Christopher C. Moore
Barry M. Farr
Healthcare-associated infections (HAIs) of the central nervous system (CNS) are a rare but serious occurrence in the modern hospital. As with other types of HAIs, infection of the CNS most often follows a procedure that provides access for microbes to bypass normal host barriers. While a majority of episodes follow neurosurgery (NS), other neuroinvasive procedures (e.g., lumbar puncture or placement of an epidural catheter) occasionally can infect the CNS [1,2]. Nosocomial CNS infections that are not due to microbial contamination during procedures generally affect neonates who possess an immature blood-brain barrier that may be more easily crossed during bacteremia and the immunosuppressed. Nosocomial CNS infections range from superficial surgical site infections (SSIs) as the result of neurosurgery to meningitis, meningoencephalitis, or focal suppurations including brain abscess, subdural empyema, or epidural abscess.
Incidence
Data from the Centers for Disease Control and Prevention's (CDC) National Nosocomial Infections Surveillance (NNIS) program, which were collected in 163 U.S. hospitals between 1986 and 1993, document 5.6 NS nosocomial CNS infections for every 100,000 patients discharged [3]. This rate is approximately half of what it was a quarter of a century ago (1/10,000 discharges) [4]. Meningitis is the most common CNS infection, accounting for 91% of the total, followed by intracranial abscesses in 8% and spinal abscesses in only 1%.
Somewhat higher rates of CNS infection have been observed among the immunosuppressed, ranging from 20 per 100,000 discharges for cancer patients [5] to 5–12 per 100 discharges among transplant patients [6]. Meningitis has constituted 71% of CNS infections among cancer patients, followed by brain abscess and encephalitis (making up 27% and 2% of episodes, respectively) [5]. Brain abscess appears to be more common among transplant patients (see Chapter 45), accounting for ~40% of CNS infections after heart and heart-lung transplants [7]. The highest rate of infection for a hospital service has been 45 per 100,000 discharges for the newborn nursery, according to NNIS data collected between 1986 and 1990 [8].
The incidence of CNS infection is relatively high among NS patients compared with other groups of patients. Among patients with American Society of Anesthesiology (ASA) scores <3, an operative duration <75th percentile, and a wound classification of “clean” or “clean contaminated,” infection rates in the NNIS program have been 0.56/100 craniotomies, 0.70/100 spinal fusions, and 3.85/100 ventricular shunts [9]. Among patients with higher ASA scores (i.e., greater severity of underlying illnesses), longer durations of the procedure, and/or contamination of the wound, higher rates were observed [9]. The most common infection following NS procedures has been superficial SSI, accounting for 60% of SSIs after craniotomy and 75% after laminectomy, according to NNIS data [3]. According to updated NNIS data from January 1992 through June 2004, the rate of SSI after craniotomy was 2.40/100 operations when patients had a risk index category of 2 or 3. This rate was reduced to 1.72/100 operations and 0.91/100 operations when the risk index category was 1 or 0, respectively. Ventricular shunt placement resulted in SSI in 5.35/100 operations with a risk factor index of 1–3 compared to 4.42/100 operations with a risk factor index
P.574
of 0 [10]. Meningitis is the second most common CNS infection after craniotomy, accounting for 22% of episodes, and it is the most common form of CNS infection after ventricular shunt placement, accounting for 76% of episodes.
Risk Factors
The most obvious risk factor for nosocomial CNS infection has been NS. Skin flora, which usually cannot be cultured from the operative site immediately after antiseptic preparation, regrow during the operation and can be cultured from a majority of operative sites just before closure [11]. As with other types of surgery, it is thus likely that most infections occur during the procedure while the wound is open, becoming contaminated from regrowth of the patient's own skin flora at the margins of the wound or occasionally by organisms from the operative team introduced on contaminated gloves or instruments or settling from the air into the wound. Infection at another body site also is a risk factor for infection of the NS wound. In an outbreak of meningitis due to Klebsiella spp., for example, colonization and/or infection of the respiratory or urinary tracts appeared to precede CNS infection [12]. Another study found that in 70% of NS patients with meningitis, there was antecedent or simultaneous isolation of the same organism from another body site [13]. A review of 15,200 NS procedures performed at one tertiary care center from January 1986 to December 2001 revealed an infection rate of 0.28% (35/12,980) after craniotomy and 1.20% (27/2,220) after ventriculostomy or ventriculoperitoneal shunt insertion, with an overall infection rate of 0.40% [14]. Another comprehensive review of 51,133 patients admitted to a NS service from 1993 and 2002 revealed 51 episodes of nosocomial meningitis, all of which were associated with NS intervention. Ventriculoperitoneal shunt procedures, either insertion or revision, accounted for 26 of the episodes. The next largest group consisted of patients undergoing surgery for an intracranial mass [15]. A total of 74% of bacterial meningitis in 61 patients aged 17–40 years identified in Taiwan had a postneurosurgical state as an underlying condition [16].
Factors that amplify the risk of postcraniotomy infection have included duration of the operation, external drainage, re-exploration, and operation through a paranasal sinus [17]. The risk factors contributing to the development of meningitis/ventriculitis after placement of a ventriculostomy have included intracerebral hemorrhage, other NS operations, drainage for >5 days, an air-vented system, irrigation of the system, and intracranial pressure >20 mm Hg [17,18]. The risk for infection of cerebrospinal fluid (CSF) shunts is increased by the duration of the procedure, thrombosis of the catheter, externalization of the shunt, inexperience on the part of the surgeon, and type of shunt (ventriculoatrial carrying a higher risk than ventriculoperitoneal shunting) [17]. A persistent CSF leak after surgery heightened the risk of infection 13-fold in one study [19]; concurrent infection at a remote site increased the risk of CNS infection 6-fold.
Korinek et al. prospectively evaluated every adult patient undergoing craniotomy in 10 NS units during a 15-month period. Of the 2,944 patients studied, 117 patients developed SSIs. Independent SSI risk factors were postoperative CSF leakage (odds ratio, 145; 95% confidence interval, 72–293) and subsequent operation (odds ratio, 7; 95% confidence interval, 4–12). Independent predictive risk factors were emergency surgery, clean-contaminated and dirty surgery, an operative time >4 hours, and recent NS. Absence of antibiotic prophylaxis was not a risk factor. The investigators also found that the NNIS risk index was effective in identifying at-risk patients [20].
Placement of an intracranial pressure monitor is associated with different rates of infection, depending on where the monitor is positioned. One study found a 7.5% infection rate with a subarachnoid screw, a 14.9% rate with a subdural cup catheter, and a 21.9% rate for a ventriculostomy catheter [21]. Another study found a 0.6% rate for epidural monitors, a 3.0% rate for a subdural bolt, and a 4.0% rate for intraventricular or parenchymal brain monitors [22].
In June 2002, a manufacturer of cochlear implants used to enhance the perception of sounds in patients with severe to profound hearing loss notified the Food and Drug Administration (FDA) of 15 reports of post-implantation bacterial meningitis in patients who received its implants. This led to a cohort study to determine the incidence of bacterial meningitis among children with cochlear implants and a nested case-control study to examine risk factors for meningitis. The incidence of all episodes of meningitis in the cohort was 239.3/100,000 person-years (95% confidence interval,156.4–350.6). Perioperative meningitis occurred at a rate of 2.1 episodes per 1,000 procedures. On multivariate modeling, the use of a positioner was significantly associated with meningitis (OR, 4.5; 95% CI, 1.3–17.9), as was inner-ear malformation with a CSF leak (OR, 9.3; 95% CI, 1.2–94.5) [23].
Gliadel wafers (1,3-bis [2-chloroethyl]-1-nitosurea) are approved for the treatment of malignant gliomas. These dime-sized disks contain carmustine, the primary chemotherapeutic agent used to treat glioblastoma multiforme. Initial studies reported an SSI rate of <5% with wafer insertion, but subsequent reports revealed infection rates of 15–23%. A 2003 review of 32 patients who received a Gliadel wafer identified 9 patients who developed an SSI. Among these 9 patients, there were four episodes of brain abscess, four of bone flap osteitis, two of epidural abscess, and one each of cellulitis and subgaleal abscess associated with wafer insertion [24].
Patients with head trauma are at increased risk of CNS infection, especially meningitis. CSF fistula raises the risk of infection in this population. A CSF leak was found to
P.575
be present in 13% of episodes of nosocomial meningitis after head trauma in one series [25]. Infection of the paranasal sinuses may be followed by CNS infection in these patients [26].
Premature birth also appears to be an important risk factor because neonates cared for in neonatal intensive care units have had the highest rates of CNS infection according to NNIS data [8]. This finding seems to be related to the high risk of bacteremia from critical care instrumentation coupled with an increased risk of secondary meningitis from bacteremia stemming from the neonate's immature blood-brain barrier. Immunosuppression is another important risk factor for nonsurgical CNS infection, usually due to hematogenous spread.
Etiologic Agents
Staphylococci and gram-negative bacilli accounted for almost 70% of CNS infections documented in the NNIS hospitals between 1986 and 1992 [3]. During this time, Staphylococcus aureus was the most common pathogen after both craniotomy and laminectomy, followed by coagulase-negative staphylococci. These organisms were followed by enterococci,Streptococcus spp., Pseudomonas aeruginosa, Acinetobacter spp., Citrobacter spp., Enterobacter spp., Klebsiella pneumoniae, Escherichia coli, miscellaneous other gram-negative bacilli, and yeast, each of which accounted for <10% of episodes. After shunt procedures, S. aureus remained the most common pathogen causing superficial SSIs, but coagulase-negative staphylococci were more typical causes of deeper SSIs; gram-negative bacilli were responsible for 19% of deeper SSIs related to shunts [3]. S. pneumoniae was the predominant pathogen in the cohort of children with bacterial meningitis as a complication of cochlear implants (15/24) and Staphylococcal spp. were predominantly isolated after infectious complications of Gliadel wafer insertion [23,24].
If all CNS infections are considered, coagulase-negative staphylococci were the most frequent pathogens, making up 31% of episodes compared with 27% for gram-negative bacilli, 11% for S. aureus, 18% for streptococcal spp., 4% for yeast, and 9% for others [3]. For meningitis, the most frequently encountered CNS infection, coagulase-negative staphylococci, accounted for 32% followed by gram-negative bacilli (29%), Streptococcus spp. (18%), S. aureus (10%), yeast (4%), and others (9%) [3]. For intracranial infections, gram-negative bacilli were the cause of 23% of episodes followed by S. aureus (19%), coagulase-negative staphylococci (17%), anaerobes (11%), fungi (8%), Streptococcus spp. (8%), viruses (4%), yeast (3%), and others (8%). Spinal abscess displayed a dramatically different distribution of etiologic agents: 67% were due to S. aureus and 33% were due to coagulase-negative staphylococci [3].
The largest study of nosocomial meningitis from a single hospital was conducted by Durand et al, who reviewed 197 episodes among 151 adult patients at Massachusetts General Hospital during a 27-year period [25]. These nosocomial episodes accounted for 40% of the total of 493 episodes of bacterial meningitis observed during the study period. The proportion of episodes that were HAIs increased during the 27-year period. In this study, gram-negative bacilli were most common, accounting for 38% of episodes, followed by S. aureus (9%), coagulase-negative staphylococci (9%), Streptococcus spp. (9%), Haemophilus influenzae (4%), Listeria monocytogenes (3%), and Enterococcus spp. (3%) [25]. The microbes responsible for nosocomial gram-negative meningitis in this study were E. coli (30%), Klebsiella (23%), Pseudomonas (11%), Acinetobacter (11%), Enterobacter (9%), Serratia(9%), Citrobacter (4%), Proteus (2%), coliform types (2%), and nonenteric types (2%) [25]. The higher proportion of gram-negative and lower proportion of staphylococcal isolates in this study than in the more recent NNIS data could be due to the fact that there were only two years of overlap between the 27-year study and the NNIS data.
In Wang's review of 15,200 operative NS procedures from 1986 to 2001, the most frequently isolated pathogen was S. aureus (13/62, 21%); 91% of episodes involved a single pathogen with coagulase-negative Staphylococcus (7/62, 11%), Pseudomonas aeruginosa (5/62, 8%), Escherichia coli (5/62, 8%) and Acinetobacter baumannii (4/62, 6%) following S. aureus in frequency [14]. A review of S. aureus meningitis in Denmark over a 16-year period (1984–1999) identified 45 episodes of meningitis and 5 episodes of brain abscesses. Forty-four of these episodes were HAIs and only 6 were community acquired. None of the isolates was methicillin resistant, and 6 were penicillin susceptible [27].
Pathogens most frequently isolated from brain abscesses have been streptococci, Enterobacteriaceae, and anaerobes, together constituting ≤70% of episodes. Among immunocompromised patients, the frequency distribution of etiologic agents is somewhat different; Toxoplasma gondii and Cryptococcus neoformans are most frequent in patients with the acquired immunodeficiency syndrome, Aspergillus and T. gondii are the most common after heart and heart-lung transplants, and Aspergillus and C. neoformans are the most typical after kidney and liver transplants [28]. Among bone marrow transplant recipients, fungi accounted for 92% of episodes in a recent study: Aspergillus in 58% of episodes and Candida in 33% [28]. A retrospective hospital-based epidemiology study identified 153 patients with brain abscess over a 15-year period (1986–2000). There were 103 community-acquired infections and 20 HAIs. Of the HAIs, 17 occurred in a postneurosurgical state. Overall Klebsiella pneumoniae and viridans streptococci were the two most prevalent pathogens, and the addition of S. aureus accounted for 47% of post-NS brain abscesses [29].
Outbreaks of CNS infection among neonates or in NS patients most often have involved aerobic gram-negative bacilli [10,30,31,32,33,34]. Such outbreaks sometimes have been linked to a healthcare provider carrying the organism [35]
P.576
and at other times to contaminated equipment, such as respirators [36] or a shaving brush used for preoperative hair removal [34]. A review of 30 adult patients with gram-negative bacillary meningitis found that the majority of episodes occurred in men and that E. coli was isolated most frequently [37]. Outbreaks due to gram-positive bacteria, such asStreptococcus spp. (both groups A and B), S. aureus, and L. monocytogenes, also have been reported [38,39,40,41,42,43,44,45,46]. Secondary spread of Neisseria meningitidis or H. influenzae within the hospital setting appears to be rare [45,46,47].
Clinical Manifestations and Diagnosis
Meningitis
The typical manifestations of meningitis—fever, headache, neck stiffness, and depressed level of consciousness—usually are present with nosocomial CNS infection, but the last three of these symptoms and signs also are frequently present in post-NS patients who do not have meningitis. These findings may stem from the underlying disease or the surgery. For this reason, changes in the degree of these symptoms and signs over time may be more important indications than their mere presence. Meningitis usually begins within 10 days of NS and almost always within a month. Fever is the most reliable single sign because usually it is not seen in postsurgical patients and because it is a component of almost all nosocomial meningitis episodes; 94% of patients with meningitis after NS had fever within the first day of illness in one study [13]. In Wang's review of 15,200 patients undergoing NS, 62 episodes of postsurgical meningitis occurred. Fever occurred in 54 of these patients and all but 13 patients had a disturbance of conscious state [14]. Fever also was a prominent finding in a review of 30 episodes of gram-negative bacillary meningitis (27/30) [37].
These usual clinical manifestations are more diagnostically useful indicators in nonsurgical patients with some exceptions, such as neonates and the immunosuppressed. Neonates with meningitis usually have fever but may fail to manifest other classic findings of meningitis, such as nuchal rigidity or a bulging fontanelle. Instead there may be a weak cry, decreased muscle tone, lack of movement, poor suck, diarrhea, vomiting, dyspnea, or apnea [48]. Likewise, the geriatric patient may not show classic symptoms or signs. High-dose corticosteroid therapy or severe neutropenia may significantly alter the clinical presentation of meningitis [6,49].
Because the clinical picture often is more difficult to interpret for nosocomial than for community-acquired meningitis, CSF analysis is correspondingly more important for confirming the diagnosis. Unfortunately, CSF abnormalities due to underlying disease and/or aseptic inflammation after NS can result in confusion, especially very early after surgery [50]. Administration of OKT3 has been associated with development of aseptic meningitis in transplant patients with negative culture results for bacteria, fungi, or viruses [51]. The CSF findings most predictive of nosocomial bacterial meningitis have been neutrophilic pleocytosis, with most who are affected having a CSF white blood cell (WBC) concentration >1,000/mm3 (and almost all having a CSF WBC concentration >100/mm3), >50% neutrophils, and hypoglycorrachia (usually <40 mg/dl). Hypoglycorrachia appears to be the most reliable indicator of infection in the absence of positive results on Gram's stain or culture [13]. CSF lactate has shown some promise as a marker of bacterial meningitis in post-NS patients. When a CSF concentration of 4.0 mmol/L was used as a cut-off value for the diagnosis, the sensitivity was 88%, the specificity was 98%, the positive predictive value was 96%, and the negative predictive value was 94% [54]. Recent guidelines for the management of bacterial meningitis have suggested that empiric antibiotic therapy should be initiated when CSF lactate concentrations are ≥4.0 mmol/L. However, it must be realized that there are many other reasons for an elevated CSF lactate concentration, including cerebral hypoxia or ischemia, anaerobic glycolysis, vascular compromise, and metabolism of CSF leukocytes [55].
Gram's stain confirmed the presence of a pathogen in ~50% and culture in 83% of nosocomial episodes in one large study [25]. After NS, Gram's stain proved Candida spp. to be the causative agent in 36% of 18 reported episodes; it is worth noting that Candida spp. meningitis resulted in neutrophilic pleocytosis in 62%, with CSF WBC counts ranging from 13 to 8,000/mm3 and a CSF glucose level of <40 mg/dl in only 12% [52]. False-positive results of Gram's stains reportedly have been due to organisms in stain reagents, on glass slides, in media used for swabs, or in tubes used to centrifuge the CSF [53,56,57,58,59]. Culture results often will be rendered negative within 24 hours of starting antibiotic therapy, but changes in glucose, protein, and WBCs usually take days to detect [60]. Antigen detection testing is seldom helpful in nosocomial meningitis but may be useful for detecting C. neoformans in immunosuppressed patients [49].
Foreign Body-Associated Infection
Infection of a shunt is classified by the CDCP as an HAI if it occurs <1 year of placement, although most episodes occur within the first 2 months. Pathogens such as S. aureus are associated with early onset infections while coagulase-negative staphylococci are associated with a later onset [61]. Fever is the most reliable symptom [62]. Infection of the proximal end of a ventricular shunt often results in symptoms of shunt obstruction (e.g., nausea, vomiting, or headache). Nuchal rigidity is present in one-third [61] of episodes. Symptoms and signs of distal infection of a shunt depend on the location of the tip. With ventriculoperitoneal shunts, peritonitis is the usual manifestation, but intestinal
P.577
obstruction, intestinal perforation, and intra-abdominal abscesses have each been reported. Aseptic inflammation around the distal end has resulted in development of a peritoneal pseudocyst [63]. A tunnel infection with inflammation along the catheter may be seen.
Ventriculopleural shunts may result in empyema with distal infection while ventriculoatrial catheters can be characterized by symptoms of endocarditis (e.g., lethargy and fever of several weeks' duration). In the cases blood cultures usually are positive, and nephritis may be detected by urinalysis and serum creatinine measurement. When a shunt infection is suspected, aspiration of shunt fluid is indicated for cytology, Gram's stain, and culture. The sensitivity of the Gram's stain is ~50% and of culture ~80% [3]. Nine of 10 patients with shunt infection will have >100 WBCs per mm3 of CSF. Glucose and protein determinations on CSF from a shunt have not proved useful [64].
In the study of children with cochlear implants, episodes of possible meningitis were associated with a CSF WBC concentration of 300 to 6,115 per cubic millimeter, and in all but one patient, there was a predominance of neutrophils. Nine episodes of bacterial meningitis were perioperative (occurring ≤30 days after surgery); 20 episodes were sporadic and occurred ≥30 days after surgery. Eight of these 20 patients had evidence of otitis media at presentation. In 11/15 patients with S. pneumoniae infection, meningitis was associated with bacteremia, and 1 patient had pneumonia. Two patients had received one dose of 7-valent pneumonococcal conjugate vaccine. One other child had received two doses of the same vaccine and had S. pneumoniae meningitis caused by serotype 10A, which is not included in the vaccine. Two children had meningitis caused by H. influenzae type b (Hib). One child was fully vaccinated against Hib and the other had received 3/4 recommended doses of Hib vaccine [23]. In the four patients who developed brain abscess after Gliadel wafer insertion, the abscesses were diagnosed 22–159 days after implantation. One patient had an unusual presentation that included focal neurological symptoms, an increase in seizure activity, and no symptoms suggestive of infection [24].
Brain Abscess
Brain abscess can occur after NS, sinus infection, sinus surgery, bloodstream infection, or penetrating head trauma (e.g., gunshot wounds) [65,66,67,68,69]. Headache, fever, and focal neurologic abnormalities are the most typical findings while seizures, nuchal rigidity, nausea, vomiting, and papilledema can each be seen in up to half of the patients. Computed tomography scanning or magnetic resonance imaging (MRI) may be useful in confirming the anatomic location and size of lesions. Stereotactic aspiration with computed tomography guidance can be used for therapeutic drainage and for obtaining fluid for cytologic and microbiologic stains and cultures to guide therapy.
Meningoencephalitis
Meningoencephalitis involves inflammation of brain parenchyma and the meninges. Meningoencephalitis has occurred in rare episodes as an HAI after corneal or dural transplants taken from cadavers and after NS using contaminated instruments or electrodes. Both rabies virus and the agent of Creutzfeldt-Jakob Disease (CJD) have been transmitted in this manner. The incubation period for rabies following transplantation was 1 month; for CJD, it was about 18 months [17]. Additionally, there is a report of West Nile Virus being transmitted to a patient via blood products. The patient developed extrapyramidal movement disorders, and MRI images revealed changes in the basal ganglia similar to changes noted in the setting of meningoencephalitis due to other flaviviruses. Although the patient received contaminated blood products, detailed molecular analysis determined that the virus that caused his initial infection and encephalitis was likely acquired naturally from a mosquito [70].
CJD is a rapidly dementing illness with prominent myoclonus. Hypokinesia, rigidity, nystagmus, tremor, or ataxia may each be present in >50% of patients. Seizures occur in 10–20%. The disease generally ends with coma and death after 7–9 months. Rabies begins with a prodrome of nonspecific symptoms (e.g., fever, headache, malaise, anorexia, nausea, vomiting, or diarrhea). The prodrome is followed after 2–20 days by an acute neurologic phase, which may be characterized by hyperactivity and disorientation (furious rabies) or by paralysis. Coma usually supervenes within 10 days of the onset of neurologic symptoms and may last for hours in untreated patients to months in treated patients. The average duration of coma is 7 days in untreated patients and 13 days in patients receiving intensive care. Until 2005, only three recoveries from rabies had been reported. A fourth recovery was recently documented after a phenobarbital coma was induced and the patient was treated with ketamine, midazolam, ribavirin, and amantadine [71].
Spinal Epidural Abscess
The classic stages of an epidural abscess are back pain, radicular pain, radicular weakness, and then paralysis. Other symptoms may include bowel and/or bladder dysfunction, sensory deficits, stiff neck, and altered mental status. Fever usually is present at the time of diagnosis. Laboratory evaluation of peripheral blood usually demonstrates leukocytosis and an elevated erythrocyte sedimentation rate. MRI with gadolinium-DTPA contrast allows the best delineation of an abscess in preparation for surgery, which is done rapidly to preserve or salvage cord function.
Subdural Empyema
Cranial subdural empyema can follow paranasal sinusitis, otitis media, penetrating trauma, or an NS procedure.
P.578
Symptoms generally begin with fever and headache followed by seizures, altered mental status, focal neurologic symptoms, nausea, and vomiting. Computed tomography scan or MRI can be used to differentiate empyema from brain abscess. Spinal subdural empyemas have occurred very rarely and show symptoms similar to spinal epidural abscess, but tenderness may be absent on physical exam [3].
Prognosis
CNS infections are regarded as being among the most serious because of the potentially disabling morbidity and mortality. Of 53 deaths among patients with CNS infection in NNIS hospitals between 1988 and 1993, 49 (92%) of the deaths were believed to have been caused or contributed to by the infection rather than a pre-existing illness [3]. The case-fatality rate for nosocomial bacterial meningitis in the study by Durand et al. was 35% compared with 25% for community-acquired bacterial meningitis in adults [25]. In the comprehensive review of 15,200 NS patients, the overall mortality of the 62 patients with meningitis was approximately 34% (21/62). Death was most often associated with sepsis (14/21). Of the 41 surviving patients, 19 were vegetative or had severe neurological deficits [14]. In the study of 30 gram-negative bacillary meningitis patients, death occurred in 11 patients. Inappropriate antibiotics were given in 8 patients, and all 8 patients died [37].
CSF shunt infections were associated with an attributable mortality rate of 23% in a study by Schoenbaum et al. [61]. Walters et al. confirmed the mortality associated with shunt infections and found a doubling of the case-fatality rate, a tripling of the number of additional surgical procedures, and significant prolongation of hospital stay among survivors [72]. The type of therapy appears to have an important effect on outcome. Antimicrobial therapy alone had a 36% success rate in treating shunt infections in one study compared with 65% for antimicrobial therapy and immediate shunt removal and 96% for shunt removal, antibiotic therapy, and ventricular aspirates or external drainage [73]. Mayhall et al. found a case-fatality rate of 100% among untreated patients with ventriculostomy infection [14]. Decline in cognitive ability has been documented following shunt infections [74]. One of the children with cochlear implant associated meningitis died, and three required removal of the implant [23].
Brain abscess is associated with a case-fatality rate of ~10% [65,67] and permanent neurologic side effects in almost 50% of the survivors [66,68]. Adverse prognostic factors include very young or very old age, ventricular rupture, delay of antimicrobial therapy, altered mental status at diagnosis, larger size and greater number of abscesses, or fungal or gram-negative bacillary pathogens [66,68,69]. Spinal epidural abscess was associated with a case-fatality rate of 13% in a review of seven case-series including 188 patients [75], but a more recent study of 43 patients reported only two deaths (5%) [76]. Paralysis was observed in 22% of patients in the review [75] and 20% of patients in the more recent series [76]. Intracranial subdural empyema is associated with a case-fatality rate of 20% to 30% and a high incidence of seizures and other side effects in survivors [77]. The most common CNS infection after NS—superficial SSI—has little effect on mortality but does prolong hospital stay [3].
Prognosis often is related to the specific cause of the CNS infection. In the study by Durand et al, case-fatality rates for the three most common HAI pathogen groups were 36% for gram-negative bacilli, 39% for S. aureus, and 0% (0 of 16) for coagulase-negative staphylococci [25]. The etiologic agent has an important effect on prognosis among the immunosuppressed, with case-fatality rates of 84% for gram-negative bacilli, 24% for S. aureus, and 37% for L. monocytogenes [6,78]. The use of voriconazole has improved survival in the setting of invasive aspergillosis. In one study, voriconazole improved survival at 12 weeks from 57.9% with amphotericin B treatment to 70.8% [79]. The type of underlying illness also affects prognosis in immunosuppressed patients with CNS infection. Case-fatality rates of 90% have been observed among patients with leukemia, compared with 77% for lymphoma and 59% for solid tumors of the head or spine [5,78].
Prevention
Because most nosocomial CNS infections are related to surgery, efforts to prevent these infections prominently include general measures for prevention of SSI, which is discussed in detail in Chapter 37. Such measures include strict attention to antiseptic preparation of the skin and aseptic technique, hair removal by depilatory or clipping rather than shaving, and minimizing the duration of operation while avoiding hemorrhage or creation of a CSF fistula, both of which can promote infection. Prophylactic antibiotics reduce the risk of HAI with craniotomy at least threefold. Either cefazolin or vancomycin is acceptable because most infections are caused by staphylococci [80]. Cefazolin would be preferred for hospitals with low rates of methicillin-resistant S. aureus (MRSA) infection (see Chapter 41) because high usage of vancomycin appears to select for vancomycin-resistant enterococci (seeChapter 15) within an institution [81]. In hospitals with high rates of MRSA infection, however, vancomycin would be the preferred agent. Due to concern for staphylococcal infection, current guidelines recommend the use of either cefazolin or vancomycin for a patient undergoing craniotomy [82].
For spinal surgery, antibiotic prophylaxis has not been standard because of the perception that infection rates are low without antibiotic prophylaxis. While some studies have documented rates <1% [83,84,85,86], others have observed
P.579
rates from 2.3% to 5.0% [87,88,89,90]. One study has shown significant prevention with antibiotic prophylaxis for patients undergoing lumbar laminectomy [91]. It is likely that large randomized trials with adequate statistical power would document benefit as has been demonstrated recently for two other clean surgical procedures with generally low infection rates, herniorrhaphy and breast surgery [92]. Gantz and Godofsky suggested that prophylactic antibiotics are already being used routinely for high-risk situations (e.g., spinal procedures involving fusion or prolonged operations, immunosuppressed patients, and implantation of hardware) [3].
Prevention of infection of a CSF shunt using antibiotic prophylaxis has been difficult to confirm despite 12 randomized trials. Only one of the 12 trials showed significant prevention, but there was very low statistical power in each trial. All but one of the 12 trials showed a trend toward benefit from prophylaxis, which was continued for 24–48 hours in 10 of the 12 trials. To have 80% power to show a statistically significant benefit in the mean reduction of infection in these 12 trials would require a sample size of 790, but the average sample size in the 12 trials was only 113. A meta-analysis of these trials verified a 48% relative reduction in the rate of infection and suggested that this reduction might be beneficial [93]. The infection rate in the treatment group in these trials averaged 6.8%, however, which led the authors of the meta-analysis to suggest that a different strategy (e.g., use of a catheter with antimicrobial or anti-adherence qualities) may be needed for more effective prevention.
Respiratory isolation of patients with suspected menin-gococcal meningitis in a private room (with clinicians wearing masks) until 24 hours after the start of effective therapy has been associated with only very rare episodes of transmission to other patients [47] or to health-care personnel; usually the latter form of transmission has been due to exceptional exposure to the patient's respiratory secretions (e.g., mouth to mouth resuscitation) [94,95,96]. Chemoprophylaxis of household or other very close contacts of patients with meningitis due to N. meningitidis or H. influenzae is indicated and may be arranged by the local public health department, which should be contacted promptly after admission of a patient with this disease. Eradication of the organism from the index patient before discharge may require additional therapy with rifampin because many regimens used for therapy of meningitis do not eliminate carriage [8].
References
P.580
P.581