACHE Acute Physiology and Chronic Health Evaluation.
Adjusted odds ratios are reported using variables of age, APACHE II score, need for blood transfusion, ventilator
dependence, and sex to compare rate of recurrence for each quartile against Quartile 1.
a
P 0.05 and
b
P 0.001 compared with Quartile 1.
*P 0.05 and
†
P 0.001 compared with Quartile 1.
CAN WE DEFINE THE IDEAL DURATION OF ANTIBIOTIC THERAPY? 427
pneumonias categorized by fixed duration of
therapy, duration after resolution of leukocy-
tosis, or duration after resolution of fever,
there was no statistically significant differ-
ence in recurrence of infection among Quar-
tiles 1–3. In all three groups, however, Quar-
tile 4 demonstrated a higher rate of recurrence
than Quartile 1 (p 0.05 for all three groups).
Ventilator-associated pneumonias (VAPs)
were then analyzed separately. There were
374 VAPs associated with either fever or
leukocytosis. Quartiles 1 and 2 were consis-
tently associated with the lowest recurrence
rates compared with Quartiles 3 and 4, with
similar mortality rates (data not shown). All
analyses were performed for general surgery,
trauma, and transplant patients separately,
with similar results.
Finally, Quartile 1 of each group was com-
pared with regard to the mean duration of
treatment and recurrence of infection. Quartile
1 of the fixed-duration group represented the
shortest duration (5.5 0.1 days), with a com-
plication rate of 19.7%, followed by fever reso-
lution (7.1 0.3 days; p 0.0001 vs. quartile 1
of the fixed-duration group) and leukocytosis
resolution (9.0 0.6 days; p 0.0001 vs. quar-
tile 1 of the fixed-duration group). The recur-
rence rates were 21.9% and 26.7% in the fever
and leukocytosis groups, respectively (p 0.05
vs. complication rate of Quartile 1 in fixed-du-
ration group).
T
ABLE
7. I
NTRA
-A
BDOMINAL
I
NFECTIONS
D
IVIDED INTO
Q
UARTILES
A
CCORDING
TO
D
URATION OF
A
NTIBIOTICS
R
ECEIVED AFTER
R
ESOLUTION OF
L
EUKOCYTOSIS
Quartile 1 Quartile 2 Quartile 3 Quartile 4
Range
All patients 0–5 days 6–10 days 11–15 days 15 days
N
All patients 130 127 124 118
Age (years)
All patients 55.3 1.5 53.1 1.5 52.8 1.5
a
51.9 1.4
a
Sex, M/F
All patients 59/71 65/62 75/49
a
58/60
APACHE II
All patients 11.6 0.6 12.6 0.6 14.1 0.6
a
14.5 0.8
a
Days abx
All patients 7.2 0.4 12.0 0.5 15.5 0.3
a
28.7 1.1
a
Any comp. (%)
All patients 19 (14.6) 23 (18.1) 32 (25.8)
a
41 (34.7)
a
Risk of recurrence
Adjusted OR
0.50 1.00 2.00 3.50
Quartile 2 vs. Quartile 1
Quartile 3 vs. Quartile 1
✽
Quartile 4 vs. Quartile 1
✽
abx antibiotics; APACHE Acute Physiology and Chronic Health Evaluation.
Adjusted odds ratios are reported using variables of age, APACHE II score, need for blood transfusion, ventilator
dependence, and sex to compare rate of recurrence for each quartile against Quartile 1.
a
P 0.05 compared with Quartile 1.
*P 0.05 compared with Quartile 1.
428 HEDRICK ET AL.
DISCUSSION
According to these data, shorter courses of
antibiotics may result in outcomes similar to
or better than those of longer courses of ther-
apy for all infections and for pneumonia and
intra-abdominal infections specifically. With
regard to the method used for determination
of treatment duration, the physiology-based
analyses seemed to result in a trend toward
longer treatment without improvement in
outcome compared with a fixed duration of
treatment.
Our results were similar to those other stud-
ies that have advocated shortened antibiotic
courses. Chastre et al. [10] published a ran-
domized, double-blind clinical trial comparing
eight and 15 days of antibiotic treatment for
VAP and found no difference in outcome. A
similar study was undertaken for community-
acquired pneumonia comparing seven and 10
days of antibiotics and failed to demonstrate a
better outcome with the longer duration of
therapy [20]. Several authors have also investi-
gated the use of shortened courses of antibi-
otics in intra-abdominal infections and have
found similar outcomes with shorter courses
[8,9].
Possible explanations for our findings relate
to prevention of the development of resistance,
failure to mask the presence of an occult infec-
tion, and overall prevention of healthcare-as-
T
ABLE
8. I
NTRA
-A
BDOMINAL
I
NFECTIONS
D
IVIDED INTO
Q
UARTILES
A
CCORDING
TO
D
URATION OF
A
NTIBIOTICS
R
ECEIVED AFTER
R
ESOLUTION OF
F
EVER
Quartile 1 Quartile 2 Quartile 3 Quartile 4
Range
All patients 0–6 days 7–11 days 12–16 days 16 days
N
All patients 128 132 144 146
Age (years)
All patients 50.8 1.6 51.0 1.5 51.1 1.3
a
49.6 1.3
a
Sex, M/F
All patients 68/60 80/52 80/64 86/60
APACHE II
All patients 11.9 0.6 11.4 0.5 14.0 0.6
a
14.7 0.7
a
Days abx
All patients 7.0 0.5 11.5 0.2 15.8 0.2
a
32.4 2.9
a
Any comp. (%)
All patients 27 (21.1) 26 (19.7) 24 (16.7) 52 (35.6)
a
abx antibiotics; APACHE Acute Physiology and Chronic Health Evaluation.
Adjusted odds ratios are reported using variables of age, APACHE II score, need for blood transfusion, ventilator
dependence, and sex to compare rate of recurrence for each quartile against Quartile 1.
a
P 0.05 compared with Quartile 1.
*P 0.05 compared with Quartile 1.
Risk of recurrence
Adjusted OR
0.50 0.75 1.50 2.50 3.50
Quartile 2 vs. Quartile 1
Quartile 3 vs. Quartile 1
Quartile 4 vs. Quartile 1
✽
CAN WE DEFINE THE IDEAL DURATION OF ANTIBIOTIC THERAPY? 429
sociated infections. Infections with resistant or-
ganisms have been associated with prolonged
antibiotic use and poorer outcomes [3,4]; there-
fore, by limiting exposure to antibiotics, resis-
tance may be prevented, and outcomes may be
improved. Additionally, superinfections or re-
current infections can oftentimes be masked
with prolonged antibiotic use owing to failure
to eradicate the infection completely; this no-
tion underlies the logic of stopping antibiotics
after a specified course of therapy even in a pa-
tient with continuing fever or leukocytosis. Fi-
nally, patients who remain in the hospital to
complete a course of intravenous antibiotics are
more vulnerable to the development of a noso-
comial infection.
The two methods examined in this study for
determining the proper duration of antibiotic
treatment have been well described in the litera-
ture. Traditionally, fixed-duration therapy has
been the standard for treating infections; how-
ever, some studies have demonstrated better out-
comes with treatment regimens based on resolu-
tion of physiologic parameters such as fever or
leukocytosis in comparison with a set duration of
antibiotic therapy [13–15]. One randomized trial
in 94 patients with complicated appendicitis, con-
ducted by Taylor et al. [15], compared the two
methods and reported similar outcomes with a
longer duration of therapy. In contrast to our orig-
inal hypothesis, the results of our study seem to
support fixed-duration therapy as opposed to a
T
ABLE
9. P
NEUMONIAS
C
ATEGORIZED INTO
Q
UARTILES
A
CCORDING TO
T
OTAL
D
URATION OF
A
NTIBIOTICS
R
ECEIVED
Quartile 1 Quartile 2 Quartile 3 Quartile 4
Range
All patients 0–7 days 8–11 days 12–15 days 15 days
N
All patients 183 183 180 216
Age (years)
All patients 50.8 1.4 49.5 1.4 51.6 1.4 49.1 1.1
a
Sex, M/F
All patients 119/64 120/63 123/57 148/68
APACHE II
All patients 15.7 0.5 15.9 0.5 16.9 0.5 18.3 0.5
a
Days abx
All patients 5.5 0.1 9.7 0.1 13.5 0.1 24.3 1.8
a
Any comp. (%)
All patients 36 (19.7) 35 (19.1) 51 (28.3) 102 (47.2)
b
Risk of recurrence
Adjusted OR
0.50 1.00 2.0 3.50
Quartile 2 vs. Quartile 1
Quartile 3 vs. Quartile 1
Quartile 4 vs. Quartile 1
†
abx antibiotics; APACHE Acute Physiology and Chronic Health Evaluation.
Adjusted odds ratios are reported using variables of age, APACHE II score, need for blood transfusion, ventilator
dependence, and sex to compare rate of recurrence for each quartile against Quartile 1.
a
P 0.05 and
b
P 0.001 compared with Quartile 1.
†
P 0.001 compared with Quartile 1.
430 HEDRICK ET AL.
physiology-based method in providing the short-
est courses of treatment (lower cost) with similar
outcomes. These results imply that, for our pa-
tient population, the host response to infection
may persist after the bacteria are eradicated, and
that ongoing therapy does not improve out-
comes. Perhaps a combination of the methods
based on a set duration of therapy that can be dis-
continued earlier if the patient defervesces or the
leukocytosis resolves may be most reasonable.
Our data are limited by the retrospective na-
ture of the study; thus, we are unable to demon-
strate a causal relationship. It is possible that
the patients deemed less sick (and hence at a
lower risk for recurrence) were treated with
shorter-course therapy and ultimately fared
better. Similarly, patients who continued to
manifest signs/symptoms of illness were likely
continued on antibiotics and may have devel-
oped recurrences more frequently. In the ab-
sence of prospectively defined treatment
groups, we are unable to control for these bi-
ases, and therefore, further prospective trials
are needed.
In addition, we may not have accounted for
all of the confounding factors that contribute to
recurrence. Although multivariate logistic re-
gression analysis was performed to adjust for
age, APACHE II score, and need for blood
transfusion or mechanical ventilation, invari-
ably, this does not control completely for the
impact of co-morbid conditions on clinical out-
T
ABLE
10. P
NEUMONIAS
D
IVIDED INTO
Q
UARTILES
A
CCORDING TO
D
URATION OF
A
NTIBIOTICS
R
ECEIVED AFTER
R
ESOLUTION OF
L
EUKOCYTOSIS
Quartile 1 Quartile 2 Quartile 3 Quartile 4
Range
All patients 0–4 5–8 9–12 12
N
All patients 86 92 84 92
Age (years)
All patients 47.5 2.0 49.3 2.0 51.7 2.0 51.9 1.7
a
Sex, M/F
All patients 55/31 62/30 54/30 64/28
APACHE II
All patients 16.9 0.8 16.0 0.7 17.2 0.7 18.8 0.9
a
Days abx
All patients 9.0 0.6 10.4 0.5 13.9 0.4 27.7 4.2
a
Any comp. (%)
All patients 23 (26.7) 19 (20.7) 32 (38.1) 36 (39.1)
a
Risk of recurrence
Adjusted OR
0.50 1.00 2.0 3.50
Quartile 2 vs. Quartile 1
Quartile 3 vs. Quartile 1
Quartile 4 vs. Quartile 1
*
abx antibiotics; APACHE Acute Physiology and Chronic Health Evaluation.
Adjusted odds ratios are reported using variables of age, APACHE II score, need for blood transfusion, ventilator
dependence, and sex to compare rate of recurrence for each quartile against Quartile 1.
a
P 0.05 compared with Quartile 1.
*P 0.05 compared with Quartile 1.
CAN WE DEFINE THE IDEAL DURATION OF ANTIBIOTIC THERAPY? 431
comes. Nonetheless, in no circumstance did a
longer treatment period appear to be associated
with even a trend toward a better outcome, de-
spite controlling for these variables. Therefore,
the likelihood of a missed benefit of long treat-
ment courses would appear to be small.
Patients who died while on antibiotics were
excluded from the analysis, as we were inter-
ested in evaluating only the duration of a com-
pleted course of antibiotic therapy. Death while
on antibiotics may result from inadequate em-
piric treatment, delay in diagnosis, or pul-
monary failure related to VAP but cannot re-
flect the impact of antibiotic duration because
the patient did not complete therapy. Within a
randomized trial, the patients who die early in
the course of therapy (from complications re-
lated to infection or from their underlying ill-
ness/injury) will theoretically be equally dis-
tributed within the two treatment groups.
However, in this retrospective study, those pa-
tients who died while still being treated with
antibiotics would automatically have been in
the short-course group, thereby artificially in-
creasing the mortality rate within this group.
Additionally, patients who manifested neither
fever nor leukocytosis were excluded, as we
were interested in evaluating the duration of
therapy in relation to normalization of these
physiologic parameters. Although unlikely, it
T
ABLE
11. P
NEUMONIAS
D
IVIDED INTO
Q
UARTILES
A
CCORDING TO
D
URATION
OF
A
NTIBIOTICS
R
ECEIVED AFTER
R
ESOLUTION OF
F
EVER
Quartile 1 Quartile 2 Quartile 3 Quartile 4
Range
All patients 0–5 6–8 9–12 12
N
All patients 146 139 149 167
Age (years)
All patients 49.0 1.5 49.8 1.6 48.9 1.5 51.0 1.3
Sex, M/F
All patients 101/45 87/52 111/38 108/59
APACHE II
All patients 17.0 0.6 16.3 0.6 16.0 0.5 18.7 0.6
Days abx
All patients 7.1 0.3 10.5 0.3 13.5 0.3 22.2 1.1
Any comp. (%)
All patients 32 (21.9) 36 (25.9) 39 (26.2) 70 (41.9)
a
abx antibiotics; APACHE Acute Physiology and Chronic Health Evaluation.
Adjusted odds ratios are reported using variables of age, APACHE II score, need for blood transfusion, ventilator
dependence, and sex to compare rate of recurrence for each quartile against Quartile 1.
a
P 0.05 compared with Quartile 1.
*P 0.05 compared with Quartile 1.
Risk of recurrence
Adjusted OR
0.40 0.60 1.00 1.40 2.00 2.80
Quartile 2 vs. Quartile 1
Quartile 3 vs. Quartile 1
Quartile 4 vs. Quartile 1
✽
432 HEDRICK ET AL.
is possible that those patients may behave dif-
ferently than the subset of patients we studied.
CONCLUSIONS
The evidence presented in the current study
supports (but does not prove) the benefits of a
shortened duration of treatment for surgical pa-
tients with infections. Within the limits of this ob-
servational analysis, longer treatment periods
were actually associated with worse outcomes,
unlike prior studies that have demonstrated
equivalence between different treatment periods.
Large randomized trials are warranted to un-
derstand the true impact of a shortened duration
of therapy and to determine if fixed-duration or
physiology-based methods for determining the
length of treatment are more beneficial.
REFERENCES
1. Goldmann DA, Weinstein RA, Wenzel RP, et al.
Strategies to prevent and control the emergence and
spread of antimicrobial-resistant microorganisms in
hospitals: A challenge to hospital leadership. JAMA
1996;275:234–240.
2. National Nosocomial Infections Surveillance (NNIS)
System report, data summary from January 1992–
April 2000, issued June 2000. Am J Infect Control
2000;28:429–445.
3. Pelletier SJ, Raymond DP, Sawyer RG, et al. Outcome
analysis of intraabdominal infection with resistant
gram-positive organisms. Surg Infect 2002;3:11–19.
4. Peres-Bota D, Rodriguez H, Dimopoulos, et al. Are
infections due to resistant pathogens associated with
a worse outcome in critically ill patients? J Infect
2003;47:307–316.
5. Rice LB. Controlling antibiotic resistance in the ICU:
Different bacteria, different strategies. Cleve Clin J
Med 2003;70:793–800.
6. Gorecki P, Schein M, Rucinski JC, et al. Antibiotic ad-
ministration in patients undergoing common surgical
procedures in a community teaching hospital: The
chaos continues. World J Surg 1999;23:429–432.
7. Mora Y, Avila-Aguero ML, Umana MA, et al. Epi-
demiological observations of the judicious use of an-
tibiotics in a pediatric teaching hospital. Int J Infect
Dis 2002;6:74–77.
8. Andäker L. Höjer H, Kihlström E, et al. Stratified du-
ration of prophylactic antimicrobial treatment in
emergency abdominal surgery: Metronidazole-
fosfomycin vs. metronidazole-gentamicin in 381 pa-
tients. Acta Chir Scand 1987;153:185–192.
9. Schein M, Assalia A, Bachus H. Minimal antibiotic
therapy after emergency abdominal surgery: A
prospective study. Br J Surg 1994;81:989–991.
10. Chastre J, Wolff M, Fagon J, et al. Comparison of 8 vs.
15 days of antibiotic therapy for ventilator-associated
pneumonia in adults. JAMA 2003;290:2588–2598.
11. Hedrick TL, Sawyer RG. Duration of antimicrobial
therapy in intra-abdominal infections. Infect Med
2004;21:506–510.
12. Mazuski JE, Sawyer RG, Nathens AB, et al. The Sur-
gical Infection Society Guidelines on antimicrobial
therapy for intra-abdominal infections: An executive
summary. Surg Infect 2002;3:161–173.
13. Lennard ES, Minshew BH, Dellinger EP, et al. Leuko-
cytosis at termination of antibiotic therapy: Its im-
portance for intra-abdominal sepsis. Arch Surg
1980;115:918–921.
14. Lennard ES, Dellinger EP, Wertz MJ, et al. Implica-
tion of leukocytosis and fever at conclusion of antibi-
otic therapy for intra-abdominal sepsis. Arch Surg
1982;195:19–24.
15. Taylor E, Dev V, Shah D, et al. Complicated appen-
dicitis: Is there a minimum intravenous antibiotic re-
quirement? A prospective randomized trial. Am Surg
2000;66:887–890.
16. Garner JS, Jarvis WR, Emori TG, et al. CDC defini-
tions for nosocomial infections, 1988. Am J Infect Con-
trol 1988;16:128–140.
17. Pelletier SJ, Crabtree TD, Gleason TG, et al. Bac-
teremia associated with central venous catheter in-
fection is not an independent predictor of outcomes.
J Am Coll Surg 2000;190:671–680.
18. Knaus WA, Draper EA, Wagner DP, Zimmerman JE.
APACHE II: A severity of disease classification sys-
tem. Crit Care Med 1985;13:818–829.
19. Harrell FE. Regression Modeling Strategies with Ap-
plications to Linear Models, Logistic Regression, and
Survival Analysis. New York. Springer, 2001.
20. Siegel RE, Alicea M, Lee A, Blaiklock R. Comparison
of 7 versus 10 days of antibiotic therapy for hospital-
ized patients with uncomplicated community-ac-
quired pneumonia: A prospective, randomized, dou-
ble-blind study. Am J Ther 1999;6:217–222.
Address reprint requests to:
Dr. Traci L. Hedrick
P.O. Box 801380
Department of Surgery
University of Virginia Health System
Charlottesville, VA 22908
E-mail: Th8q@virginia.edu
