Healthcare-Associated Central Nervous System Infections
Jeffrey M. Tessier
W. Michael Scheld
Healthcare-associated infections related to the central nervous system (CNS) are a relatively infrequent but important category of hospital-acquired infections. These infections span a spectrum from superficial wound infections, to ventricular shunt infections, to deep-seated abscesses of the brain parenchyma. The patient populations affected are equally diverse, involving neonates, children, and adults, with occurrence on nearly all medical and surgical services.
Healthcare-associated infections of the CNS are usually serious, if not life threatening, and are frequently associated with a poor outcome (1, 2, 3 and 4,5,6, 7, 8, 9, 10, 11, 12 and 13). These healthcare-associated infections present many challenges in diagnosis, and many controversies exist regarding effective prophylaxis and proper management. In addition, the identification of a particular infection as healthcare-associated may not be clear-cut; thus, overlaps and ambiguities concerning acquisition are unavoidable. Fortunately, a heightened awareness has fostered declining rates of infection. In spite of improving techniques and new preventive strategies, the threat is constant, and the stakes remain painfully high. The first part of this chapter focuses on the clinical and epidemiologic aspects of infections related directly to neurosurgical and neuroinvasive procedures as well as infectious processes that invade the CNS secondarily from other sites. The second part of this chapter discusses prevention and control of these infections.
RISK FACTORS
General Risk Factors
Not surprisingly, the patients at greatest risk for acquiring healthcare-associated CNS infections are neurosurgical patients. Patients with surgical site infections (SSIs) are drawn almost entirely from this population. These patients are subjected to procedures that traverse the scalp, violate meningeal coverings, impinge upon the paranasal sinuses, implant foreign bodies, and expose tissues to hematogenous sources of infection. Infection in this setting is often facilitated by the presence of a cerebrospinal fluid (CSF) leak that occurs when the dura is disrupted and the subarachnoid space communicates with the skin, nasal cavity, paranasal sinuses, or middle ear (14, 15, 16, 17, 18 and 19). This group includes adult and pediatric patients undergoing common neurosurgical and neuroinvasive procedures such as craniotomy, spinal fusion, laminectomy, insertion of halo pins, burr hole placement, and implantation of ventricular shunts and reservoirs. Less common procedures include stereotactic brain biopsy, hypophysectomy, paranasal sinus surgery, acoustic neuroma resection, temporary ventricular drainage, placement of intracranial monitoring devices, nerve stimulator placement, lumbar puncture, spinal anesthesia, myelography, and skull/spinal fixation.
Patients who have suffered accidental head trauma are another population at increased risk to develop meningitis. These individuals have sustained trauma or fractures to the basilar skull and facial bones, facilitating the formation of a CSF fistula. This posttraumatic condition substantially increases the likelihood of CNS infection, particularly bacterial meningitis (20, 21 and 22). In one series, a CSF leak was a predisposing factor in approximately 9% of cases of healthcare-associated bacterial meningitis (5).
The majority of healthcare-associated CNS infections reported from the National Nosocomial Infections Surveillance (NNIS) system at the Centers for Disease Control and Prevention (CDC) occurred in newborn nurseries and on surgical services (Table 27-1). All other hospital services account for a small but still substantial number of cases. Patients from this smaller population generally have a parameningeal source of infection that is either contiguous (e.g., sinusitis) or occult (e.g., unsuspected CSF leak), reactivation of latent infection, or an infection that has hematogenously seeded the CNS from a distant site. Patients with malignancies (especially lymphoma and leukemia), organ transplants, and other immunocompromised hosts frequently fall into this last category.
Risk factors for SSIs can be classified into host factors and surgical factors. Examples of host factors include age, sex, American Society of Anesthesiologists (ASA) physical status classification, underlying diseases such as diabetes mellitus, nutritional status, presence of other remote infections, and duration of preoperative stay. Surgical factors include whether the procedure was emergent or elective, hair removal technique, surgeon, use of perioperative antibiotics, duration of surgery, type of operation, site of surgery, and whether gloves were punctured (23) (see Chapter 21 on SSIs.) One study showed that when
patients underwent a neurosurgical procedure, the presence of a postoperative CSF leak was associated with a 13-fold increase in the infection risk (24). Also, a non-CNS concurrent infection increased the infection risk six times, whereas use of perioperative antibiotics was associated with a decrease in the infection rate of about 20%. Three other risk factors—paranasal sinus entry, placement of a foreign body, and use of postoperative drains—were associated with an increased risk of infection, although these associations were not statistically significant. Factors not associated with an increased risk of infection included obesity, surgical reexploration, use of the operative microscope, steroid administration, and acute therapy for seizures. Length of surgery was also not a factor associated with an increased risk of infection. A prospective study of postoperative neurosurgical infections demonstrated a validated five-category classification system for neurosurgical infections based on specific definitions. It found that infection rates were highest for contaminated cases (contamination known to occur, 9.7%), followed by dirty cases (established sepsis at the time of surgery, 9.1%), clean-contaminated (risk of contamination of operative site during surgery, 6.8%), clean with temporary or permanent foreign body (6.0%), and clean (no identifiable risk factors present, 2.6%). In this study, surgery lasting longer than 4 hours was associated with an infection rate of 13.4% (25).
patients underwent a neurosurgical procedure, the presence of a postoperative CSF leak was associated with a 13-fold increase in the infection risk (24). Also, a non-CNS concurrent infection increased the infection risk six times, whereas use of perioperative antibiotics was associated with a decrease in the infection rate of about 20%. Three other risk factors—paranasal sinus entry, placement of a foreign body, and use of postoperative drains—were associated with an increased risk of infection, although these associations were not statistically significant. Factors not associated with an increased risk of infection included obesity, surgical reexploration, use of the operative microscope, steroid administration, and acute therapy for seizures. Length of surgery was also not a factor associated with an increased risk of infection. A prospective study of postoperative neurosurgical infections demonstrated a validated five-category classification system for neurosurgical infections based on specific definitions. It found that infection rates were highest for contaminated cases (contamination known to occur, 9.7%), followed by dirty cases (established sepsis at the time of surgery, 9.1%), clean-contaminated (risk of contamination of operative site during surgery, 6.8%), clean with temporary or permanent foreign body (6.0%), and clean (no identifiable risk factors present, 2.6%). In this study, surgery lasting longer than 4 hours was associated with an infection rate of 13.4% (25).
TABLE 27-1 Healthcare-Associated CNS Infections by Hospital Service in NNIS Hospitals 1986 to 1992 | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
In addition to neonates (see Chapter 52) and patients undergoing neurosurgery, patients undergoing invasive diagnostic or therapeutic procedures that penetrate the CNS are at risk for developing a healthcare-associated CNS infection (see Section VIII). A subgroup of neurosurgery patients at high risk for healthcare-associated CNS infections includes those with ventricular shunts. Since most shunt infections (70%) have an onset within 2 months of surgery, it is likely that the infecting microorganism is introduced during surgery or in the postoperative period (10). Risk factors for shunt infections are discussed in Chapters 49 and 65. The rate of infection varies with the neurosurgeon (26,27). The efficacy of prophylactic antibiotics in preventing shunt infection is controversial and is discussed below (see Prevention). Patients undergoing diagnostic or therapeutic procedures that penetrate the CNS, such as the installation of dyes or drugs, are more likely to develop healthcare-associated meningitis (28). Although such infections occur infrequently in the present era, they should be considered in the appropriate setting.
Device-Related Risk Factors
Infection is a well-recognized complication of ventriculostomy catheters used for monitoring and drainage (29). Aucoin et al. (30) noted that the rate of infection was associated with the type of monitor used. The lowest infection rate was associated with the subarachnoid screw (7.5%), followed by a rate of 14.9% for the subdural cup catheter and a 21.9% rate for the ventriculostomy catheter. An intracranial monitoring technique, the Camino intraparenchymal fiberoptic catheter system, is associated with an infection rate of 2.5% (31). The method of ventriculostomy insertion using the tunneled technique has been associated with the lowest rates of infection (29). Use of prophylactic antibiotics did not reduce significantly the risk of infection. In a study by Mayhall et al. (9) of ventriculostomy-related infections, risk factors significantly associated with infection included an intracerebral hemorrhage with intraventricular hemorrhage, a neurosurgical operation, ICP of 20 mm Hg or higher, ventricular catheterization for longer than 5 days, and irrigation of the system. The incidence of infection was not related to insertion location when the intensive care unit was compared with the operating room. Infection rates were also not reduced by the use of nafcillin prophylaxis. Several additional studies have confirmed the direct relationship between the duration of ventricular catheters and infection risk (32, 33, 34, 35 and 36). Additional risk factors associated with ventriculitis include sepsis, pneumonia, urinary tract infection, depressed skull fracture requiring surgery, craniotomy, CSF leakage around the device, drain blockage, reinsertion related to catheter malfunction, and intraventricular hemorrhage. To reduce the risk of ICP monitor-related infections, it is recommended that the device be inserted using aseptic technique, that the device be removed as soon as possible and preferably before 5 days, and that a closed system be maintained. A randomized, controlled trial of external ventricular drainassociated infection compared regular exchange of the drain every 5 days with clinically indicated exchanges and found no difference in the rate of infection between these groups (37). The use of prophylactic catheter exchange and extending the duration of catheterization to 10 days has been proposed, but more data are needed (29). The type of ICP monitor device used influences the rate of infection, with epidural tunneled monitors having the lowest rates.
SOURCES OF INFECTION
Sources of Infecting Microorganisms
Nonsurgical Infections Healthcare-associated CNS infections can be classified into those infections unrelated to surgery and postsurgical infections. In patients with
nonsurgical-related infections, the microorganisms can compose a patient’s endogenous flora, such as coagulasenegative staphylococci (CoNS), or arise from an exogenous source, such as from a contaminated solution or device (28). Gram-negative bacilli are usually responsible for infections related to contaminated solutions or devices (38). Microorganisms can gain access to the CSF by hematogenous spreading of an infectious agent, spread to the CSF from contiguous foci, such as an infected sinus, or via a communication of the CSF with the flora of the skin, sinuses, or other mucosal surfaces (39,40). CSF leakage can be obvious in a patient with rhinorrhea or otorrhea, or occult if the subarachnoid space communicates with a paranasal sinus. Rarely, neoplasms erode into the subarachnoid space and produce a fistula. Microorganisms can also gain access to the CSF by direct inoculation of the agent in a patient having a lumbar puncture, especially if a substance is injected. Microorganisms acquired in this manner are usually gramnegative rods (41,42). It is extremely unusual to develop meningitis following a lumbar puncture unless a solution is injected into the CSF.
nonsurgical-related infections, the microorganisms can compose a patient’s endogenous flora, such as coagulasenegative staphylococci (CoNS), or arise from an exogenous source, such as from a contaminated solution or device (28). Gram-negative bacilli are usually responsible for infections related to contaminated solutions or devices (38). Microorganisms can gain access to the CSF by hematogenous spreading of an infectious agent, spread to the CSF from contiguous foci, such as an infected sinus, or via a communication of the CSF with the flora of the skin, sinuses, or other mucosal surfaces (39,40). CSF leakage can be obvious in a patient with rhinorrhea or otorrhea, or occult if the subarachnoid space communicates with a paranasal sinus. Rarely, neoplasms erode into the subarachnoid space and produce a fistula. Microorganisms can also gain access to the CSF by direct inoculation of the agent in a patient having a lumbar puncture, especially if a substance is injected. Microorganisms acquired in this manner are usually gramnegative rods (41,42). It is extremely unusual to develop meningitis following a lumbar puncture unless a solution is injected into the CSF.
Infection is a well-recognized complication of chronic epidural catheters and intracerebroventricular devices (43) used for control of pain in patients with AIDS or malignancy (44) (see Chapter 60). Hayek et al. studied patients with noncancer pain and found a higher infection rate (5.51 infections per 1,000 catheter-days) among patients using tunneled epidural catheters (TECs) for neuropathic pain compared to those with TECs used to treat somatic pain (2.43 infections per 1,000 catheter-days) (45). Staphylococci accounted for 11 of 23 positive epidural space or catheter tip cultures, supporting the hypothesis that most of the TEC infections were due to skin flora migration and colonization of these catheters. Other complications include meningitis and epidural abscess, but prolonged surgery during catheter placement has been found to be the only factor associated with catheter infection (46). Infection may also complicate the use of an Ommaya reservoir (47). Repeated access of these devices may permit colonizing skin flora such as Staphylococcus aureus, S. epidermidis, or diphtheroids to produce ventriculitis and meningitis. The source of the infecting microorganisms may also be the hands of the hospital personnel accessing the device, although powder contamination from gloves has also been implicated (48).
Neurosurgical Infections Although many sources of contamination of a neurosurgical operation have been described, it is usually impossible to document with certainty the source for a given SSI. Probably most infections occur at the time of surgery from either direct inoculation of residual flora of the patient’s skin or from contiguous spread from infected host tissue. Direct inoculation of microorganisms can also occur occasionally from the hands of surgical team members via a tear in a glove. Rarely, the source of infection is traced to contaminated surgical material such as a solution, device, or instrument. In two neurosurgical patients with postoperative Bacillus cereus meningitis, the source of the microorganisms was found to be heavily contaminated linen (49). Occasionally, during the postoperative period, an SSI results from direct inoculation of microorganisms. Airborne contamination at the time of surgery, either from the patient or from operating room personnel, accounts for some neurosurgical infections (1,50). Lastly, a postoperative infection rarely results from hematogenous seeding of a wound from an infected intravenous line or other remote infection.
INCIDENCE AND DISTRIBUTION
Healthcare-associated infections of the CNS (excluding wound or SSIs) are relatively uncommon, accounting for approximately 0.4% of all healthcare-associated infections (R. Gaynes, personal communication to Nelson Gantz). Meningitis accounts for 91% of these infections, followed by intracranial suppurations (8%) and isolated spinal abscess (1%) (R. Gaynes, personal communication to Nelson Gantz). When infection rates are examined using data reported from 163 hospitals participating in the NNIS system, 0.56 CNS infections per 10,000 hospital discharges occurred from 1986 through early 1993 (R. Gaynes, personal communication to Nelson Gantz). Comparable rates over the past 25 years have shown a slow decline from approximately one infection per 10,000 hospital discharges to the present lower rate (54). While these numbers are relatively small, it must be noted that CNS infections directly related to neurosurgical procedures (SSIs) are not reflected in these numbers. The majority of healthcare-associated CNS infections occurring in this setting are designated under the larger category of SSIs (22% of all healthcare-associated infections) by the CDC National Healthcare Safety Network (NHSN) system surveillance criteria (see below) (401). Certain healthcare-associated CNS infections may represent a greater proportion of specific types of infection. For example, a retrospective study of acute bacterial meningitis in adults over a 27-year period at the Massachusetts General Hospital found 40% of 493 total episodes to be healthcareassociated in origin (5).
Healthcare-associated surgical site and CSF infections among neurosurgical patients are a primary focus of this chapter. Table 27-2 shows the distribution of SSIs complicating neurosurgical procedures and illustrates the significant proportion of deep infections that occur in relation to the surgical site; these data are derived from the NNIS reporting period 1986-1992. Infection rates as reported in the general neurosurgical literature are often difficult to interpret and compare for a variety of reasons, including differences in definitions, methodology, reporting techniques, and use of prophylactic antibiotics. Not uncommonly, postoperative infections unrelated to the surgical site or CNS are included in the rate calculation (2). An overview of infection rates associated with neurosurgery from some of the more rigorously performed (although nonstandardized) studies over the last 30 years is shown in Table 27-3. Taking into account some of the problems mentioned above, most hospital series report infection rates of <5%. When individual neurosurgical procedures are compared, differences in infection rate become more apparent. The incidence of all CNS infection following typically clean craniotomy may vary from <1% to nearly 9%,
whereas the rates following laminectomy range from 0.6% to 5%. Postoperative meningitis after clean craniotomy has a reported incidence of 0.5% to 2% when perioperative antibiotics are given (55,56,72, 73 and 74). Without antibiotic prophylaxis, other studies have found rates ranging from 2% to 7% (74, 75 and 76). A more recent large prospective study of infections after craniotomy among 2,944 patients found an overall SSI rate of 4%, with meningitis representing approximately 48% of these infections (77).
whereas the rates following laminectomy range from 0.6% to 5%. Postoperative meningitis after clean craniotomy has a reported incidence of 0.5% to 2% when perioperative antibiotics are given (55,56,72, 73 and 74). Without antibiotic prophylaxis, other studies have found rates ranging from 2% to 7% (74, 75 and 76). A more recent large prospective study of infections after craniotomy among 2,944 patients found an overall SSI rate of 4%, with meningitis representing approximately 48% of these infections (77).
TABLE 27-2 Surgical Site Infections Following Neurosurgical Procedures | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Infection rates for selected neuroinvasive procedures are shown in Table 27-4. Again, differences in methodology, definition, and duration of follow-up greatly affect the reported rates. Analysis of infection rates following ventricular shunt surgery is particularly complex. Depending on the use of a case rate (occurrence per patient) or an operative rate (occurrence per procedure) of infection and the duration of follow-up, an extremely wide variation in incidence may be seen. Perhaps, when in 1916 Cushing (107) stated, “There has never been any infection, even of a stitch in the scalp, in something over 300 cranial operations in the writer’s series,” he underestimated the situation. A procedure-oriented risk factor analysis is covered in a later section, and additional details are discussed elsewhere in this text (see Chapters 49, 60, and 65).
TABLE 27-3 Infection Rates in Selected Neurosurgery Trials | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
TABLE 27-4 Infection Rates in Selected Neuroinvasive Procedures | |||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| |||||||||||||||||||||||||||||||||||||||||
Examination of SSIs reported from NNIS system hospitals between 1992 and 2004 shows infection rates in uncomplicated procedures with minimum risk factors to be 0.91/100 operations for craniotomies, 1.04/100 operations for spinal fusion, 0.88/100 operations for laminectomies, and 4.42/100 operations for ventricular shunts (108) The last rate is the third highest among all operative procedures (108). These surveillance rates, by definition, include both superficial and deep infections related to the operative site (109,110). The addition of one or more complications (surgical risk factors) will increase most of the figures to varying degrees (111).
The incidence of both community- and hospital-acquired CNS infections in immunocompromised hosts has been estimated to range from <1% to over 10%, depending on the host population (112, 113, 114 and 115). Classic studies at the Memorial Sloan-Kettering Cancer Center in the early 1970s revealed an incidence of CNS infections approximating 0.02% of total hospitalizations (116). These infections occurred most commonly in lymphoma patients (33%), followed by neurosurgical patients (30%) and leukemic patients (20%). Overall, meningitis accounted for the majority of infections (71%), followed by brain abscess (27%) and encephalitis (2%). Of note, intracerebral abscess in leukemic patients was responsible for 70% of CNS infections in this group. It has been postulated that conventional incidence figures may significantly underestimate the actual magnitude of CNS infections in this population (112). Other studies have shown similar patterns in cancer patients, with perhaps a higher incidence of CNS infection in transplant recipients estimated at 5% to 12% (114,115). One retrospective study of bone marrow transplant recipients found symptomatic neurologic complications, predominantly infectious (23% of complications), among 16% of patients (117). CNS infections were more common among allogeneic compared to autologous transplants and included cerebral toxoplasmosis, viral encephalitis, and fungal infections. Brain abscess was found to be a common complication in one study of heart and heart-lung transplant recipients, accounting for 35% to 44% of CNS infections (113,118). These abscesses are often caused by fungi, particularly Aspergillus species, among liver transplant recipients (119). Bacterial meningitis in the febrile neutropenic patient is often indolent in presentation and masked by the early use of broad-spectrum antibiotics. Disseminated fungal infections are not uncommon in the compromised host and are frequently difficult to diagnose; Candida is reported to involve the CNS in up to 50% of cases (120,121). Although the absolute number of healthcare-associated infections in this population cannot easily be determined, the proportion is likely to be high, as many occur after multiple or prolonged hospitalizations and are caused by typical healthcare-associated pathogens.
TYPES OF HEALTHCARE-ASSOCIATED CENTRAL NERVOUS SYSTEM INFECTIONS
Healthcare-associated infections related to the CNS may be broadly divided into two major categories (Table 27-5): postsurgical infections and nonsurgical infections, including those related to neuroinvasive or neurodiagnostic procedures. The first category consists of SSIs (109). Infections of this type may occur following craniotomy, ventriculostomy, and spinal column surgery. Rarely, SSIs complicate other neurosurgical operations, such as peripheral nerve surgery and carotid endarterectomy. SSIs are further classified as superficial or deep incisional SSIs, using the fascial plane as divider. Deep surgical infections unrelated to soft tissues are classified as organ/space SSIs by the aforementioned CDC criteria (109). These infections may present as a local and/or diffuse infectious process. Local suppurative infections complicating neurosurgical procedures include the following: parenchymal brain abscess, subdural empyema, epidural abscess,
discitis, subgaleal collection, and osteomyelitis of the cranium or spine. Diffuse infection of the subarachnoid space defines meningitis or ventriculitis if the process is related to a prior ventriculostomy and essentially remains localized. This latter distinction is somewhat arbitrary. Meningoencephalitis is an infrequent diffuse healthcareassociated CNS infection generally due to prions or viruses transferred during neuroinvasive procedures or via organ transplantation (122, 123, 124, 125, 126, 127 and 128).
discitis, subgaleal collection, and osteomyelitis of the cranium or spine. Diffuse infection of the subarachnoid space defines meningitis or ventriculitis if the process is related to a prior ventriculostomy and essentially remains localized. This latter distinction is somewhat arbitrary. Meningoencephalitis is an infrequent diffuse healthcareassociated CNS infection generally due to prions or viruses transferred during neuroinvasive procedures or via organ transplantation (122, 123, 124, 125, 126, 127 and 128).
TABLE 27-5 Healthcare-Associated CNS Infections | ||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||
Nonsurgical infections constitute a smaller, but equally important, class of healthcare-associated CNS infections. These infections are acquired by a variety of routes that include spread from a contiguous focus, posttraumatic/CSF leak, and neuroinvasive procedures, as well as hematogenous spread. Meningitis, brain abscess, subdural empyema, and epidural abscess all may occur in this setting.
DEFINITIONS, DIAGNOSTIC CRITERIA, AND CLINICAL PRESENTATION
It is essential for the purposes of identification, surveillance, and management that healthcare-associated infections be defined and diagnosed with as much sensitivity and specificity as possible. Unfortunately, factors such as colonization and aseptic inflammation prevent the establishment of gold standards and place many conditions within a spectrum of disease. Recognition of an infection as healthcare associated is often not straightforward, and CNS infections are no exception. Doubt over hospital versus community acquisition of an infection is a constant problem compounded by the ubiquity of the major pathogens. The time course that defines specific healthcareassociated infections is neither consistently defined, easy to determine, nor universally accepted. Although the CDC outlines strict definitions and diagnostic criteria, the length of hospitalization prior to an infection being classified as healthcare associated is not specified, with the exception that such infections should not be considered HAIs if they are present or incubating at the time of admission to the acute care setting (109). For SSIs related to implantable devices, healthcare-associated infection may be diagnosed up to 1 year after surgery, according to CDC criteria (109). Some experts consider 60 days a more reasonable length of time for healthcare-associated ventricular shunt infections, as the majority of infections occur within this period (129). In addition, the diagnosis of infection ultimately may be left to the discretion of the attending physician and is inherently subjective. A prospective study by Taylor et al. (130) demonstrated that 40% of neurosurgical wound infections were diagnosed using nonstandardized criteria by the surgeon. The potential effect on infection rates is obvious. Ventricular shunt infections illustrate several of these problems. CSF profiles may be nondiagnostic, the microorganism involved may be from the normal flora, and the infection may become evident weeks after hospital discharge. This section integrates the CDC definitions with additional clinical criteria to facilitate proper identification and diagnosis of healthcare-associated infections related to the CNS. The CDC surveillance definitions for healthcare-associated surgical site and specific CNS infections have been previously published (109).
Surgical Site and Related Surgical Infections
Studies dealing with SSIs in neurosurgical patients have used a variety of both strict and less stringent diagnostic criteria for identification (2,4,12,25,57,58,131, 132, 133, 134, 135 and 136). Commonly, these infections are classified in the surgical literature as either superficial or deep. Superficial neurosurgical infections are considered to be limited by the cranial or lumbodorsal fascia. Deep wound infections encompass soft tissue infections below the fascia, including discitis, osteomyelitis, and bone flap infections. However, infections below the dura (ventriculitis, meningitis, brain abscess) have been included under this heading as well (24,56,131). To improve surveillance and clarify potential overlap in reporting, the CDC definitions include the category of organ/space SSI to cover additional sites adjacent to the operative site. Specific organ/space SSIs related to neurosurgery include the following: meningitis, ventriculitis, disc space infection, osteomyelitis, intracranial abscess, and spinal abscess (109). With the exception of infections related to implantable devices, infection occurs within 30 days of the operative procedure. Since the organ/space SSI category includes several non-soft tissue infections, the definitions are relatively liberal. Diagnosis of some of these infections is covered in subsequent sections, as they also occur unrelated to surgical procedures. More detail on SSIs in general may be found elsewhere in the text (Chapter 21).
Incisional Surgical Site Infections
From a practical point of view, the diagnosis of incisional SSIs is usually made clinically. Neurosurgical site infections must be promptly identified because of the propensity to spread to deeper spaces (137). Superficial incisional SSIs tend to be diagnosed at an early stage, usually within the first postoperative week (59,138,139). Generally, the area is swollen and erythematous with local tenderness. Purulent discharge and/or microorganisms isolated from drainage or a wound aspirate complete the picture. Temperature and the white blood count (WBC) are not uniformly elevated; the erythrocyte sedimentation rate (ESR) and CRP may be increased (2,136). Deep incisional SSIs present later postoperatively with a course that may be insidious or progressive. The average time between surgery and the diagnosis of a deep infection in spinal surgery may vary from 10 to 15 days, with the range extending several weeks (136,140). A relatively normal appearance of the overlying surgical site contributes to this delay in many cases (140). Elevations of temperature, WBC, ESR, and CRP, as well as the presence of fever/chills or hyperglycemia in diabetic patients, while clearly nonspecific signs, are not infrequently seen (136,138,140). Patients often complain of increased pain at the surgical site (141).
Infections of bone flaps following craniotomy are well described and account for up to one half of infections following this procedure (1,56,60,142), though a more recent large series described bone flap osteitis in 12% of postcraniotomy SSIs (77) By definition, infection involves either the free (devitalized) or the osteoplastic bone flap following a supratentorial craniotomy. These infections may be obviously symptomatic with high fever, scalp tenderness, and suppuration (4,143) or more indolent with a persistent fistula (2). In one series, 12 of 13 bone flap
infections were diagnosed within 30 days of surgery (139); Korinek found a median time to diagnosis of bone flap osteitis of 27 days (77). Sequential nuclear scanning with technetium 99 may have enhanced diagnostic accuracy for cranial flap osteomyelitis, especially to rule out this infection (143). Indium 111-labeled leukocyte scanning is a useful technique (144,145). Plain skull radiographs are helpful, if positive, but lack sufficient sensitivity to be useful routinely (60). The use of magnetic resonance imaging (MRI) is invaluable in establishing a diagnosis, while CT findings are nonspecific and may not help establish a diagnosis of infection (146). In general, a cranial bone flap infection is diagnosed clinically with either radiographic or microbiologic confirmation (4). A subgaleal abscess occasionally occurs adjacent to a scalp surgical site. In this case, a localized collection forms in the space between the galea of the scalp and the pericranium. Scalp tenderness, erythema, fever, and regional adenopathy may be seen. Osteomyelitis or intracranial spread of infection can occur secondarily if the underlying skull integrity has been compromised. Diagnosis of most deep incisional SSIs may be established clinically, via culture of a deep aspirate, or, rarely, with the assistance of radiologic studies. Evaluation of a soft tissue fluid collection with sonography or CT scan can be helpful.
infections were diagnosed within 30 days of surgery (139); Korinek found a median time to diagnosis of bone flap osteitis of 27 days (77). Sequential nuclear scanning with technetium 99 may have enhanced diagnostic accuracy for cranial flap osteomyelitis, especially to rule out this infection (143). Indium 111-labeled leukocyte scanning is a useful technique (144,145). Plain skull radiographs are helpful, if positive, but lack sufficient sensitivity to be useful routinely (60). The use of magnetic resonance imaging (MRI) is invaluable in establishing a diagnosis, while CT findings are nonspecific and may not help establish a diagnosis of infection (146). In general, a cranial bone flap infection is diagnosed clinically with either radiographic or microbiologic confirmation (4). A subgaleal abscess occasionally occurs adjacent to a scalp surgical site. In this case, a localized collection forms in the space between the galea of the scalp and the pericranium. Scalp tenderness, erythema, fever, and regional adenopathy may be seen. Osteomyelitis or intracranial spread of infection can occur secondarily if the underlying skull integrity has been compromised. Diagnosis of most deep incisional SSIs may be established clinically, via culture of a deep aspirate, or, rarely, with the assistance of radiologic studies. Evaluation of a soft tissue fluid collection with sonography or CT scan can be helpful.
Organ/Space Surgical Site Infections
Discitis (infection of the intervertebral disc space) is a relatively uncommon but potentially serious postoperative complication of spinal surgery (147, 148, 149, 150 and 151). The fact that almost 20 years of surgery passed before this infection was recognized illustrates the difficulties encountered in diagnosis (152). Patients typically present with worsening back pain and muscle cramping 1 to 8 weeks after surgery and initial improvement of preoperative symptoms (58,61,153,154). In a series of 111 cases of discitis described by Iversen et al. (155), back pain appeared at an average of 16 days postoperatively. Occasionally, overt infection occurs immediately after surgery (61,156). Patient examination may disclose pain with lumbar range of motion, paraspinal muscle spasm, and/or an abnormal straight leg raising test (58,61,151,157). Neurologic deficits are unusual and should raise suspicion for an epidural abscess. Fever is variably present, and the superficial surgical site frequently appears normal. Most notable is the severe and persistent low back pain out of proportion to the findings on physical examination. Routine laboratory studies such as the WBC are generally unremarkable, with the exception of the ESR (61,155,157). Following spine surgery, the ESR rises rapidly (peak 90-110) and falls steadily to near-normal levels within several weeks (158,159). A significantly elevated ESR more than 2 weeks postoperatively correlates positively with disc space infection (62,158, 159 and 160). Others have found this test less valuable, especially with early infections (155). CRP, an acute-phase reactant, may be useful as a diagnostic tool when followed serially in patients postoperatively. A prospective study of 348 consecutive patients undergoing spinal surgery had CRP measured on days 1, 3, and 5 postoperatively; these values demonstrated a characteristic increase and fall in 96% of patients experiencing a benign clinical course, with mean values of 14.9, 15.4, and 7.9 mg/dL on days 1, 3, and 5, respectively (161). However, 4.6% (16 patients) displayed an abnormal CRP response with a second increase, and five of these patients were ultimately diagnosed with a postoperative spinal infection, though none with diskitis. The sensitivity, specificity, positive, and negative predictive values for the abnormal CRP response in this patient population were 100%, 96.8%, 31.3%, and 100%, respectively.
Several radiographic modalities are helpful in establishing the diagnosis of discitis. Plain films are of little utility in the early weeks, as most decreases in disc height are expected postoperatively. More characteristic findings occur weeks to months later with blurring of the end plate and irregularity and lytic destruction of the subchondral surface (162). Osteomyelitis of the adjacent vertebrae may occur in advanced cases. These findings are visualized in greater detail with CT scans (163). Currently, MRI with gadolinium enhancement has become the procedure of choice for the so-called failed back syndrome following spinal surgery (164). Early changes on MRI may distinguish disc space and vertebral body infection from the normal postoperative spine with a high degree of accuracy (154,165, 166, 167 and 168). Nuclear imaging is of limited value because of the high level of background positivity (62). Sequential technetium 99 and gallium 67 scans improve sensitivity but require at least 48 hours to complete (169). Although somewhat controversial, diagnosis of infectious discitis should be confirmed by biopsy despite a consistent clinical and radiographic picture. Tissue sampling allows discrimination between septic and aseptic (chemical or avascular discitis) processes and facilitates directed antibiotic therapy. Peripheral blood cultures are rarely positive for the offending microorganism (58,170). Percutaneous needle aspiration of the affected disc space under fluoroscopic or CT guidance is the method of choice. Ideally, antibiotics should be withheld until after the procedure is complete. The results of the Gram’s stain and/or culture are diagnostic in up to 70% of cases, and histologic examination may indicate a septic picture in cases lacking positive microbiology (157,160).
Isolated vertebral osteomyelitis is very uncommon following laminectomy and related procedures. When present, it is usually associated with progressive infection of the contiguous disc space (spondylodiscitis) (154, 170, 171 and 172). Clinical presentation and diagnosis are virtually the same as outlined above for discitis.
Meningitis
The diagnosis of healthcare-associated meningitis requires a high index of clinical suspicion and support from CSF analysis. Excluding ventricular shunt infections, most cases of meningitis following neurosurgery are diagnosed in the early postoperative period. Several series have shown that the majority of cases develop within 10 days of surgery, and virtually all are diagnosed within 28 days (3,6,7,60,72,173). Healthcare-associated meningitis unrelated to surgical procedures has a more variable time course. Posttraumatic bacterial meningitis associated with a CSF leak may occur days to years after the initial injury (21,174). Although some of these infections may develop in the hospital, acquisition of the infecting microorganism likely has occurred in the community environment (21,22,175). Since the CDC definitions do not specify a period during or after hospitalization that distinguishes healthcare-associated
from community-acquired infection, evidence for hospital acquisition must be sought (109). In a review of 197 episodes (157 patients) of healthcare-associated meningitis by Durand et al. (5), 97% of patients were diagnosed more than 48 hours after admission or within 1 week of discharge (5). Interestingly, 41 episodes (10 patients) in this study were recurrent during the same hospitalization. Other studies indicate a similar pattern of presentation (7,21). We consider it reasonable to view nonsurgical healthcareassociated meningitis as developing several days after hospitalization and unrelated to an obvious communityacquired infection. Unfortunately, these distinctions are not always easy to make.
from community-acquired infection, evidence for hospital acquisition must be sought (109). In a review of 197 episodes (157 patients) of healthcare-associated meningitis by Durand et al. (5), 97% of patients were diagnosed more than 48 hours after admission or within 1 week of discharge (5). Interestingly, 41 episodes (10 patients) in this study were recurrent during the same hospitalization. Other studies indicate a similar pattern of presentation (7,21). We consider it reasonable to view nonsurgical healthcareassociated meningitis as developing several days after hospitalization and unrelated to an obvious communityacquired infection. Unfortunately, these distinctions are not always easy to make.
The standard clinical signs and symptoms suggestive of meningitis are often of little help in diagnosing healthcare-associated infection. Fever appears to be the most ubiquitous finding in all healthcare-associated cases (3,6,7,72,173). Neurosurgical patients commonly demonstrate an altered level of consciousness, neck stiffness, and headache reflecting some combination of their underlying disease and the surgical procedure itself in the absence of infection. These relatively nonspecific findings may become more useful if a change over time is noted or a new fever develops. Findings indicative of meningeal irritation are more useful in nonsurgical patients, especially when combined with fever and a change in mental status. Aseptic meningeal inflammation is a common postoperative condition that may further confound the diagnosis. Clinical parameters have been consistently unable to distinguish aseptic from bacterial meningitis (176,177). The use of corticosteroids may blunt the signs and symptoms of inflammation in both surgical patients and compromised hosts (114,178). Neutropenic hosts cannot mount an inflammatory response, and the resultant symptoms are often minimal (178). Low-grade fever, lethargy, and/or headache may be the only clues in these patients (115). Concurrent medical conditions or extremes of age often modify the typical clinical presentation (6,179,180). Finally, the administration of perioperative antibiotics may alter the natural course of clinical responses and laboratory findings (see below).
The signs and symptoms of posttraumatic bacterial meningitis are often similar to those seen in acute bacterial meningitis (181). However, as with the neurosurgical patient, clinical findings may be more difficult to interpret in the patient with considerable head trauma. CSF infection should be considered when there has been any change in neurologic status, or when fever or neck stiffness is noted that was not present initially (21,182). For these patients at increased risk, it is important to establish evidence of CSF leakage when meningitis is a concern. In a retrospective study of 860 patients with moderate-to-severe head trauma, 12 (1.39%) developed meningitis, with 58% of these patients presenting with clinically apparent rhinorrhea (183). The most common signs of a CSF leak are rhinorrhea, otorrhea, hemotympanum, Battle’s sign (mastoid ecchymosis), and cranial nerve palsies (22,184). Detection of CSF rhinorrhea is critical and may be performed at the bedside using a glucose oxidase reagent strip to detect increased glucose in nasal secretions, with the caveat that blood, especially when visible in the nasal fluid, may produce a falsely positive test (185). Unfortunately, a negative result does not rule out the presence of a fistula (186). Identification of beta(2)-transferrin in nasal secretions using immunofixation or electrophoresis has shown promise as a useful indicator of CSF leakage (187,402, 403 and 404). A fluorescein dye test can also be used to identify suspected cases of CSF otorrhea and localize the source (188). Radiographic studies are the procedures of choice to document and localize CSF leakage. CT scanning and MRI are superior to plain films in diagnosing basilar skull fractures and identifying fistulae (189,190). Radioisotope cisternography using 111In- or technetium-99m-labeled diethylenetriamine pentaacetic acid (DTPA) is highly sensitive, but specificity is a problem and localization is poor (191,406). A combination of different imaging modalities may be required to accurately localize the site of a CSF leak (405); high-resolution CT combined with a fluorescein injection study may offer the best characteristics currently (191). Considering the diagnostic subtleties associated with healthcare-associated meningitis, examination of the CSF assumes a critical role.
Analysis of CSF obtained from hospitalized patients at risk for developing meningitis is often difficult. Neurosurgical patients commonly have abnormal CSF profiles secondary to underlying disease (tumor), procedures, intracranial bleeding, and seizure activity. Perioperative antibiotics will influence the results of cultures of CSF. Nonsurgical patients are likely to be receiving concurrent antibiotics for other infections. Compromised patients may have blunted inflammatory reactions or abnormal CSF profiles from noninfectious processes (e.g., carcinomatous or leukemic meningitis). Despite these limitations, the results are often revealing, and examination of the CSF should be performed routinely in all suspected cases (407).
The CDC definition for healthcare-associated meningitis does not specify abnormal values for routine CSF parameters. As with community-acquired bacterial meningitis, most cases of healthcare-associated meningitis are associated with an increased CSF white cell count, neutrophilic pleocytosis, elevated protein, and depressed glucose (2,3,5,7,20,72,137,148,181,192,193). Neurosurgical patients with culture-proven meningitis generally have more than 100 WBCs/mm3 with over 50% neutrophilia (7,72,137,173). In the series by Berk and McCabe (173), all patients were noted to have over 100 WBCs/mm3, with the majority having more than 1,000 cells/mm3 (median 2,500). In 72 episodes of culture-negative healthcare-associated meningitis described by Durand et al. (5), 97% of patients had more than 300 WBCs/mm3, and 96% had more than 50% neutrophils. Since an intracerebral bleed or a subarachnoid hemorrhage allows both WBCs and RBCs to enter the CSF, a correction formula may be used to better approximate the number of abnormal white blood cells (194). A CSF protein level > 100 mg/dL and a glucose level <40 mg/dL are present in the majority of healthcare-associated cases (5,7,72,137,173,177). Unfortunately, several studies have found no significant difference in cell counts and other CSF parameters in (early) postoperative patients with septic versus aseptic meningitis (176,177). In these patients, a significantly lowered glucose level (<20 mg/dL) might be the best indicator of an infectious etiology in the absence of culture data (5). The administration of muronomonab (OKT3) to organ transplant recipients during rejection has been associated with the development of aseptic meningitis (195,196).
Routinely, the CSF should be Gram-stained and set up for bacterial culture. In immunocompromised patients, fungal, mycobacterial, and viral studies may be indicated as well. The yield on Gram-stained CSF is lower than in community-acquired cases and approximates 50% overall (5,78). Although a positive culture remains the gold standard, it is impossible to make this requirement for healthcare-associated cases if the clinical data and CSF profile are otherwise supportive. In one large retrospective study, a positive culture was obtained in 83% of healthcare-associated cases and a comparable percentage of communityacquired cases (5). Since concurrently positive cultures are often obtained from sites outside the CNS, cultures from blood, adjacent wounds, and urine are suggestive in the appropriate setting (3,7,60,63,72,137,173).
Clearly, the diagnostic value of CSF sampling, under any circumstance, can be greatly influenced by the administration of intravenous antibiotics. The effect of antibiotics prior to lumbar puncture is most marked on the Gram’s stain and culture with little alteration of the other standard parameters (197,198). A negative Gram’s stain and culture will commonly occur after 24 hours of appropriate therapy (199). The CSF glucose and white cell count usually remain abnormal for at least several days (194). When combined with the baseline abnormal CSF of the craniotomy patient or the tempered inflammatory reaction of the neutropenic host, the effect of prior antibiotics on diagnosis is substantial, and second-line tests assume greater importance. Latex agglutination to detect the capsular polysaccharide of Cryptococcus neoformans is a highly efficacious test in immunocompromised patients (115,178). Broad-range polymerase chain reaction (PCR) to amplify the 16S ribosomal RNA sequences specific to bacterial pathogens offers a promising avenue to supplement Gram’s stain and bacterial culture, particularly in patients who have received antimicrobial therapy prior to CSF sampling (200).
Final mention should be made concerning the role of neuroimaging in the diagnosis of bacterial meningitis. Although contrast enhancement of meninges may be seen on CT or MRI early in the course of illness, these findings are nonspecific and contribute little to establishing the diagnosis (164). A better use of these modalities is to exclude other CNS pathology or to diagnose intracranial complications of meningitis (201).
CEREBROSPINAL FLUID SHUNT INFECTIONS
A variety of temporary and permanent prosthetic devices are used to access, drain, divert, and monitor the CSF. These devices may be internalized for chronic use or externalized for use in the acute setting. Internalized devices consist of shunts (ventriculoperitoneal, ventriculoatrial, ventriculoureteral, lumboperitoneal), and reservoirs (lumbar, ventricular). Externalized devices facilitate drainage (ventriculostomy, lumbar drain, external shunt) or measure ICP when the device (intraventricular, epidural, subdural) is connected to a transducer. Insertion of a ventriculoperitoneal shunt is the most common surgical procedure performed for the long-term control of hydrocephalus. Infections complicating these devices may occur at any site or compartment traversed by the prosthesis. Proximal infections include meningitis, ventriculitis, empyema, abscess, and infection involving the surgical site (wound infection, cellulitis, osteomyelitis). Distal infections include tunnel infections along the catheter tract, bacteremia, pleuritis, peritonitis, and related intra-abdominal infections. Infections of temporary devices are almost always healthcare associated, because their insertion and use requires hospitalization. The current CDC guidelines define infection secondary to an implantable device as healthcare associated if it occurs within 1 year of the operative procedure and the two appear to be related (109). Such a designation must often be based subjectively on the type of infection, clinical setting, and responsible microorganism. Because of the clustering of shunt infections within 60 days of implantation (10,64,79,80,202,203), shorter periods have been suggested to designate a shunt infection as healthcare associated (129). Because of the considerable overlap among infections of different CNS prosthetic devices, this discussion can focus on the diagnosis of CSF shunt infections as the prototype for this group. Certain specific infections potentially related to CSF shunts have already been covered in detail earlier in this chapter (SSIs) or are covered in later sections (intracranial suppurations).
The most important risk factor for the development of CNS shunt infection is the level of training of the neurosurgeon, with neurosurgical trainees having a higher rate of infection (27). Variables such as year of placement of the shunt, age of the patient, length and time of the operation, and exact placement of the distal drain do not increase the risk of infection (202,204). Additionally, elevated CSF protein content does not appear to increase the risk of shunt infection (205).
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
