Introduction

Normal pressure hydrocephalus (NPH) was classically defined by Solomon Hakim in his seminal thesis in 1964 and then later more formally with Raymond Adams in 1965 with the clinical triad of psychomotor retardation, gait apraxia and imbalance, and urinary incontinence [1,2]. When an identifiable risk factor such as inflammation, hemorrhage, infection, or traumatic brain injury has been implicated in the diagnosis of NPH it is often referred to as secondary NPH as opposed to idiopathic NPH (iNPH), the term used when no clear etiology is identified. A working diagnosis for NPH is made when then this constellation of symptoms is associated with clinical signs and radiographic evidence of ventricular system enlargement in the context of normal cerebrospinal fluid (CSF) opening pressures [3].

CSF diversion using the most common methods of ventriculoperitoneal or ventriculoatrial shunting has emerged as a useful tool in the management of this condition, carrying the potential for dramatic improvement in quality of life and reversal of the debilitating symptoms that so often characterize this disease process [3–5]. Several systematic reviews have reported improvement of symptoms in 30–50% of patients with idiopathic NPH and 50–70% of those with secondary NPH, underscoring the rising practice of CSF diversion in this disorder [5–7]. With increasing improvements in diagnostic testing and shunting equipment, the success rates for shunting procedures seem to have increased over time [8]. In addition, the incidence and prevalence of normal pressure hydrocephalus appear to be increasing as well. The incidence of shunt-responsive NPH has been reported in the past to be around 2.2 per million persons per year. With improvements in preoperative evaluations, heightened awareness, and an aging population, it is thought to be increasing [9]. Brean et al., in 2008, suggested that in Vestfold County, Norway, NPH had a prevalence of 21.9, and an incidence of 5.5 per 100 000 individuals based on the numbers of patient referrals with the appropriate diagnostic and clinical criteria. In addition, they noted that the highest prevalence of patients identified in their study was found in the 70- to 79-year-old age group (181.7/100 000), and second highest in the over 80 age group (93.3/100 000) [10,11]. While recognizing that this is a fairly homogeneous local population, it should be noted that those age groups are currently, and are projected to continue to be, among the fastest growing segments of the US population, based on the 2010 Census Data [12].

Due to the elderly demographic in whom this syndrome has been most often described, particularly between the sixth and eighth decades of life, establishing an accurate diagnosis distinguishing the syndrome from other potential comorbid medical conditions in the differential remains challenging. Ancillary tests such as high volume CSF spinal taps, external continuous lumbar drainage trials, and CSF outflow resistance studies have provided some tools with positive predictive values as to who will benefit from CSF shunting [13]. Unfortunately these tests have also been demonstrated to have low negative predictive values, and thus likely to fail to identify some patients who actually may benefit from shunting [14,15]. Meanwhile, neurosurgeons who offer a CSF diversion procedure to potential NPH patients are doubly challenged to evaluate the risk-to-benefit ratio of the procedure from the standpoint of the consequences of shunt-related complications associated with the procedure itself [9]. There is controversy surrounding the potential for complications from shunting. That, coupled with the variable results following CSF diversions which have been noted historically, has led some to believe that the overall morbidity rate for shunting NPH patients may even outweigh the benefits [9]. In addition, many studies have limited the analysis of risk and benefit to the early period following surgery, within the first months following the procedure. There are, however, important contributions that attempt to include the long-term results and possible complications following shunting. This chapter hopes to compile, compare, and place into context the incidence and the relevance of shunt-related complications, with special emphasis on infections. The chapter will review the types of late complications, their likelihood, and some clinical examples. We review here the late infection rates and identify key pathogens and associated risk factors that predispose to shunt infections. And lastly we review current strategies aimed at reducing the likelihood of shunt complications and shunt infection rates and discuss interventions at patient follow-up that can assist in maximizing benefit from shunt procedures, and help in maintaining symptomatic relief in patients following shunt surgery for NPH.

Types of shunt complications

CSF shunting and particularly ventriculoperitoneal shunting is a common procedure used to treat a variety of CSF flow dynamic pathologies including obstructive, communicating, and normal pressure hydrocephalus from a variety of etiologies. As with any treatment modality, and especially with any surgical procedure, there are inherent potential risks to treatment, which in the case of shunt surgery include infection, possible shunt-induced complications, and failure of the therapy to provide relief. In general there are several types of specific long-term complications associated with shunting procedures in addition to those more likely seen in the NPH population in particular. These include underdrainage of CSF, overdrainage of CSF, mechanical failures of the hardware such as alterations in valve resistance over time, malposition of the tubing, and frank breakage, CSF leak, subdural hemorrhage or hygroma, seizure, and infection.

Underdrainage and overdrainage

The symptoms of underdrainage tend to be those of return of the NPH triad. With either underdrainage or mechanical failure, the identification of these patients occurs when their symptoms return after a period of initial improvement following shunting. Patients who demonstrate initial improvement in gait, urinary continence, and cognition, who then deteriorate with return of their initial symptoms complex either partially or completely, are to be suspected of mechanical failure or underdrainage. Underdrainage conditions arise when shunt systems develop increased resistance across the valve over time, disconnection or kinking of the system components, or migration of the tubing out of an appropriate position for proper drainage and/or reabsorption. With the development of variable resistance valves that allow adjustment over time, patients are able to have their diversion rates changed based on symptoms of over- and underdrainage. It is not clear that any of these opportunities have decreased the need for subsequent revisions in this patient population over time as the revision rates remain fairly consistent across the studies over the years, but they may be associated with an improvement in general in the success rate in maintaining patients’ symptomatic improvement [5,8].

The symptoms of overdrainage are also well described. Patients may complain of positional headaches or dizziness, imbalance, visual difficulties, or auditory changes. This clinical situation may need to be managed by changing the resistance in the shunt system, either by revision or adjustment. Adjustment of a programmable valve to a higher resistance, or revision of the valve system to a higher pressure, antisiphon device, or a valve with a decreased fixed flow rate may be warranted. Overdrainage does put patients at an increased risk for subdural hematoma or hygroma formation. Patients with these conditions present with new onset of headaches, unilateral or bilateral, mental status changes, motor or sensory impairments, or seizures. Imaging studies, either MRI or CT, can document the occurrence of extra-axial fluid collections. Management of the acute subdural hygroma/hematoma can be quite challenging. Acute intervention may be required and may run the gamut from simple adjustment of the shunt resistance, revision or occlusion of the shunt system, to burr hole evacuation, craniotomy and evacuation, and rarely complete craniectomy followed later by cranioplasty. Staged procedures may be necessary to complete the recovery of the more complicated clinical situations. Close follow-up is likely required with serial CT scans and/or MRI scans to assess for recovery and to gauge the advisability and timing of returning the patient back to a shunt in working order.

Imaging

Evaluation of the CSF diversion system may be accomplished by a combination of imaging and laboratory analyses. X-rays of the tubing system assess for position and continuity. A radiolucent gap along the path of the shunt is an indication of a possible disconnect. CT scans of the brain assess for shunt position and associated extra-axial fluid collection intracranially such as subdural hematoma or hygroma. CT scan of the abdomen will evaluate for tubing location, the possibility of intra-abdominal pseudocyst formation within the peritoneum, a clue to shunt infection, and extraperitoneal pseudomeningocele and/or coiling of the tubing suggestive of malposition of the catheter (as in Figure 18.1). CSF analysis via either a shunt reservoir aspiration or via a lumbar puncture may be needed to address the possibility of infection. MRI of the brain can document the proper location of the catheter within the ventricular system, can evaluate for extra-axial fluid, persistent periventricular white matter edema associated with underdrainage, and exuberant dural enhancement which may be indicative of an overdrainage condition (as in Figures 18.2 and 18.3) [16].

Figure 18.1 Abdominal CT scan in a patient with a ventriculoperitoneal shunt. Note migration of the abdominal catheter into the extraperitoneal space and a surrounding CSF collection.

Figure 18.2 Contrast MRI shows exuberant dural enhancement suggestive of intracranial hypotension in a patient with NPH, a ventriculoperitoneal shunt, and a programmable valve (note the right parietal magnetic artifact). This finding is reversible. See Figure 18.3.

Figure 18.3 Contrast MRI in same patient as in Figure 18.2. This is the follow-up study 3 months after adjustment of the programmable valve. Normal dural enhancement suggestive of intracranial normotension is noted.

Shunt complication rates

Many studies have tried to address the overall complication rates for shunting procedures. Reddy et al. recently performed a retrospective review of 1015 ventriculoperitoneal shunts performed at their institution from 1961 to 2010 [17]; 70% were adults, 12% were NPH. In their study population, they identified 2239 total shunt procedures, of which 1224 were revision surgeries. It was noted, however, that in their adult patients only 39% of the surgeries were for revision, substantially less than for the general shunt population [17]. Others have demonstrated that 70–80% of patients with shunts, adult and children, and multiple diagnoses, will require at least one revision at some point in time [17–19]. Infections compromise a significant subset of shunt complications. They are associated with an increased risk of seizures, reduced intellectual performance, increased long-term mortality rate, and neurological disability and account for longer hospitalizations [20,21]. Studies that have looked at the overall complication rates of ventriculoperitoneal shunt have identified that despite the near universal use of preoperative and postoperative antibiotics, around 1–25% become infected [22,23]. A more specific number has been identified in two larger prospective trials with defined variables for infection in pediatric populations, which reported rates ranging from 8% to 10% [24,25]. The recent review by Reddy et al. in 2012 also looked at long-term infection rates in their mixed adult and pediatric group of 1015 patients whose hydrocephalus was managed with a ventriculoperitoneal shunt. They demonstrated a per patient overall infection rate of 10.5%; however, only 2.4% were in patients with NPH [17].

Ventriculoperitoneal shunt complications and infections in NPH patients

CSF shunting for NPH patients has become increasingly common, and may be underutilized as a treatment modality [11]. Cognitive decline associated with NPH is a known cause of preventable dementia, and the at-risk population is increasing in number. Ancillary tests such as external continuous lumbar drainage trials, high volume CSF taps, and CSF outflow resistance studies have increased the preoperative positive predictive values as to who will benefit from CSF shunting [13]. As more is understood about the pathophysiology of this disorder, the decision as to which preoperative indicators might help predict those who will benefit from shunting will aid surgeons in assisting patients in their decisions about the benefits versus risks of placement of a shunt. The risks associated with surgery in this patient population have been studied over time. Table 18.1 presents the data from multiple recent series that attempted to give an estimate of the rate of complication associated with CSF diversion procedures in the population of patients with NPH.

Table 18.1 Complications from shunting in patients with NPH^a

^a Cells left blank if authors did not report the indicated data. See text for additional analysis.

Vanneste et al. reviewed their experience with shunting procedures from 1980 to 1989, and reviewed the literature from 1966 to 1980 [9]. They described up to a 27% overall complication rate within the first 2 months, but the long-term postsurgical morbidity and mortality from shunting in their NPH populations was 3% in patients with secondary NPH and 9% in patients they felt had true idiopathic NPH. These results included complications associated with all patients’ comorbidities, and were not specifically limited to neurologic sequelae [9]. Boon et al. in 1997 [26] and 1998 [27] in the Dutch Normal-Pressure Hydrocephalus Study performed a prospective study enrolling 101 patients with NPH who underwent placement of a ventriculoperitoneal shunt and then underwent follow-up evaluations for one year. They noted that their population had extensive comorbidities, 48 past and current neurological and 190 non-neurological conditions. They observed 5 deaths in the first month from comorbidities, and 10 died over the next 11 months from comorbidities that included cerebral and cardiac ischemia and pneumonia. These data indicate the significance of comorbidities in the patient population at risk [26]. They also noted an increased risk of subdural effusions in patients receiving a low pressure shunt system (71%) compared to medium pressure systems (34%) [27].

Hebb and Cusimano performed their meta-analysis in 2001, which analyzed the literature to include all clinical trials of shunting in patients with NPH from 1966 through 2000 [28]. They identified 35 independent studies and a total patient population of 883. Their results are also included in the data presented in Table 18.1. The overall incidence of complications of any sort, acute or late, was 38% (of which 22% were shunt revisions), and they noted a 6% rate of long-term neurologic morbidity or mortality [28]. Since then, additional studies have added insight to help clarify long-term complication rates from CSF diversion procedures in patients with normal pressure hydrocephalus. Savolainen et al., in 2002, published their series of 51 patients, all less than 75 years of age, who were treated for NPH between 1993 and 1995, and thenfollowed for 5 years [4]. Their study used specific established criteria for proceeding with CSF diversion. Half of their patients underwent placement of a ventriculoperitoneal shunt and half did not. At 5-year follow-up eight patients with a shunt and nine patients without a shunt had died of complications associated with their comorbidities, none of which were shunt related according to the authors [4]. McGirt et al., in 2005, reviewed their single institution population of 132 patients with normal pressure hydrocephalus treated between 1993 and 2003, who underwent shunting procedures. They describe a 33% revision rate, a 2% late complication rate, and 1% mortality [29]. Marmarou et al. also in 2005 reviewed their series of 151 patients who were evaluated for idiopathic NPH between 1993 and 2003 of which 102 were treated with shunt surgery and then followed for 12 months [30]. They noted that 24 patients had complications overall, of which 10 (9.8%) were major complications including infection, subdural hematoma, and a venous sinus thrombosis. All but the patient with the sinus thrombosis had partial or complete recovery [30]. Pujari et al., in a 2008 retrospective analysis, reviewed the revision rates for their series of 55 patients who had 3 or more years of follow-up after an initial shunt procedure for the treatment of NPH [5]. They reported a revision rate of 53%, in this group, half of these after the first year of follow-up, the majority of revisions being for shunt malfunction. However, they noted that revisions, sometimes multiple over the study period in some patients, were still associated with a 74% incidence of symptomatic improvement. They concluded that with attention to follow-up, and monitoring for possible shunt obstructions or dysfunctions, there is a better chance of sustaining symptomatic improvement in this patient population [5]. Klinge et al. published in 2012 their results from the European multicenter study on idiopathic normal pressure hydrocephalus [8]. The series consisted of 115 patients who had undergone ventriculoperitoneal shunt placement for idiopathic NPH and who had completed their 1-year follow-up. The authors noted an overall complication rate requiring further surgery of 15% and 1% infection rate. They noted a complication rate not requiring surgery of 13%. They also noted that with the use of an adjustable shunt valve system they were able to make a total of 76 adjustments over the year in 36 patients (31%). Such attention to adjustments may help to minimize the complication rate due to over- or underdrainage [8]. For comparison, Bloch and McDermott performed a retrospective analysis of patients with idiopathic normal pressure hydrocephalus treated only with a lumboperitoneal shunt [31]. The review consisted of 33 patients in this report. They were followed annually for an average of 19 months (1–94). Nine patients (27%) required at least one revision. Two patients developed shunt infections. No patients developed subdural hematomas. The authors suggest that a shunt in the lumboperitoneal configuration is not more likely to need further surgery than a shunt in the ventricular configuration [31].

Shunt infections

Infection-related complications of CSF diversion for hydrocephalus, while not as common as shunt malfunction, are potentially serious developments in this population. In patients with shunts placed for any cause, those that develop infections suffer almost twice the risk of death and are three times more likely to undergo multiple shunt revision surgeries as patients who don’t develop infections [32,33]. In Hebb and Cusimano’s review, overall infection rate was 5% with a range of 0–18% [28]. Recent reports from large long-term outcome studies on patients who underwent CSF diversion for normal pressure hydrocephalus show similar infection rates [34]. Zemack and Romner described their experience implanting adjustable shunts in 218 NPH patients with an infection rate of 6.4% [34]. McGirt et al. included 132 patients who underwent shunt implantations for normal pressure hydrocephalus with an overall infection of 6.7% [29]. Pujari et al. reported a shunt infection rate of 10% in their series of 55 NPH patients [5].

Also as noted above, results from Reddy et al. in their large single institution study showed that first time shunt infection rates in adults are lower than those in children. Interestingly the same study showed shunts implanted to treat normal pressure hydrocephalus were associated with a lower infection rate (2.4%) in comparison to shunts implanted to treat other hydrocephalus etiologies (10.5%) [17]. Aside from age, other described risk factors for the development of shunt infections include placement of a shunt after previous shunt infections and the neurosurgeon’s experience [35,36].

Treatment strategies aimed at reducing shunt infection rates

Several strategies have emerged over the years aimed at curbing shunt-related infection rates. The use of preoperative and postoperative antibiotics, advances in surgical technique, surgeon experience in shunt placement, and developments in shunt technology have reduced shunt-related infection but unfortunately have not eliminated them [41,43].

The development of antibiotic-impregnated shunt (AIS) catheters became possible when early in vitro and in vivo studies demonstrated that antibiotics incorporated into silicone catheters withstood sterilization procedures and provided local antimicrobial activity [44–46]. The clinical introduction of AIS catheters was based on sound empiric rationale, and the shunt catheters were initially evaluated in the pediatric population where benefit was noted [43,47]. Subsequently there have been numerous clinical studies that have demonstrated varying degrees of efficacy and safety in the literature. Many of these studies have either been too small or underpowered and predominantly retrospective in nature. Parker et al., in 2011, performed a systematic literature review and meta-analysis of all 11 studies to that date that directly compared the infection rates between AIS and non-AIS catheter cohorts [41]. The studies were performed in pediatric populations, adult populations, and a mixture of the two. In the 11 studies, 5613 shunt procedures were identified. There was an overall infection rate of 5.4%, but there was a significant reduction in shunt-related infections using AIS catheters versus non-AIS catheters (3.3% vs. 7.2%) and furthermore the use of AIS catheters was not associated with an increased incidence of antibiotic-resistant organisms. Additionally, a subset analysis of adult and pediatric populations revealed a similar reduction in infections in both subsets when the AIS catheters were used: adults (0.9% vs. 5.8%) and pediatric (5.0% vs. 11.2%) [41].

The use of AIS catheters has predominately been studied in pediatric patients, where it has demonstrated a significant reduction in the absolute and relative risk of shunt infections. Several investigators have been able to suggest similar efficacy in adults [48,49]. Farber et al. in 2011 reviewed their institutional experience from 2004 to 2009 having uniformly converted to AIS catheters in 2006 [49]. They analyzed infection rates within the first year after shunt placement in adult cohorts taken from each period. The mean age of the study population was 60 ± 18 years. Five hundred patients were compared in this study (250 AIS vs. 250 non-AIS catheter placements). Hydrocephalus had various etiologies in their cohorts, but 76% of the patients underwent surgery for NPH and they were reasonably distributed between cohorts (73% in the AIS vs. 78% in the non-AIS cohorts). The overall infection rate was 2.6%, and the incidence was decreased in AIS patients (1.2%) compared to non-AIS patients (4.0%) (p = 0.0492). The most common organism in this group was Staphylococcus epidermidis [49]. Parker et al., in their meta-analysis of AIS versus non-AIS cohort studies from 2002 to 2010 reviewed 5613 shunt procedures, with infections noted in 301 cases (5.4%) [41]. The most common pathogen in both AIS and non-AIS catheter groups was Staphylococcus aureus (29.5% in the AIS and 30% in the non-AIS groups). Other pathogens identified were S. epidermidis (13.3%), Enterococcus spp. (3.7%), and Pseudomonas spp. (2.0%) [41] (Table 18.2).

Table 18.2 Organism prevalence in shunt infections in the era of antibiotic-impregnated shunt (AIS) catheters (see text for further analysis)

Although prospective, randomized trials are needed to confirm the direct association of reduced infection rates and AIS catheters, the available data are compelling. Some have speculated that reluctance to adopt AIS catheters has been due to the initial indirect cost associated with their use, as they may be $400 more than conventional catheter systems [50]. However, Attenello et al. performed a cost analysis of their treatment of patients with shunt infections at their institution [50]. In their review of 608 shunt surgeries, they demonstrated a 2.4-fold decreased likelihood of shunt infection in patients who had an AIS catheter implanted compared to the patients without. Given that their average hospitalization cost to treat a patient with a shunt infection was $48 454 they were able to realize a direct cost saving of $402 133 per 100 shunts placed by using AIS catheters, including the added cost of the catheter itself. The scope of the problem of shunt infections was indicated by Patwardhan and Nanda, who analyzed the Nationwide Inpatient Sample Database for the year 2000, a period before the widespread availability of AIS catheters, and who were able to estimate the cost of treating patients with implanted ventricular shunts [51]. They noted 5574 admissions for patients needing ventricular shunt procedures, of which 7.2% were for underlying infection. Given the database represents a sample size of 20% of the US admissions, the data project the annual admissions for shunt infections alone at over 2000 patients. They found an average cost per patient for treating all admitted patients with shunt complications, infection or otherwise, at $35 816, which is similar to the Attenello figure of $48 454, and would translate into an annual direct cost of over $70 million in the period prior to the institution of AIS catheters [50,51]. These figures presented are represented as estimates, but they are probably gross underestimates of the total cost of shunt infections as they only address fixed hospital costs. They do not take into account any economic costs attributable to shunt infections such as lost time, lost educational opportunity, lost wages, and costs associated with extended care, all of which are fixtures of the increased acuity associated with patients hospitalized with shunt infections.

Conclusion

The use of CSF diversion in NPH has come a long way from its humble initial applications. As we review the literature we can get the sense that the long-term complication rates from placement of ventriculoperitoneal shunts in patients with NPH are improving. The ability to maintain patients in symptomatic relief with shunts is improving. The risk of infection from shunt placement is falling. These successes are a result of a combination of better awareness of the condition we are trying to treat, improved equipment available for the treatment, in particular programmable valves and antibacterial impregnated catheters, increased opportunities to intervene in follow-up to help maintain patients’ symptom-free status, and vigilance by the healthcare providers to continue with ongoing follow-up so as to make needed adjustments and recommend surgical revision when indicated. Still, moving forward into the future, the risk-to-benefit ratio for shunting in patients with NPH can be substantially improved by realizing the opportunity to address both risk and benefit. Better understanding of the most appropriate and predictive preoperative evaluations will allow the offering of surgery particularly to those patients most likely to improve their symptoms with shunting. The population at risk with this disease is at present likely underserved, and at the same time growing in number. Public awareness about NPH has dramatically increased and many patients and their families now inquire about the disorder with the hope of possible intervention. On the other side of the analysis, the decision to shunt patients with NPH adds a known set of risks to the therapy for this patient population, one which often comes with significant comorbidities. Improvement in the risk profile will occur in the future from continued evolution of the process of patient management: accurate perioperative assessments, improved attention to safety and quality at the time of surgical placements, advances in hardware, and vigilant long-term follow-up evaluations, intervening when necessary to maintain the patient with optimal CSF diversion and by extension optimal symptomatic relief.