Académique Documents
Professionnel Documents
Culture Documents
ORIGINAL ARTICLE
Sciences, Section of Clinical Microbiology and Infectious Diseases, Uppsala University, Uppsala, Sweden
Introduction
Abstract
Objectives. Surgical site infections (SSIs) may be devastating
for the patient and they carry high economic costs. Studies
of SSI after neurosurgery report an incidence of 111%.
However, patient material, follow-up time and definition of SSI
have varied. In the present study we prospectively recorded
the prevalence of SSI 3 months after standard intracranial
neurosurgical procedures. The incidence, impact and risk
factors of SSI were analysed. Methods. We included patients
admitted during 2010 to our unit for postoperative care after
standard neurosurgical procedures. SSI was defined as evident
with positive cultures from surgical samples or CSF, and/or
purulent discharge during reoperation. Follow-up was done
after 3 and 12 months and statistics was obtained after 3
months. The predictive values on the outcome of demographic
and clinical factors describing the surgical procedure were
evaluated using linear regression. Results. A total of 448
patients were included in the study and underwent a total of
466 procedures. Within 3 and 12 months, 33 and 88 patients,
respectively, had died. Of the surviving patients, 20 (4.3% of
procedures) developed infections within 3 months and another
3 (4.9% of procedures) within 12 months. Risk factors for SSI
were meningioma, longer operation time, craniotomy, dural
substitute, and staples in wound closure. Patients with SSI had
significantly longer hospital stay. Multivariate analysis showed
that factors found significant in univariate analysis frequently
occur together. Discussion. We studied the prevalence of SSI
after 3 and 12 months in a prospective 1-year material with
standard neurosurgical procedures and found it to be 4.3%
and 4.9%, respectively. The analysis of the results showed that
a combination of parameters indicating a longer and more
complicated procedure predicted the development of SSI.
Our conclusion is that the prevention of SSI has to be done
at many levels, especially with patients undergoing long
surgical procedures.
Correspondence: Sami Abu Hamdeh, Department of Neuroscience, Uppsala University, Section of Neurosurgery, Akademiska Sjukhuset, Uppsala 751 85,
Sweden. E-mail: sami.abu.hamdeh@akademiska.se
Received for publication 6 February 2013; accepted 11 August 2013
270
Patients
We included patients admitted to our intermediate-care
post-operative ward (NIMA) during 2010 after having been
operated for intracranial tumours, chronic subdural haematomas or reconstructive cranioplasty. The surgical procedures for the tumours included removal or resection through
craniotomies and craniectomies, and biopsies through burr
holes. Petrosectomies, transnasal approaches and spine
cases were excluded as well as patients treated pre- or postoperatively in the neurointensive care unit. The reason for
the exclusion of these patients was to facilitate the creation
of a database for continuous supervision of SSI rates in our
department, where we wanted a homogenous group and a
foreseeable number of patients.
A total of 448 patients admitted to NIMA during 2010 met
our criteria and were included in the study. Of the included
patients 268 (54.9%) were males. Median age was 64 years
(range 692). Fourteen patients underwent two or more
procedures. The majority of operations in the present study
were performed for intracranial tumours (Table I) (54.5% of
procedures, with high-grade glioma being the most common (21.7%)) followed by burr-hole craniectomy for chronic
subdural haematoma (29.4%). Of all procedures 37.6% were
performed in an emergency setting, and 62.4% were elective.
Mean operating time was 3 h 23 min. Virtually all patients
received antibiotic prophylactics before surgery (99.1%).
Cloxacillin was the antibiotic of choice in most cases (92.1%)
followed by Clindamycin (5.4%), used in the presence or
suspicion of penicillin allergy. Forty patients (8.9%) had a
record of diabetes and 220 were on steroid treatment before
undergoing surgery (49.1%). After 3-month follow-up 33
patients had deceased (7.4%), and after 12 months a total of
88 patients had died (19.6%).
Data collection
Data were collected using the electronic patient journal system. We collected demographic data as gender, age and diagnosis, as well as factors suspected to influence the risk for
wound infections. This included the occurrence of diabetes,
steroid treatment, antibiotic prophylactics, type of antibiotics, blood replacement therapy, type of surgical procedure,
duration of surgery, used operating theatre, number of people in the operating room during surgery, inserted surgical
implants, use of wound drainage, technique for skin closure
and hair removal.
Table I. This table shows the number of patients with each diagnosis
and those that developed a SSI.
No. of
Infected
Percent of
Diagnosis
operations
patients
infections (%)
SDH
Glioma
Meningioma
Cranioplasty
Metastasis
Chiari
Colloid cyst
Vascular
Other tumours
Miscellaneous
140
120
83
38
25
11
6
9
30
4
3
3
10
3
1
1
1
1
0
0
2.1
2.5
12
7.9
4
9
16.7
11.1
0
0
271
Statistics
Statistica 10.0 (Stat Soft, Inc. Tulsa, OK) was used for descriptive and analytical statistics. Categorical variables were
compared using Fishers exact test and continuous variables
using t-test.
We included a fairly large number of variables in the analysis. This created a high dimensionality making the material
difficult to overview. Therefore we used principal components analysis with Non-linear Iterative Partial Least Squares
(NIPALS) to illustrate our material. NIPALS is a mathematical procedure that aims to represent a set of (possibly correlated) multivariate variables with the aid of a smaller number
of uncorrelated variables known as principal components.
Using NIPALS we defined two principal components: p1 and
p2 (Fig. 1). The figure shows the influence of each variable on
p1 and p2, respectively. This is called the variables loading
factors and is plotted on the x-axis for p1 and on the y-axis for
p2. Figure 1 demonstrates that some variables are clustered
(placed close to each other). Clustered variables influence
the model in similar ways, which also indicates they are correlated. We analysed the risk factors further using a generalised linear model with a multinomial ordinal response and
a logistic link function. The best subset of predictive variables
for SSI was selected using Aikakes Information Criterion.9
A p value 0.05 was considered statistically significant.
Ethics
The Uppsala University Regional Ethical Review Board for
clinical research granted permission for the study.
Results
Prevalence of infections
Twenty cases were complicated by surgical site infection
after 3 months of follow-up and after 12 months of follow-up,
three more cases of surgical site infection were identified.
Out of the 23 cases complicated by SSI, three patients eventually died within the 12 months of follow-up. Incidence
of SSI was 4.3% of all procedures (4.5% of patients) after
272
Fig. 1. In this figure we used Principal components analysis with NIPALS in order to illustrate our material. Some variables such as dural substitute,
diagnosis of meningioma and skin closure are clustered. This means that they influence the model in similar ways, which also indicates that they
are correlated.
Fig. 2. (a) Contrast-enhanced CT of patients operated for parietotemporal meningioma. Images demonstrate fluid accumulation
in postoperative cavity and extradurally with contrast-enhancing
capsule, as well as underlying brain oedema. (b) Contrast-enhanced
CT images of patients operated with cranioplasty of own bone after
hemicraniectomy for malignant middle cerebral artery infarction.
Images demonstrate large epidural fluid accumulation with irregular
contrast-enhancing capsule as well as evidence of osteomyelitis.
Table III. This table lists the results from the univariate analysis of risk
factors from SSI.
Mean
operations
Mean infected
p value
CNS
Propionebacterium
Staphylococcus Aureus
Enterobacter aerogenes
CNS Propionebacterium
CNS Propionebacterium enterococcus
No positive culture
Age
Theatre staff
Length of stay
Operation time
8
6
5
1
1
1
1
34.8
26.1
21.7
4.3
4.3
4.3
4.3
273
62 years
8.4 persons
9 days
3 h 23 min
Due to the loss of 88 patients who died within 12 months, statistical analysis was performed after 3 months. Tables III and
IV list variables that were considered as possible risk factors
for the development of SSI. Individual factors that were found
to be statistically significant (p 0.05) were prolonged length
of hospital stay, prolonged operation time, craniotomy as the
method of surgery, use of dural substitute, and use of staples
for skin closure. Among the diagnoses, operated meningioma was found as a statistically significant risk factor for
the development of SSI (Table V). No significant difference
was found between operating theatres.
Variables found statistically significant in univariate
analysis were included in multivariate logistic regression. A
significant correlation (Spearmans R) was found between
length of operation time and length of hospital stay. Therefore length of stay was not included in analysis. Results of
multivariate modelling are seen in Table VI. The best subsets
of variables predicting the development of SSI included dural
substitute, operation time, method for skin closure, and surgical method. To illustrate how these are correlated we also
did a principal component analysis (see methods) (Fig. 1).
This figure shows that staples for skin closure, long surgery
time, dural substitute and meningioma are clustered, indicating that they frequently occur together.
Table IV. This table lists the results from the univariate analysis of risk factors from SSI.
No. of operations
No. infected
% Infected
Admission
Time of surgery
Diabetes
Steroids
Blood transfusion
Hair removal
Wound drainage
Method of surgery
Dural substitute
Wound closure
n.s
n.s
0.02
0.03
Discussion
Gender
61 years
9.3 persons
13 days
4 h 12 min
258 Men
208 Women
175 Acute
291 Elective
368 Work hours
124 On call
39 Yes
427 No
222 Yes
244 No
54 Yes
411 No
239 Complete
226 Partial
378 Yes
88 No
192 Craniectomy
274 Craniotomy/
36 Yes
430 No
158 Staples
307 Sutures
15 Men
11 Women
9 Acute
17 Elective
21 Work hours
5 On call
2 Yes
24 No
13 Yes
13 No
4 Yes
22 No
17 Complete
9 Partial
19 Yes
7 No
6 Craniectomy
20 Craniotomy
8 Yes
18 No
14 Staples
12 Sutures
3.2%
2.4%
1.9%
3.6%
4.5%
1.1%
0.4%
5.2%
2.8%
2.8%
0.9%
4.7%
3.7%
1.9%
4.1%
1.5%
1.3%
4.3%
1.7%
3.9%
3.0%
2.6%
p value
n.s
n.s
n.s
n.s
n.s
n.s
n.s
n.s
0.04
0.0003
0.03
Long hospital stay, long duration of the operation, craniotomy as method of surgery, dural substitute, wound closure with
staples and the diagnosis of meningioma all significantly increased the risk of SSI, n.s. Not significant.
127 Operations
82 Operations
129 Operations
38 Operations
57 Operations
4 Infections
10 Infections
6 Infections
3 Infections
3 Infections
0.9%
2.3%
1.4%
0.7%
0.7%
n.s
0.02
n.s
n.s
n.s
Risk factors
We found in the univariate analysis that the surgical method
craniotomy, use of dural substitute, diagnosis of meningioma, and wound closure with staples were significantly
more common in cases with SSI. Longer duration of the surgical procedure was also associated with increased risk for
infections. The findings regarding long surgery and implants
are consistent with other studies.
Parameters such as dural substitute, diagnosis of meningiomas, and longer surgery frequently occur together as
illustrated with principal components analysis (Fig. 1). It is
therefore difficult to point out single factor as responsible.
We think instead that these parameters indicate a more complicated surgical situation, which increases the possibilities
for the wound to be exposed to bacteria. It seems that this
situation is more common in surgery of meningiomas in our
material. On the other hand, the fact that these variables
were selected into the same multivariate model (Table VI)
indicates that each has some independent predictive power.
Common sense precautions such as reducing door openings,
using as non-traumatic techniques as possible and monitoring air quality should be applied, especially during lengthy
procedures.
Other factors such as hair removal, wound drain, age,
gender, number of people in the operating room, diabetes,
steroids and blood transfusion were not related to increased
risk for SSI. We found that more than 99% of patients
received prophylactic antibiotics in accordance with the
departments protocol. The importance of prophylactic antibiotics has been demonstrated earlier.5,11,13 As described in
detail by the Surgical Infection Prevention Project it is not
enough simply to administer antibiotics; it is necessary to
match the type of antibiotic used with the most common
cutaneous pathogens, and to optimise timing, redosing and
discontinuation.
Impact of SSI
Patients who developed SSI required readmission, reoperation and antibiotic therapy in order to control the infection.
In 13 cases further surgery was necessary for cranial reconstruction after the infection had been controlled. One patient
developed hemiparesis after presenting with subdural empyema. This illustrates the high socio-economic cost of SSI,
which was estimated in earlier studies, as well as the patient
suffering SSI causes.5
Conclusion
In this study we measured the 3- and 12-month prevalence of SSI after 466 craniotomies and craniectomies
Table VI. This table shows the best subsets of variables predicting SSI. Dural substitute stands out as important, in combination with surgery
duration and method for skin closure.
Degree of
Variable 1
Variable 2
Variable 3
Variable 4
freedom
AIC
L.Ratio Chi2
p value
Dural substitute
Dural substitute
Dural substitute
Dural substitute
Method of surgery
Method of surgery
Method of surgery
Method of surgery
Operation time
Skin closure
Skin closure
Method of surgery
Skin closure
Skin closure
Method of surgery
Operation time
Skin closure
Skin closure
Dural substitute
Dural substitute
Dural substitute
Dural substitute
Operation time
Method of surgery
Skin closure
Skin closure
Operation time
Operation time
Method of surgery
Method of surgery
Operation time
Operation time
1
2
2
3
2
3
3
4
1
2
1
2
3
2
1
170.0146
170.2706
170.3912
171.0302
171.7102
172.2108
172.3870
172.9546
173.8141
173.9941
174.8053
175.6570
175.9274
176.8035
176.8366
7.64083
9.38483
9.26417
10.62518
7.94517
9.44461
9.26838
10.70081
3.84134
5.66132
2.85007
3.99835
5.72802
2.85187
0.81881
0.006
0.01
0.01
0.02
0.02
0.03
0.03
0.04
n.s
n.s
n.s
n.s
n.s
n.s
n.s
References
1. OKeeffe AB, Lawrence T, Bojanic S. Oxford craniotomy infections
database: A cost analysis of craniotomy infection. Br J Neurosurg
2012;26:2659.
2. McClelland S III, Hall WA . Postoperative central nervous system
infection: incidence and associated factors in 2111 neurosurgical
procedures. Clin Infect Dis 2007;45:559.
275