Outcomes of Surgical, Percutaneous Endoscopic, and Percutaneous Radiologic Gastrostomies

Objectives  To evaluate and compare outcomes and complications in patients having undergone gastrostomy by surgical (SG), percutaneous endoscopic (PEG), or percutaneous radiological (PRG) procedure.

Design  Retrospective analysis.

Setting  University-based tertiary care center.

Patients  Of 82 patients who met inclusion criteria, 14 patients (median age, 40 years) received a surgical tube placement (SG), in 24 patients (median age, 55 years) a PEG procedure was performed, and in 44 patients (median age, 57 years) the tube was placed under fluoroscopic guidance (PRG). Indications for gastrostomy were similar in all groups, representing mainly cancer of the oropharyngeal, head and neck region (51 [61%]) as well as the upper gastrointestinal tract (6 [8%]), neurological disorders (15 [18%]), and others (10 [13%]).

Main Outcome Measures  Catheter function rates, major and minor procedure-related complications, and survival.

Results  Median follow-up was 17.2 months. Ten patients (71%) died in the SG group 7 to 855 days (median, 67 days) after the procedure, 7 patients (29%) died 5 to 263 days (median, 103 days) after PEG placement, and 30 patients (68%) died within 3 to 621 days (median, 112 days) after PRG, of their underlying disease or disease-related complications; 1 procedure-related death occurred 6 days after radiological tube placement. We observed a rate of minor complications of 43% (6 patients), 33% (8), and 36% (16) and a major complication rate of 14% (2 patients), 17% (4), and 11% (5) in the SG, PEG, and PRG groups, respectively. Tube function rates at 1 year were 67% (9 patients) and 68% (20) in the SG and PEG groups, respectively, and 10% lower (39) in the PRG group, although the difference was not statistically significant.

Conclusions  There is no major difference between SG, PEG, and PRG concerning procedure-related complications. Tube function tends to be inferior after radiological tube placement.

PROVIDING ENTERAL nutrition in patients who cannot ingest nutrients for one reason or another, but who have an otherwise functioning gut, is the most common indication to perform gastrostomy. As a less common reason, gastrostomy is used for patients who require prolonged gastric decompression because of chronic gastrointestinal tract obstruction. In addition, feeding tubes are a common method for long-term enteral feeding. In comparison with parenteral nutrition, feeding tubes are much easier to handle and less expensive, and infectious complications are less common.

Surgical gastrostomy was initially suggested by Egeberg, a Norwegian surgeon, in 1837,¹ whereas the first successful gastrostomy was carried out in 1876, almost 40 years later, by Verneuil in Paris, France.^1,2 Mortality and morbidity rates of gastrostomy performed by laparotomy vary, and complication rates of 3% to 60% have been described.^2,3 Therefore, in 1980, Gauderer et al^4-7 introduced a new technique into clinical practice as they created a feeding gastrostomy by means of a percutaneous technique under endoscopic guidance. Another method, the percutaneous nonendoscopic or radiological gastrostomy, was first performed in 1981 by Preshaw,⁸ a Canadian surgeon; with this technique the tube is placed with fluoroscopic guidance rather than by means of endoscopy or surgery.^8,9 A laparoscopic technique was introduced in the early 1990s.^10-12 Meanwhile, percutaneous endoscopic gastrostomy (PEG) has almost entirely displaced surgical gastrostomy (SG) in clinical practice, because the procedure is carried out more easily and without general anesthesia in this usually elderly, high-risk patient population; in addition, it avoids the mortality and morbidity associated with laparotomy.^9,13 Despite the minor invasiveness of endoscopic placement of percutaneous feeding tubes,^14-16 major and minor complications remain a topical problem.^17-20

The purpose of this descriptive and retrospective investigation was to determine success rates, outcome, minor and major complication rates, and procedure-related mortality in patients having undergone PEG placement, percutaneous radiological gastrostomy (PRG), and SG in a university teaching hospital.

Eighty-three patients underwent gastrostomy from January 1, 1994, to July 31, 1997, at the Departments of General Surgery and Radiology, University Clinic of Vienna, Vienna, Austria. Medical records of the patients' hospital stay were complete in 82 patients. Information concerning delayed complications and later follow-up was obtained from the corresponding outpatient records. This allowed a detailed analysis to be performed. Median follow-up was 17.2 months.

Fourteen patients underwent open SG, whereas in 24 patients an endoscopic gastrostomy (PEG) was performed, and 44 patients received a radiological tube placement (PRG). All SGs as well as the PEGs in critically ill patients were performed in the operating room, whereas the remaining PEGs and PRGs took place in an endoscopy suite. All procedures were performed by surgeons or radiologists with adequate experience of the respective techniques.

Operative gastrostomy was done in either the Stamm version (infants and adolescents) or the Witzel procedure (adults). An upper midline skin incision was made and the linea alba divided. Two or 3 purse-string sutures were placed around the proposed entrance site on the anterior surface of the stomach, and a stab wound was placed through the center of the inner purse-string suture (Stamm). A Foley catheter, 14F (Willy Rüsch AG, Kernen, Germany), was drawn through the abdominal wall through a stab wound at a selected site and then inserted into the stomach. Next, some tacking sutures were placed in the stomach around the gastrostomy tube and through the anterior abdominal wall near the exit site of the catheter. When these sutures were tied and the balloon of the catheter filled with fluid (0.9% sodium chloride), the stomach was fixed to the abdominal wall, and risk of leakage was minimized. Finally, an anchoring suture was placed in the skin at the exit site of the tube and secured around it to fix it in place. The gastrostomy tube was then flushed with 0.9% sodium chloride and tested for possible leaks.

With the Witzel procedure, the abdominal incision, the placement of the catheter in the stomach, and the location of the tube exit site on the skin were the same as for the Stamm gastrostomy except that only a single purse-string suture was placed around the entrance site on the anterior surface of the stomach. After a Foley catheter, 24F (Willy Rüsch AG), was introduced through the abdominal wall from a stab wound, it was inserted into the stomach through a stab wound that was made through the center of the purse-string suture. By suturing the wall of the stomach over the tube for a distance of 4 to 6 cm, a serosal tunnel was created, whereby the last suture was tied around the feeding tube to secure it in place. Finally, the stomach was sutured to the abdominal wall as in the Stamm gastrostomy. A suture was placed in the skin at the catheter exit site and secured around the tube to anchor it in place after the balloon of the catheter was filled with 0.9% sodium chloride. Also with the Witzel procedure, the gastrostomy tube was flushed and tested for possible leaks.

For the endoscopic technique (Compat-PEG, 22F, Sandoz Nutrition Corp [now Novartis Nutrition Corp], Minneapolis, Minn), the gastroscope was passed into the stomach, and after an upper gastrointestinal tract endoscopy was performed, the stomach was sufficiently inflated with air and the appropriate puncture site was located by diaphanoscopy. An approximately 6-mm incision was made with the patient under local anesthesia, and after the puncture cannula was advanced into the stomach under endoscopic control, a guide wire was inserted through the cannula and grasped with the biopsy forceps, and drawn out together with the gastroscope. The proximal end of the wire was attached to the PEG tube's (22F) fixation loop and pulled into the stomach on the distal end of the wire. Then the tube was pulled through the abdominal wall until a silicone disk stopped at the inner gastric wall. The tube was then inserted in a preformed site of an external bolster and secured by wrapping a separate pull tie around the bolster collar so that a firm connection between the gastric and abdominal wall was ensured. The correct position of the tube was once again checked endoscopically and the tube flushed with 0.9% sodium chloride and tested for possible leaks.

When PRG (Cook Inc, Bloomington, Ind) was to be performed, a nasogastric tube was introduced to inflate the stomach with approximately 500 mL of air before the procedure. After administration of local anesthesia, the lower part of the body of the stomach was punctured under fluoroscopic guidance, a guide wire introduced, and the tract enlarged with a series of dilators before the gastrojejunostomy catheter (Friction Lock Malecot, 16F; Cook Inc) was inserted. A contrast medium was injected to identify the catheter in its correct position and to test the tube for possible leakage. To prevent inadvertent straightening of the catheter, a suture was tied completely around it and a dressing was applied to the site in a routine manner.

During all the procedures, no antibiotics were administered to the patients routinely except to all patients undergoing SG and to all patients in the PEG group with cancer and esophageal perforation. Clear fluids were given 12 to 24 hours after placement of the feeding tubes, and, if tolerated, either continuous or bolus enteral feeding was started after 24 hours for patients who underwent PEG and PRG and after the first postoperative bowel movements for patients who had SG. Abdominal radiography was performed in all patients undergoing PRG and in patients who suffered from abdominal discomfort and/or pain after a PEG procedure. After each feeding as well as after each administration of medication, the gastrostomy tubes and catheters were flushed with 20 mL of lukewarm water.

Differences in the incidence of minor and major complications between the groups were tested by the Fisher exact test. Function rates of tubes and survival rates of patients were estimated according to the Kaplan-Meier method. The statistical analysis was performed by the Institute of Medical Statistics, University Clinic of Surgery, Vienna.

Of the 82 cases reviewed, there were 14 SGs, 24 PEGs, and 44 PRGs. The patients' age ranged from 1 day (2 patients) to 90 years; more than 60% of the patients were older than 50 years (Figure 1). There were slightly more males than females (51 vs 31) (Table 1). Two newborns with esophageal atresia, of whom 1 was born prematurely with a weight of 1120 g and who additionally suffered from an atrial septal defect, were 1 day of age each, 1 child with an esophageal stricture after acid ingestion was 2 years old, and 1 patient suffering from cerebral paresis (severe epilepsy) was 6 years of age when gastrostomy was performed; all others were aged 10 to 90 years.

The indications for placing an enteral access device were similar for all groups. Oropharyngeal, head and neck cancer (n=51, 61%) was the most common indication for the procedures (SG, 6 [43%]; PEG, 9 [38%]; PRG, 36 [82%]). Neurological disorders, including cerebral trauma (n=15, 18%), were the second leading indication (SG, 2 [14%]; PEG, 10 [42%]; PRG, 3 [7%]), followed by other malignant and nonmalignant diseases, such as esophageal cancer, gastric cancer, esophageal perforation, esophageal stricture or atresia, and cachexia (Table 2).

Successful tube placements were carried out in 82 patients. Of the procedures, 58 (71%) were performed with the patient under local anesthesia (1% lidocaine), in 7 (8%) an additional sedative (midazolam) was administered, whereas the remaining 17 procedures (21%) (14 SGs) were performed with the patient under general anesthesia (Table 3). One dose of prophylactic antibiotics was given before or during the procedure to all patients who had SG and to all patients in the PEG group with cancer and esophageal perforation.

The occurrence of pneumoperitoneum was observed in 5 patients having undergone the PEG procedure (21%) and in 8 patients after PRG (18%) and was diagnosed by means of abdominal x-rays (Table 4). This kind of post-PEG pneumoperitoneum was benign and self-limited, resolved within 1 week, and did not require laparotomy.^21,22

The indications for tube replacements as well as the replacement intervals for the SG, PEG, and PRG groups are summarized in Table 4. In approximately 4 (28%) of the patients who underwent SG, in 7 (30%) of the patients who had PEG, and in 16 (37%) of the patients who had PRG, replacement of the gastrostomy tube was necessary because of malfunction caused by blockage or clogging, fracture or cracks, and tube leaks (leakage of feedings). The replacement interval ranged from 6 to 41 days (median, 16 days) for SG, 7 to 62 days (median, 15 days) for PEG, and 3 to 155 days (median, 22 days) for PRG. Of the 27 patients who had their original tube replaced, 13 patients required a second replacement because of one of the described reasons. The interval between the first and second replacement was 15 days (SG), 22 to 30 days (median, 27 days) (PEG), and 13 to 170 days (median, 40 days) (PRG). During all these procedures, no replacement-related complication occurred.

Summarizing all minor and major complications, the patients' complication rates were 37% and 13%, respectively (Table 4). The detailed evaluation of the various complications according to the time of occurrence (≤14 days, >14 days) is shown in Table 5.

The overall mortality rate from all causes was 47 (57%) of 82 patients; 10 (71%) for SG, 7 (29%) for PEG, and 30 (68%) for PRG (Table 4). All deaths were related to the underlying diseases or disease-related complications except for 1 procedure-related death of a patient in the PRG group who died 6 days after the procedure because of aspiration and consecutive multiorgan failure, and 1 premature neonate who died of intracerebral bleeding 1 week after SG.

The overall procedure-related complications were divided into early (within 14 days) and late (after 14 days) complications and classified as minor and major (Table 5). Minor complications included dislodged tubes or catheters, tube leaks, wound infection requiring wound care, and peristomal proud flesh and fever.

Aspiration, bleeding, infection with the necessity of administering systemic antibiotic therapy, and urgency in performing surgical intervention (acute abdomen, peritonitis, or gastric perforation) were major complications.

All complications of each group are listed in Table 5. Within 14 days, 6 dislodged tubes (7%) were seen in 82 patients, 2 (14%) of 14 in the SG group, 1 (4%) of 24 in the PEG group, and 3 (7%) of 44 in the PRG group. The number of dislodged tubes after 14 days was 6 (7%) of 82. In this later phase, no tube displacement was seen in the SG group, 1 (4%) of 24 in the PEG group, and 5 (11%) of 44 in the PRG group. The explanation for this distribution is that the PRG group consisted mainly of elderly and confused patients who pulled on their tubes.

Regarding the early complications (≤14 days), tube leaks were found in only 1 (1%) of 82 patients, whereas after 14 days, tube leaks occurred in 7 (9%) of 82 patients. Three (4%) of 82 patients acquired an infection requiring wound care within the first 2 weeks compared with 7 (9%) of 82 patients in whom the infection appeared later than 14 days.

Aspiration as an early major complication was seen in 1 patient in the PRG group (2%). No bleeding occurred in any patient during the procedures.

In 2 (8%) of the 24 patients in the PEG group, a local infection was seen within the first 2 weeks and an antibiotic was administered. Another patient with cancer of the esophagogastric junction invading the esophagus, in whom a surgical gastrostomy was performed, developed a systemic infection with positive blood culture (requiring antibiotics) 3 months after the operation. In this case the gastric tube was removed and a stent was implanted into the esophagus.

Laparotomy had to be performed because of peritonitis in 3 patients (7%) at 24, 48, and 70 hours after having undergone the PRG procedure. Dislodged tubes were the main reason for peritonitis, and in 1 patient a tear-off of the T-bolster was observed. In another 4 patients (5%), surgical intervention by means of laparotomy had to be done 16 days after PRG, 48 days after SG, and 22 and 31 months after PEG. Two patients developed peritonitis after tube displacements, whereas the other 2 patients had an acute abdomen caused by a tube leak. In 1 patient in the PEG group, transgastric migration of the fixation plate in the stomach was seen, probably because of too high traction on the outer tube and consequent necrosis of the stomach wall. All cases were surgically repaired successfully with uneventful recovery.

No statistical difference was observed among the various groups concerning minor and major complication rates before and after 14 days of occurrence.

Fourteen tubes placed temporarily because of dysphagia caused by neurological disorders (stroke, bleeding) or in patients who had undergone an operation for esophageal atresia or perforation could be removed uneventfully because of the patients' recovery (4 tubes in the SG group [29%], 5 tubes in the PEG group [21%], and 5 tubes in the PRG group [11%]). These tubes were in place for a median of 68, 78, and 82 days in the SG, PEG, and PRG groups, respectively. One newborn patient had to undergo laparotomy 8 days after tube removal because of dehydration and fluid imbalance caused by a gastrocutaneous fistula (Table 6).

At the time of investigation, 25 (31%) of 82 patients still were alive with functioning tubes (1 SG [7%], 12 PEG [50%], and 12 PRG [27%]). Forty-three patients died with their feeding tubes in place; 4 patients died after tube removal.

The detailed survival rates are shown in Figure 2. Of 47 patients who died, 43 patients died because of their underlying disease (tumor progression, neurological disease, cachexia) and 3 patients because of multisystem organ failure (1 procedure-related death after aspiration and 2 patients who developed multiple organ failure after esophageal perforation with persistent mediastinitis). One premature neonate with esophageal atresia and atrial septal defect died 7 days after surgery of intracerebral bleeding (Table 7).

Function rates after first tube placement were virtually identical for the SG and PEG groups and 10% lower in the PRG group (Figure 3). After repeated tube placement, SG tubes performed slightly better than PEG tubes, while PRG tubes functioned markedly worse, although these differences did not reach statistical significance (Figure 4). The main reason for tube replacement was irreversible blockage (Table 4).

Gastrostomy is the preferred method of administering enteral nutrition in patients with impaired ability to eat. Many authors firmly support the advantages of enteral over parenteral feeding²³ because disorders of intestinal motility are avoided, mucosal atrophy is prevented,²⁴ and structural gastrointestinal tract integrity is maintained. For more than a century, surgically placed feeding tubes were the only possibility for feeding patients who could not maintain sufficient oral intake, until PEG and PRG procedures gained widespread popularity in surgery and gastroenterology during the last 15 years. The percutaneous approach to gastrostomy tube placement is generally recognized to be lower in cost and reduces procedure time as well as the necessity for general anesthesia. Because it makes laparotomy unnecessary, this approach is also considered safer than the surgical placement. Comparing PEG and PRG, the endoscopic technique benefits from direct inspection of the esophagus and stomach and of the needle puncture of the stomach. In addition, guide wire and catheter looping in the peritoneal cavity do not occur. Besides the possibility of performing a tube placement in patients in whom endoscopic approach to the stomach is not possible because of pharyngoesophageal stenosis, the PRG procedure has the advantage of introducing an uncontaminated catheter through the abdominal wall.

Various improvements in gastrostomy procedures have been proposed.²⁵ Edelman et al^10,11 published a new promising laparoscopic approach in 1991. One of the main advantages of this technique¹² is its application in patients in whom endoscopy is not possible because of impassable obstructions from tumors in the pharyngoesophageal region or facial trauma. In addition, it offers the benefits of intra-abdominal inspection and therefore reduces the chances of injuries during the procedure.

Compared with PEG, SG is associated with a higher morbidity and mortality rate in the literature.^26-28 On the other hand, some studies do not confirm these results.^29,30 Furthermore, PEG and PRG placements are accompanied by a certain risk of morbidity and mortality.^31-33 Rather low survival rates in patients who received an SG or PRG tube are explained by the fact that these groups consisted mainly of patients with advanced cancer of the upper gastrointestinal tract who died of their underlying disease.

Procedure-related rates of both minor and major complications of 3% to 60%, as reported by different authors,^2,3,20,31,34 are consistent with our findings of 43% minor complications for SG, 33% for PEG, and 36% for PRG, as well as a major complication rate for SG, PEG, and PRG of 14%, 17%, and 11%, respectively. When the incidence and types of complications are reviewed in detail, surgical tube placement appears to be associated mainly with minor complications such as tube leaks and infections of the cutaneous insertion site, while 6 of 7 cases of peritonitis requiring laparotomy occurred in the groups with interventional tube placement. One surgical patient was reoperated on for peritonitis after an uneventful postoperative course 48 days after SG placement. The intraoperative findings showed an ulcerlike lesion of the anterior stomach wall distant from the intact tube implantation site.

The SG and PEG tubes show virtually identical function rates. The high blockage or clogging rate of PRG tubes compared with SG and PEG tubes in our series most likely results from the smaller diameters of the tube used for PRG (16F vs 24F for the Witzel procedure and vs 22F for PEG). Especially after tube replacement (Figure 4), inferior tube function rates were observed. This observation, together with the limited practicability of administering minced homemade food through these tubes, precludes the use of PRG tubes in patients for whom home care is planned.

As described by many authors, the percutaneous approach for placing a feeding tube is associated with less expenditure of costs, time, personnel, analgesics, and hospital stay compared with operative gastrostomy. Because of these facts, this technique was generally accepted and performed during the last 2 decades with satisfactory results. The determination to use either the percutaneous or surgical approach should therefore be based on the individual necessities of the patient.

Although the indications for surgical tube placement have diminished for the described reasons, our series shows that the outcomes of SG are not at all inferior to those of PEG and PRG. Severe complications requiring an emergency operation occur in both interventional groups and jeopardize the proposed advantages of the interventional technique. Function rates of PRG tubes make them a less-attractive choice for individuals being cared for on an outpatient basis.

In our opinion, SG is still the method of choice when a laparotomy is performed for any other reason (as part of another operative procedure). Furthermore, SG should be performed when an apposition of the stomach to the abdominal wall is uncertain, or when gastroscopy or laparoscopy is not possible or several attempts have failed. It also has its place as the second-line treatment after tube complications with other methods require surgical management.

Reprints: Bela Teleky, MD, Department of Surgery, University Clinic of Vienna, Währinger Gürtel 18-20, A-1090 Vienna, Austria.