British Journal of Anaesthesia 108 (3): 485–90 (2012)
Advance Access publication 16 December 2011 . doi:10.1093/bja/aer416
Meningitis or epidural abscesses after neuraxial block for
removal of infected hip or knee prostheses
K. Gritsenko 1,2*, D. Marcello 1, G. A. Liguori 1, K. Jules-Elysée 1 and S. G. Memtsoudis 1
1
2
Department of Anesthesiology, Hospital for Special Surgery, 535 East 70th Street, New York, NY 10021, USA
Weill Cornell Medical College, 1300 York Avenue, New York, NY 10065, USA
* Corresponding author. E-mail: [email protected]
Editor’s key points
† Infection is often
considered a
contraindication to
neuraxial anaesthesia.
† The authors conducted a
retrospective analysis of
474 patients undergoing
removal of an infected
joint prosthesis with
neuraxial anaesthesia.
† None of the patients
receiving neuraxial
anaesthesia developed
epidural abscess before
discharge from hospital.
Background. Infection, whether localized or systemic, can be a relative contraindication to
neuraxial anaesthesia. Data correlating neuraxial anaesthesia and the development of
meningitis or epidural abscess in this setting are limited.
Methods. Retrospective chart review was performed on 710 medical records of patients
admitted between 1998 and 2009 for removal of potentially infected total hip and total
knee prostheses. Ultimately, 474 patients were identified as being infected. Factors that
predisposed a patient to an immunocompromised state, and signs and symptoms of
infection in the pre-, intra-, and postoperative stages were documented. Bacteraemic
patients were reviewed for signs of neuraxial infection. The endpoint of follow-up was
development of complications before hospital discharge.
Results. All 474 patients had removal of the infected prosthesis under neuraxial
anaesthesia. Mean patient age was 65.5 yr (58% .65 yr) and mean length of hospital
stay was 21 days. Patient characteristics included concurrent disease (65%), steroid use
(5.3%), preoperative antibiotic use (50.8%), signs of inflammatory process (84%),
bacteraemia (4.2%), and documented positive intraoperative joint cultures (88%). Using
clinical standards for diagnosis of central neuraxial infection, patients developed
infectious complications (incidence of 0.6% on 95% confidence interval), although three
patients had findings attributable to anaesthesia, including epidural haematoma, psoas
abscess, and back pain.
Conclusions. Based on clinical criteria, our findings suggest that the incidence of central
nervous system infection after neuraxial anaesthesia in patients with infected hip and
knee prostheses is low after neuraxial block.
Keywords: epidural abscess; meningitis bacterial; neuraxial anaesthesia complications;
postoperative complications; prosthesis-related infections
Accepted for publication: 19 October 2011
A feared complication of joint arthroplasty is prosthesis infection with a yearly incidence of 0.5–1% for hip, 0.5–2%
for knee, and ,1% for shoulder arthroplasty.1 – 5 Removal
of an infected prosthesis is often required, followed by a
course of i.v. antibiotics. Anaesthesia for prosthesis
removal can be performed using either general or regional
technique. The presence of a joint infection (which might
be systemic at the time of presentation for removal) is a
relative contraindication to neuraxial anaesthesia based
on the concern that subarachnoid or dural puncture
could cause spread of the infection to the intrathecal or
epidural space, leading to meningitis or epidural abscess
formation.6 7
A review of the literature on the incidence of meningitis or
epidural abscess in patients with potentially localized or systemic infections8 concluded that serious central neuraxial
infections such as arachnoiditis, meningitis, or abscess after
spinal or epidural anaesthesia are rare. This conclusion was
based on a number of large-scale studies evaluating the
incidence of these complications in uninfected patients.
The authors concluded that central neuraxial blocks should
not be performed in patients with untreated systemic infections, and that patients with systemic infection can safely
undergo spinal anaesthesia provided that prior antibiotic
therapy is initiated. Only two studies, however, were included
that assessed this population,9 10 but since their review, one
other relevant study has been published.11 In all three
reports, there were no cases of epidural abscess or meningitis, but the sample sizes in these studies were 319, 531, and
46 patients, respectively.
Moen and colleagues12 reported on neurological complications occurring after neuraxial block in a very large group
& The Author [2011]. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved.
For Permissions, please email: [email protected]
BJA
Gritsenko et al.
of patients from Sweden (1 260 000 spinals and 450 000 epidurals) and found 29 cases of meningitis and 13 cases of epidural abscess. None of these patients had documented
infections before undergoing neuraxial block.
It is common practice at our institution to perform epidural or spinal anaesthesia for infected lower extremity prosthesis removal. We designed a retrospective study to test the
hypothesis that the incidence of clinically evident neuraxial
infection in this setting is increased. The primary outcome
was occurrence of an epidural abscess or meningitis. Secondary outcomes included patient and clinical factors (i.e.
co-morbidities, clinical signs and symptoms, antibiotic use)
associated with these cases.
Methods
Analysis
Follow-up
Enrolment
After IRB approval, medical records spanning a 12 yr period
(1998 –2009) were retrospectively reviewed for all patients
admitted to the hospital for the removal of a unilateral hip
and knee infected joint prosthesis. Patients were identified
utilizing ICD-9 code for prosthesis infection. Final inclusion
criteria identified patients who underwent removal of an
infected total hip or knee prosthesis because of confirmed
infection after receiving neuraxial anaesthesia. Infections
Assessed for eligibility (n=710)
records identified by
matching ICD-9 codes for
removal of joint prosthesis and
removal of infected joint
prosthesis
Lost to follow-up when
transferred to outside
institutions (n=8)
Charts analysed (n=474)
Excluded from analysis (n=0)
Fig 1 Consort flow diagram for selection of charts for review.
486
were confirmed with positive intraoperative cultures. If
patients presented with infectious symptoms (i.e. swollen
joint, fever, erythema) had negative intraoperative cultures
and documented recent antibiotic use, they were also
counted as confirmed infectious cases.
Patients who did not undergo prosthesis removal or were
considered aseptic were excluded. Aseptic patients
were defined as patients with no recent antibiotic use
before admission and negative postoperative wound
cultures. Additionally, patients who were transferred to
another institution shortly after their surgery (at most, a
day after surgery) were excluded because follow-up information regarding potential infectious complications was unavailable. Finally, patients were excluded if they did not
receive neuraxial anaesthesia. If general anaesthesia was
performed, the rationale for this anaesthetic decision was
recorded when available.
Significant co-morbidities (diabetes mellitus, prior myocardial infarction, chronic renal insufficiency, rheumatoid
arthritis, hypertension) were documented. In addition, infectious indicators including signs and symptoms of systemic
infection such as fever (institutionally defined as a temperature .37.58C), chills, hypotension (defined as a systolic arterial pressure ,100 mm Hg), and an elevated leucocyte count
228 charts excluded for
not meeting eligibility
criteria:
Lack of prosthesis
removal (n=65)
General anaesthesia
(n=80)
Aseptic joint (n=83)
BJA
Incidence of meningitis or epidural abscess
(defined as .10.7 cells nl21) and also results of intraoperative and postoperative cultures were documented. The perioperative use of antibiotics up to 6 months before
admission and characteristics of intraoperative infectious
organisms were also documented. The primary outcome of
a neuraxial infectious complication was defined as a clinical
diagnosis of meningitis or epidural abscess in the inpatient
postoperative period before discharge. Upon completion of
chart review, descriptive analysis was conducted with
Microsoft Excel and SPSS 14.0. All chart review and analysis
was performed at the Hospital for Special Surgery in New
York, NY, United States.
Results
Seven hundred and ten patients admitted to the hospital
between 1998 and 2009 were identified for the removal of
a unilaterally infected hip or knee joint prosthesis. Exclusion
criteria are described in Figure 1. The reasons for the use of
general anaesthesia included patient preference (3/80), coagulopathy (20/80), and anatomic contraindications
such as prior posterior spinal fusions or severe spinal stenosis
(9/80). In 48 cases, no clear reasons were provided by the
anaesthesiologists as to why general anaesthesia was
selected. Upon further review, these charts revealed that
no patients developed neuraxial infectious complications.
Comparison of these 80 patients with those who underwent
neuraxial block showed no major differences in terms of
patient profiles, and preoperative and postoperative infectious indicators, except for a higher incidence of fever in
the neuraxial group (P¼0.019). The remaining 474 patient
charts were included in the analysis. The patient profiles
and significant co-morbidities are presented in Table 1.
Table 1 Patient profile upon admission
Neuraxial
anaesthesia
(n5474) (%)
General
anaesthesia
(n580) (%)
ASA class: I/II/III/IV 2/215/236/14
0/19/57/4
Average age (yr)
65.5 (range: 12-95)
65.2 (range: 32-88)
Number of patients
over 65 yr old
273 (57.6)
42.0 (52.5)
Length of stay (days) 21 (range: 3-151)
22.6
Diabetes mellitus
74 (15.6)
18.0 (22.5)
Rheumatoid
arthritis
49 (10.3)
10.0 (12.5)
Steroid use
25 (5.3)
5.0 (6.3)
Hypertension
253 (53.4)
36.0 (45)
Chronic renal
insufficiency
10 (2.1)
0 (0)
Myocardial
infarction
Average BMI
Obese (defined as
BMI .35)
29 (6.1)
9 (11.3)
Localized infection was confirmed in all 474 patients by
one of the following determinants: positive intraoperative
joint culture (87%, 414/474), positive Gram stain (56%, 264/
474), and joint pus expressed within the surgical site (50%,
236/474). Patients with negative cultures were included if
there was recent antibiotic use that could explain the negative joint culture results (9.3%, 44/474; 19/44 had positive
Gram stains, 11 of whom had joints that expressed pus; in
8/44 cases, joint pus was the only additional confirmation
of infection). In addition, 17% of the confirmed infected
patients were infected with two or more organisms. The
most common organisms that grew out of intraoperative
joint cultures are listed in Table 2. The pre- and postoperative
infectious indicators are presented in Tables 3 and 4. Preadmission antibiotics were started in 241 (51%) of patients.
Bacteraemia was noted in 4.2% (20/474) of patients. Eighteen of these patients received antibiotics in the operating
theatre and one patient who did not receive an intraoperative antibiotic had taken antibiotics before operation.
Patients received spinal, epidural, combined spinal/epidural,
or general anaesthesia with neuraxial anaesthesia
(Table 5). Of note, two patients were converted to general anaesthesia after unsuccessful attempts to insert a spinal
catheter.
Of 474 patients evaluated, none had a clinically evident
epidural abscess or meningitis (incidence of 0.6% on 95% confidence interval). Documented postoperative complications
included one patient each with epidural haematoma, psoas
abscess, and back pain. The epidural haematoma occurred in
a 43-yr-old woman with no significant medical history or
pre-admission antibiotic and steroid use. Upon admission,
she was afebrile and had a normal leucocyte count
and normal coagulation. After receiving a combined spinal/
Table 2 Most common intraoperative culture results for patients
receiving neuraxial anaesthesia. *One patient’s joint culture grew
both MRSA and MRSE
n
Gram positive
5
Methicillin-sensitive S. aureus (MSSA)
110
Methicillin-resistant S. aureus (MRSA)
7
Methicillin-resistant Staphylococcus
epidermidis (MRSE)
16
Staphylococcus epidermidis
110
Coagulase-negative Staphylococcus species
51
Enterococcus faecalis
38
Streptococcus viridians
19
Propionibacterium acnes
13
Gram negative
Pseudomonas aeruginosa
29.6
91 (19.2)
28.9
13.0 (16.3)
450* (total)
Staphylococcus unspecified
Escherichia coli
Other organisms
Candida albicans
53 (total)
17
8
3 (total)
3
487
BJA
Gritsenko et al.
Table 3 Preoperative infectious indicators. SD, standard deviation
Neuraxial
anaesthesia
(n5474) (%)
General
anaesthesia
(n580) (%)
33 (41.3)
P-value
241 (50.8)
Joint swelling
362 (76.4)
42 (52.5)
0.69
Joint erythema
129 (27.2)
9 (11.3)
0.59
0.25
Fever
75 (15.8)
4 (5.0)
0.02
Chills
34 (7.2)
5 (6.3)
0.76
9 (1.9)
3 (3.8)
0.51
Average white
blood cell count
(cells nl21)
8.93 (SD ¼3.43)
9.26 (SD ¼3.93)
General
anaesthesia
(n580) (%)
P-value
Puss found in joint 236 (49.8)
26 (32.5)
0.26
Positive
postoperative
joint cultures
414 (87.3)
66 (82.5)
0.79
Positive blood
culture
20 (4.2)
13 (16.3)*
0.24
Back pain
1 (0.2)
0 (0)
Headaches
0 (0)
0 (0)
n/a**
Epidural
haematoma
1 (0.2)
0 (0)
0.69
Meningitis
0 (0.2)
0 (0)
n/a**
0.69
Epidural abscess
0 (0)
0 (0)
n/a**
Psoas abscess
1 (0.2)
0 (0)
0.69
epidural anaesthesia, the patient underwent removal of an
infected total hip. Her joint did not express pus and no antibiotics were used intraoperatively. Operative wound cultures
grew Clostridium septicum. Several days later, after complaining of severe back pain, postoperative MRI revealed an epidural
collection, with compression of the thecal sac. The INR at the
time of diagnosis was 1.8 (secondary to warfarin therapy).
The patient was taken to the operating theatre for neurosurgical decompression of the lumbar spine and the haematoma
was evacuated; intraoperative cultures were negative. Five
days after the procedure, repeat MRI showed no significant
reaccumulation of the haematoma.
The patient who experienced a psoas abscess was an
84-yr-old female with a history of rheumatoid arthritis and
myocardial infarction. She was transferred from another
488
%
63.9
Epidural
58
11.8
Spinal
87
18.4
Failed neuraxial (convert to GA)
2
0.42
GA combined with spinal
7
1.5
GA combined with epidural
GA combined with CSE
5
1.1
10
2.1
0.51
Table 4 Postoperative infectious indicators. *For two patients,
intraoperative blood culture only confirmation of infection. Note
that both had preoperative aspirations that yielded organisms,
but only one-half used antibiotics before operation. **For n/a: no
P-value calculated due to standard deviation¼0
Neuraxial
anaesthesia
(n5474) (%)
n
305
Combined spinal/epidural (CSE)
Pre-admission
antibiotic use
Hypotension
(systolic AP
,100 mm Hg)
Table 5 Anaesthesia type for patients receiving neuraxial
anaesthesia. GA, general anaesthesia.
hospital with fever and an elevated leucocyte count (12.1
cells nl21). Five days before transfer to our institution, she
had developed a retroperitoneal bleed while on warfarin
therapy (INR 2.0). She did not report chills or joint swelling,
and had not used steroids or antibiotics. She underwent
spinal anaesthesia for removal of her total knee replacement,
as her INR had normalized by the day of surgery. Intraoperative cultures, including blood, were positive for methicillinsensitive Staphylococcus aureus. Concern for a potential
abscess precipitated by back pain initiated a work up, including brain and spinal CT and MRI. These tests revealed only a
psoas abscess, which was surgically drained and resolved.
The patient who reported back pain was a 48-yr-old
female with hypertension and a long history of chronic
lower back pain secondary to lumbar spinal stenosis. Upon
hip prosthesis removal using combined spinal/epidural
anaesthesia, pus was noted and joint cultures grew
methicillin-resistant S. aureus. Evaluation of her spine with
MRI was negative.
Discussion
The findings suggested by this retrospective analysis cautiously indicate that the incidence of in-hospital central
nervous system infection after administration of neuraxial
anaesthesia in patients with an infected hip or knee prosthesis is exceedingly low. This finding is consistent with the
few series published in patients with distant systemic infections undergoing neuraxial blockade. Bader and colleagues9
evaluated 319 parturients with chorioamnionitis, eight of
whom had documented bacteraemia, and none developed
a spinal infection. Goodman and colleagues10 looked at
517 patients with the same pathology (13 of whom had bacteraemia), and again, none of the patients developed a
spinal or epidural abscess. More recently, Kotzé and colleagues11 studied the incidence of infectious complications
after thoracic epidural anaesthesia in 50 paediatric
patients with empyema, and none developed a central
nervous system infection. The number of patients enrolled
in our study is relatively small, but similar to these prior
reports. It was important to study this patient population
since they are generally older, sicker, and represent a group
of patients which may become more prevalent in the
Incidence of meningitis or epidural abscess
future as the incidence of joint replacement surgery is significantly increasing.
We had two major complications: a psoas abscess and
epidural haematoma. There was one infectious complication
in the setting of retroperitoneal haematoma and bacteraemia, leading to psoas abscess. This occurred after the
patient had a spinal anaesthetic placed via the midline approach. Psoas abscess has been described after epidural anaesthesia. Lee and colleagues13 reported a case occurring 6–
7 months after epidural anaesthesia, accompanied by vertebral osteomyelitis and discitis. In the light of the time lag
between the procedure and the event and the fact that cultures grew Tuberculosis mycobacterium, they were thought to
be unrelated. There was also a case report of iliopsoas
abscess after removal of an indwelling epidural catheter
that had been implanted for 13 days.14 Alpantaki and colleagues15 also reported a case of an epidural and psoas abscess
occurring 1 month after epidural anaesthesia. Vertebral
osteomyelitis and epidural and psoas haematomas were
diagnosed simultaneously, which might have been a continuation of the same process.16
As demonstrated by these case reports, psoas abscess can
occur in the setting of neuraxial anaesthesia. Given that our
patient had a retroperitoneal haematoma with concomitant
bacteraemia, this is the most likely cause. A spinal source,
however, cannot be ruled out. Psoas abscess has also been
reported to occur from a local or contiguous source or from
haematogenous dissemination.16 The epidural haematoma
occurred 5 days after epidural anaesthesia with normal
coagulation parameters. When the haematoma was diagnosed, INR was elevated due to warfarin treatment. Moen
and colleagues12 found a higher incidence of spinal haematoma in elderly patients compared with younger obstetric
patients. Of note, 11 of the 33 patients with haematoma had
a coagulopathy and documented thromboprophylaxis
therapy.
Preoperative inclusion criteria were based on identification
of patients with documented active infection requiring prosthetic joint removal. Criteria associated with bacterial
seeding and spread of infection include comorbid diseases,
immunocompromised states related to steroid use, HIV,
age .65, concurrent infection, and markers of acute inflammation based on perioperative localized signs and symptoms, physical exam, and cultures.6 8 Approximately 50%
of patients in this study had pus on removal of the joint,
and 88% had positive intraoperative Gram stains that provided evidence of potential bacterial seeding; several
patients also had microbiological evidence of infection
based on preadmission joint aspirations. Subsets of patients
had documented bacteraemia (4.2%) or steroid use (5.3%).
Given no documented neuraxial complications, no specific
conclusion can be drawn regarding association between
comorbidities or immunocompromised states and the
incidence of infectious complications.
Antibiotic use was analysed in the pre-, intra-, and postoperative periods. Approximately 50% of patients had antibiotic use before admission. Interestingly, patients who
BJA
received antibiotics before operation still grew organisms in
their intraoperative cultures (29%) and/or had positive
Gram stain results (27%), which could indicate improper antibiotic coverage, insufficient length of treatment, resistant
species, non-compliance, or multiple infectious entities.
Thus, the potential for infectious spread, and thus infectious
complications, potentially grew more likely. This raises the
question whether pre-, intra-, or postoperative antibiotic
use might be a factor in preventing infection or specifically,
to prevent meningitis or epidural abscess formation. The
American Society of Regional Anesthesia recommendations
are that patients with evidence of systemic infection can
safely undergo spinal anaesthesia, provided antibiotic
therapy has been initiated with a positive response
to therapy. Approximately 16% (75/474) of our patients presented with fever, and 63% (47/75) of these patients used
antibiotics preadmission, potentially indicating inadequate
response to therapy. These data, although limited, add to
the literature of this controversial issue.
The development of meningitis or an epidural abscess can
take from several hours to several days, respectively.12 In this
study, the average length of hospital stay was 21 days, with a
range of 3–151 days. Although the observation period was
likely adequate to capture all relevant endpoints, it remains
a limitation that some patients could have developed a
central neuraxial infection after discharge.
At our institution, patients undergoing removal of infected
joints usually undergo spinal (short procedure) or combined
spinal/epidural anaesthesia (anticipated long surgery),
unless the INR is above 1.2. A total of 80 patients from the
study underwent general anaesthesia. In approximately
50% of these cases, the exact reason for choosing general
over a neuraxial technique could not be determined from
the chart. Hence, the unavailability of the decision process
by which the anaesthesiologist chose general over regional
anaesthesia could represent a possible selection bias. This
subgroup of patients arguably could represent more severe
cases of systemic infection. However, these patients were
similar to those who underwent neuraxial anaesthesia in
terms of patient profile, pre- and postoperative infectious
indicators, except for a higher incidence of fever in the neuraxial anaesthesia group.
A final limitation is that the primary endpoint (epidural
abscess or meningitis) was not formally evaluated in all
patients. Specifically, radiographic studies and cerebrospinal
fluid samples were not obtained for all subjects. Therefore, it
is possible that a subclinical abscess or meningitis might
have been present but not detected. Given the pathophysiology of these morbidities, subclinical presentation is possible,
but unlikely.
In conclusion, the results of this limited case series lend
support to the use of neuraxial anaesthesia in patients presenting for removal of infected hip or knee prosthesis. Based
on a sample of only 474 patients, issues of safety cannot be
assured. However, as the number of joint replacement procedures are increasing, these data should be useful in assisting
with the anaesthetic management of this population.
489
BJA
Declaration of interest
None declared.
Funding
This work was supported by the Department of Anaesthesiology Research and Education Fund, Hospital for Special
Surgery, New York, NY, USA.
References
1 Matthews PC, Berendt AR, McNally MA, Byren I. Diagnosis and
management of prosthetic joint infection. Br Med J 2009; 338:
1378– 83
2 Baddour LM, Sexton DJ. In: Sexton DJ, ed. Treatment of Prosthetic
Joint Infections. UpToDate, 2011
3 Widmer AF. New developments in diagnosis and treatment of
infection in orthopedic implants. Clin Infect Dis 2001;
33(Suppl. 2): S94– 106
4 Berbari EF, Hanssen AD, Duffy MC, et al. Risk factors for prosthetic
joint infection: case– control study. Clin Infect Dis 1998; 27:
1247– 54
5 Sperling JW, Kozak TK, Hanssen AD, Cofield RH. Infection after
shoulder arthroplasty. Clin Orthop Relat Res 2001; 382: 206– 16
6 Darouiche RO. Spinal epidural abscess. N Engl J Med 2006; 355:
2012– 20
490
Gritsenko et al.
7 Berman R, Eisele J. Bacteremia, spinal anesthesia, and development of meningitis. Anesthesiology 1978; 48: 376–7
8 Horlocker TT, Wedel DJ. Regional anesthesia in the immunocompromised patient. Reg Anesth Pain Med 2006; 31: 334– 45
9 Bader AM, Gilbertson L, Kirz L, Datta S. Regional anesthesia in
women with chorioamnionitis. Reg Anesth 1992; 17: 84 –6
10 Goodman EJ, DeHorta E, Taguiam JM. Safety of spinal and epidural anesthesia in parturients with chorioamnionitis. Reg
Anesth Pain Med 1996; 21: 436– 41
11 Kotzé A, Hinton W, Crabbe DC, Carrigan BJ. Audit of epidural
analgesia in children undergoing thoracotomy for decortication
of empyema. Br J Anaesth 2007; 98: 662– 6
12 Moen V, Dahlgren N, Irestedt L. Severe neurological complications
after central neuraxial blockades in Sweden 1990–1999. Anesthesiology 2004; 101: 950– 9
13 Lee BB, Ngan Kee WD, Griffith JF. Vertebral osteomyelitis and
psoas abscess occurring after obstetric epidural anesthesia. Reg
Anesth Pain Med 2002; 27: 220– 4
14 Aota Y, Onari K, Suga Y. Iliopsoas abscess and persistent radiculopathy: a rare complication of continuous infusion techniques of
epidural anesthesia. Anesthesiology 2002; 96: 1023– 5
15 Alpantaki K, Papoutsidakis A, Katonis P, Hadjip Avlou A. Vertebral
osteomyelitis, epidural and psoas abscess after epidural catheter
use. Acta Orthop Belg 2007; 73: 670–3
16 van den Berge M, de Marie S, Kuipers T, Jansz AR, Bravenboer B.
Psoas abscess: report of a series and review of the literature.
Neth J Med 2005; 63: 413– 6