Since the description of the "pull" technique for percutaneous endoscopie gastrostomy (PEG) by Gauderer et al1 in 1980, this procedure has become a well-accepted technique for feeding patients who cannot or do not eat orally, and is recommended for long-term enterai feeding.2 The technique is performed by puncture of the stomach through the abdominal wall under endoscopie control, with the assumption that the needle enters the stomach directly.
With increasing use of the PEG tube, it is expected that side effects and late complications will be seen more frequently. Health care professionals who care for patients with PEG tubes must become familiar with the various presentations and appropriate treatment of such events. A very rare complication is misplacement of the PEG through or into the colon, which can cause symptoms immediately or later after PEG replacement. We report 6 cases of misplacement of the PEG into the colon and review the literature concerning this rare complication.
MATERIALS AND METHODS
An 84-year-old demented woman, who was a resident of a nursing home, underwent PEG tube insertion due to her refusal to eat. Five weeks later, it was replaced, without gastroscopy, because of a tear in the tube. Two days later, fecal content was noticed in and around the PEG tube. A fistulogram showed the PEG tube to be in the colon (Figure 1), without any leakage into the peritoneum or connection to the stomach. The tube was removed and a nasogastric tube inserted for feeding. A new PEG was inserted 3 weeks later. After 2 years of follow up, the patient is still PEG-tube fed, without any remarkable events.
A 68-year-old male nursing home resident, demented as a result of a past hemorrhagic stroke, underwent PEG insertion. Ten days later, after selfextubation, a Foley catheter was blindly introduced into the gastrocutaneous fistula at the bedside. The Foley catheter was replaced 2 days later by a silicone PEG replacement tube. On admission to our department, fecal content was noticed in the tube, and a fistulogram showed that the tube was in the colon, with no signs of peritoneal leak or any gastrocolic fistula. The skin around the tube was infected. The tube was removed and the patient was fed through a nasogastric tube.
A 73-year-old demented, community-dwelling male patient had a replacement gastrostomy tube inserted percutaneously without endoscopy through a fistula from a previous PEG, which had been inserted 14 months earlier. Immediately after the first feeding through the reinserted gastrostomy tube, the caregiver noticed severe light-brown diarrhea, which subsequently occurred after every tube feeding. On admission to hospital, a fistulogram through the feeding tube revealed misplacement of the tube into the sigmoid colon. There were no signs of peritonitis or gastrocolonic fistula. The tube was replaced by a nasogastric feeding tube for a month, which was subsequently replaced by a new endoscopically inserted percutaneous gastrostomy tube. At 1-year follow up, the patient was tolerating tube feeding without any further problems.
A 67-year-old schizophrenic bedridden patient was a resident of a nursing home. His past medical history included several episodes of bowel obstruction secondary to megacolon, which resolved with conservative treatment. A PEG was inserted for the first time 3 years before his current admission, after severe aspiration pneumonia. The patient was admitted to hospital 3 additional times because of tube expulsion or obstruction; each tune, the tube position was checked by gastroscopy. The fifth PEG tube insertion was carried out de novo under endoscopie control after a period of 2 months, in which there had been an attempt to feed the patient orally. One month later, the patient complained of severe hunger, although he had been fed with adequate amounts through the PEG tube. He had severe diarrhea with white stools very soon after every meal, which did not respond to a low-residue diet. He was hospitalized after a month for investigation of these complaints. On admission, the patient was alert and begging for food. He was severely cachexie. He had no abdominal tenderness. The patient had hypoalbuminemia (2.9 g/dL), hypocholesterolemia (102 mg/dL), and a total lymphocyte count of 1100/mm3. A fistulogram through the feeding tube revealed misplacement of the tube into the sigmoid colon. The tube was replaced by a surgically inserted Janeway gastrostomy in an attempt to prevent further expulsion of the tube. Two-and-a-half years' follow-up have been uneventful.
A 75-year-old man underwent coronary bypass and aortic valve replacement with mitral valve repair. He was also known to have had reflux esophagitis, chronic obstructive lung disease, peripheral vascular disease, hypothyroidism, and chronic constipation.
The postoperative course was complicated by reintubation, sepsis, and volvulus of the sigmoid colon. Tracheostomy was performed on the 17th day postsurgery, and a PEG was inserted 42 days after operation. Two weeks after PEG insertion, a watery fecal discharge was noticed from the PEG, but the abdominal examination was normal. Plain abdominal x-ray showed distention of the small bowel and colon. Two days later, a clinical diagnosis of colocutaneous fistula was made. The diagnosis was confirmed a day later by a fistulogram that showed the tip of the catheter in the colon, with a thin fistula from the colon to the stomach. The patient was fed through a nasogastric tube while the tube in the colon was left in place for a further 26 days for drainage, when it was removed uneventfully. The general condition of the patient improved, allowing weaning from mechanical ventilation, rehabilitation of oral feeding, and spontaneous closure of the tracheostomy and the colocutaneous fistula.
An 83-year-old man was admitted with pneumonia, which resolved with antibiotic treatment. The patient had longstanding end-stage Parkinson's disease and 3 years earlier had had a cerebrovascular accident, which left him with a right hemiparesis and motor dysphasia. He had chronic constipation and was bedridden. An initial PEG had been inserted 1 year before the current admission because of eating difficulties. Six months later, the tube was changed due to erosion of the material. During the attempt to pull the tube out, it came out without the inner bumper, and a new tube was immediately inserted through the existing fistula. Two days later, the patient developed gastric obstruction with profound vomiting of gastric content, presumably secondary to occlusion by the inner bumper, which had become impacted in the pylorus. With conservative treatment, the obstruction resolved and the bumper was found in the stool a day later. The feeding tube functioned well until the current admission. Due to technical problems, the tube was changed again through the existing fistula, without any technical difficulties. The following day, a fecal discharge was seen in the tube, and a fistulogram demonstrated that it was located in the colon, with a communication to the stomach. The tube was pulled out and a nasogastric feeding tube inserted. In order to assess whether another attempt to insert a PEG was possible, computerized tomography of the abdomen was carried out, which showed intestinal loops covering the stomach, leaving no obvious point suitable for direct puncture of the stomach (Figure 2). The patient continued to receive nasogastric tube feeding.
We conducted a search of MEDLINE and PubMed, looking for colocutaneous fistula associated with PEG, limited to the English language and only to those reports pertaining to adult patients.
Between January 1, 2001 and December 31, 2005, when the 6 cases described above occurred, a total of 2384 PEGs were inserted in the 4 hospitals where our patients originated. The literature search yielded 22 reports of misplacement of a PEG into the colon, resulting in a colocutaneous fistula.3"24 Our review excluded reports of gastrocolonic fistula without direct communication between the colon and the skin26 because those cases could be caused by other mechanisms and not by penetrating the colon during tube insertion. The clinical characteristics of the 22 cases, together with our 6 cases, are presented in Table I. Of the 28 patients, 20 were men and 7 women (the gender of 1 patient was not specified). The average age was 63.2 years (range, 21-89 years). Only 8 of the 28 patients had undergone previous abdominal surgery or had relevant abdominal pathology. At least 2 patients had hypomotility of the bowel secondary to Parkinson's disease. Figure 3 demonstrates the course of the patients from PEG insertion through the symptomatic period (with its variability) until diagnosis. In 11 cases, the symptoms of colocutaneous fistula developed without the tube's having been replaced. In 1 patient, the symptoms developed after removal of the tube, after the patient resumed oral feeding. Brown malodorous fluid exuded from the PEG site. Sixteen patients developed symptoms of colocutaneous fistula after tube replacement for various reasons (extrusion, 3 patients; obstruction, 2; leak, 1; deterioration of the tubes, 7; reason not reported, 2). For those without tube replacement, the time that elapsed from initial insertion until symptoms developed ranged from 3 to 75 days. In those patients with tube replacement, the time from initial tube insertion to its replacement ranged from 10 days to 15 months, during which time there were no symptoms of colocutaneous fistula. Symptoms after tube replacement started immediately, up to 6 weeks after tube last insertion. The time interval from beginning of symptoms to diagnosis ranged from immediate diagnosis to a delay of 6 months.
Two groups of symptoms were noted. Severe watery or white-colored diarrhea occurred in 14 patients (colonie symptoms). Eleven patients had fecal discharge around or in the tube; some of those also had abdominal pain or other upper gastrointestinal tract symptoms. Four patients had both types of symptoms. Three patients did not show any of these symptoms, and the diagnosis was made during replacement of the tube (1 during endoscopie gastroscopy, 1 during surgical gastrostomy, and 1 was diagnosed because of intestinal sounds heard while the old tube was being pulled out). Another patient was diagnosed by computed tomography during a workup for fever. In 2 reports, there were no details of the symptoms.
Fistulograms of 12 patients demonstrated tract continuity from the skin through the colon and into the stomach. In 2 cases, there was involvement of small bowel. In 13 patients, the fistulogram demonstrated a tract between the skin and the colon, without any connection to the stomach.
The approach to treatment of this complication varied. In 10 cases, the treatment was surgical, 3 had laparoscopy as a diagnostic and treatment procedure, 5 had a surgical gastrostomy to gain more reliable access to the stomach, 1 was operated to address the colocutaneous fistula, and 2 had both closure of the fistula and a surgical gastrostomy. Colonoscopic clipping was performed in 1 patient. In 14 patients, the tube was removed without any invasive procedure. Details of the treatment were not provided in 3 cases.
Over a period of 5 years, we observed 6 patients admitted to our acute geriatric department with misplacement of a PEG tube into the colon. Five different gastroenterologists in 4 different hospitals had inserted the original PEG in these patients. None of them developed symptoms or signs of peritonitis, and a gastrocolonic fistula was demonstrated by contrast material injected through the tube in only 2 patients, probably due to its rapid closure. A surgical gastrostomy was performed in 1 of our patients (case no. 4), whereas removal of the tube and reinsertion of a new one by the "pull" technique was successful in 2 patients. In 2 patients, a nasogastric tube was left in situ, in one (case no. 2) because of his poor general condition secondary to an unrelated disease, to which he succumbed a month later. In the other patient (case no. 6), computed tomography revealed that a further percutaneous procedure was not possible, due to interposition of the bowel between aie stomach and the abdominal wall. The last patient (case no. 5) was able to resume oral feeding as part of general rehabilitation, and the tube was removed without surgery. Follow-up of 1-30 months in the 5 patients was unremarkable.
The first case report of misplacement of a gastrostomy tube into the colon was by Saltzberg et al,3 in a patient hi whom the penetration of the catheter through the colon on its way to the stomach was unrecognized for several months until the feeding catheter was removed, when the replacement tube could only be advanced into the colon. Bui et al6 discussed the possible pathologic mechanisms in 2 other patients, one of whom presented with fecal content in the feeding tube after PEG replacement, whereas the other presented with small bowel obstruction, secondary to accidental jejunostomy. Yamazaki et al15 reported a patient in whom replacement of a PEG tube resulted in watery diarrhea due to its placement in the colon. After closure of the colocutaneous fistula, another attempt to perform a new PEG resulted in a second gastrocolocutaneous fistula. Subsequently, the patient underwent open surgical gastrostomy. In their report, Yamazaki et al15 reviewed 11 cases of colocutaneous fistula, 5 of whom had undergone previous abdominal surgery. They recommended open abdominal gastrostomy in such patients to prevent this complication.
The mechanism of misplacement of the PEG tube into the colon was similar in all our 6 patients. Presumably, during the initial procedure, the needle passed through a segment of colon on its way into the stomach. In 3 of them, the tip of the tube did not proceed beyond the colon during blind reinsertion through an established fistula. In the other 2 cases, the original tube that had been inserted into the stomach under gastroscopic visualization was not changed but had imperceptibly retracted from the stomach to the colon. One of these patients was diagnosed 5 weeks later because of severe diarrhea, hunger, and malnutrition with weight loss. The tube misplacement in the other patient was diagnosed 2 weeks after insertion because of a fecal discharge through the tube, which lasted 2 days. case 6 in our series is even more surprising because at the first reinsertion of the tube, it was successfully inserted into the stomach; however, on the second replacement it was inserted into the colon.
Gastrocolic fistula has been described by some authors as a rare complication of PEG, including the large series of Ponsky et al.26 Of 227 PEGs in adult patients, only 1 had a gastrocolic fistula, which developed from erosion of the gastric wall by pressure of the tube bumper. No evidence of retrograde regression of the tube or any clue to penetration through the colon during the initial insertion was described in that case. Therefore, any report of gastrocolic fistula that did not provide details of the pathophysiology of penetration of the colon during the initial insertion was excluded from our review.
Although penetration of a PEG tube into the colon has been described several times, it is still a rare complication with which gastroenterologists and geriatricians may not be familiar. Therefore, it is not surprising to see that there was a lag period of up to 6 months from the onset of symptoms until diagnosis.16 Yamazaki et al15 recommended open gastrostomy in all patients with previous abdominal surgery. We question this recommendation for 2 reasons. This complication seems to be quite rare, even in the population of postabdominal surgery PEG patients; hence, the benefits of PEG far outweigh the risk of colocutaneous fistula. Furthermore, previous abdominal surgery or abdominal pathology is not a very significant risk factor for colocutaneous fistula. In our review, only 8 of the 28 patients had previous abdominal surgery or relevant abdominal pathology. Nevertheless, the gastroenterologist has to keep in mind the possibility of transcolonic insertion of the needle, particularly when there is a clinical history or evidence of megacolon, subphrenic transposition of the colon, or previous abdominal surgery. The gastroenterologist must also adhere to the safety procedures described by Gauderer et al1 and later in more detail by Ponsky and Gauderer27 to clearly visualize the transillumination of the gastroscope through the abdominal wall and the imprint of a finger pressed on the skin by the endoscope in the stomach. Another practical method suggested by Strodel et al28 is to insert a small-gauge anesthetic needle slowly through the skin and directly into the stomach while aspirating on the syringe attached to the needle. If air is aspirated into the syringe before the needle pierces the gastric mucosa, which should be visualized with the endoscope, then there is air-filled bowel, small or large, in between the stomach and the abdominal wall. Unfortunately, most of the case reports in our review had no details concerning these safety methods, and therefore we could not evaluate their efficacy in preventing misplacement of PEG tubes into the colon.
An alternative approach was suggested by Marcy et al,29 who described gastrocolic fistula after percutaneous fluoroscopic-guided gastrostomy (PFGG). They suggested the use of Ultrasound or computed tomography for any PFGG placement in patients suspected of having complex abdominal anatomy, and one could adopt this recommendation also for PEG insertion. Levine et al12 recommended fluoroscopic examination with contrast material injection into every replaced PEG tube because tube replacement lends itself to the risk of misplacement into the colon. This approach will diagnose misplacement that occurs during tube replacement but obviously will not demonstrate late regression of the tube to the colon. Our practice is to conduct fluoroscopic contrast examination only when there is a clinical suspicion of misplacement.
Misplaced tubes in the colon may be asymptomatic and undiagnosed until blind reinsertion of a replacement tube through the existing fistula. The absence of clinical signs of peritoneal leak of gastric secretions or food does not exclude the presence of a cutaneocologastric fistula. Spontaneous or unnoticeable withdrawal to the colon of a tube that had originally been positioned in the stomach is even rarer and must be kept in mind whenever there is clinical suspicion, even in a newly inserted PEG tube.
The various approaches suggested for the management of colocutaneous fistula due to misplacement of PEG tube range from conservative extubation to invasive exploration of the colon. Gyokeres et al21 questioned whether the colonie clip used by Kiin et al to close the colonie fistula was necessary at all. In the current review, 12 patients were treated surgically (including 1 from our series) and 16 were treated conservatively (including 1 colonoscopic clipping). The total number of the patients in this review is too small to draw any conclusions, but in our limited experience the conservative approach may be the appropriate initial management, especially in patients with a well-established fistula, when there is no danger of leak of colonie content to the peritoneum when the tube is removed.
Insertion of a PEG tube into the colon is a rare complication that can be asymptomatic and undiagnosed until the tube is replaced. It may present as severe diarrhea soon after feeding or by fecal discharge in or around the tube. Previous upper abdominal surgery or abdominal pathology may be a risk factor in some cases. There are no obvious recommendations in preventing this complication. The gastroenterologist should use all the safety techniques known to minimize the risk of penetrating the bowel on the way to the stomach. In case of uncertainty, he should delay the procedure and use an ultrasound or computed tomography to ensure its safety. Conservative management of a misplaced PEG tube may be limited to removing the tube, although some patients reported in the literature were treated surgically. Some had surgery to address the colocutaneous fistula, but the majority of those who had surgery were operated simply to gain more reliable access to the stomach for enterai access and to avoid another episode of colon injury. Clinicians caring for PEG patients should be familiar with the various clinical presentations.