Trial di Controllo Randomizzato sull' uso della gentamicina topica nella riduzione delle infezioni sternali dopo l'intervento di cardiochirurgico

1. Local Gentamicin Reduces Sternal Wound
Infections After Cardiac Surgery: A Randomized
Controlled Trial
Örjan Friberg, MD, Rolf Svedjeholm, MD, PhD, Bo Söderquist, MD, PhD,
Hans Granfeldt, MD, Tomas Vikerfors, MD, PhD, and Jan Källman, MD, PhD
Departments of Cardiothoracic Surgery, Clinical Microbiology, and Infectious Diseases, Örebro University Hospital, Örebro,
Department of Cardiothoracic Surgery, Linköping University Hospital, Linköping, Sweden
Background. Sternal wound infections remain a major
cause of morbidity after cardiac surgery. Vancomycin is
often the only effective antibiotic available for their
treatment but its use for routine prophylaxis is inadvis-
able for ecological reasons. Local application of gentami-
cin produces high antibiotic concentrations in the
wound. We aimed to determine whether this treatment
could have an additive effect on the incidence of sternal
wound infections when combined with routine
prophylaxis.
Methods. Two thousand cardiac surgery patients were
randomized to routine prophylaxis with intravenous
isoxazolyl-penicillin alone (control group) or to this pro-
phylaxis combined with application of collagen-gentam-
icin (260 mg gentamicin) sponges within the sternotomy
before wound closure. Endpoint was any sternal wound
infection within 2 months postoperatively. Evaluations
were double-blind and made on an intention-to-treat
basis.
Results. Evaluation was possible in 967 and 983 pa-
tients in the control and treatment groups, respectively.
The incidence of sternal wound infection was 4.3% in the
treatment group and 9.0% in the control group (relative
risk 0.47; 95% confidence interval 0.33–0.68; p < 0.001).
Early reoperation for bleeding was more common in the
treatment group (4.0% vs 2.3%, p ⴝ 0.03). No difference in
postoperative renal function was noted.
Conclusions. Local collagen-gentamicin reduced the
risk for postoperative sternal wound infections. Further
studies are warranted to confirm these results, particu-
larly with regard to deep infections.
(Ann Thorac Surg 2005;79:153–62)
© 2005 by The Society of Thoracic Surgeons
In cardiac surgery, sternal wound infection (SWI) con-
tinues to be one of the most serious postoperative
complications [1–4]. Coagulase-negative staphylococci
(CoNS) have become the most common causative agents
of SWI, followed by Staphylococcus aureus and others,
such as gram-negative rods [2, 5–7]. Due to the shift
towards CoNS infections, the presentation is often de-
layed and more insidious, while treatment has become
more demanding often requiring radical surgical revision
with or without the use of muscle or omental flaps to
achieve cure [1].
Prophylaxis with intravenous (IV) ␤-lactam antibiotics
(cephalosporins or in Sweden most commonly isox-
azolyl-penicillins), during surgery and with continuation
for up to 48 hours postoperatively, is routinely practiced
[8, 9]. However, most CoNS are methicillin resistant, and
are thus resistant to all ␤-lactam antibiotics due to the
presence of the mecA gene [10]. Therefore, vancomycin is
often the only effective antibiotic available for treatment
of sternal infections caused by CoNS, but its use for
routine prophylaxis is strongly discouraged because of
the risk of selection of resistant bacteria with consequent
ecological and therapeutic consequences [11]. Also, van-
comycin is intrinsically a less active antistaphylococcal
agent than are ␤-lactam antibiotics [10]. Hence, alterna-
tive approaches to prevent postoperative sternal infec-
tions need to be explored.
With local application of gentamicin, bound to bovine
collagen in padlike sponges (Collatamp-G; Schering-
Plough, Stockholm, Sweden), there is evidence that the
concentration of gentamicin in the mediastinal fluid
reaches levels high enough to be effective against bacte-
ria that are normally considered resistant (including most
CoNS) [12]. The aim of this study was to determine
whether locally administered gentamicin in addition to
routine IV prophylaxis with isoxazolyl-penicillin could
reduce the total incidence of SWI after cardiac surgery.
Patients and Methods
Patients
The study was conducted in two cardiothoracic centers in
Sweden during the period September 2000 to September
2002. The centers are single referral centers and the sole
providers of cardiac surgery in their catchment regions of
about 0.6 and 1.0 million inhabitants, respectively.
Accepted for publication June 11, 2004.
Address reprint requests to Dr Friberg, Department of Cardiothoracic
Surgery, Örebro University Hospital, SE-701 85 Örebro, Sweden; e-mail:
orjan.friberg@orebroll.se.
© 2005 by The Society of Thoracic Surgeons 0003-4975/05/$30.00
Published by Elsevier Inc doi:10.1016/j.athoracsur.2004.06.043
CARDIOVASCULAR

2. Eligible for inclusion in the study were all patients
undergoing cardiac surgery through median sternotomy,
including operations on the ascending aorta. Exclusion
criteria were: known allergy to gentamicin; pregnancy or
breastfeeding; treatment with aminoglycosides during
the last 2 weeks before surgery; and expected difficulty in
fulfilling the follow-up requirements, for linguistic or
other reasons.
The trial was approved by the ethics committees at
both participating centers, and written informed consent
was obtained from all patients. In accordance with the
ethics committees’ approval, emergency patients were
included and informed immediately after surgery. If they
did not agree to participate they were excluded from
further evaluation.
Design and Treatment
The study was randomized, prospective, and double-
blind (ie, the patients and those physicians and nurses
who collected all the postoperative data and classified the
infections were blinded regarding what treatment the
patients had received).
An external statistician who was not involved in the
enrolment or assessment of the patients performed the
randomization. The allocation sequence was computer-
generated with a variable block size of 8, 12, or 16 patients
and was stratified for center. Assignments were con-
cealed in sealed envelopes. At the end of the operation,
immediately before sternal closure, the envelope was
opened and the patient was allocated to a treatment
group (routine IV prophylaxis combined with sternal
application of collagen-gentamicin; see below) or a con-
trol group (IV prophylaxis alone). No documentation
concerning the group to which the patient had been
randomized was left in the patient’s file.
Collatamp-G consists of a flat absorbable bovine col-
lagen sponge with gentamicin sulfate. A 10 ⫻ 10 ⫻ 0.5-cm
sponge contains 280 mg collagen and 130 mg gentamicin
(200 mg gentamicin sulfate). The treatment group re-
ceived two such sponges in the wound immediately
before closure. The sponges were cut into appropriate
sizes and put between the sternal halves. More than two
layers of Collatamp-G were avoided so as not to compro-
mise sternal healing and stability, and any left-over
sponge were put behind the sternum at the proximal or
distal end.
In the control group the wound was closed in a
conventional way. The primary end-point was any SWI
within 2 months postoperatively.
Surgical Technique
A policy of strict hygiene routines to prevent infections
has been adopted at both centers including avoidance of
unnecessary personnel rotation in the operating rooms.
One center had laminar air flow ventilation in the oper-
ating rooms, the other had upward displacement venti-
lation. Most, but not all, surgeons used double gloving.
The operations were performed with a standard tech-
nique. Cardiopulmonary bypass (CPB) was used in all
patients except in 9% of the coronary bypass (CABG)
operations, which were carried out off pump (Table 1).
Internal mammary arteries were harvested with a pedicle
(not skeletonized). The sternum was sutured with 6 to 8
sternal wires, and the fascia, subcutis and skin were
closed with running, monofilament, absorbable sutures.
The surgeon who performed the heart operation, usually
a senior surgeon, also closed the wound. Antibiotic
prophylaxis was given IV, in accordance with the proto-
col of each center, starting immediately before the skin
incision and then with 1 to 2 repetitions during surgery,
with cloxacillin at one center and dicloxacillin at the
other, at the doses 2 g and 1 g, respectively. This was
continued every 8th hour for 24 to 48 hours postopera-
tively, depending on whether the drainage tubes had
been removed. Patients with penicillin intolerance re-
ceived clindamycin 600 mg at similar intervals. Preoper-
ative topical nasal antibiotics were not used. Preoperative
oral mouthwash twice, the night and the morning before
surgery, with chlorhexidine solution was implied at one
of the centers (Örebro).
An approach towards early reoperation at clinical signs
of deep infection was applied at both centers.
Follow-Up and Data Collection
During the postoperative stay in the surgical department,
all reoperations and SWIs were recorded. Serum creati-
nine was measured preoperatively, on postoperative
days 1 and 3, and then on clinical indications. All major
complications during the postoperative stay were
recorded.
After discharge the patients were contacted by tele-
phone 2 months postoperatively and were asked to
answer a structured list of standardized questions. Those
who could not be reached by phone were sent a letter
with a questionnaire containing the same questions. Any
reported symptom of impaired wound healing or possi-
ble wound infection resulted in a check of medical
records from all postoperative contacts with medical
services. Data on clinical status, bacterial cultures, labo-
ratory findings, and surgical or antibiotic treatment were
recorded.
Four infections diagnosed after more than 2 months
were discovered (60 to 73 days). These were also included
in the analysis because the randomization had not been
broken. Mortality data were double-checked against the
national Swedish population register.
Classification of Infections
All classifications were made by the same person, a
senior surgeon (Ö.F.), before the randomization code was
broken. A second opinion was obtained from another
surgeon or from a specialist in infectious diseases in
doubtful cases.
Criteria for definition and classification of surgical site
infections according to Centers of Disease Control and
Prevention [11] were used with minor modifications
(Table 2). Depths 1 and 2 were considered as superficial
infections, and depths 3 and 4 were considered as deep
infections.
Strict definition and classification of an infection some-
154 FRIBERG ET AL Ann Thorac Surg
LOCAL GENTAMICIN AND STERNAL WOUND INFECTIONS 2005;79:153–62
CARDIOVASCULAR

3. Table 1. Baseline Data
Control
(n ⫽ 967)
Treatment
(n ⫽ 983)
All Other Operations
During Study Perioda
(n ⫽ 792)
Number of patients in each center (%)
Linkoping University Hospital 464 (48.0) 474 (48.2) 429 (54.2)
Orebro University Hospital 503 (52.0) 509 (51.8) 363 (45.8)
Sex
Number of male (%) 735 (76.0) 753 (76.6) 525 (66.3)
Age in years, median (range) 68 (20–87) 68 (25–87) 70 (16–88)
Risk factors
BMI median (range) 26.6 (14.8–46.1) 26.3 (15.6–42.8)
Serum creatinine (␮mol/l) median (range) 94 (50–1084) 96 (45–801)
Preop dialysis, number (%) 9 (0.9) 5 (0.5)
Diabetes, insulin/drug treated, number (%) 174 (18.0) 180 (18.3) 154 (19.4)
COPD, number (%) 58 (6.0) 52 (5.3) 55 (6.9)
Left ventricular ejection fraction ⬍35%, number (%) 65 (6.7) 50 (5.1) 37 (4.7)
Preop medication: number (%)
Oral steroid treatment 47 (4.9) 22 (2.2)
Immunosuppression 12 (1.2) 9 (0.9)
Preop anticoagulation, any 570 (58.9) 622 (63.3)
Dalteparin or heparin (within 24 h) 338 (35.0) 347 (35.3)
Warfarin (within 3 days) 9 (0.9) 9 (0.9)
Clopidogrel (within 3 days) 11 (1.1) 9 (0.9)
Aspirin (within 7 days) 465 (48.1) 536 (54.5)
Dipyridamole 6 (0.9) 3 (0.3)
Preop hospital stay, days, median (range) 1 (0–40) 1 (0–41)
Emergency operation, number (%) 23 (2.4) 23 (2.3) 97 (12.2)
Type of surgery, number (%)
CABG 698 (72.2) 732 (74.5) 555 (70.1)
Off pump (% of CABG) 63 (9.0) 65 (8.9) 51 (9.2)
Valve 135 (14.0) 131 (13.3) 108 (13.6)
Aortic aneurysm 18 (1.9) 11 (1.1) 48 (6.0)
Congenital malformation 5 (0.5) 6 (0.6) 4 (0.5)
CABG ⫹ other (mostly valve) 100 (10.3) 95 (9.7) 66 (8.3)
Other 11 (1.1) 8 (0.8) 11 (1.4)
Use of mammary artery, number (%)
LIMA only
all operations 679 (70.2) 701 (71.3)
CABG(% of CABG) 615 (88.1) 641 (87.6)
RIMA only
all operations 4 (0.4) 6 (0.6)
CABG(% of CABG) 3 (0.4) 5 (0.6)
Bilateral IMA
all operations 10 (1.0) 9 (0.9)
CABG(% of CABG) 9 (1.3) 9 (1.2)
Times, minutes
Operation, duration median (range) 195 (65–632) 195 (60–778)
CPB, duration median (range) 88 (19–365) 87 (30–480) 91 (22–619)
a
Consists of 742 patients never included and the 50 patients excluded from the study; only data retrievable from the centres’ clinical databases are shown
for these.
BMI ⫽ body mass index (kg/m2
); CABG ⫽ coronary artery bypass grafting; COPD ⫽ chronic obstructive pulmonary disease; CPB ⫽
cardiopulmonary bypass; LIMA/RIMA ⫽ left/right internal mammary artery.
155
Ann Thorac Surg FRIBERG ET AL
2005;79:153–62 LOCAL GENTAMICIN AND STERNAL WOUND INFECTIONS
CARDIOVASCULAR

4. times proved difficult. Therefore, the infections were
classified as “definite” or “probable.” Infections classified
as definite were those with typical symptoms and signs
and a positive culture. All probable and definite infec-
tions were considered as “infection” in the following
calculations unless otherwise stated.
Five samples for bacterial culture, of which at least two
were tissue samples, were taken intraoperatively at re-
operations for deep infection. To be classified as definite
mediastinitis or sternitis (depth 4), two of these five
cultures had to be positive with the same bacterial
isolate. In infected patients handled at referring hospi-
tals, cultures were taken on clinical indications alone.
Statistical Analysis
The sample size was chosen so as to detect an approxi-
mate 50% reduction in the incidence of any SWI, and the
incidence in the control group was assumed to be at least
4%. One thousand eighteen patients in each group pro-
vide a power of 80% for distinguishing between infection
rates of 4% and 1.9% in the two groups (two-sided type I
error of 5%). Therefore the 1000 patients in each group
were considered to be appropriate because the incidence
of 4% in the control group was assumed to be somewhat
underestimated.
Subgroup analyses with regard to bacterial etiology,
depth of infection and established preoperative risk
groups (diabetes and high body mass index [BMI]) [13–
20] were conducted. The sample size calculations did not
account for these subgroup analyses but since the true
incidence proved higher than assumed the analyses were
considered justifiable.
All analyses were made on an intention-to-treat basis,
ie, patients who died and those reoperated on for bleed-
ing or other reasons during the follow-up period were
included in the analysis. An interim analysis was per-
formed by an external statistician after completion of
follow-up of 1000 patients.
For categorical variables the relative risk with 95%
confidence intervals was calculated and two-sided P
values were calculated with Pearson’s ⌾2
test or, if
expected frequencies fell below 5, Fisher’s exact test. For
continuous variables the Mann-Whitney rank sum test
was used. For the statistical analyses, Statistica 6.1 (Stat-
Soft Inc, Tulsa, OK), SPSS 11.5 (SPSS Inc, Chicago, IL)
software, and Microsoft Excel (Microsoft, Redmond, WA)
were used.
Role of the Funding Source
The sponsors had no role in the study design, in the
collection, analysis and interpretation of data, in the
writing of the report, or in the decision to submit the
paper for publication.
Results
Patients
During the study period from October 2000 to August
2002, a total of 2742 patients underwent cardiac surgery at
the two centers. Two thousand eligible patients were
identified and preliminarily included (Fig 1). Of these, 8
patients were excluded at the end of the operation for
medical reasons (eg, sternotomy could not be primarily
closed because of hemodynamic instability or the ster-
num was reopened immediately for other reasons) and 1
was discovered to have been erroneously included (met
exclusion criteria). Thirty-five patients whose operation
was an emergency declined further participation. Thus,
1956 patients entered the study. Six of these could not be
reached at the follow-up.
Fig 1. Trial profile.
Table 2. Classification of Infections
Depth Localization Description
1 Cutis e.g., infected crusts
Superficial
2 Subcutis Involving subcutaneous tissue but not reaching down to
sternal fixation wires
3 Presternal Infections reaching below the superficial fascia, involving
sternal wires.
Deep
4 Sternal bone or mediastinum Unstable sternal fixation with signs of osteomyelitis or positive
bacterial cultures from mediastinum or mediastinal abscess.
156 FRIBERG ET AL Ann Thorac Surg
LOCAL GENTAMICIN AND STERNAL WOUND INFECTIONS 2005;79:153–62
CARDIOVASCULAR

5. Table 1 shows baseline data of the remaining 1950
patients and of all other patients operated on during the
study period (including those excluded during the trial
process).
Outcome
The total incidence of SWI (primary end-point) was 9.0%
in the control group and 4.3% in the treatment group (p ⬍
0.001). There was a reduction in all major microbiological
types of infections (Table 3, Fig 2). Significantly fewer
patients in the treatment group received antibiotics or
needed surgical revision of the sternal wound during the
follow-up.
There was a risk reduction in all depths that was
statistically significant for superficial infections (Table 3)
and for deep infections in high-risk groups (diabetes,
BMI ⬎ 25; Table 4).
Twenty-four of 129 SWIs (17 in the control and 7 in the
treatment group; p ⫽ 0.036) were diagnosed before dis-
charge from cardiothoracic center. The median time from
operation to diagnosis of SWI was 20 days (range 3 to 73
days) in the treatment group and 15 days (range 3 to 62
days) in the control group.
All patients, except one (belonging to the control
group), with symptoms of sternal dehiscense that re-
quired refixation were classified as having a SWI.
General Outcome
There were no significant differences in mortality, post-
operative rise in creatinine or need for dialysis between
the groups (Table 3). The total 60-day mortality of the 129
patients with SWI was 1.5% (2 patients with deep SWI,
both in the treatment group).
There were significantly more reoperations for bleed-
ing in the treatment group (4.0% vs 2.3%). Data retrieved
from the clinical databases of the centers showed that
3.5% of the 792 patients not included in the study had had
reoperations for bleeding. Four of 39 patients reoperated
on for bleeding in the treatment group and 1 of 22 such
patients in the control group had postoperative SWI.
The reduction of infections was similar when excluding
patients who were reoperated for bleeding (n ⫽ 61),
Table 3. Outcome
Outcome
Control
(n ⫽ 967)
no. (%)
Treatment
(n ⫽ 983)
no. (%) RR (95% CI) p Value
Primary end-point
All SWIs 87 (9.0) 42 (4.3) 0.47 (0.33 to 0.68) ⬍0.001
Treatment
Antibiotic treatment 174 (18.0) 114 (11.6) 0.64 (0.52 to 0.80) ⬍0.001
Surgically treated SWIa
38 (3.9) 21 (2.1) 0.54 (0.32 to 0.92) 0.021
Bacterial etiology
Staphylococcus aureus 20 (2.1) 8 (0.8) 0.39 (0.17 to 0.89) 0.020
Coagulase-negative staphylococci 33 (3.4) 11 (1.1) 0.33 (0.17 to 0.65) ⬍0.001
Gram-negative bacteria 4 (0.5) 1 (0.1) — 0.21
Other bacteria or multiple species 15 (1.6) 10 (1.0) 0.66 (0.30 to 1.45) 0.29
Missing or negative bacterial samples 15 (1.6) 13 (1.3) 0.85 (0.41 to 1.78) 0.67
Subclassifications
Probable SWI 32 (3.3) 15 (1.5) 0.46 (0.25 to 0.84) 0.010
Definite SWI 55 (5.7) 27 (2.7) 0.48 (0.31 to 0.76) 0.001
All superficial SWI 55 (5.7) 19 (1.9) 0.34 (0.20 to 0.57) ⬍0.001
Depth 1 14 (1.4) 8 (0.8) 0.56 (0.24 to 1.33)
Depth 2 41 (4.2) 11 (1.1) 0.26 (0.14 to 0.41)
All deep SWI 32 (3.3) 23 (2.3) 0.71 (0.42 to 1.20) 0.20
Depth 3 17 (1.8) 10 (1.0) 0.58 (0.27 to 1.26)
Depth 4 15 (1.6) 13 (1.3) 0.85 (0.41 to 1.78)
General outcome
Hospital mortality 10 (1.0) 11 (1.1) 1.08 (0.46 to 2.54) 0.85
Total 60-day mortality 17 (1.8) 19 (1.9) 1.10 (0.57 to 2.10) 0.77
Early reoperation for bleeding 22 (2.3) 39 (4.0) 1.74 (1.04 to 2.91) 0.032
Postop hemodialysisb
10 (1.0) 10 (1.0) 0.98 (0.41 to 2.35) 0.97
Postop rise in serum creatinine, median ␮mol/L 5.1 7.8 0.28
Mechanical ventilation more than 48 hours 25 (2.6) 31 (3.2) 0.82 (0.49 to 1.38) 0.45
a
Any surgical revision of the wound; b
patients on preoperative hemodialysis not included.
CI ⫽ confidence interval; RR ⫽ relative risk; SWI ⫽ sternal wound infection.
157
Ann Thorac Surg FRIBERG ET AL
2005;79:153–62 LOCAL GENTAMICIN AND STERNAL WOUND INFECTIONS
CARDIOVASCULAR

6. reoperated for other reasons (n ⫽ 7), or who died within
2 months (n ⫽ 36) from the analysis (leaving 1850 pa-
tients, 927 and 923 in the control and treatment groups,
respectively). In this group there were 62 (6.7%) versus 25
(2.7%) superficial infections (p ⬍ 0.001), and 31 (3.3%)
versus 19 (2.1%) deep infections (p ⫽ 0.088) in the control
and treatment groups, respectively.
The mean lengths of postoperative stay until discharge
were 6.7 and 7.1 days in the treatment and control
groups, respectively, with a median of 6 days in both
groups.
Adverse Reactions
No case of verified or suspected allergic or other adverse
reaction to gentamicin or bovine collagen was identified.
Comment
This is the first randomized controlled trial of local
gentamicin prophylaxis in cardiac surgery. Local appli-
cation of collagen-gentamicin between the sternal halves
was associated with a 53% relative reduction (4.7% abso-
lute reduction) in the incidence of SWI after sternotomy
on an unselected cardiac surgery patient population. No
significant impairment in postoperative renal function or
other adverse effects of the treatment were noted.
A limitation of our study is the absence of a placebo
arm with collagen without gentamicin. However, the
patients and those assessing the outcome were blinded
regarding treatment. Since the proposed hemostatic ef-
fect of collagen did not reduce reoperation rate for
bleeding, it is conceivable that the reduction in SWI was
due to the gentamicin component.
Definition and classification of surgical site infections
are, despite classification systems, occasionally subjec-
tive. In this study infections were therefore classified as
definite or probable. However, the risk reductions in
these two classes were almost equal and – scientifically
most important – there were significant reductions in the
definite infections, in staphylococcal (including CoNS)
infections, in surgical revisions and in antibiotic
treatment.
The reported incidences of postoperative SWI varies
considerably, not only because of differences in defini-
tions and classification of infections but also because of
variations in follow-up between centers and notably
Fig 2. Incidence of wound infections and their
bacterial etiology. □ ⫽ control; o ⫽ treatment.
(SWI ⫽ sternal wound infections; CoNS ⫽
coagulase negative Staphylococci.)
Table 4. Incidence of Sternal Wound Infections in Risk Groups With Diabetes or Increased BMI
Risk Factor(s)
Number
(% of study
population) Depth
SWI/Total in Group (%)
RR (95% CI) p Value
Control Treatment
Diabetesa
354 (18.2) Superficial 13/174 (7.47) 3/180 (1.67) 0.22 (0.06 to 0.78) 0.0086
Deep 17/174 (9.77) 7/180 (3.89) 0.40 (0.17 to 0.94) 0.028
BMI ⬎ 25b
1299 (66.6) Superficial 42/651 (6.45) 14/648 (2.16) 0.33 (0.18 to 0.61) ⬍0.001
Deep 29/651 (4.45) 16/648 (2.47) 0.55 (0.30 to 1.01) 0.050
Diabetes or BMI ⬎ 25 1385 (71.0) Superficial 46/687 (6.70) 14/698 (2.01) 0.43 (0.24 to 0.78) ⬍0.001
Deep 30/687 (4.37) 17/698 (2.44) 0.56 (0.31 to 1.00) 0.047
a
Treated with insulin or oral medication; b
WHO definition of overweight.
BMI ⫽ body mass index (kg/m2
); CI ⫽ confidence interval; RR ⫽ relative risk; SWI ⫽ sternal wound infection.
158 FRIBERG ET AL Ann Thorac Surg
LOCAL GENTAMICIN AND STERNAL WOUND INFECTIONS 2005;79:153–62
CARDIOVASCULAR

7. because most published series are retrospective [5, 13, 16,
19, 21–23]. The 9.0% infection rate in our control group is
certainly higher than many surgeons would anticipate.
However, it should be emphasized that the majority of
the infections detected in the present study were not
deep infections, as shown in Table 3. Furthermore, in this
study, 4 of 5 infections were diagnosed after discharge.
This underlines the importance of thorough follow-up. In
a recent study with complete 90-day follow-up, the total
incidence of SWI was 9.0% [24], which is similar to that
found in our control group. In two earlier studies with
complete 2-months follow-up, assessing the ASEPSIS
wound score [25, 26], the incidence of any sternal wound
infection was 8.0% and 7.3%, respectively.
The clinical characteristics of CoNS infections also
deserve consideration. They have led to more insidious
infections that have required extensive antibiotic treat-
ment and not uncommonly delayed extensive surgical
revision, consuming substantial healthcare resources [1].
Hence, over the last decade a policy of active surveillance
and early aggressive surgical revision was developed at
both centers, which has resulted in earlier detection, less
complicated revisionary surgery, and reduced length of
antibiotic treatment [27]. We suggest that the mortality of
only 1.5% during 60 days of follow-up in patients with
SWI in the present study is in accordance with this
policy.
We applied the criteria for definition and classification
of Surgical Site Infections according to Centers of Dis-
ease Control and prevention [11]. This is a systematic and
frequently used classification, made for applying on any
surgical site infection. It is focused on the depth, but not
on the clinical severity, of the infection. An evaluated
clinical wound score system, such as the ASEPSIS score
[25], would have given additional information, but to
apply this to almost 2000 patients was beyond our
resources.
Randomized studies of prophylaxis with local collag-
en-gentamicin have reported a reduction of surgical site
infections in “dirty” abdominal surgery [28–31] and in
prosthetic repair of groin hernias [32]. There is one report
of successful use of collagen-gentamicin for treatment of
mediastinitis [33].
Intravenous aminoglycosides have been used for pro-
phylaxis in cardiac surgery, usually combined with ␤-lac-
tam antibiotics [9]. However, no significant additional
effect of the IV combination, compared to cephalosporins
alone, has been reported [34–36].
In a recent meta-analysis IV ␤-lactam antibiotics were
superior to glycopeptides, such as vancomycin, for pre-
vention of SWI after cardiac surgery, although glycopep-
tides were superior for prevention of infections caused by
methicillin-resistant gram-positive bacteria [37]. Topical
vancomycin has been shown to reduce the incidence of
sternal infection in one randomized study [38].
Local application of gentamicin produces high local
antibiotic concentrations that may have an effect on
bacteria that are normally considered to be resistant [12].
Although gentamicin is generally used for gram-negative
infections, it does have a spectrum of bactericidal activity
for many gram-positive organisms, including staphylo-
cocci [39]. Synergism between aminoglycosides and
␤-lactam antibiotics has also been shown [40]. Further
detailed analyses of the bacteriological findings, includ-
ing antibiotic susceptibility patterns, are required.
The gentamicin-collagen sponges were placed be-
tween the sternal halves. We hypothesized that this is
where it is most important to achieve high antibiotic
concentrations in order to prevent osteomyelitis. A lower
incidence of SWI was found in the treatment group in all
four depths but the study was not powered to detect
significant differences in subgroups. Even so, the reduc-
tion in superficial infections was significant and patients
with established risk factors for infection (diabetes or
increased BMI) revealed a significant reduction in deep
infections. Further studies are warranted to confirm these
results. From a clinical perspective, the significant differ-
ence in the incidence of infections that required surgical
treatment is important.
Bovine collagen is rapidly absorbed and is theoretically
an attractive vehicle for antibiotics. The product is con-
sidered safe regarding BSE and is approved for commer-
cial use in, for example, the countries within the Euro-
pean Union. We found no evidence of allergic reactions
or adverse local effects on sternal healing. On the other
hand, we found a higher incidence of early reoperation
for bleeding in the treatment group. The medical records
from the reoperation of the patients in the treatment
group provided no obvious explanation, nor was any
obvious, repeated or unusual source of the bleeding
reported. Notably, the incidence of reoperation for bleed-
ing in the treatment group was close to that seen in our
routine practice and among the patients not included in
the study. Thus, although this issue needs to be moni-
tored in the future, available data, rather than pointing at
any untoward effect, seem to suggest a lower than ex-
pected reoperation rate in the control group. Neverthe-
less, patients reoperated on for bleeding did not become
infected to an extent that significantly influenced the
overall result in an untoward direction.
A potential drawback of the local gentamicin treatment
is the risk of selecting gentamicin-resistant bacterial
clones over time. However, aminoglycosides have al-
ready been used extensively for many years. The phar-
macokinetics of local gentamicin, with an early, high
peak in the local gentamicin concentration, low serum
concentrations and then rapid disappearance of the drug
reduces the selection of resistant bacteria and may, in
this respect, even be favorable compared to normal IV
use [12].
Given the unique role of vancomycin for antibiotic
treatment of overt sternal wound infections and the
inadvisability of using it for routine prophylaxis in con-
sideration of the ecological drawbacks, local gentamicin
provides the potential advantage of reducing the total
exposure of the environment to vancomycin.
Strict blood glucose control has in recent years been
identified as a means of reducing SWI [41]. This was not
yet fully in clinical practice during the study period and
159
Ann Thorac Surg FRIBERG ET AL
2005;79:153–62 LOCAL GENTAMICIN AND STERNAL WOUND INFECTIONS
CARDIOVASCULAR

8. further studies are required to establish if local gentami-
cin may provide additional benefit in practices employ-
ing such strategies.
In conclusion, our study indicates a significant risk
reduction in postoperative SWI in cardiac surgery. Fur-
ther studies are warranted to confirm this result partic-
ularly with regard to deep infections. Such studies should
also address the question whether routine use may be
justified or whether high-risk groups that may have
particular benefit from this approach should be
identified.
The study was financed by grants from the Research Committee
of Örebro County Council and from Schering-Plough, who also
provided free Collatamp-G. We are indebted to Professor Len-
nart Bodin, Örebro University, for valuable statistical help and
advice; Lars-Göran Dahlin for valuable advice and for help with
the classifications; and Berit Anselmius, Jessica Cederblad,
Anders Fransson, and Maria Smedeus for all their efforts with
patient administration and data collection.
References
1. Mossad SB, Serkey JM, Longworth DL, Cosgrove DM 3rd,
Gordon SM. Coagulase-negative staphylococcal sternal
wound infections after open heart operations. Ann Thorac
Surg 1997;63:395–401.
2. Tegnell A, Aren C, Ohman L. Coagulase-negative staphylo-
cocci and sternal infections after cardiac operation. Ann
Thorac Surg 2000;69:1104–9.
3. Hall RE, Ash AS, Ghali WA, Moskowitz MA. Hospital cost of
complications associated with coronary artery bypass graft
surgery. Am J Cardiol 1997;79:1680–2.
4. Taylor GJ, Mikell FL, Moses HW, et al. Determinants of
hospital charges for coronary artery bypass surgery: the
economic consequences of postoperative complications.
Am J Cardiol 1990;65:309–13.
5. Stahle E, Tammelin A, Bergstrom R, Hambreus A, Nystrom
SO, Hansson HE. Sternal wound complications–incidence,
microbiology and risk factors. Eur J Cardiothorac Surg
1997;11:1146–53.
6. Gardlund B, Bitkover CY, Vaage J. Postoperative mediasti-
nitis in cardiac surgery – microbiology and pathogenesis.
Eur J Cardiothorac Surg 2002;21:825–30.
7. Wilson AP, Gruneberg RN, Treasure T, Sturridge MF. Staph-
ylococcus epidermidis as a cause of postoperative wound
infection after cardiac surgery: assessment of pathogenicity
by a wound-scoring method. Br J Surg 1988;75:168–70.
8. Markewitz A, Schulte HD, Scheld HH. Current practice of
peri- and postoperative antibiotic therapy in cardiac surgery
in Germany. Working Group on Cardiothoracic Surgical
Intensive Care Medicine of the German Society for Thoracic
and Cardiovascular Surgery. Thorac Cardiovasc Surg 1999;
47:405–10.
9. Parry GW, Holden SR, Shabbo FP. Antibiotic prophylaxis for
cardiac surgery: current United Kingdom practice. Br Heart J
1993;70:585–6.
10. Chambers HF. Methicillin resistance in staphylococci: mo-
lecular and biochemical basis and clinical implications. Clin
Microbiol Rev 1997;10:781–91.
11. Mangram AJ, Horan TC, Pearson ML, Silver LC, Jarvis WR.
Guideline for prevention of surgical site infection, 1999.
Hospital Infection Control Practices Advisory Committee.
Infect Control Hosp Epidemiol 1999;20:250–78.
12. Friberg O, Jones I, Sjoberg L, Soderquist B, Vikerfors T,
Kallman J. Antibiotic concentrations in serum and wound
fluid after local gentamicin or intravenous dicloxacillin
prophylaxis in cardiac surgery. Scand J Infect Dis 2003;35:
251–4.
13. Olsen MA, Lock-Buckley P, Hopkins D, Polish LB, Sundt
TM, Fraser VJ. The risk factors for deep and superficial chest
surgical-site infections after coronary artery bypass graft
surgery are different. J Thorac Cardiovasc Surg 2002;124:
136–45.
14. Szabo Z, Hakanson E, Svedjeholm R. Early postoperative
outcome and medium-term survival in 540 diabetic and 2239
nondiabetic patients undergoing coronary artery bypass
grafting. Ann Thorac Surg 2002;74:712–9.
15. Bitkover CY, Gardlund B. Mediastinitis after cardiovascular
operations: a case-control study of risk factors. Ann Thorac
Surg 1998;65:36–40.
16. El Oakley R, Paul E, Wong PS, et al. Mediastinitis in patients
undergoing cardiopulmonary bypass: risk analysis and mid-
term results. J Cardiovasc Surg (Torino) 1997;38:595–600.
17. Gummert JF, Barten MJ, Hans C, et al. Mediastinitis and
cardiac surgery–an updated risk factor analysis in 10,373
consecutive adult patients. Thorac Cardiovasc Surg 2002;50:
87–91.
18. Milano CA, Kesler K, Archibald N, Sexton DJ, Jones RH.
Mediastinitis after coronary artery bypass graft surgery. Risk
factors and long-term survival. Circulation 1995;92:2245–51.
19. Ridderstolpe L, Gill H, Granfeldt H, Ahlfeldt H, Rutberg H.
Superficial and deep sternal wound complications: inci-
dence, risk factors and mortality. Eur J Cardiothorac Surg
2001;20:1168–75.
20. The Parisian Mediastinitis Study Group. Risk factors for
deep sternal wound infection after sternotomy: a prospec-
tive, multicenter study [see comments]. J Thorac Cardiovasc
Surg 1996;111:1200–7.
21. Hall JC, Hall JL, Edwards MG. The time of presentation of
wound infection after cardiac surgery. J Qual Clin Pract
1998;18:227–31.
22. Loop FD, Lytle BW, Cosgrove DM, et al. Sternal wound
complications after isolated coronary artery bypass grafting:
early and late mortality, morbidity, and cost of care. Ann
Thorac Surg 1990;49:179–86.
23. Culliford AT, Cunningham JN Jr, Zeff RH, Isom OW, Teiko
P, Spencer FC. Sternal and costochondral infections follow-
ing open-heart surgery. A review of 2,594 cases. J Thorac
Cardiovasc Surg 1976;72:714–26.
24. Jonkers D, Elenbaas T, Terporten P, Nieman F, Stobberingh
E. Prevalence of 90-days postoperative wound infections
after cardiac surgery. Eur J Cardiothorac Surg 2003;23:97–
102.
25. Wilson AP, Treasure T, Sturridge MF, Gruneberg RN. A
scoring method (ASEPSIS) for postoperative wound infec-
tions for use in clinical trials of antibiotic prophylaxis. Lancet
1986;1:311–3.
26. Wilson AP, Livesey SA, Treasure T, Gruneberg RN, Stur-
ridge MF. Factors predisposing to wound infection in cardiac
surgery. A prospective study of 517 patients. Eur J Cardio-
thorac Surg 1987;1:158–64.
27. Tegnell A, Isaksson B, Granfeldt H, Ohman L. Changes in
the appearance and treatment of deep sternal infections. J
Hosp Infect 2002;50:298–303.
28. Guzman Valdivia Gomez G, Guerrero TS, Lluck MC, Del-
gado FJ. Effectiveness of collagen-gentamicin implant for
treatment of “dirty” abdominal wounds. World J Surg 1999;
23:123–6.
29. Rutten HJ, Nijhuis PH. Prevention of wound infection in
elective colorectal surgery by local application of a gentam-
icin-containing collagen sponge. Eur J Surg Suppl 1997;578:
31–5.
30. Rosen HR, Marczell AP, Czerwenka E, Stierer MO, Spoula
H, Wasl H. Local gentamicin application for perineal wound
healing following abdominoperineal rectum excision. Am J
Surg 1991;162:438–41.
31. Gruessner U, Clemens M, Pahlplatz PV, Sperling P, Witte J,
Rosen HR. Improvement of perineal wound healing by local
160 FRIBERG ET AL Ann Thorac Surg
LOCAL GENTAMICIN AND STERNAL WOUND INFECTIONS 2005;79:153–62
CARDIOVASCULAR

9. administration of gentamicin-impregnated collagen fleeces
after abdominoperineal excision of rectal cancer. Am J Surg
2001;182:502–9.
32. Musella M, Guido A, Musella S. Collagen tampons as
aminoglycoside carriers to reduce postoperative infection
rate in prosthetic repair of groin hernias. Eur J Surg 2001;
167:130–2.
33. Leyh RG, Bartels C, Sievers HH. Adjuvant treatment of deep
sternal wound infection with collagenous gentamycin. Ann
Thorac Surg 1999;68:1648–51.
34. Hall JC, Christiansen K, Carter MJ, et al. Antibiotic prophy-
laxis in cardiac operations. Ann Thorac Surg 1993;56:916–22.
35. Kaiser AB, Petracek MR, Lea JWT, et al. Efficacy of cefazolin,
cefamandole, and gentamicin as prophylactic agents in car-
diac surgery. Results of a prospective, randomized, double-
blind trial in 1030 patients. Ann Surg 1987;206:791–7.
36. Kriaras I, Michalopoulos A, Michalis A, et al. Antibiotic
prophylaxis in cardiac surgery. J Cardiovasc Surg (Torino)
1997;38:605–10.
37. Bolon MK, Morlote M, Weber SG, Koplan B, Carmeli Y,
Wright SB. Glycopeptides are no more effective than beta-
lactam agents for prevention of surgical site infection after
cardiac surgery: A meta-analysis. Clin Infect Dis 2004;
38:1357–63.
38. Vander Salm TJ, Okike ON, Pasque MK, et al. Reduction of
sternal infection by application of topical vancomycin. J Tho-
rac Cardiovasc Surg 1989;98:618–22.
39. Karlowsky JA, Jones ME, Draghi DC, Thornsberry C, Sahm
DF, Volturo GA. Prevalence and antimicrobial susceptibili-
ties of bacteria isolated from blood cultures of hospitalized
patients in the United States in 2002. Ann Clin Microbiol
Antimicrob 2004;3:7.
40. Coker AO. A study of synergism between cloxacillin and
gentamicin on resistant staphylococci (penicillinase produc-
ing and gentamicin resistant). East Afr Med J 1989;66:141–7.
41. Furnary AP, Zerr KJ, Grunkemeier GL, Starr A. Continuous
intravenous insulin infusion reduces the incidence of deep
sternal wound infection in diabetic patients after cardiac
surgical procedures. Ann Thorac Surg 1999;67:352–60.
INVITED COMMENTARY
One of the most contentious issues in surgery is the use
of prophylactic topical antibiotics to prevent wound in-
fection [1]; cardiac surgery is no exception. Despite reg-
ular use of prophylactic intravenous antibiotics, postop-
erative mediastinitis occurs in a significant number of
patients undergoing open heart surgery through median
sternototmy. This major complication plagues the most
commonly used incision in cardiac surgery, and is asso-
ciated with a mortality rate as high as 47%. Staphylococcus
aureus or S epidermidis are isolated in as many as 70% of
the cases. Therefore, prophylactic instillation of anti-
gram-positive antimicrobials during wound closure may
reduce the risk of mediastinitis. This concept has indeed
been tested in a previous prospective, randomized, con-
trolled study; Vander Salm and associates [2] found that
topical vancomycin applied during wound closure after
median sternotomy was associated with a significant
reduction in the rate of sternal wound infection (from
3.6% in the control group to 0.45% in the treatment
group). We found sub-MIC levels of vancomycin in
patient’s serum after topical application of 1 g vancomy-
cin during sternotomy closure [3]. This, at least in theory,
may encourage the emergence of vancomycin-resistant
pathogens. Therefore finding a non–vancomycin antimi-
crobial that effectively prevents mediastinitis will be a
welcome addition to the field.
In this issue of The Annals, Friberg and associates
report the results of a prospective, randomized clinical
trial of using prophylactic collagen-gentamicin sponges
placed between the sternal edges at the time of wound
closure after mediastinotomy in 2,000 cases undergoing
open heart surgery. The incidence of surgical site infec-
tion was significantly reduced from 9% in the control
group to 4.3% in the treatment group. These apparently
higher rates of surgical wound infection were attributed
to the fact that this is a prospective randomized study
with longer and near complete follow-up. Friberg and
colleagues have used their own modification of the Cen-
ters for Disease Control (CDC) definition of wound
infection, where “. . . sternal wire involvement with infec-
tion was the cut-off between superficial and deep wound
infection . . . .” Based on this definition, the incidence of
deep wound infection was also reduced from 3.3% in the
control group to 2.3% in the treatment group.
The CDC definition for surveillance of surgical site
infections [1] identifies the following three categories of
surgical site infection: superficial incisional surgical site
infection, deep incisional surgical site infection, and
organ/space surgical site infection. This classification
alone may be adequate for incisions with a reasonable
demarcation between the superficial and the deep com-
ponents of the wounds, as well as well-defined organ and
space compartments, such as thoracotomy and laparot-
omy incisions. In cases of mediastinitis, however, the
sternum is considered an organ in the CDC definition,
yet it represents the deep layer of the wound and is
always involved in the disease process. It is almost
impossible to determine whether the infection is con-
fined to the pectoral fascia, the sternal edges, the sternal
wires, the retrosternal space, or the mediastinal spaces
alone. Infections at the suprastenal notch and around the
xiphoid cartilage are also difficult to define as deep-,
organ-, or space-site infection.
We have previously devised a classification for post-
median sternotomy deep wound infection [4]. In this
classification, all deep sternal wound infections were
classified according to the time of first presentation, and
according to whether the patient had any previous at-
tempts of surgical treatment for mediastinitis. The clas-
sification also takes into account the factors that carry a
proven risk of developing mediastinits such as diabetes,
obesity, and the use of oral steroids or immunosuppres-
sive agents.
Friberg and coworkers acknowledge that the applica-
tion of other wound score systems was beyond their
resources; they also highlight other limitations of their
study including the lack of a collagen sponge only—
without gentamicin—group, and the significantly higher
161
Ann Thorac Surg FRIBERG ET AL
2005;79:153–62 LOCAL GENTAMICIN AND STERNAL WOUND INFECTIONS
© 2005 by The Society of Thoracic Surgeons 0003-4975/05/$30.00
Published by Elsevier Inc doi:10.1016/j.athoracsur.2004.08.051
CARDIOVASCULAR