Neurosurgery 1992-98
Neurosurgeons, in keeping with general
March 1994, Volume 34, Number 3
409 Operative Sepsis in Neurosurgery: A Method of
Classifying Surgical Cases
Clinical Study
AUTHOR(S):
Narotam, Pradeep K., M. Med., F.C.S. (SA) Neuro;
van Dellen, James R., F.R.C.S., Ph.D.;
du Trevou, Michael D., M.Med., F.C.S. (SA);
Gouws, Eleanor, B.Sc.(Hons)
Department of Neurosurgery, University of Natal and
Wentworth Hospital (PKN, JRvD, MDdT) and
Department of Biostatistics, Medical Research
Council (EG), Durban, South Africa
Neurosurgery 34; 409-416, 1994
ABSTRACT: NEUROSURGICAL OPERATIONS
HAVE traditionally been classified along the lines of
general surgical procedures. A prospective study,
during an 18-month period, was undertaken in 2249
patients undergoing neurosurgical procedures to
establish and evaluate a method of classifying
surgical cases by the use of specific neurosurgical
criteria. Patients were placed in one of five categories
according to the level and type of contamination at
the time of surgery. Infection included all abnormal
wounds and was documented as deep when infection
occurred beneath the galea (subgaleal pus, osteitis,
abscess/empyema, ventriculomeningitis) and as
superficial if only the scalp (including wound
erythema) was involved. A statistically significant
difference in the sepsis rate was found in the different
categories (P < 0.0001). Of the 342 "dirty cases,"
9.1% of patients developed further wound sepsis.
Concomitant cerebrospinal fluid fistulae (44%),
second operations (11.8%), and patients with
penetrating injuries (9.2%) were the major factors
implicated in sepsis in the "contaminated" category
(9.7%). In the "clean contaminated" category, a sepsis
rate of 6.8% was found. Prolonged surgery (longer
than 4 hours) was also implicated in higher infection
rates (13.4%). This study strongly supports the
separation of patients who have foreign materials
implanted (sepsis rate = 6.0%) from "clean" patients,
essentially cases categorized as having no known risk
factors that may affect sepsis, in whom a sepsis rate
of 0.8% was found (P < 0.001). Importantly, surgery
for the repair of so-called "clean" neural tube defects
in neonates requires separate consideration. An
infection rate of 14.8% existed in this subgroup. A
uniform system of reporting wound abnormalities is
also proposed.
KEY WORDS: Neurosurgical sepsis; Sepsis;
Wound infections
The determination of accurate operative sepsis rates,
especially in neurosurgery, where infection can have
devastating sequelae, is an important part of patient
management (28). A uniform method of reporting
sepsis would also have many benefits in comparing
Downloaded from https://academic.oup.com/neurosurgery/article-abstract/34/3/409/3771249
by Bla The Librarian, user
on 30 May 2018
results (24,28).
surgery trends, have traditionally classified operative
cases as "clean," "dirty," or "contaminated" (29,30). In
1964, the National Academy of Sciences-National
Research Council (3) suggested that operative wounds
should be classified according to the level of intrinsic
contamination at the site of surgery and introduced an
additional category--"clean contaminated." These
procedures involved entry into body cavities known
to contain microorganisms. This four-category
classification is suitable for general surgical
procedures and allows uniform reporting of sepsis
rates (3,13). With neurosurgical procedures, however, it
has certain shortcomings because of an increasing use
of implanted foreign materials either temporarily
(e.g., subdural pressure monitors, which are known to
double the risk of infection) or permanently (e.g.,
shunts, acrylic, or wire) (2,20,31). In 1984, Vlahov et al.
(28)
, using the National Academy of Sciences-National
Research Council format, grouped their neurosurgical
cases into the above four categories but only reported
on their experience with "clean" cases. In their report,
they supported the need to introduce a further
category--"clean with foreign body" (14,28).
There has still been no uniformity in the placement
of cranioplasty and shunt procedures in
classifications (6,15,24,26,28,31). In some series, shunts
Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.
were considered separately from clean cases, whereas
cranioplastic procedures were included (4,12,14,15,28,30).
We propose that to categorize a neurosurgical case as
"clean," there must be absolutely no factor present
known to increase the risk of infection, including the
presence of a foreign body.
Unfortunately, the diagnosis of sepsis in various
series has not been uniform, which has complicated
comparative analysis (2,6,14,15,17,19,20,22,28,29,31). Wright
(30)
defined neurosurgical wound sepsis as "purulence,
meningitis or osteitis," established beyond doubt by
the presence of pathogenic bacteria. Malis (17)
classified the wounds in his report as septic only if a
positive culture was obtained from the discharging
wound or if meningitis or meningism was present,
together with cerebrospinal fluid (CSF) leukocytosis
and wound inflammation. These stringent criteria
have been adopted in various series reporting on
wound infection or on the use of prophylactic
antibiotics (10-12,17). Wound swabs may not always
yield a pathogenic organism, especially when
different or multiple bacterial isolates are found and
skin commensals may be falsely implicated. Clearly,
then, the incidence of wound infections can be
underreported. The University of Maryland Hospital
later added the further criterion of "physician's
diagnosis," because any abnormal wound is of
concern to the neurosurgeon (28).
PATIENTS AND METHODS
A method of classifying surgical procedures by the
use of criteria relevant to neurosurgery was devised
(adapted from Vlahov et al. [28]) to identify patients at
the highest risk of developing sepsis. Over an 18-
month period, all patients undergoing surgery were
prospectively entered into the study. The patients
were classified into one of the five categories
 according to their potential of developing sepsis
(Table 1).
Dirty
The "dirty" category comprised patients with sepsis
already present at the time of surgery, viz., brain
abscess, subdural empyema, ventriculitis, meningitis,
osteitis, and purulent skin infection.
Contaminated
Although no frank sepsis was present,
contamination of the operative site was known to
have occurred, namely, compound skull fractures,
open scalp lacerations at the operative site (older than
4 h), patients with CSF leakage, and finally, a
subsequent operation at the same surgical incision
within a period of 4 weeks.
Clean contaminated
Known risk factors in the "clean contaminated"
group were entry into the paranasal sinuses, cranial
base fractures, breaches in standard surgical
techniques, and surgery taking longer than 2 hours.
Clean with foreign body
"Clean with foreign body" cases would meet all
other criteria for clean surgery but would have
permanently or temporarily implanted foreign
materials present, namely, shunts, intracranial
pressure monitors, reservoirs, and ventricular drains.
Large, metallic foreign materials such as Halifax
clamps, Caspar plates, and Luque rectangles were
also included. Methyl methacrylate ("acrylic"),
because it was not entirely homogenous, was
included. Smaller metallic foreign materials such as
aneurysm clips and ligaclips were excluded. Sutures
(vicryl) and monofilamentous materials (stainless
steel wire) used for standard neurosurgical
procedures and routine wound closure were also
excluded.
Clean
The entry of a case into the "clean" category was
by the exclusion of all the above factors. Surgery was
usually elective, under ideal operating conditions,
with closed suction bellows drainage placed in the
subgaleal space for a period not exceeding 24 to 48
hours.
Hair was routinely shaved preoperatively, and the
scalp was examined for wounds or local sepsis.
Savlon, followed by Hibitane in alcohol, was used as
the standard method of skin preparation for surgery.
All surgery was performed in one of two
neurosurgical operating rooms by standard
techniques. In addition, Cephamandole (15 mg/kg)
was used perioperatively in all clean, clean
contaminated, and clean with foreign body cases.
Penicillin and sulfadiazine were used as prophylactic
antibiotics in "open" trauma patients.
The patients were monitored for sepsis
postoperatively, either until the time of discharge
back to the primary referral hospital or after 4 weeks
had elapsed if they were still in the hospital or had
returned for follow-up visits. (All septic cases at
Downloaded from https://academic.oup.com/neurosurgery/article-abstract/34/3/409/3771249
by Bla The Librarian, user
on 30 May 2018
peripheral hospitals would be referred to the
neurosurgical unit because Wentworth is the only
neurosurgical referral unit in the region.) Deep
postoperative sepsis was diagnosed when there was
ventriculitis, meningitis, a brain abscess, osteitis, or
subgaleal pus (24,31). Meningitis was diagnosed if an
organism was identified by staining or culture, or if
the CSF revealed polymorphonuclear leukocytosis (>
20/mm3) associated with a low CSF sugar (< 2.0
mmol/L) (6,23). Superficial sepsis was graded as mild
(wound erythema), moderate (superficial
inflammation of whole wound, serous discharge, or
localized infection), or severe (wound breakdown,
sinuses, fistulae, cellulitis, or infection of more than
one-third of the wound). All abnormal wounds,
regardless of whether organisms were cultured or not,
were considered to be septic (16).
All samples of wound discharge were transported
for later laboratory testing in Stuarts' medium. CSF
samples were submitted for all patients suspected of
harboring meningitis or ventriculitis. Infections
elsewhere in the body, a possible source for
septicemia, were also monitored. Statistical analyses
used were the χ2 test, Fisher's exact test, and the
Bonferroni adjustment for pairwise comparisons with
the appropriate adjustment of significance levels (P =
0.05 per number of comparisons).
Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.
RESULTS
During the 18-month period, 2249 surgical
procedures were performed. Overall, 92 deep wound
infections (4.1%) and 51 superficial wound infections
(2.3%) were recorded (Table 2). As expected, a
higher risk of infection was found with an increasing
degree of contamination. The proposed classification
was evaluated by use of the χ2 test, which revealed
that there was a significant difference within the five
categories (P < 0.0001). On pairwise comparison, this
difference was the most marked between the clean
and the other four groups (P < 0.005). The
contaminated group had higher infection rates than
both the clean contaminated and the clean with
foreign body (χ2, P < 0.05) groups. The differences
between the clean with foreign body and clean
contaminated groups were not significant.
The data obtained from the different categories
were also analyzed by use of the criteria proposed by
Wright (30) and those proposed by Malis (17), which
have been used in several studies on antibiotic
prophylaxis. These were compared with the criteria
that we used in the above analyses and that had been
modified from the recommendations of Lowbury (16)
(Table 3). Significantly lower overall sepsis rates of
3.8 and 3.7%, respectively, were obtained with the
criteria of Malis (17) and Wright (30), compared with a
6.4% infection rate with our criteria.
Dirty cases
Of the 342 patients classified as dirty, further
septic complications occurred in 31 (9.1%) (Table 2).
Ventriculomeningitis (15 cases) was the major cause
of deep infection. There were also two brain
abscesses, and one case each of osteitis and subgaleal
infection. Moderate to severe superficial skin
 infections occurred in 12 patients. Positive cultures
were obtained in 22 patients (71%).
Contaminated cases
When surgery was performed in patients in whom
wound contamination was known to have occurred, it
posed a higher risk of infection (sepsis rate = 9.7%)
(Table 2). Distinct differences occurred in the five
subgroups (χ2, P < 0.0001; Bonferroni adjustment, P
< 0.0083) (Table 4).
Infections occurred in 14 (11.8%) of the 119
patients who underwent subsequent operations.
Ventriculomeningitis (eight patients) was the most
common complication. Brain abscess, osteitis, and
subgaleal infections were less frequent. Moderate to
severe skin infections occurred in three patients.
Cerebrospinal fistulae were present in 25 patients
at the time of surgery. Six of these patients developed
ventriculomeningitis, and five had superficial wound
infections (44%). On pairwise comparison, this was a
significant risk factor and weighed heavily in this
category. Patients with a penetrating head injury (n =
87), namely, stab or gunshot wounds, showed a high
incidence of deep wound infections (8.1%), of which
50% were cases of ventriculomeningitis.
Clean contaminated
A 6.8% infection rate occurred in the clean
contaminated category (Table 2). Although there
were no significant differences in the subgroups,
prolonged surgery (i.e., longer than 4 hours duration)
had a tendency toward a higher infection rate (13.4%)
(Table 5).
Clean with foreign body
The risk of infection was doubled when a foreign
body was introduced to what would previously have
been regarded as a clean case (Table 2). This was
highly significant (χ2, P < 0.05; Bonferroni
adjustment, P < 0.005). The sepsis rate for subdural
intracranial pressure monitoring was low (2.5%).
Ventricular catheters, either for temporary CSF
diversions (even when used for up to 72 to 96 hours)
or for intracranial pressure monitoring, exhibited a
slightly higher infection rate (5.9%) than subdural
catheters, but this was not significant (P > 0.05) (2)
(Table 6).
Clean cases
After the exclusion of all known factors that
influence sepsis, 25% of the operative cases fell into
the clean category (Table 2). Although one should
expect a zero infection rate, it was, in fact, 2.6%! On
further analysis, almost all infections occurred in
neonates who had surgery for the repair of a closed
myelomeningocele, lipomyelomeningocele, or
encephalocele. In this subgroup, the infection rate
was 14.8% (Table 7). Clearly, this has to be identified
as an additional risk factor. With the exclusion of
these neural tube defect repairs, there were only four
patients with wound abnormalities (sepsis rate =
0.8%). Wound erythema occurred in one patient who
underwent laminectomy and in three patients with
cranial surgery, one of whom had a positive culture
Downloaded from https://academic.oup.com/neurosurgery/article-abstract/34/3/409/3771249
by Bla The Librarian, user
on 30 May 2018
(Staphylococcus aureus).
BACTERIOLOGY
The plating of suspected infected material was
performed within 12 hours from the Stuarts' transport
medium. The bacteriological data for the cases
classified as dirty were analyzed separately, because
this group had already established infection. Positive
cultures were performed from samples from 88 of the
septic patients from the other four categories--a 62%
yield. Staphylococcus aureus was the most common
isolate in the clean with foreign body category (n =
25) and the contaminated category (n = 21).
Nosocomial infections, or superinfections, by
"hospital" microorganisms, viz., Pseudomonas (n =
27), Klebsiella (n = 21), and Acinetobacter (n = 10)
species were similar in the various categories
(Table 8).
EFFECT OF CONCURRENT INFECTIONS ON
WOUND SEPSIS
Dirty cases needed to be excluded from these
analyses, because in this category, more than one
third of the central nervous system infections were
secondary to otorhinological disease or occurred after
trauma. In the patients with wound infections, 13.4%
had concurrent infections in other body systems
Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.
(chest infections, urinary tract infection, etc.).
Concurrent infections occurred in 17.4% of patients
without wound infections (Table 9). This difference
was not significant (χ2, P = 0.27).
DISCUSSION
We have formulated a method of prospectively
analyzing neurosurgical procedures according to
factors previously reported and those that are likely
to increase the risk of infection by the use of specific
criteria (14,20,24,28,30,31). We believe that the 2249
patients in the period of review represent the work of
a busy general unit, treating a wide spectrum of
neurosurgical disorders including a high proportion
of septic and trauma cases (15 and 45%,
respectively). Elective spinal and cranial surgery in
patients of all ages constituted only 40% of the
workload.
The selective exclusive classification and reporting
of wound infections by Wright (30) and by Malis (17)
have gained some acceptance (10,12,15,17,19,22,30).
Certainly, an infection rate of 0% is easily achievable
if stringent exclusive criteria are used (17).
Furthermore, despite the refined methods of specimen
collection, microbiological testing may not always
reveal the causative organism, even though the
wound may be clinically infected. This may require
further surgery, prolong the hospital stay, and
therefore, add to the morbidity. To obviate this
problem, Vlahov et al. (28) suggested the use of
"wound infection as diagnosed by the surgeon." To
remove this bias in the reporting of superficial wound
infections, we adopted the system as proposed by
Lowbury (16), whereby all abnormal wounds were
considered to be infected.
When we examined our infection rates using the
criteria of Malis (17) and of Wright (30), naturally, a
 lower infection rate in all categories could be posted.
However, in our opinion, this can result in as much as
40% underreporting of possible infection, which does
not meet the entry requirements of a pathological
organism. In our series of the 10 patients with wound
erythema, which although generally may not have
been regarded as infected (17) (only one patient had a
positive culture) nevertheless constituted an
abnormal wound (16). Therefore, a "wound
abnormality rate," although very inclusive, may lead
to a more unbiased system of reporting.
Pathological organisms could be isolated in only
62% of our patients, despite an excellent and tried
method of specimen collection. A yield of 70 to 80%
of bacterial growth is generally as high as can be
expected. The high yield of S. aureus in the cultures
and the pattern of nosocomial isolates are similar to
results from other reported series (2,6,9,10,12,15,22,31).
The introduction of computed tomographic
scanning; initial, consecutive, and early surgical
intervention; and newer antibiotics have reduced the
complication rates for cerebral abscesses, empyemas,
and osteitis. The risk of further sepsis in dirty
neurosurgical cases is therefore lower than in general
surgery, where sepsis rates of up to 40% have been
reported because of continuing contamination from
the bowel (13).
In patients with contamination as a result of
trauma, a significantly higher infection rate was
found in patients with CSF fistulae (44%) than in the
other subgroups. The high infection rate of 44% in
this subgroup of patients is not unexpected because
CSF pathways were exposed to outside
contamination. This is similar to the experience of
Aarabi (1) in the Iran-Iraq war and of Meirowsky et al.
(18)
in the Vietnam conflict. The overall rate of wound
sepsis in our patients with a compound skull fracture
or an open scalp wound was 3.1% which compares
favorably with results from other reported series
(Glasgow, 3.5%; Rotterdam, 5%; Pelonomie, 5.3%;
and Brisbane, 7%) (25).
In various series reporting on "clean" neurosurgical
procedures, subsequent operations were strongly
implicated in sepsis (7,20,30). In Wright's (30) review of
clean craniotomies, the sepsis rate was found to have
increased if the same wound was explored again
within a period of 4 weeks. Second operations for
gliomas are at the highest risk of developing sepsis
(24)
. Clearly, second episodes of surgery expose a
surgical wound to increased contamination and
therefore should be categorized as contaminated
cases. This is supported by the high infection rate in
this category (11.8%) and is justified by the finding
that there was no significant difference, on pairwise
testing, within the contaminated category itself.
Several studies have also shown a direct link
between the development of postoperative sepsis and
the length of surgery (5,10,20,24,30). Vlahov et al. (28),
after examining their neurosurgical wound infection
rate, revealed a low infection rate for procedures
taking less than 2 hours, with a steady increase up to
4 hours (24,30). Earlier, Wright (30) reported a similar
trend in his series for craniotomies. Extended
exposure of the wound can lead to devitalization of
Downloaded from https://academic.oup.com/neurosurgery/article-abstract/34/3/409/3771249
by Bla The Librarian, user
on 30 May 2018
tissue and an increased risk of contamination (5,28).
Our experience with prolonged surgery is similar to
that of other reported series, with a 5.6% infection
rate for surgical procedures of 2 to 4 hours duration
and 13.4% in a smaller number of patients where the
surgical time was more than 4 hours, although this, in
our series, did not reach statistical significance (10,20,
24,30)
.
The similar patterns of bacterial isolates in the
contaminated group and in the clean with foreign
body group support the notion that the introduction
of a foreign body leads to increased contamination,
especially by the colonization of catheters (2,31). The
widespread use of permanent foreign materials in
neurosurgical practice compounds the issue. A high
infection rate was obtained (41.2%) in patients who
underwent procedures involving the implantation of
acrylic (n = 17) (Table 6). By pairwise comparisons,
the use of acrylic forms a significant risk and
therefore justifies its exclusion from the clean
category. Even within this category of clean with
foreign body, acrylic was a major contributor to
sepsis. This might have been because, in most of
these patients, infected bone flaps had been
previously removed after extensive surgery for
subdural empyema or for contamination as a result of
trauma, even though it is our policy only to perform
Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.
cranioplasties 2 years after the initial removal of the
bone flaps.
Our infection rate of 6.3% for shunt procedures is
similar to that of most reported series (21), despite the
fact that a large number of CSF shunts had been
inserted in patients with postmeningitic
hydrocephalus, tuberculous meningitis, or
hydrocephalus associated with myelomeningoceles.
This protocol would allow for the rapid identification
of risk factors, should an increasing rate of infection
be identified in the clean category or in the clean with
foreign body category; then, appropriate infection
control strategies can be instituted. There is an
increasing use of spinal instrumentation and the
insertion of metallic rods, plates, and screws on a
permanent basis in neurosurgical practice. With this
sepsis protocol, the effect of these materials on
wound abnormalities can be further evaluated by the
use of additional subcategories in the broad category
of "clean with foreign body."
Another example is that we identified that the
repair of neural tube defects in neonates is an
additional risk factor when the clean category was
evaluated. A high incidence of wound complications
(22.4%) has been reported in the past in patients who
undergo repair of myelomeningoceles or
encephaloceles and has been ascribed to poor wound
healing (5,8,11). Our sepsis rate in this group was
14.8%. Associated hydrocephalus, with or without
ventriculitis, or abnormal skin at the site of surgery
may be an associated aggravating factor (5,11). These
patients therefore fall into a different risk group and
need to be analyzed separately. After the exclusion of
this group from our clean category, the sepsis rate for
"clean" cases fell to 0.8% (Table 7). The prevailing
sepsis rate in our unit, before the introduction of this
protocol, ranged from 2 to 5% for previously
 classified "clean" cases (6). of myelomeningocele in Puerto Rico. J
Neurosurg 72:726-731, 1990.
CONCLUSION 6. Bullock R, van Dellen JR, Ketelbey W,
We have validated, in a prospective study, a Reinach SG: A double-blind placebo-
classification for neurosurgical operations, using controlled trial of peri-operative prophylactic
specific criteria according to risk of infection, and antibiotics for elective neurosurgery. J
have justified the inclusion of all abnormal wounds in Neurosurg 69:687-691, 1988.
the category by wound infection to remove bias in the 7. Cairns H: Bacterial infection during
reporting of operative wound infections. This would intracranial operation. Lancet 1:1193-1198,
allow for valid comparison of yearly infection rates 1939.
within a department and between different centers. 8. Charney EB, Weller SC, Sutton LN, Bruce
There may also be implications for the more selective DA, Schut LB: Management of the newborn
use of prophylactic antibiotics in certain categories of with myelomeningocoele; time decision-
neurosurgery that are at the highest risk of making process. Pediatrics 75:58-64, 1985.
developing infection, i.e., subsequent operations, 9. Cruse PJE, Foord R: A five year prospective
prolonged surgery, the presence of CSF fistulae, the study of 23,619 surgical wounds. Arch Surg
repair of neural tube defects, and the implantation of 107:206-209, 1973.
foreign bodies. Furthermore, the categorization of 10. Djindjian M, Lepresle E, Homs JB: Antibiotic
clean by excluding known factors, also identified by prophylaxis during prolonged clean
other investigators to increase infection, appears to be neurosurgery--results of a randomized double-
justified. The introduction of a foreign body certainly blind study using oxacillin. J Neurosurg
deserves to be considered in a category of its own, as 73:383-386, 1990.
does surgery for the repair of neural tube defects. 11. Foltz EL, Kronmal R, Shurtleff DB: To treat
Theoretically, clean cases should not develop or not to treat: A neurosurgeon's perspective
infection. If they do, rigorous attempts need to be to myelomeningocoele. Clin Neurosurg
made to identify and address the precipitating factors. 20:147-163, 1973.
12. Geraghty J, Feely M: Antibiotic prophylaxis

ACKNOWLEDGMENTS
We thank Dr. Jerzy Golek for assistance in
preparation of the sepsis protocol, members of the
staff of the Department of Neurosurgery for
compiling data, Mrs. Cyndy McKeown and Mr. S.
Naidoo for processing data, and Mrs. M. du Plooy for
the preparation of the manuscript.
Received, February 25, 1993.
Accepted, September 27, 1993.
Reprint requests: Dr. P.K. Narotam, Section of
Neurosurgery, Dept. of Surgery, University of
Arizona Health Sciences Center, Tucson, AZ 85724.
REFERENCES: (1-31)
1. Aarabi B: Causes of infection in penetrating
head wounds in the Iran-Iraq war.
Neurosurgery 25:923-926, 1989.
2. Aucoin PJ, Kotilainen HR, Gantz NM,
Davidson R, Kellogg P, Stone B: Intracranial
pressure monitors--epidemiologic study of
risk factors and infection. Am J Med 80:2369-
2376, 1986.
3. Ad Hoc Committee of the Committee on
Trauma, Division of Medical Sciences,
National Academy of Sciences, National
Research Council: Post-operative wound
infections. The influence of ultraviolet
irradiation of the operating room and various
other factors. Ann Surg 160[suppl]:1-192,
1964.
4. Balch RL: Wound infections complicating
neurosurgical procedures. J Neurosurg 26:41-
45, 1967.
5. Bru RH, Rodriguez R, Ramirez MV, Gonzalez
R, Martinez V: Experience in the management
Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.
in neurosurgery. A randomized controlled
trial. J Neurosurg 60:724-726, 1984.
13. Guglielmo BJ, Hohn DC, Koo PJ, Hunt TK,
Sweet RL, Conte JE Jr: Antibiotic prophylaxis
in surgical procedures. A critical analyses of
the literature. Arch Surg 118:943-954, 1983.
14. Haines SJ: Systematic antibiotic prophylaxis
in neurologic surgery. Neurosurgery 6:355-
361, 1980.
15. Haines SJ, Goodman ML: Antibiotic
prophylaxis of postoperative neurosurgical
wound infection. J Neurosurg 56:103-105,
1982.
16. Lowbury EJL: Control of hospital infection,
in Lowbury EJL, Ayliffe GAJ, Feddes AM,
Williams JD (eds): A Practical Handbook.
London, Chapman & Hall, 1981, pp 30-31.
17. Malis LI: Prevention of neurosurgical
infections by intraoperative antibiotics.
Neurosurgery 5:334-343, 1979.
18. Meirowsky AM, Caveness WF, Dillon JD,
Rish BL, Mohr JP, Kistler JP, Weiss GH:
Cerebrospinal fluid fistulae complicating
missile wounds of the brain. J Neurosurg
54:44-48, 1981.
19. Mollman HD, Haines SJ: Risk factors for post-
operative neurosurgical wound infections. A
case control study. J Neurosurg 64:902-907,
1986.
20. Quadery LA, Medlery AV, Miles J: Factors
affecting the incidence of wound infections in
neurosurgery. Acta Neurochir (Wien) 39:133-
141, 1973.
21. Quigley MR, Reigel DH: Factors relating to
cerebrospinal fluid shunt infections. Neurosci
J 2:16, 1991.
22. Rasmussen S, Ohrstrom JK, Westergaard L,

Downloaded from https://academic.oup.com/neurosurgery/article-abstract/34/3/409/3771249

by Bla The Librarian, user
on 30 May 2018
 Kosteljanetz M: Post-operative infections of
osteoplastic compared with free-bone flaps.
Br J Neurosurg 4:493-496, 1990.
23. Ross D, Rosegay H, Pons V: Differentiation
of aseptic and bacterial meningitis in post-
operative neurosurgical patients. J Neurosurg
69:669-674, 1988.
24. Tenney JH, Vlahov D, Salcman M, Ducket
TB: Wide variation in risk of wound infection
following clean neurosurgery. Implications
for peri-operative antibiotic prophylaxis. J
Neurosurg 62:243-247, 1985.
25. vanden Heever CM, vander Merwe DJ:
Management of depressed skull fractures.
Selective conservative management of non-
missile injuries. J Neurosurg 71:186-190,
1989.
26. Van Ek B, Dijkmans B, Van Dulkan H, Van
Furth R: Antibiotic prophylaxis in
craniotomy. A prospective double-blind
controlled study. Scand J Infect Dis 20:630-
639, 1988.
27. Van Saene HKF, Curran RA, Stoutenbeek CP,
Percival A: What the intensivist should know
about bacteriology. Review. Die Suider-
Afrikaanse Tydskrif vir Kritieke Sorg 6:5-
8, 1990.
28. Vlahov D, Montgomery E, Tenney JH, Kahn-
Eisenberg S: Neurosurgical wound infections:
Methodological and clinical factors affecting
calculation of infection rates. J Neurosurg
Nursing 16:128-133, 1984.
29. Winston KR: Hair and neurosurgery.
Neurosurgery 31:320-329, 1992.
30. Wright L: A survey of possible etiologic
agents in postoperative craniotomy infections.
J Neurosurg 25:125-132, 1966.
31. Young RF, Lawner PM: Peri-operative
antibiotic prophylaxis for prevention of post-
operative neurosurgical infections. J
Neurosurg 66:701-705, 1987.
COMMENTS
The authors have examined in a prospective study
the importance of considering the implantation of a
foreign body in a clean neurosurgical wound in
preoperatively predicting the risk of postoperative
wound infection. After carefully reviewing the
literature on postoperative neurosurgical wound
infection, this seemed to be a logical proposal, even
14 years ago (1). However, logical proposals need to
be validated empirically, and the authors have
performed a service to all neurosurgeons by doing so.
The "take home" message is clear: the implantation of
a foreign body beyond the size of an aneurysm clip or
a few wire sutures increases the risk of postoperative
wound infection from levels traditionally associated
with clean neurosurgical operations (1% or less) to
those associated with "clean contaminated" or
"contaminated" wounds (5 to 10%). This occurs
despite the use of intraoperative antibiotic
prophylaxis.
The authors have dealt with one of the major
problems of such clinical studies of wound infection:
Downloaded from https://academic.oup.com/neurosurgery/article-abstract/34/3/409/3771249
by Bla The Librarian, user
on 30 May 2018
the definition of wound infection. They have
preferred a broad definition that includes many
wounds that show a degree of erythema disturbing to
the surgeon but not associated with a positive wound
or cerebrospinal fluid culture. The relative rates of
infection in their classification groups should not be
affected by this definitional issue, because the
definitions appear to have been applied equally
across the groups. However, the absolute infection
rates can be markedly affected, as is shown when the
authors apply the definitions of Malis and Wright (2,
3)
, producing lower absolute infection rates across all
categories. The absolute infection rates reported by
the authors, therefore, should not be used as a
standard for comparison unless their liberal definition
of wound infection is also used.
No one, including the authors, has dealt adequately
with the question of what amount of foreign material
is required to place an operation in the category of
"clean with foreign body." Neurosurgeons are used to
placing routinely multiple wire sutures, aneurysm
clips, and now, small metallic plates and screws in
wounds. These have not generally been considered to
represent the "implantation of a foreign body." When
does one cross the line? One aneurysm clip is not a
"foreign body," but how about two or three or five?
What about the use of small amounts of acrylic to fill
Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.
burr holes in cosmetically important sites? Is one burr
hole filled with acrylic more of a risk than one or two
aneurysm clips? At this time, this is more a matter of
interest than of action, because we do not have a
proven plan of action that changes the infection-
preventive measures we use when the implantation of
a foreign body is carried out. Should such preventive
measures be developed, however, it would be most
important to know when they should be applied.
Therefore, the question of the volume and type of
foreign material implanted may be a useful area for
further observation or research.
Stephen J. Haines
Minneapolis, Minnesota
REFERENCES: (1-3)
1. Haines SJ: Systemic antibiotic prophylaxis in
neurosurgery. Neurosurgery 6:355-361,
1980.
2. Malis LI: Prevention of neurosurgical
infections by intraoperative antibiotics.
Neurosurgery 5:334-343, 1979.
3. Wright L: A survey of possible etiologic
agents in postoperative craniotomy infections.
J Neurosurg 25:125-132, 1966.
There is merit in revisiting well-established
guidelines that are accepted in day-to-day practice to
determine whether there is a need for modifications.
The authors of this communication have undertaken
the onerous task of structuring a prospective protocol
to analyze the relationship of postoperative sepsis to
a variety of wound conditions. They have
accumulated sufficient numbers to achieve statistical
 significance. Their addition of a fifth category to the
operative wound classification of the National
Academy of Sciences-National Research Council,
designated as "clean with foreign body," is consistent
with the widely held opinion that the implantation of
intracranial devices--particularly cerebrospinal fluid
shunts and cranioplasty plates--is associated with a
greatly increased risk of infection (2).
Perhaps an asterisk should be used to designate
those cases involving the repair of neural tube defects
to distinguish them from otherwise "clean" cases.
Another circumstance deserving of mention is the
heightened hazard of infection when operating a
second time on malignant brain gliomas in subjects
who have had radiotherapy after the initial procedure
(1,3)
. The challenge of any attempt at systematization
is how to deal with the exceptions. The suggested
changes in this schema represent a step in the right
direction.

Norman H. Horwitz
Washington, District of Columbia
REFERENCES: (1-3)

1. Ammirati M, Galicich JH, Arbit E, Liao Y:

Reoperation in the treatment of recurrent

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.
intracranial malignant gliomas. Neurosurgery
21:607-614, 1987.
2. Leedom JM, Holtom PD: Infectious
complications, in Apuzzo MLJ (ed): Brain
Surgery: Complication Avoidance and
Management. New York, Churchill
Livingstone, 1993, vol 1, pp 124-144.
3. Young B, Oldfield EH, Markesbery WR,
Haack D, Tibbs PA, McCombs P, Chin HW,
Maruyama Y, Meacham WF: Reoperation for
glioblastoma. J Neurosurg 55:917-921, 1981.

Downloaded from https://academic.oup.com/neurosurgery/article-abstract/34/3/409/3771249

by Bla The Librarian, user
on 30 May 2018
 Table 1. Classification of Neurosurgical Operations

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.
Table 2. Analysis of Wound Infection after Surgical
Proceduresa

Table 3. Incidence of Postoperative Sepsis with

Various Criteria

Downloaded from https://academic.oup.com/neurosurgery/article-abstract/34/3/409/3771249

by Bla The Librarian, user
on 30 May 2018
 Table 4. Analyses of Sepsis in Contaminated Cases

Table 5. Analyses of Infections in Clean

Contaminated Category

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.
Table 6. Analyses of Sepsis in Clean with Foreign
Body Category

Table 7. Analyses of Clean Cases

Downloaded from https://academic.oup.com/neurosurgery/article-abstract/34/3/409/3771249

by Bla The Librarian, user
on 30 May 2018
 Table 8. Analyses of Bacterial Isolates Based on
Intrinsic Pathogenicity in the Various Categories (27)

Redistribution of this article permitted only in accordance with the publisher’s copyright provisions.
Table 9. Influence of Concurrent Infectionsa

Downloaded from https://academic.oup.com/neurosurgery/article-abstract/34/3/409/3771249

by Bla The Librarian, user
on 30 May 2018