Veterinary Surgery

33:542550, 2004

A Prospective Study of Postoperative Surgical Site Infections

in Dogs and Cats
SIMONE EUGSTER, DrMedVet, PETER SCHAWALDER Prof DrMedVet,
FREDERIC GASCHEN, PD, DrMedVet, Diplomate ECVIM-CA & ACVIM, and PATRICK BOERLIN,

Prof DrMedVet

ObjectiveTo assess postoperative surgical site infection (SSI) rate and to identify associated
predictive factors.
Study DesignProspective clinical study.
AnimalsDogs and cats that had surgery (1010 interventions) during 58 weeks from April 1999 to
June 2000.
MethodsData sheets were completed by clinicians. Patients were controlled for clinical evidence
of SSI at suture removal. Two definitions of SSI (infection and infection/inammation) were
developed specifically for this study and used for statistical analysis. Logistic regression models were
built in order to identify significant predictive factors for SSI.
ResultsWounds with infection/inammation occurred in 5.8% and infected wounds in 3%
of patients. The outcome infection was associated with 3 major risk factors (duration of surgery,
increasing number of persons in the operating room, dirty surgical site) and 1 protective factor
(antimicrobial prophylaxis). The outcome infection/inammation was associated with 6 significant factors (duration of anesthesia, duration of postoperative intensive care unit stay, wound
drainage, increasing patient weight, dirty surgical site, and antimicrobial prophylaxis).
ConclusionsSSI frequency in companion animals is comparable with the frequency observed in
human surgical patients. Several significant predictive factors for SSI in small animals surgery were
identied.
Clinical RelevanceBaseline information for SSI surveillance in our hospital and for comparison
with other studies was dened. The factors identied may help to predict infections in surgical
patients and to take adequate preventive measures for patients at risk.
r Copyright 2004 by The American College of Veterinary Surgeons
Key words: infection, surgery, wound, antimicrobial prophylaxis, cat, dog.

remains a substantial cause of morbidity and mortality

among hospitalized patients.14
To our knowledge, few studies have evaluated risk
factors for postoperative SSI in small animals511 and
most available information on risk factors associated
with SSI originates from human surgical patients. Factors generally considered as definite patient-associated
determinants for SSI are extremes of age, morbid obesity,

INTRODUCTION

OSOCOMIAL INFECTIONS in general, and

postoperative surgical site infections (SSIs) in particular, are a growing concern in veterinary hospitals.
Surveillance and control programs of SSI are well established in human hospitals. Despite all preventive measures, it is estimated that SSI range from 2% to 5%, and

From the Divisions of Surgery and Internal Medicine, Companion Animal Hospital, Department for Clinical Veterinary Medicine
and the Institute for Veterinary Bacteriology, University of Bern, Switzerland.
Address reprint requests to Dr. P. Boerlin, DrMedVet, Department of Pathobiology, University of Guelph, Guelph, ON, Canada
N1G 2W1. E-mail: pboerlin@uoguelph.ca.
Dr. Boerlins current address is Department of Pathobiology, Ontario Veterinary College, University of Guelph, Canada.
Submitted February 2003; Accepted June 2004
r Copyright 2004 by The American College of Veterinary Surgeons
0161-3499/04
doi:10.1111/j.1532-950X.2004.04076.x

542

EUGSTER ET AL.

remote infection, American Society of Anesthesiologists

preoperative assessment scores (ASA scores) of
3, and
prolonged preoperative hospitalization.14,1214 A clear
correlation between the 4 categories of wound contamination (clean, clean-contaminated, contaminated, and
dirty) and SSI rate has been reported.2,10,12,15 Predictably,
infection risk increases with increasing wound contamination.16 However, because of wide variation in infection
rates associated with different operative procedures within the same wound category, it is generally accepted that
a risk stratication based on wound categories alone is
insufcient.
Culver et al.3 developed a risk index score to predict a
surgical patients risk of acquiring SSI, which performs
significantly better across a broad range of surgical procedures. This risk index score ranges from 0 to 3 and
results from a number of the following risk factors: (a)
ASA score of 3, 4, or 5; (b) intervention classied as either
contaminated or dirty; (c) intervention lasting over T
hours, where T depends upon the surgical procedure being performed. Definite procedure-related factors associated with SSI are preoperative hair removal the day
before surgery, duration of surgery, and antibiotic prophylaxis.1,4,79,12,17,18 Several other factors have remained
controversial as determinants of SSI and include: diabetes
mellitus, malnutrition and low serum albumin, malignancy, immunosuppressive therapy, steroids, multiple operative procedures during the same anesthesia, tissue trauma,
foreign material, drains, and emergency surgery.1,12
Our objective was to dene the postoperative SSI rate
in our companion animal clinic to obtain baseline information for surveillance and for comparison with other
clinics. We also wanted to identify risk factors associated
with SSI in dogs and cats.

MATERIALS AND METHODS

Study Design
Data were systematically collected from dogs and cats that
had a surgical procedure between April 26, 1999, and June 9,
2000. All interventions were included except dental and ophthalmic surgery, which were performed in a separate building
and under different conditions than general surgery. Feline
castration was also excluded, because they are not routine in
our clinic and wound closure and postintervention follow-up
are usually different from other interventions. Procedures like
esophagostomy or chest tube placement that required a small
incision but needed subsequent wound closure or were performed in patients subsequently hospitalized for several days
were included. A data sheet was included within the medical
record of every surgical patient when the animal was prepared
for anesthesia. The 1st part of the record covering the pre- and
intraoperative period was completed by an anesthesiologist or
technician monitoring the patient during anesthesia. The at-

543

tending clinician completed the 2nd part of the record covering the postoperative period until patient discharge. Missing
information was obtained retrospectively from the anesthesia
protocols and surgery reports.

Surgical Preparation
The same aseptic preparation protocol was used for all
patients. After induction, surgical sites were clipped and
scrubbed for 5 minutes with a povidone iodine detergent, then
the patient was moved from the induction room to the operating room, and the surgical site was sprayed with a povidone iodine solution just before draping. Patients had surgery
in one of 2 surgical suites: 1 suite equipped with a single surgery table was reserved for clean procedures, and the other
suite had 2 surgical tables where both clean and non-clean
procedures were performed. Interventions not requiring entirely sterile conditions were occasionally performed in the
induction room. Cefalexin (20 mg/kg intravenously; Rilexines , Virbac, Carros, France) was administered prophylactically at induction for most clean and clean-contaminated
procedures. Additional doses were administered every 3 hours
until the end of surgery. If infection was present (procedures
classied as contaminated or dirty), antimicrobial agents specific for that infection were administered.
Preparation of the surgeon and assistants included scrubbing for 5 minutes with a 4% chlorhexidine detergent and
wearing sterile impervious disposable paper gowns, latex surgical gloves, masks, and caps. Operating room personnel wore
masks, caps, and washable cotton gowns at all times.

Data Collection
Recorded patient-related data was: dog or cat, sex (male,
male castrated, female, female spayed), age in years (age o1
year was coded as 0), body weight (kg), nutritional status
(subjectively classied as obese, normal, thin, or cachectic),
case origin (referred or directly admitted to the clinic), preoperative hospital stay, and entire hospital stay in days. Procedure data recorded were: interval preparation surgery in
minutes (duration between time aseptic preparation was nished and time of the 1st incision), duration of surgery in
minutes (interval between 1st skin incision and nal closure),
type of anesthesia (general or local), patient ASA score (this
score is representative of the patients physical status before
surgery and ranges from 1 to 5, with 1 being an otherwise
healthy young patient and 5 being a moribund patient not
expected to survive surgery more than 1 day,19 emergency
(yes/no), daytime (surgery performed during emergency hours
or during normal hospital working hours), traumatic insult
(yes/no), number of surgical procedures performed on the
same patient during one anesthesia, operating room, maximum number of persons in the room during surgery, presence
of students (yes/no), wound category (clean, clean-contaminated, contaminated, dirty) classied by the attending surgeon
according to the criteria established by the National Research
Council (1964),20 placement of implant (yes/no), placement of
drain (yes/no; drains were mostly open drains covered with a

544

SURGICAL SITE INFECTIONS IN DOGS AND CATS

sterile bandage inserted in a dedicated additional incision),

presence of other disease which could potentially inuence
wound healing (e.g., endocrinopathy, neoplasia, polytrauma)
or infection (either clearly purulent infection or infections
conrmed by bacteriological analysis) at a remote site (yes/
no), use of pre- and intraoperative prophylactic antimicrobial
agents from 2 hours before 1st skin incision to wound closure
(yes/no), antimicrobial therapy before or after the procedure
(yes/no, excluding pre- and intraoperative prophylactic antibiotic administration), administration of dexamethasone during perioperative period (yes/no), duration of postoperative
stay in the intensive care unit (ICU) in days (durations of o24
hours for routine postoperative recovery were not considered
as stay in the ICU), duration of anesthesia in minutes (interval
between induction to extubation).

Outcome
A hospital surgeon examined the patients at suture removal
for clinical evidence of inammation or SSI. If the animal left
the clinic before suture removal, the referring veterinarian was
informed about the study and received a detailed list of criteria
for SSI evaluation. The hospital clinician or the referring veterinarian was contacted directly or by telephone to obtain
information about the wound condition at suture removal. If
the referring veterinarian did not remove the sutures, the
owner was contacted and interviewed directly by the principal
investigator. Signs of infection registered for the local wound
area were redness, swelling, pain, heat, serous wound drainage, purulent wound drainage, wound dehiscence, abscessation, and stulation.

Definition of SSI
Various definitions have been used for postoperative SSI in
veterinary medicine.511 Therefore, based on the signs of infection listed above, we used 2 different definitions for postoperative SSI matched as closely as possible to those reported
previously. A wound was classied as infected if purulent

drainage, an abscess, or a stula was recorded. A wound was

classied as infected/inamed when it was infected or
when
3 of the following signs were present simultaneously:
redness, swelling, pain, heat, serous discharge, and wound
dehiscence. Rectal temperature measurements during the
postoperative period were not included in these 2 denitions,
because this variable was frequently not recorded. Moreover,
2 previous veterinary studies did not detect an association
between increased postoperative rectal temperature and development of a local SSI.10,11

Statistical Methods
All statistical analyses were performed using computer
software (Statistix for Windowss , Analytical Software,
Tallahassee, FL, and MedCalc 5s , MedCalc Software, Mariakerke, Belgium). When animals had several interventions at
different dates, each intervention was recorded as a separate
surgery and the interventions were considered independently
for statistical evaluation. Associations between qualitative
variables were tested using w2 tests or Fishers exact tests as
appropriate. Associations of quantitative data with the outcomes (infection and infection/inammation) were tested
using MannWhitney rank sum test. The Armitage test for
trend was used for the analysis of associations between ASA
scores or wound categories and outcomes. A logistic regression analysis was performed using a manual stepwise backward elimination procedure starting with all the risk factors
having a P-value  25% in the univariate analysis. The elimination procedures were stopped when the P-value for all
variables in the model was o10%. The Walters hierarchical
coding method was used for the ordinal variables ASA scores
and wound category.21 To maximize the power of the univariate analyses, all available subjects for which the corresponding dependent and independent variables had been
recorded were used. Consequently, the number of patients
analyzed and reported in the results section may vary from
one variable to another (Tables 13). For the multivariate
analyses, only patients with complete data records for the

Table 1. Distribution of Patients and Intervention Characteristics (Qualitative)

Species (n 945) (%)
Sex (n 944) (%)
Nutritional status (n 836) (%)
Origin (n 945) (%)
Emergency hours intervention (n 836) (%)
Traumatic insult (n 836) (%)
ASA score (n 814) (%)
Other disease (n 836) (%)
Wound category (n 836) (%)
Type of anesthesia (n 836) (%)
Implant (n 836) (%)
Drain (n 834) (%)
Operation room (n 836) (%)
Undergraduate students (n 833) (%)
Preoperative antimicrobial treatment (n 834) (%)
Antimicrobial prophylaxis (n 836) (%)
Postoperative antimicrobial treatment (n 835) (%)
Immunosuppression or cytostatics (n 836) (%)

Cats: 29.7
Females: 28.6
Obese: 13.4
Direct: 35.6
Yes: 9.2
Yes: 32.4
1: 25.9
Yes: 16.7
Clean: 70.8
General: 99.8
Yes: 33.9
Yes: 8.3
Clean room: 39.8
Present: 50.7
Yes: 25.1
Yes: 93.2
Yes: 77.5
Yes: 8.3

Dogs: 70.3
Spayed females: 21.6
Normal: 81.0
Referred: 64.4
No: 90.8
No: 67.6
2: 38.8
No: 83.3
Cleancontaminated: 13.4
Local: 0.2
No: 66.1
No: 91.7
Second room: 57.1
Absent: 49.3%
No: 74.9
No: 6.8
No: 22.5
No: 91.7

Males: 33.6
Thin: 5.3

Castrated males: 16.2

Cachectic: 0.4

3: 25.8

4 and 5: 9.5

Contaminated: 10.4

Dirty: 5.4

Induction room: 3.1

545

EUGSTER ET AL.
Table 2. Distribution of Patients and Intervention Characteristics (Quantitative)
Median (Average)
Age of dogs in years (n 659)
Age in years for cats (n 270)
Weight of dogs in kg (n 621)
Weight of cats in kg (n 249)
Duration of preparation in minutes (n 834)
Duration of surgical intervention in minutes (n 835)
Duration of anesthesia in minutes (n 869)
Number of simultaneous surgical procedures (n 872)
Number of persons in operation room (n 759)
Duration of preoperative stay in days (n 835)
Duration of postoperative stay in days (n 820)
Duration of stay in ICU in days (n 832)
Duration of hospital stay in days (n 820)
Time between surgical intervention and suture removal in days (n 930)

variables used at the start of the backward elimination procedure were used for the analysis.

RESULTS
Patient Population
Data from 1010 procedures performed on 735 animals
were studied. Complete records including data on signs of
inammation or infection were obtained from 712
(70.5%) interventions. The distribution of the variables
recorded and used for statistical analysis (Tables 1 and 2)
and use of antimicrobial agents (Table 3) are provided.
SSIs and Associated Factors
Suture removal was performed by a hospital surgeon
(52% of patients), referring veterinarian (38%), or the
owner (10%). Surgical sites were classied as infected/
inamed (55 of 946; 5.8%) or as infected (28; 3.0%)
Variables having an association with the development of
SSI are listed in Table 4. Factors significantly associated
with infection/inammation (Po.05) had the same
trend for infections, but mostly at a non-significant
level. This lack of power most probably resulted from the
relatively low number of cases with infection. Heavier
body weight was significantly associated with postoperative SSI in dogs (P .009) but not in cats. A significant

5.0
2.0
27.0
4.0
20
90
165
1
5
1
2
1
3
11

(5.1)
(3.4)
(25.7)
(3.9)
(22)
(98)
(171)
(1.1)
(5.5)
(1.6)
(3.2)
(1.2)
(4.7)
(11.5)

MinimumMaximum

1st3rd Quartiles

0.115.0
0.217.0
1.285.0
1.29.0
0245
7380
20480
16
218
029
030
017
041
229

2.08.0
1.05.0
11.035.0
3.04.5
1525
60130
115220
11
46
01
14
01
16
1012

increase for risk of infection with increasing wound contamination in the infection/inammation group was
observed (Table 5). A similar trend was visible for the
infected wounds but because of the low number of
positive cases (n 23), this association was not statistically signicant. An increasing ASA score was also
significantly associated with both infection and infection/inammation (Table 6). Pre- and postoperative administrations of antimicrobials were associated with an
increased frequency of SSI. Antimicrobial prophylaxis
seemed to give some protection against surgical wound
infections but the very low numbers of animals without
prophylaxis together with the relatively low frequency of
infection did not result in a statistically significant effect
at the univariate level.
Several other statistically non-significant trends were
also apparent. For instance, the duration of the postoperative stay in the ICU, as well as duration of patient
surgical preparation, of surgery, and anesthesia seemed
associated with higher frequencies of postoperative SSI.
The number of persons in the operating room and the
presence of another disease can be considered similarly.
Finally, duration of postoperative hospitalization and
total hospitalization were both significantly associated
with postoperative infections/inammation at the univariate level (Po.030). Median duration of postoperative stay with no infection was 2 days (mean 3 days; 1st
quartile, 1 day; 3rd quartile, 4 days), whereas median

Table 3. Use of Antimicrobial Agents in Relation with Wound Category

Wound Category
Antimicrobial Use
Preoperative
Perioperative (prophylaxis)
Postoperative

Clean

CleanContaminated

Contaminated

Dirty

15.6% (n 590)
92.2% (n 592)
69.7% (n 591)

28.6% (n 112)
96.4% (n 112)
92.9% (n 112)

60.9% (n 87)
95.4% (n 87)
98.9% (n 87)

71.1% (n 45)
93.3% (n 45)
100% (n 45)

n, number of patients in the corresponding category with available data on antimicrobial use.

546

SURGICAL SITE INFECTIONS IN DOGS AND CATS

Table 4. Factors Associated with Postoperative Surgical Site Infections

(Univariate Level)

Table 6. Percentage of Infections in Relation to ASA Scores

ASA Scores

P-Value
Infection/
Infection Inammation
Weight in dogs (risk)w
Weight in cats (risk)w
Surgical site category (see Table 5)z
ASA score (see Table 3)y
Other simultaneous disease/infection (risk)z
Antimicrobial pretreatment (risk)z
Pre- and intraoperative antimicrobial
prophylaxis (protective)z
Antimicrobial postoperative treatment (risk)z
Operating roomz
Number of persons present during
intervention (risk)w
Duration of patient preparation (risk)w
Duration of anesthesia (risk)w
Duration of surgical intervention (risk)w
Duration of postoperative stay in ICU (risk)w
Placement of a drain (risk)z

.051k
.436k
.161k
.043k
.223k
.038k
.202k

.009k
.127k
.002k
.050k
.303
.131k
.377k

.266
.080k
.107k

.028k
.332
.824k

.114k
.251
.160k
.542
.100k

.152k
.193k
.183k
.079k
o.001k

ASA, anesthesiologists preoperative assessment.

For definition, see Materials and Methods section.
wMannWhitney test.
zw2-test.
yArmitage test for trend.
zFishers exact test; Only variables with P-values equal or smaller than
.25 are included in the list.
kThese variables were used as starting variables for the backward
elimination procedures in the logistic regression models.

duration of postoperative stay for patients with an infection or with infection/inammation was 4 days
(mean 6 days; 1st quartile, 1 day; 3rd quartile 8 days for
infection and mean 7 days; 1st quartile, 1 day; 3rd
quartile, 10 days; for infection/inammation). No association or trend was observed between development of
SSI in animals and the other factors studied.

A 1st multivariate model was built for the outcome

infection using the starting variables indicated in Table
4 and the 712 interventions with complete data sets for
Table 5. Percentage of Wounds with Signs of Infection in Relation to
Wound Categories
CleanContamiClean Contaminated nated Dirty
(%)
(%)
(%)
(%) P-Valuew
2.0
4.9

Infection/inammation

II
(%)

III
(%)

IV
(%)

V
(%)

P-Valuew

1.5
4.4

2.9
5.6

3.4
7.9

5.6
8.3

0.0
0.0

.043
.050

ASA, anesthesiologists preoperative assessment.

For definitions, see Materials and Methods.

wP-value obtained with Armitage test for trend using the categories
IV (since only one case had a score of V, the categories IV and V were
pooled together).

these variables. Animal species was forced into the model

(kept in the model at all stages of the procedure despite
lack of statistical significance) because it was considered
an important factor with strong inuences on the characteristics of the surgical interventions. The ASA scores
were also forced into the model because in human medicine they are known to be a major risk indicator for
development of SSI. The variables remaining at the end
of the backward elimination procedure and the corresponding odds ratios for the nal model are listed in
Table 7. Ninety-seven percent of the cases were correctly
classied by this model and the HosmerLemeshow statistic was 3.76 (P .88), thus indicating a good t of the
model to the actual observations.
A 2nd multivariate model was built for the outcome
infection/inammation using the starting variables listed in Table 1 and the 778 interventions with complete
data sets for these variables. As for the 1st model, the
animal species and ASA scores were forced into the
model. Despite their lack of significance at the univariate
level, antimicrobial prophylaxis and personnel were also
included as starting variables for the backward elimination procedure, because of their high significance in the

Table 7. Variables Associated with Surgical Site Infections in a

Logistic Regression Model

Multivariate Analysis

Infection
Infection/inammation

Infection

I
(%)

3.5
4.5

4.6
9.1

For definition, see Materials and Methods.

wP-value obtained with Armitage test for trend.

6.7
17.8

.161
.002

Predictor
Duration of surgical intervention
(minutes)
Number of persons present
during intervention
Dirty wound
Antimicrobial prophylaxis

Odds Ratio
(95% CI)

P-Value

1.01 (1.001.02)

.001

1.30 (1.031.64)

.026

5.56 (1.3922.24
0.15 (0.030.88)

.015
.035

The variables animal species and anesthesiologists preoperative assessment scores were forced into the model but remained non-significant
at the end of the elimination procedure and are not shown in this table.
For each additional minute of surgical intervention, the animal is
1.01 times more likely to have an infection and for each additional
person in the operating room, the animal is 1.3 times more likely to have
an infection.

EUGSTER ET AL.
Table 8. Variables Associated with Postoperative Surgical site Infection/Inammation in a Logistic Regression Model
Odds Ratio
(95% CI)

Predictor
Dirty wound
Antimicrobial prophylaxis
Duration of anesthesia (minutes)
Duration of postoperative
stay in ICU (days)
Presence of drain
Weight of animal (kg)

P-Value

3.56 (1.279.98)
0.26 (0.090.82)
1.004 (1.001.01)
1.16 (1.031.30)

.016
.021
.067
.011

2.64 (1.056.65)
1.03 (1.011.05)

.039
.014

The variables animal species and anesthesiologists preoperative assessment scores were forced into the model but remained non-significant
at the end of the elimination procedure and are not shown in this table.
For each additional minute of anesthesia, the animal is 1.004 times
more likely to have an infection/inammation, for each additional day
spent in the intensive care unit, the animal is 1.16 times more likely to have
an infection/inammation, and for each additional kilogram of weight,
the animal is 1.03 more likely to have an infection/inammation.

1st model. The remaining variables and the corresponding odds ratios for the nal model at the end of the procedure are summarized in Table 8. The Hosmer
Lemeshow statistic for this model was 9.64 (P .29)
and is indicative of a not completely satisfactory t of the
model to the actual observations; however, 94% of the
cases were correctly classied by this model.
DISCUSSION
Infection Rate
The 5.8% rate for surgical sites identied as infected/
inamed, we found, is similar to previous reports.6,7,9,10
Using a case definition including purulent discharge and
signs of local inammation, Vasseur et al.10 reported an
infection rate of 5.1% in a retrospective study. If we
consider only surgical sites with an infection, our infection rate was 3.0%, which is lower than 5.5% reported
in a prospective study by Brown et al.7 using a similar
definition (wounds with purulent discharge). Similarly,
the infection rate observed here for clean-contaminated
wounds is lower than reported by Nicholson et al.9 These
discrepancies are probably indicative of different patient
populations and/or of different inclusion criteria, or of a
possible bias introduced by patients with incomplete
records in our study.
Risk and Protection Factors for SSIs
To control for confounding effects between the numerous variables studied at the univariate level, multivariate logistic regression models were built for the
2 outcomes infection and obvious infection. Using
this comprehensive approach, important changes were

547

observed in the predictive value of different potential risk

and protective factors. ASA scores were forced into these
models, because they represent definite predictive indicators for SSI in humans and are part of a relatively accurate risk stratication index system for SSI in human
surgery across a broad range of surgical procedures.1,3,22
Moreover, results from our univariate analysis showed
that an increasing ASA score was significantly related to
a higher risk for both infection and infection/inammation. Four major predictive variables for SSI were
identied in our logistic regression model when using the
definition of infection as an outcome.
Our results concur with studies of human SSI, where
duration of intervention represents an important risk
factor.1,3,4,22 In our model risk of infection increased
1.01 times for each additional minute of surgery, which
corresponds to doubling the risk of infection approximately every 70 minutes. Relatively similar results were
reported in another prospective veterinary study where
the risk for postoperative wound infection was estimated
to double if the procedure took 90 minutes compared
with a surgical time of 60 minutes.7 However, another
report did not nd any significant association between
development of an infection and duration of the intervention.11 The lower sample size (112 patients) compared
with our study (712) and Brown et al. (1255) may have
contributed to this lack of significance. Thus, duration of
surgical intervention must be regarded as an important
predictive factor for SSI not only in human, but also in
small animal surgery.
For each additional person in the surgical suite, the
risk of SSI was 1.3 times higher. Reports on this factor
are sparse, and it is not explicitly reported as a major risk
factor for SSI. To our knowledge, no other veterinary
study has examined this variable. It is possible that the
number of persons in the operating room could be an
indicator of another risk factor, although except for
ASA-scores, no correlation with other variables recorded
was observed. Because ASA-scores were forced into the
multivariate models, their effect was accounted for and
would not be expected to be additionally reected again
in the number of persons in the operating room. In addition, no clear trend was visible between the type of
interventions associated with large or small numbers of
persons in the room, and the effect of surgery complexity
is partially accounted for in our nal model by the duration of the operation. Finally, the consequences of an
increasing number of persons on contamination of the
wound in a clean surgical suite with numerous aseptic
manipulations are quite obvious.4,23 Therefore, a limitation on personnel in the operating room is still a valid
recommendation for a reduction of SSI frequency.4
The presence of a surgical site classied as dirty was
a highly predictive risk factor for infection as well as

548

SURGICAL SITE INFECTIONS IN DOGS AND CATS

for infection/inammation in our multivariate analyses. No association between postoperative SSI and the
other wound classes was identied with the multivariate
regression model. Nevertheless, univariate analysis demonstrated an increasing surgical site class as a significant
risk factor for infection/inammation and showed a
clear but not significant trend for infection. Our results
are in full agreement with previous human and veterinary
data with regard to this risk factor,2,10,12 but, as previously discussed, risk stratication based on wound categories alone remains a moderately predictive factor for
SSI because of its failure to account for intrinsic patient
risk.3, 7,10,24
Despite the relatively small number of patients not
administered antimicrobials in our study, the multivariate
statistical models for infection as well as for infection/
inammation showed that pre- and intraoperative antimicrobial prophylaxis represents an important protective factor against SSI. Patients administered
antimicrobial prophylaxis were on average 67 times less
likely to develop SSI than patients without prophylaxis.
Administration of prophylactic antibiotics seems routine
for clean-contaminated procedures and for most clean
interventions in human surgery.14,18 The objective of
prophylactic treatment is to reduce intraoperative bacterial contamination below the critical level needed to induce an infection and is not intended to prevent
postoperative contamination.
In their guidelines for prevention of SSI, Mangram
et al.14 recommend the use of pre- and intraoperative
antimicrobial prophylaxis for all clean-contaminated procedures and for clean procedures when SSI would be a
major threat for the patient, as well as for those procedures where a prosthetic joint or any intravascular prosthetic material will be inserted. Few studies on use of
antimicrobial prophylaxis for clean procedures have been
reported in veterinary medicine. Vasseur et al.10 reported
a significantly lower infection rate if prophylactic antibiotics were administrated and surgical duration exceeded
90 minutes, but not for a shorter surgical time. For elective orthopedic surgery, Whittem et al.11 also reported a
significantly higher infection rate in dogs administered
saline solution rather than prophylactic antibiotics 30
minutes before surgery. In another prospective report
with a standardized protocol for perioperative administration of antimicrobial agents, no significant association
between administration of antimicrobial prophylaxis and
the SSI rate was identied.7 However, as mentioned
above, the populations under investigation in these studies may differ substantially, in particular with respect to
patients with clean surgical sites.
The association of dirty surgical sites and antimicrobial prophylaxis with the outcome infection/inammation in our 2nd logistic regression model were slightly

weaker than in the 1st model. This reects the fact that
the case definition for infection/inammation was
probably less specific than the case definition for infection. Duration of anesthesia rather than duration of
surgery was significantly associated with the outcome in
the 2nd model. Because both variables are clearly correlated (Pearsons correlation coefcient 0.86), it may be
difcult to decipher the exact role of each factor. Our
results related to duration of anesthesia are in close
agreement with those of Beal et al.5 and coworkers where
there was a 30% increase in risk of SSI in clean wounds
for each additional hour of anesthesia. There are some
obvious biological reasons for the role of a prolonged
anesthesia as a predisposing factor for SSI. These include
perioperative hypothermia leading to impairment of
phagocytic function, and hypotension or hypoxia, which
may result in reduced perfusion and oxygenation of
wound tissue with similar consequences and diminished
tissue regeneration.14,25,26
In our study, 3.1% of the patients hospitalized in ICU
developed an infection and 6.1% developed an infection/inammation. From the 2nd logistic regression
model for each additional day spent in ICU, patients
were 1.16 times more likely to develop an infection/inammation. In human hospitals  20% of nosocomial
infections occur in ICUs, and surgical ICU patients are at
the greatest risk.27 One study yielded an SSI rate of 1% of
admissions in ICU patients whereas in an earlier study
SSI rate was 16.6%, both higher than the overall human
SSI rate (25%).27,28 Because of the debilitation of ICU
patients and increased contamination pressure by often
multiresistant microorganisms through invasive devices,
nosocomial infections are a general problem in ICUs.
Such problems are already present in veterinary ICUs
and may have to be controlled more tightly in veterinary
medicine also.29,30 Use of a drain is associated with an
increased risk of wound infection because foreign materials reduce the number of microorganisms required for
infection by 10,000-fold.14,31,32 However, drains are also a
source of local inammation, which may have been classied as infection/inammation in our study. Therefore, any association of drain and infection identied here
has to be interpreted with caution and may be less strong
than suggested by an odds ratio of 2.6.
Heavier animals were seemingly significantly more at
risk for infection than smaller animals. Previous human
and veterinary studies have reported that morbid obesity
and malnutrition predispose to SSI.1,4,10,14 However, the
nutritional status of our patients was not associated with
infection and does not account for an association between weight and infection/inammation. Possible differences in behavior between large and small breeds or
more frequent inammation signs in large breeds (misinterpreted as infection) because of a more intense

EUGSTER ET AL.

mechanical stress on surgical wounds in these heavy,

stronger animals may be responsible for the observed association between weight and infection/inammation.
Some variables significantly associated with the development of infection in the univariate analysis were the
increasing duration of postoperative hospital stay and
shortening of the interval between surgery and suture removal. These factors are clear indicators of the increased
care needed by patients with SSI and of the associated
costs.
Our results allow a 1st estimation of the SSI rate in our
companion animal hospital and are broadly comparable
with data reported from other veterinary hospitals. We
were able to dene some highly predictive factors for SSI
in dogs and cats. The logistic regression models we developed may represent a useful tool for surgeons in making decisions on the necessity of additional preventive
measures in patients at a high risk for developing postoperative SSI. Surveillance programs should be implemented to follow the evolution of nosocomial infections
in veterinary medicine. Our methodology using chart reviews and postdischarge surveillance has been shown as
one of the most effective methods for detection of SSI in
human hospitals.33 However, it is very tedious, labor intensive, and difcult to implement for routine surveillance, particularly in veterinary medicine. Unfortunately,
surveillance based on microbial analysis submissions is
much less sensitive. Thus, an approach between effective
surveillance using simplied charts combined with elective microbial analysis specifically adapted to veterinary
clinics still needs to be found. Our study represents an
additional step in this direction.

5.

6.

7.

8.

9.

10.
11.

12.
13.

14.

15.

ACKNOWLEDGMENTS
The authors wish to thank all clinicians and technicians of
the Companion Animal Hospital, who patiently completed
the data record sheets during the study. We are also very
thankful to S. McEwen and R. Reid-Smith for reading the
manuscript and for their helpful suggestions.

16.

17.
18.

REFERENCES
1. Kluytmans J: Surgical infections including burns, in Wenzel
RP (ed): Prevention and Control of Nosocomial Infections.
Baltimore, MD, Williams & Wilkins, 1997, pp 840865
2. Cruse PJE, Foord R: The epidemiology of wound infection: a
10-year-prospective study of 62,939 wounds. Surg Clin
North Am 60:2740, 1980
3. Culver DH, Horan TC, Gaynes RP, et al: Surgical wound
infection rates by wound class, operative procedure, and
patient risk index. Am J Med 91(Suppl 3B): 152S157S,
1991
4. Wong ES: Surgical site infections, in Mayhall CG (ed): Hospital Epidemiology and Infection Control of Nosocomial

19.

20.

21.

549

Infections. Baltimore, MD, Lippincott, Williams & Wilkins, 1996, pp 155173

Beal MW, Brown DC, Shofer FS: The effects of perioperative
hypothermia and the duration of anesthesia on postoperative wound infection rate in clean wounds: a retrospective
study. Vet Surg 29:123127, 2000
Billings L, Vasseur PB, Fancher C, et al: Wound infection
rates in dogs and cats after use of cotton muslin or disposable impermeable fabric as barrier material: 720 cases
(19831989). J Am Vet Med Assoc 197:889892, 1990
Brown DC, Conzemius G, Shofer F, et al: Epidemiologic
evaluation of postoperative wound infections in dogs and
cats. J Am Vet Med Assoc 210:13021306, 1997
Heldmann E, Brown DC, Shofer F: The association of propofol usage with postoperative wound infection rate in clean
wounds: a restrospective study. Vet Surg 28:256259, 1999
Nicholson M, Beal M, Shofer F, et al: Epidemiologic evaluation of postoperative wound infection in clean-contaminated wounds: a retrospective study of 239 dogs and cats.
Vet Surg 31:577581, 2002
Vasseur PB, Levy J, Dowd E, et al: Surgical wound infection
rates in dogs and cats. Vet Surg 17:6064, 1988
Whittem TL, Johnson AL, Smith CHW, et al: Effect of perioperative prophylactic antimicrobial treatment in dogs undergoing elective orthopedic surgery. J Am Vet Med Assoc
215:212216, 1999
Cruse PJE, Foord R: A ve-year prospective study of 23,649
wounds. Arch Surg 107:206210, 1973
De Boer AS, Mintjes-de Groot AJ, Severijnen AJ, et al: Risk
assessment for surgical site infections in orthopedic patients. Inf Control Hosp Epidemiol 20:402407, 1999
Mangram AJ, Horan TC, Pearson ML, et al: Guideline for
prevention of surgical site infection 1999. Inf Control Hosp
Epidemiol 20:247278, 1999
Horan TC, Gaynes RP, Martone WJ, et al: CDC definitions
of nosocomial surgical site infections, 1992: a modication
of CDC definitions of surgical wound infections. Inf Control Hosp Epidemiol 13:606608, 1992
Greene CE: Surgical and traumatic wound infections, in
Greene CE (ed): Infectious Diseases of the Dog and Cat (ed
2). Philadelphia, PA, Saunders, 1998, pp 343347
Romatowski J: Prevention and control of surgical wound infection. J Am Vet Med Assoc 194:107114, 1989
Trilla A, Mensa J: Preoperative antibiotic prophylaxis, in
Wenzel RP (ed): Prevention and Control of Nosocomial
Infections. Baltimore, MD, Williams & Wilkins, 1997, pp
867881
Roizen MF: Routine perioperative evaluation, in Miller RD
(ed): Anesthesia (ed 2). New York, Churchill Livingstone,
1986, pp 225254
Ad hoc committee of the Committee on Trauma, Division of
Medical Sciences. National Research Council: postoperative wound infections: the inuence of ultraviolett irradication of the operating room and other factors. Ann Surg
160(Suppl): 23, 1964
Walter SD, Feinstein AR, Wells CK: Coding ordinal independent variables in multiple regression analyses. Am J
Epidemiol 125:319323, 1987

550

SURGICAL SITE INFECTIONS IN DOGS AND CATS

22. Garibaldi RA, Cushing D, Lerer T: Risk factors for postoperative infection. Am J Med 91(Suppl 3B): 158S163S, 1991
23. Alexander JW, Dellinger EP: Surgical infections and choice
of antibiotics, in Sabiston DC (ed): Textbook of Surgery
(ed 14). Philadelphia, PA, Saunders, 1991, pp 221236
24. Ehrenkranz NJ: Surgical wound infection occurrence in clean
operations, risk stratication for interhospital comparisons. Am J Med 70:909914, 1981
25. Greif R, Akca O, Horn EP, et al: Supplemental perioperative
oxygen to reduce the incidence of surgical wound infection.
N Engl J Med 342:161166, 2000
26. Van den Bogaard AEJM, Weidema WF: Antimicrobial
prophylaxis in veterinary surgery. J Am Vet Med Assoc
186:990992, 1985
27. Wallace WC, Cinat M, Gornick WB, et al: Nosocomial infections in the surgical intensive care uni: a difference between
trauma and surgical patients. Am Surg 65:987990, 1999
28. Hartenauer U, Diemer W, Gaehler R, et al: Nosokomiale
Infektionen in der operativen Intensivmedizin: Ergebnisse

29.

30.

31.

32.

33.

einer 5-jaehrigen prospektiven Erhebung. Anaesth Intensivther Notfallmed 25:93101, 1990

Boerlin P, Eugster S, Gaschen F, et al: Transmission of opportunistic pathogens in a veterinary teaching hospital. Vet
Microbiol 82:347359, 2001
Francey T, Gaschen F, Nicolet J, et al: The role of
Acinetobacter baumannii as a nosocomial pathogen for
dogs and cats in an intensive care unit. J Vet Intern Med
14:177183, 2000
Rosin E, Dow SW, Daly WR, et al: Surgical wound infection
and use of antibiotics, in Slatter DH (ed): Textbook of
Small Animal Surgery (ed 2). Philadelphia, PA, Saunders,
1993, pp 8495
Wendelburg K: Surgical wound infection, in Bojrab M (ed):
Disease Mechanisms in Small Animal Surgery (ed 2). Philadelphia, Lea & Febiger, 1993, pp 5465
Roy MC, Perl TM: Basics of surgical site infection surveillance. Infect Control Hosp Epidemiol 18:659668,
1997