Percutaneous endoscopie gastrostomy tube (PEG) placement has provided a safe and effective means of securing enteral access for patients requiring long-term nutrition support since it was first described by Gauderer et al1 in 1980. It is the preferred mode of enterai tube placement.2,3 One subset of patients requiring long-term nutrition support includes those with severe intracranial disorders who no longer have the mental or neural capacity to safely take an oral diet. Oftentimes, patients with processes such as hydrocephalus or severe head trauma require ventricular drainage procedures such as ventriculoperitoneal (VP) shunting to decrease intracranial pressure. Subsequent PEG placement is theoretically dangerous as it requires a breach of the gastrointestinal tract with possible peritoneal contamination. Should enteral bacteria colonize the shunt's distal catheter, a devastating meningitis could ensue.
Previous studies have attempted to better elucidate the infectious risk for patients with VP shunts who require PEG placement. Due to the relative infrequency of patients requiring both procedures with PEG placement occurring after shunting, these studies have been small, with between 5 and 23 patients.4-8 The results have been inconclusive, demonstrating an infection rate ranging from 0% to 50%. Because surgeons and gastroenterologists who perform PEG placement are likely to encounter patients with indwelling VP shunts, this issue requires further investigation. This study was therefore designed to review outcomes of patients undergoing PEG placement in the setting of an indwelling VP shunt. Given the serious risk of foreign body infection in the setting of potential enterai spillage, we hypothesized that the infection rate would be higher for shunt patients undergoing PEG placement than the shunt population at large.
The records of all adult patients at a single university hospital who underwent PEG placement in the presence of an indwelling VP shunt were reviewed. Using the medical center's Clinical Data Repository, patients who had undergone both a gastrostomy tube placement and a ventricular drainage procedure between July 1995 and March 2004 were identified. Preliminary chart reviews identified the subset of patients who had operative notes verifying PEG placement after VP shunt placement without interval shunt removal. These patients were included in the study, and a more detailed chart review was performed evaluating the number of days between procedures, patient age at procedures, concurrent surgeries, antibiotic usage, white blood cell counts, revisions of either procedure, and the development of infection.
All shunts were placed by neurosurgeons and PEGs were placed by either surgeons (general or thoracic) or gastroenterologists. The push method was used under gastroscopic guidance. A central nervous system infection was defined by a positive isolate from cerebrospinal fluid (CSF) culture in the presence of increased leukocyte count, elevated protein, or decreased glucose in the CSF as per Centers for Disease Control (CDC) criteria.9 Study approval was granted by the Human Investigations Committee of the University of Virginia, and the need for informed consent was waived. Unless otherwise specified, values are presented as means
Over an approximate 9-year period, 39 patients with indwelling VP shunts underwent PEG placement at our medical center. Of these patients, 18 (46%) were men. The mean age at both VP shunt and PEG placement was 59 ± 3 years (range, 20-88 years for both procedures). The mean number of days between VP shunt and PEG placement was 43.1 ± 18, whereas the median number of days between procedures was 9, with a range from 2 to 564 days (Figure 1). Thirty patients (77%) had both procedures during the same hospitalization. General or thoracic surgeons placed 77% of the PEGs, whereas gastroenterologists placed 23%. The most frequent indication for VP shunt placement was hydrocephalus (most often attributed to a mass or bleed) in 27 patients (69%), followed by normal pressure hydrocephalus in 7 patients (18%). Five patients (13%) received a shunt for other causes. The indications for PEG placement most often included need for long-term enterai access, inability to tolerate oral intake, or dysphagia. Seven patients (18%) met another indication or did not have a recorded indication for PEG placement. No patients had evidence of PEG placement primarily for gastric decompression.
Thirteen patients required VP shunt revision, some for more than 1 reason. Eight patients required revision for shunt malfunction, 3 for suspected abdominal infection (2 acquired meningitis and are described in more detail below; the third was found at autopsy to have a subphrenic abscess that was considered too mature to have been related to the recent PEG placement), 4 for suspected ventricular infection (2 with meningitis listed below, 1 with the ventricular shunt tip eroding through the skin, and 1 with positive CSF cultures before PEG placement), and 3 for other reasons. Four patients required PEG-related interventions, including 2 PEG removals, 1 PEG change, and 1 blockage removal. Twenty-two patients underwent additional operative procedures during their hospitalizations, including percutaneous tracheostomy (n = 5), tracheostomy (n = 6), other neurosurgical procedures (n = 15), and other procedures (n = 3). There were no recorded thoracic or primary abdominal operations. Four patients died during their hospitalizations, none of which were directly attributed to a shunt infection.
At the time of PEG placement, 72% of patients were receiving antibiotics, 17 patients (44%) for reasons other than operative prophylaxis and 11 patients (28%) for prophylaxis. Of those patients receiving prophylactic antibiotics, 9 (82%) received cefazolin and 2 (18%) received clindamycin. Of all patients undergoing PEG placement after VP shunt placement, the peak white blood cell count during the first 7 days after PEG placement was 11,800 ± 700 cells/µL (normal range, 4500-10,000 cells/µL), occurring 1.9 ± 0.3 days after PEG placement.
Two patients (5%) developed meningitis after PEG placement, both of whom required shunt removal. The first patient underwent cystoperitoneal shunting (from a frontal porencephalic cyst) and required PEG placement during a separate hospitalization 24 days later. Approximately 2 months after the PEG placement, numerous cultures (distal shunt catheter tip, CSF, brain tissue, intra-abdominal abscess, and abdominal shunt site) grew Staphylococcus aureus. This patient had been receiving ciprofloxacin at the time of PEG placement. The second patient underwent PEG placement 5 days after VP shunting. Approximately 15 months later, CSF cultures grew Enterococcus faecalis, although there was no growth from the peritoneal catheter tip. This patient had received prophylactic cefazolin at the time of PEG placement. Both patients were discharged from the hospital alive, although the second patient had significant neurologic decline.
In this review of 39 patients with indwelling VP shunts requiring PEG placement, 2 patients acquired meningitis 2 and 15 months post-PEG. Whether these infections are directly attributable to the PEG placement is unknown. Because the second patient acquired an infection over 1 year after PEG placement and did not have a positive peritoneal catheter culture, it is less likely that his infection was directly related to PEG placement, although E faecalis is more likely than S aureus to be of enteric origin.
Similar studies have been carried out in the past, with various conclusions. Cantor and Miskovitz,4 Baird and Salasidis,5 and Graham et al6 demonstrated 0% infection rates in studies that included 5, 6, and 15 patients who underwent PEG after VP shunt placement and thus concluded that performing these procedures in tandem is safe. In a retrospective study of 16 patients requiring both procedures, Taylor et al7 documented a 50% infection rate and concluded that this practice is unsafe and should be avoided. Sane and colleagues8 studied children requiring PEG placement after VP shunting and found a 9% infection rate in 23 patients. Although these studies have offered important insight into the problem of PEG placement in patients with VP shunts, there are no trends in outcomes or consistent conclusions, likely due to the small enrollments.
Patients with indwelling peritoneal shunts are at risk for infection even without PEG placement. The available neurosurgical literature presents infection rates from less than 2%-5% for ventricular shunting procedures10,11 and 17% for neurosurgical procedures in general.12 When compared with our infection rate of 5% for shunted patients after PEG placement, there does not appear to be increased risk with PEG placement.
Antibiotic prophylaxis was not routinely used with our patients, yet 72% of patients received antibiotics (either prophylactic or for prior indication). Peristomal wound infections have been associated with gastrostomy tube placement,13 and the risk is decreased with antibiotic administration.14,15 Peritonitis after PEG is less common than superficial infections, occurring in anywhere from 0.5% to 13% of patients.13,16,17 A metaanalysis of trials investigating the role of antibiotic prophylaxis demonstrated a benefit,18 and prophylaxis is recommended. However, Sturgis and colleagues19 did not show a decrease in peristomal wound infections with a single dose of prophylactic antibiotics. Instead, a significant decrease in infections was found for patients who had been receiving extended antibiotic therapy before PEG placement. Both of the patients in this series who acquired infections received antibiotics at the time of PEG placement (one prophylactic and one receiving prior therapy), questioning the efficacy of prophylaxis in this setting. Because the majority of patients were receiving antibiotics, we cannot conclude that there was no benefit. Because the majority of isolates from PEG-related infections are Staphylococcus species,11,12,14,15,18 antistaphylococcal antibiotics should be used when prophylaxis is desired.
The infections seen in this series occurred 2 and 15 months post-PEG. It is unknown whether there is a "safe" period after shunting to place a PEG. Reversing the order of the procedures and waiting until a PEG track is mature before inserting a VP shunt might be ideal; however, this is impossible for most patients. On the other hand, early PEG placement might be safer if shunt infection depends on the presence of a biofilm on the shunt. Biofilms, first described by Zobell20 in 1943, form within days of implanting a foreign body and provide the substrate to which bacteria adhere.21 Thus, patients immediately postshunting who are at a high likelihood of requiring long-term enterai nutrition might benefit from early PEG placement.
Although these results are somewhat reassuring, this study was retrospective and thus subject to reviewer interpretations and limited by what was recorded in the medical record. Only those infections that were treated at the hospital where both procedures were performed were captured. Due to the large catchment area and referral patterns of this university hospital, it seems unlikely that infections would have been treated elsewhere. However, it is possible that some infections were missed. Although our results suggest that PEG placement in the setting of a VP shunt is safe, larger, prospective, multi-institutional studies need to be carried out to better stratify this risk and hopefully better elucidate the role of prophylactic antibiotics in this particular setting and the ideal timing between the 2 procedures.