Downloaded from www.ajronline.org by UCSF LIB & CKM/RSCS MGMT on 12/19/14 from IP address 169.230.243.252. Copyright ARRS. For personal use only; all rights reserved
725
Fluoroscopically Guided Percutaneous Gastrostomy Gastroenterostomy: Analysis Consecutive
Marshall E. Hicks1 R. Stephen Surratt1 Daniel Picu& M. Victoria Marx1’2 Elvira V. Lang1’3
Received
July 24, 1 989; accepted
after
revision
Presented at the annual meeting of the American Roentgen Ray Society, New Orleans, May 1989. I Mallinckrodt Institute of Radiology, Washington University School of Medicine, 51 0 S. Kingshighway
Blvd., St. Louis, MO 631 10. Address quests to M. E. Hicks. 2Present address:
Department
University
1500 E. Medical
Ann Arbor, 3Present Stanford
of Michigan, address:
Radiology, Center
544-0725
Roentgen
Ray Society
154:725-728,
April 1990
For over 1 00 years, surgical gastrostomy and gastrojejunostomy were the standard methods of direct enteral access. This began to change, however, with the development of percutaneous endoscopic gastrostomy in 1979 [1 -3]. Despite the lower morbidity and mortality with endoscopic placement, the search for a simpler, safer means of gastroenteric access has continued, resulting in the recent development of fluoroscopically guided percutaneous gastrostomy. Early series of fluoroscopically guided percutaneous gastrostomies contained small numbers of patients, but favorable morbidity and mortality rates were reported when comparisons were made with endoscopic placement [4-7]. Recently, larger series have begun to emerge, substantiating the results of the initial reports [812]. We report our experience with percutaneous gastrostomy and gastroenterostomy in 1 58 consecutive patients, comparing three different types of tubes used during a 2-year period.
Materials
and
Methods
Dr.,
Department of Radiology, University School of Medicine, Stanford,
0361 -803X/90/1
Cases
reprint re-
MI 48109-0030.
CA 94305-51 05. C American
of
of 158
We reviewed our experience with 158 consecutive patients who underwent either percutaneous gastrostomy or percutaneous gastroenterostomy during a 2-year period. The catheters used included Foley catheters (36), Cope-type gastric catheters (86), or Carey-Alzate-Coons gastrojejunostomy catheters (36). Gastrojejunostomy tubes were placed in patients with gastroesophageal reflux or aspiration, gastric atony, or partial gastric obstruction. Ninety percent of the tubes were placed for feeding purposes. The technical success rate was 100%. Thirty-day follow-up was obtained in 89%. Thirty-day mortality was 26%, reflecting the substantial number of debilitated patients. No deaths were directly related to tube placement. Major morbidity was 6% and included hemorrhage, peritonitis, tube migration, and sepsis. Minor morbidity was 12%. There was no difference in 30-day mortality or feeding tolerance between the tube types (p < .05). Patients with Foley catheters had more complications necessitating surgical intervention and an increased incidence of tube changes required within 30 days. These were the only statistically significant differences between the tubes (p < .05). Our results show that percutaneous gastrostomy is a safe and effective means of gastroenteric feeding or decompression. Because of the fewer complications and ease of insertion, the Cope type of gastrostomy tube has become our preferred catheter for percutaneous feeding or decompression. AJR
November 20, 1989.
and
We reviewed percutaneous
performed.
the medical
records
of the first
gastrostomy
or gastrojejunostomy.
The procedure
was not attempted
1 61 consecutive One hundred
patients
fifty-eight
referred
procedures
in three cases (two patients
for were
with overlying
colon and one with massive gastroesophageal reflux in whom we were unable to maintain adequate inflation of the stomach for safe percutaneous puncture). The study group of 158 patients included 95 women (60%) and 63 men (40%). The mean age was 67 years; the
HICKS
726
median
age was 71 years. Four patients (2.5%) had ascites. Three (1 .9%) had histories of prior gastric surgery; this included two patients who had gastrojejunostomies for palliative treatment of
Downloaded from www.ajronline.org by UCSF LIB & CKM/RSCS MGMT on 12/19/14 from IP address 169.230.243.252. Copyright ARRS. For personal use only; all rights reserved
patients
pancreatic for peptic
used
carcinoma ulcer one
one
patient
T fasteners
patients
(4.4%),
in seven
dialysis,
and
disease. uncooperative
who
had
(Meditech,
including
patient,
one patient
and
[1 3]. At the time of tube placement,
a Billroth
I operation
Watertown, one
MA)
were
on peritoneal
patient
with
ascites
72 patients (46%) or their families
had requested that no resuscitative measures be undertaken in the event of cardiopulmonary arrest. Of the 90% of tubes placed for feeding purposes, the majority were for patients with neurologically impaired swallowing mechanisms (1 04 patients). Other feeding indications included patients with tumors of the ears, nose, or throat (1 0 patients), esophageal obstruction
(nine
patients),
and
miscellaneous
(1 9 patients). Only of the stomach or small
indications
10% of the tubes were for decompression bowel in patients with bowel obstruction.
In the first 36 patients who underwent gastrostomy, 1 6- or 18French Foley catheters were placed (Bard, Covington, GA). In the remaining 86 patients undergoing gastrostomy, 12- or 1 4-French
Cope-type
catheters
jejunostomy
were placed (Cook, Bloomington,
catheters
Carey-Alzate-Coons
were
placed
in 36 patients.
gastrojejunostomy
IN). GastroAll 36 had the
catheter(Cook).
Candidates
for these gastroenteric feeding tubes included patients with documented aspiration, gastric atony, or partial gastric obstruction. Local anesthetic was used in all cases. IV sedation with midazolam and morphine sulfate was used routinely. No patient received general anesthesia. Procedures were performed under fluoroscopic guidance, with the exception of one case in which CT was used because the patient had extensive bowel gas in the vicinity of the stomach. No additional imaging techniques were used to identify liver or colon. Antibiotics were not given routinely. Glucagon (1 mg, Lilly, Indianapohs, IN) was given IV to inhibit gastric motility. All catheters were placed by using the Seldinger technique with air insufflation of the stomach via an indwelling nasogastric tube. After puncture with an 1 8-gauge Teflon-sheathed needle (Cook), an 0.038in. (1 -mm) Amplatz extra-stiff guidewire (Cook) was passed into the stomach. We used a method for puncture described and refined by Wills
and
dilatation
Oglesby of the
Cope
[14]. tract.
Foley
catheters
catheters
and
were
placed
after serial
gastrojejunostomy
cathe-
ters were placed through a peel-away sheath of the appropriate size. When T fasteners were used, they were placed before catheter insertion.
Catheters
Inc., Somerville,
Tube feedings
were
secured
to the skin
with
0-prolene
(Ethicon,
NJ).
were initiated
on the day after placement
for gas-
trostomy tubes, unless a complication precluding feeding developed. Gastrojejunostomy patients routinely began receiving feedings on the same day as tube placement. The nasogastric tube usually was removed the day after the procedure. The chi square and Van der Waerden one-way tests were used to compare the tube types for differences in patient age, sex, code status, morbidity, mortality, feeding tolerance, and required tube changes and to identify risk factors with respect to increased morbidity and mortality (p < .05).
Results
The technical success rate was 1 00% in all 1 58 cases in which the procedure was attempted. More than one puncture was required in 10 cases (6%). This was usually because the stomach deflated during placement of the catheter. In all instances the additional puncture was successful.
ET AL.
AJR:154,
April 1990
Thirty-day follow-up was obtained in 1 41 patients (89%). The 30-day mortality rate calculated from these 1 41 patients was 26%. Twenty-seven of the 36 patients who were dead at 30 days died of unrelated causes. Although there were no cases where tube placement was definitely related to the death of a patient, there were three cases (2%) in which a complication occurred from tube placement that negatively altered the clinical course of the patient and may have contributed to the patient’s death. In an additional six patients (4%), no obvious precipitating cause of death was recorded in the physician’s record note section of the chart; however, in all, multiorgan failure was believed to be the cause of death. All six of these patients were assigned the status of “no code.” Three of these patients died less than 48 hr after uncomplicated placement of a gastrostomy tube. The other three also had uncomplicated procedures and were tolerating tube feedings before they died. Morbidity was divided into major and minor. Major morbidity was defined as a complication that (1 ) required an additional interventional procedure, (2) prolonged the hospital stay, (3) required a transfusion, or (4) required IV antibiotics. Major morbidity consisted of five cases of hemorrhage (3%), two cases of peritonitis (1 %), two cases of tube migration (1 %), and one case of sepsis (1 %). Three of the five patients with hemorrhage recovered uneventfully. One of these patients was on steroids for lupus cerebritis at the time of tube placement and experienced gastrointestinal bleeding after the procedure. Diffuse gastritis was found with endoscopy after tube placement. The patient was transfused with 3 units of packed RBCs and the bleeding resolved. The gastritis healed with tapering the administration of the steroids.
In another
patient,
there
was
a small
amount
of bleeding
with the initial puncture at the time of catheter placement. This was probably from perforation of a vein in the posterior wall of the stomach. The hemoglobin dropped 2 g/dl over a 2-day
period,
and
the
nasogastric
tube
aspirate
was
positive
for occult blood, indicating a small amount of persistent hemorrhage. This patient was transfused with 2 units of packed RBCs and the bleeding ceased. In a third patient, another slow gastrointestinal hemorrhage was manifested by a melanotic stool and a decrease in hematocrit from 39% to 27% over a 3-day period. No difficulties were noted in tube placement. This patient experienced no further bleeding after transfusion with 2 units of packed RBCs. The remaining two patients with bleeding complications were classified as having possible tube-related deaths, although both had ongoing multisystem failure. One of these two patients was on Coumadin for atrial fibrillation with a prothrombin time greater than 40 sec at the time of gastrojejunostomy tube placement. Gastrointestinal hemorrhage ensued after catheter placement. Although the bleeding subsided within 1 2 hr after the procedure with correction of the prothrombin time by transfusion of 1 0 units of fresh frozen plasma and 6 units of packed RBCs, the patient died 5 days after tube placement. She had chronic renal failure and congestive heart failure; the acute cause of death was respiratory failure. The other patient had an occult hemorrhage; the hematocrit dropped from 34% to 20% in the 2 weeks
PERCUTANEOUS
AJR:154, April 1990
after tube placement.
Downloaded from www.ajronline.org by UCSF LIB & CKM/RSCS MGMT on 12/19/14 from IP address 169.230.243.252. Copyright ARRS. For personal use only; all rights reserved
may
have
been
caused
The patient
had a prolonged
by the occult
hemorrhage
GASTROSTOMY
ileus, which or ongoing
urosepsis. The family requested that no supportive measures be given and the patient died 1 6 days after catheter placement. One of the two patients included in the peritonitis complication group was a patient in acute renal failure with altered mental status. This made clinical evaluation of this complication
difficult,
although
peritonitis
was
suspected.
This
was
GASTROENTEROSTOMY
727
Patients who were classified as no code at the time of tube placement were at risk for death within the 30-day follow-up period (p < .05); however, the no-code status did not entail a higher incidence of morbidity. Other risk factors found not to be statistically significant for an increase in morbidity or mortality included sex, age, difficulty with placement, or anatomic considerations such as ascites or prior gastric surgery.
the
only case in which an additional puncture at the time of placement was associated with morbidity. The patient’s family requested that no further supportive therapy be given and the patient died 6 days after tube placement. The other case of peritonitis was proved and occurred after deflation of the balloon on the Foley gastrostomy catheter 1 day after tube placement. This patient required a Stamm gastrostomy and lavage and recovered uneventfully. Of the two cases of tube migration, one also had a Foley catheter that eroded into the lesser sac 1 2 days after tube placement. A resultant abscess required surgical evacuation with a jejunostomy tube placement. The other tube migration complication was caused by a Cope catheter that was partially outside the stomach. An additional percutaneous gastrostomy tube was placed and the patient recovered uneventfully. The final major morbidity complication was in a patient with Proteus sepsis from a wound infection. This patient recovered with
AND
IV antibiotics.
Minor morbidity included skin infection (seven patients), transient increase in fever or WBCs (seven patients), transient ascitic leak (two patients), minor skin bleeding (two patients), and catheter migration (one patient). All minor complications were either self-limiting or required minimal wound care. Of the patients in whom tubes were placed for feeding indications, 88% tolerated the feeding without difficulty. Aspiration or significant reflux was seen in only 5% of patients. Diarrhea resulting in termination of feedings or morbidity was observed in only 3% of patients. Other feeding difficulties included prolonged ileus and large gastric residuals (4%). Tube changes within 30 days involved 1 5% of the tubes. When exchange for a like tube was necessitated, the reason was either a defective hub on the tube, a deflated balloon (on Foley catheters), or a tube clogged because of improper flushing or feeding. This occurred with five Foley catheters, three Cope catheters, and four gastrojejunostomy catheters. Five successful and two unsuccessful conversions of gastrostomy catheters to gastrojejunostomy catheters were performed. Two gastrojejunostomy catheters were converted to gastrostomy tubes, both in patients who became less at risk for aspiration because of improvement in clinical status. Both patients tolerated gastrostomy feedings. No statistically significant difference was demonstrable between the three tube types with regard to patient age, sex, code status, major or minor morbidity, 30-day mortality, or feeding tolerance. There were two statistically significant differences between the tube types. Foley catheters were used in the two cases in which complications resulted in surgery, and a higher rate of tube changes was required for the Foley catheters in the 30-day follow-up period.
Discussion
The reported procedure-related mortality for surgical gastrostomy ranges from 1 .8% to 6.0% [1 5, 1 6]. Mortality related to percutaneous endoscopic gastrostomy placement generally is reported as less than 2% [1 7, 1 8]. Procedure-related morbidity ranges from 1 3% to 23% with surgically placed tubes and from 1 0% to 20% with endoscopic placement [3, 1 5, 1 9]. The largest published fluoroscopically guided percutaneous
gastrostomy
series
reported
a procedure-related
mortality of 0.8% and overall morbidity of 6.0% [12]. In an attempt to define our population of patients
more
clearly,
large
the
patients
were
classified
by code
status.
The
number of no-code patients reflects the significant number of debilitated and terminally ill patients included in our study group. To our knowledge, no other study has used this gross indicator of health to discern differences in outcome from the procedure. Even though the patients were not randomized, our study provides data to compare three different types of tubes. The only statistically significant differences between the tube types were an increased incidence of morbidity requiring surgery in the Foley catheter group and more frequent tube changes required in the Foley catheter group in the 30-day follow-up period. Although Foley catheters were the initial type of gastrostomy tube placed fluoroscopically at this institution, the two complications requiring surgery in the Foley catheter group (deflation of the balloon with peritonitis and erosion of the tube through the posterior stomach) appear more related to the nature of the tube than to inexperience with placement. For these reasons, we have abandoned the Foley catheter for use as a gastrostomy tube. At least three of the major morbidity complications in our series were probably preventable. The association of corticosteroid use and ulcer disease is a subject of debate. The patient in this study on steroids for lupus cerebritis with a gastrointestinal hemorrhage from gastritis should nevertheless serve as a reminder to anticipate this as a possible complication of gastrostomy in patients at risk for ulcer disease or gastritis. The patient on Coumadin with an elevated prothrombin time also had gastrointestinal hemorrhage, which was an avoidable and unfortunate complication of gastrostomy. The third preventable complication represents the only case where a Cope catheter migrated. This resulted either from improper securing of the fixation string or improper placement during formation ofthe loop. “Twirling” the catheter after formation of the loop during fluoroscopic observation ensures that the catheter is intraluminal and not intraperitoneal or partially formed in the wall of the stomach. In addition,
Downloaded from www.ajronline.org by UCSF LIB & CKM/RSCS MGMT on 12/19/14 from IP address 169.230.243.252. Copyright ARRS. For personal use only; all rights reserved
728
HICKS
the two complications requiring surgery may have been preventable if Cope catheters had been used instead of Foley catheters. As we found no difference in feeding tolerance or requirementfor tube change within 30 days (between Cope catheters and gastrojejunostomy catheters), we have not altered our practice of placing a gastrostomy tube rather than gastrojejunostomy tube for routine feeding access. We reserve gastrojejunostomy tubes for patients with known gastroesophageal reflux or documented aspiration, gastric atony, or partial gastric obstruction. The orientations of the muscular layers in the stomach wall are advantageous to percutaneous gastrostomy placement. Three distinct layers run in different directions: an inner oblique layer, a middle circular layer, and an outer longitudinal layer. This arrangement is probably the reason we were able to puncture and dilate the stomach wall without peritoneal leakage even if the access route was lost at the time of initial tube placement. This occurred in 1 0 cases; an additional puncture proved adequate for access in all cases. In only one case was additional puncture and tube placement associated with morbidity; as mentioned previously, this was not a statistically significant factor in morbidity. The muscular coat of the stomach is also probably responsible for the fact that it has proved unnecessary to appose the gastric wall and the anterior abdominal wall. Our series serves as further evidence that routine use of T fasteners for this purpose is not required. We have used T fasteners only in selected instances. We have found them useful for uncooperative patients when it is feared that the tube might be removed inadvertently. They are also helpful for conversion of a gastrostomy tube to a gastrojejunostomy tube within 7 days after placement. Because gastrocutaneous tracts are usually mature at 7 days, we do not use T tacks for conversion of gastrostomy tubes to gastrojejunostomy tubes more than 1 week after placement [1 8, 20]. Finally, we have used them in patients with ascites or those on peritoneal dialysis. The indications for gastrostomy or gastrojejunostomy and the classification of complications vary among institutions and make comparisons between reported series difficult. In addition, there are differences in patient populations and referral patterns. Our definition of major morbidity included some complications that would have been classified as minor complications in other studies, such as tube migration or infection requiring antibiotics or tube removal [1 2, 1 7, 21 , 22]. However, we believe that our classification reflects a definite distinction between levels of care necessary for patient support in the morbidity phase of the complication. The method of classification of complications must be a consideration when comparing results of different series of gastrostomy placement, whatever the means of insertion. In summary, our results compare favorably with surgical and percutaneous endoscopic gastrostomy as a safe and effective means of gastroenteric feeding or decompression.
ET AL.
AJR:154,
April 1990
We compared Foley catheters, Cope catheters, and gastrojejunostomy catheters for differences in morbidity, mortality, feeding tolerance, and tubal patency. Foley catheters had a higher frequency of morbidity requiring surgery and required more tube changes in the 30-day follow-up period. Therefore, we prefer the Cope type of gastrostomy tube for initial gastrostomy tube placement unless a gastrojejunostomy tube is indicated.
REFERENCES 1 . Preshaw RM. A percutaneous tube. Surg Gynecol
2. Gauderer
Obstet
method for inserting a feeding gastrostomy
1981;1 52:658-660
MW, Ponsky JL, Izant RJ. Gastrostomy
percutaneous
endoscopic
technique.
J Pediatr
without laparotomy:
Surg
3. Ponsky JL, Gauderer MW, Stellato TA. Percutaneous tomy:
review
of 150 cases.
Arch
Surg
4. Tao HH, Gillies RR. Percutaneous
a
1980; 1 5 :872-875
endoscopic
gastros-
1983;1 18:913-914
feeding gastrostomy.
1983;141:
AJR
793-794
5. Wills JS, Oglesby JT. Percutaneous
gastrostomy.
Radiology
1983;149:
449-453
6. Wills JS, Oglesby JT. Percutaneous Radiology 1985;154:71-74 7. Ho CS, Gray
RR, Goldfinger
M, Rosen
gastrostomy:
further
lE, McPherson
experience.
R. Percutaneous
gastrostomy for enteral feeding. Radiology 1985;156:349-351 8. vanSonnenberg E, Wittich GR, Cabrera OA, et al. Percutaneous tomy
and
gastrojejunostomy.
2. Clinical
experience.
AJR
581 -586 9. Alzate GD, Coons HG, Elliott J, Carey PH. Percutaneous
gastros1986;1 46:
gastrostomy
for
jejunal feeding: a new technique. AJR 1986;147:822-825 1 0. Gray RR, St. Louis EL, Grosman H. Percutaneous gastrostomy and gastrojejunostomy. Br J Radio! 1987;60: 1067-1 070 1 1 . Ho CS, Yee ACN, McPherson R. Complications of surgical and percutaneous nonendoscopic gastrostomy: review of 233 patients. Gastroentero!ogy 1988;95: 1206-1 210
12. Halkier BK, Ho CS, Yee ACN. Percutaneous Seldinger
technique:
review
of
252
feeding gastrostomy
patients.
Radiology
with the 1989;171:
359-362 13.
Brown AS, Mueller PR, Ferucci nylon T-fastener for fixation 1986;158:543-545
JT. Controlled percutaneous gastrostomy: of the anterior gastric wall. Radiology
gastrostomy. Radiology 1988:167: 41-43 Ruge J, Vazquez RM. An analysis of the advantages of Stamm and percutaneous endoscopic gastrostomy. Surg Gynecol Obstet 1986;
14. Wills JS, Oglesby JT. Percutaneous 15.
162: 13-16 16. Wasiljew BK, Vjiki GT, Beal JM. Feeding gastrostomy: mortality.
Am J Surg
complications
and
EP. Percutaneous
en-
1982;143:194-195
17. Larson DE, Burton DD, Schroeder
KW, DiMagno
doscopic gastrostomy: indications, success, in 314 consecutive patients. Gastroenterology
complications,
and mortality
1987:93:48-52 TA, Aszodi A. Percutaneous approaches to enteral alimentation. Am J Surg 1985:149: 102-1 05 19. Stiegmann G. Goff J, VanWay C, Perino L, Peariman N, Norton L. Operative versus endoscopic gastrostomy. Preliminary results of a prospective randomized trial. Am J Surg 1988:155:88-92 18. Ponsky
20.
21
.
JL, Gauderer
MW, Stellato
vanSonnenberg E, Wittich GR, Brown LK, et al. Percutaneous gastrostomy and gastroenterostomy. 1 . Techniques derived from laboratory evaluation. AIR 1986;146:577-580 Grant JP. Comparison of percutaneous endoscopic gastrostomy with Stamm gastrostomy. Ann Surg 1988:207:598-603
22. Sangster W, Cuddington gastrostomy.
Am J Surg
GD, Bachulis 1988:155:677-679
BL. Percutaneous
endoscopic
725
Fluoroscopically Guided Percutaneous Gastrostomy Gastroenterostomy: Analysis Consecutive
Marshall E. Hicks1 R. Stephen Surratt1 Daniel Picu& M. Victoria Marx1’2 Elvira V. Lang1’3
Received
July 24, 1 989; accepted
after
revision
Presented at the annual meeting of the American Roentgen Ray Society, New Orleans, May 1989. I Mallinckrodt Institute of Radiology, Washington University School of Medicine, 51 0 S. Kingshighway
Blvd., St. Louis, MO 631 10. Address quests to M. E. Hicks. 2Present address:
Department
University
1500 E. Medical
Ann Arbor, 3Present Stanford
of Michigan, address:
Radiology, Center
544-0725
Roentgen
Ray Society
154:725-728,
April 1990
For over 1 00 years, surgical gastrostomy and gastrojejunostomy were the standard methods of direct enteral access. This began to change, however, with the development of percutaneous endoscopic gastrostomy in 1979 [1 -3]. Despite the lower morbidity and mortality with endoscopic placement, the search for a simpler, safer means of gastroenteric access has continued, resulting in the recent development of fluoroscopically guided percutaneous gastrostomy. Early series of fluoroscopically guided percutaneous gastrostomies contained small numbers of patients, but favorable morbidity and mortality rates were reported when comparisons were made with endoscopic placement [4-7]. Recently, larger series have begun to emerge, substantiating the results of the initial reports [812]. We report our experience with percutaneous gastrostomy and gastroenterostomy in 1 58 consecutive patients, comparing three different types of tubes used during a 2-year period.
Materials
and
Methods
Dr.,
Department of Radiology, University School of Medicine, Stanford,
0361 -803X/90/1
Cases
reprint re-
MI 48109-0030.
CA 94305-51 05. C American
of
of 158
We reviewed our experience with 158 consecutive patients who underwent either percutaneous gastrostomy or percutaneous gastroenterostomy during a 2-year period. The catheters used included Foley catheters (36), Cope-type gastric catheters (86), or Carey-Alzate-Coons gastrojejunostomy catheters (36). Gastrojejunostomy tubes were placed in patients with gastroesophageal reflux or aspiration, gastric atony, or partial gastric obstruction. Ninety percent of the tubes were placed for feeding purposes. The technical success rate was 100%. Thirty-day follow-up was obtained in 89%. Thirty-day mortality was 26%, reflecting the substantial number of debilitated patients. No deaths were directly related to tube placement. Major morbidity was 6% and included hemorrhage, peritonitis, tube migration, and sepsis. Minor morbidity was 12%. There was no difference in 30-day mortality or feeding tolerance between the tube types (p < .05). Patients with Foley catheters had more complications necessitating surgical intervention and an increased incidence of tube changes required within 30 days. These were the only statistically significant differences between the tubes (p < .05). Our results show that percutaneous gastrostomy is a safe and effective means of gastroenteric feeding or decompression. Because of the fewer complications and ease of insertion, the Cope type of gastrostomy tube has become our preferred catheter for percutaneous feeding or decompression. AJR
November 20, 1989.
and
We reviewed percutaneous
performed.
the medical
records
of the first
gastrostomy
or gastrojejunostomy.
The procedure
was not attempted
1 61 consecutive One hundred
patients
fifty-eight
referred
procedures
in three cases (two patients
for were
with overlying
colon and one with massive gastroesophageal reflux in whom we were unable to maintain adequate inflation of the stomach for safe percutaneous puncture). The study group of 158 patients included 95 women (60%) and 63 men (40%). The mean age was 67 years; the
HICKS
726
median
age was 71 years. Four patients (2.5%) had ascites. Three (1 .9%) had histories of prior gastric surgery; this included two patients who had gastrojejunostomies for palliative treatment of
Downloaded from www.ajronline.org by UCSF LIB & CKM/RSCS MGMT on 12/19/14 from IP address 169.230.243.252. Copyright ARRS. For personal use only; all rights reserved
patients
pancreatic for peptic
used
carcinoma ulcer one
one
patient
T fasteners
patients
(4.4%),
in seven
dialysis,
and
disease. uncooperative
who
had
(Meditech,
including
patient,
one patient
and
[1 3]. At the time of tube placement,
a Billroth
I operation
Watertown, one
MA)
were
on peritoneal
patient
with
ascites
72 patients (46%) or their families
had requested that no resuscitative measures be undertaken in the event of cardiopulmonary arrest. Of the 90% of tubes placed for feeding purposes, the majority were for patients with neurologically impaired swallowing mechanisms (1 04 patients). Other feeding indications included patients with tumors of the ears, nose, or throat (1 0 patients), esophageal obstruction
(nine
patients),
and
miscellaneous
(1 9 patients). Only of the stomach or small
indications
10% of the tubes were for decompression bowel in patients with bowel obstruction.
In the first 36 patients who underwent gastrostomy, 1 6- or 18French Foley catheters were placed (Bard, Covington, GA). In the remaining 86 patients undergoing gastrostomy, 12- or 1 4-French
Cope-type
catheters
jejunostomy
were placed (Cook, Bloomington,
catheters
Carey-Alzate-Coons
were
placed
in 36 patients.
gastrojejunostomy
IN). GastroAll 36 had the
catheter(Cook).
Candidates
for these gastroenteric feeding tubes included patients with documented aspiration, gastric atony, or partial gastric obstruction. Local anesthetic was used in all cases. IV sedation with midazolam and morphine sulfate was used routinely. No patient received general anesthesia. Procedures were performed under fluoroscopic guidance, with the exception of one case in which CT was used because the patient had extensive bowel gas in the vicinity of the stomach. No additional imaging techniques were used to identify liver or colon. Antibiotics were not given routinely. Glucagon (1 mg, Lilly, Indianapohs, IN) was given IV to inhibit gastric motility. All catheters were placed by using the Seldinger technique with air insufflation of the stomach via an indwelling nasogastric tube. After puncture with an 1 8-gauge Teflon-sheathed needle (Cook), an 0.038in. (1 -mm) Amplatz extra-stiff guidewire (Cook) was passed into the stomach. We used a method for puncture described and refined by Wills
and
dilatation
Oglesby of the
Cope
[14]. tract.
Foley
catheters
catheters
and
were
placed
after serial
gastrojejunostomy
cathe-
ters were placed through a peel-away sheath of the appropriate size. When T fasteners were used, they were placed before catheter insertion.
Catheters
Inc., Somerville,
Tube feedings
were
secured
to the skin
with
0-prolene
(Ethicon,
NJ).
were initiated
on the day after placement
for gas-
trostomy tubes, unless a complication precluding feeding developed. Gastrojejunostomy patients routinely began receiving feedings on the same day as tube placement. The nasogastric tube usually was removed the day after the procedure. The chi square and Van der Waerden one-way tests were used to compare the tube types for differences in patient age, sex, code status, morbidity, mortality, feeding tolerance, and required tube changes and to identify risk factors with respect to increased morbidity and mortality (p < .05).
Results
The technical success rate was 1 00% in all 1 58 cases in which the procedure was attempted. More than one puncture was required in 10 cases (6%). This was usually because the stomach deflated during placement of the catheter. In all instances the additional puncture was successful.
ET AL.
AJR:154,
April 1990
Thirty-day follow-up was obtained in 1 41 patients (89%). The 30-day mortality rate calculated from these 1 41 patients was 26%. Twenty-seven of the 36 patients who were dead at 30 days died of unrelated causes. Although there were no cases where tube placement was definitely related to the death of a patient, there were three cases (2%) in which a complication occurred from tube placement that negatively altered the clinical course of the patient and may have contributed to the patient’s death. In an additional six patients (4%), no obvious precipitating cause of death was recorded in the physician’s record note section of the chart; however, in all, multiorgan failure was believed to be the cause of death. All six of these patients were assigned the status of “no code.” Three of these patients died less than 48 hr after uncomplicated placement of a gastrostomy tube. The other three also had uncomplicated procedures and were tolerating tube feedings before they died. Morbidity was divided into major and minor. Major morbidity was defined as a complication that (1 ) required an additional interventional procedure, (2) prolonged the hospital stay, (3) required a transfusion, or (4) required IV antibiotics. Major morbidity consisted of five cases of hemorrhage (3%), two cases of peritonitis (1 %), two cases of tube migration (1 %), and one case of sepsis (1 %). Three of the five patients with hemorrhage recovered uneventfully. One of these patients was on steroids for lupus cerebritis at the time of tube placement and experienced gastrointestinal bleeding after the procedure. Diffuse gastritis was found with endoscopy after tube placement. The patient was transfused with 3 units of packed RBCs and the bleeding resolved. The gastritis healed with tapering the administration of the steroids.
In another
patient,
there
was
a small
amount
of bleeding
with the initial puncture at the time of catheter placement. This was probably from perforation of a vein in the posterior wall of the stomach. The hemoglobin dropped 2 g/dl over a 2-day
period,
and
the
nasogastric
tube
aspirate
was
positive
for occult blood, indicating a small amount of persistent hemorrhage. This patient was transfused with 2 units of packed RBCs and the bleeding ceased. In a third patient, another slow gastrointestinal hemorrhage was manifested by a melanotic stool and a decrease in hematocrit from 39% to 27% over a 3-day period. No difficulties were noted in tube placement. This patient experienced no further bleeding after transfusion with 2 units of packed RBCs. The remaining two patients with bleeding complications were classified as having possible tube-related deaths, although both had ongoing multisystem failure. One of these two patients was on Coumadin for atrial fibrillation with a prothrombin time greater than 40 sec at the time of gastrojejunostomy tube placement. Gastrointestinal hemorrhage ensued after catheter placement. Although the bleeding subsided within 1 2 hr after the procedure with correction of the prothrombin time by transfusion of 1 0 units of fresh frozen plasma and 6 units of packed RBCs, the patient died 5 days after tube placement. She had chronic renal failure and congestive heart failure; the acute cause of death was respiratory failure. The other patient had an occult hemorrhage; the hematocrit dropped from 34% to 20% in the 2 weeks
PERCUTANEOUS
AJR:154, April 1990
after tube placement.
Downloaded from www.ajronline.org by UCSF LIB & CKM/RSCS MGMT on 12/19/14 from IP address 169.230.243.252. Copyright ARRS. For personal use only; all rights reserved
may
have
been
caused
The patient
had a prolonged
by the occult
hemorrhage
GASTROSTOMY
ileus, which or ongoing
urosepsis. The family requested that no supportive measures be given and the patient died 1 6 days after catheter placement. One of the two patients included in the peritonitis complication group was a patient in acute renal failure with altered mental status. This made clinical evaluation of this complication
difficult,
although
peritonitis
was
suspected.
This
was
GASTROENTEROSTOMY
727
Patients who were classified as no code at the time of tube placement were at risk for death within the 30-day follow-up period (p < .05); however, the no-code status did not entail a higher incidence of morbidity. Other risk factors found not to be statistically significant for an increase in morbidity or mortality included sex, age, difficulty with placement, or anatomic considerations such as ascites or prior gastric surgery.
the
only case in which an additional puncture at the time of placement was associated with morbidity. The patient’s family requested that no further supportive therapy be given and the patient died 6 days after tube placement. The other case of peritonitis was proved and occurred after deflation of the balloon on the Foley gastrostomy catheter 1 day after tube placement. This patient required a Stamm gastrostomy and lavage and recovered uneventfully. Of the two cases of tube migration, one also had a Foley catheter that eroded into the lesser sac 1 2 days after tube placement. A resultant abscess required surgical evacuation with a jejunostomy tube placement. The other tube migration complication was caused by a Cope catheter that was partially outside the stomach. An additional percutaneous gastrostomy tube was placed and the patient recovered uneventfully. The final major morbidity complication was in a patient with Proteus sepsis from a wound infection. This patient recovered with
AND
IV antibiotics.
Minor morbidity included skin infection (seven patients), transient increase in fever or WBCs (seven patients), transient ascitic leak (two patients), minor skin bleeding (two patients), and catheter migration (one patient). All minor complications were either self-limiting or required minimal wound care. Of the patients in whom tubes were placed for feeding indications, 88% tolerated the feeding without difficulty. Aspiration or significant reflux was seen in only 5% of patients. Diarrhea resulting in termination of feedings or morbidity was observed in only 3% of patients. Other feeding difficulties included prolonged ileus and large gastric residuals (4%). Tube changes within 30 days involved 1 5% of the tubes. When exchange for a like tube was necessitated, the reason was either a defective hub on the tube, a deflated balloon (on Foley catheters), or a tube clogged because of improper flushing or feeding. This occurred with five Foley catheters, three Cope catheters, and four gastrojejunostomy catheters. Five successful and two unsuccessful conversions of gastrostomy catheters to gastrojejunostomy catheters were performed. Two gastrojejunostomy catheters were converted to gastrostomy tubes, both in patients who became less at risk for aspiration because of improvement in clinical status. Both patients tolerated gastrostomy feedings. No statistically significant difference was demonstrable between the three tube types with regard to patient age, sex, code status, major or minor morbidity, 30-day mortality, or feeding tolerance. There were two statistically significant differences between the tube types. Foley catheters were used in the two cases in which complications resulted in surgery, and a higher rate of tube changes was required for the Foley catheters in the 30-day follow-up period.
Discussion
The reported procedure-related mortality for surgical gastrostomy ranges from 1 .8% to 6.0% [1 5, 1 6]. Mortality related to percutaneous endoscopic gastrostomy placement generally is reported as less than 2% [1 7, 1 8]. Procedure-related morbidity ranges from 1 3% to 23% with surgically placed tubes and from 1 0% to 20% with endoscopic placement [3, 1 5, 1 9]. The largest published fluoroscopically guided percutaneous
gastrostomy
series
reported
a procedure-related
mortality of 0.8% and overall morbidity of 6.0% [12]. In an attempt to define our population of patients
more
clearly,
large
the
patients
were
classified
by code
status.
The
number of no-code patients reflects the significant number of debilitated and terminally ill patients included in our study group. To our knowledge, no other study has used this gross indicator of health to discern differences in outcome from the procedure. Even though the patients were not randomized, our study provides data to compare three different types of tubes. The only statistically significant differences between the tube types were an increased incidence of morbidity requiring surgery in the Foley catheter group and more frequent tube changes required in the Foley catheter group in the 30-day follow-up period. Although Foley catheters were the initial type of gastrostomy tube placed fluoroscopically at this institution, the two complications requiring surgery in the Foley catheter group (deflation of the balloon with peritonitis and erosion of the tube through the posterior stomach) appear more related to the nature of the tube than to inexperience with placement. For these reasons, we have abandoned the Foley catheter for use as a gastrostomy tube. At least three of the major morbidity complications in our series were probably preventable. The association of corticosteroid use and ulcer disease is a subject of debate. The patient in this study on steroids for lupus cerebritis with a gastrointestinal hemorrhage from gastritis should nevertheless serve as a reminder to anticipate this as a possible complication of gastrostomy in patients at risk for ulcer disease or gastritis. The patient on Coumadin with an elevated prothrombin time also had gastrointestinal hemorrhage, which was an avoidable and unfortunate complication of gastrostomy. The third preventable complication represents the only case where a Cope catheter migrated. This resulted either from improper securing of the fixation string or improper placement during formation ofthe loop. “Twirling” the catheter after formation of the loop during fluoroscopic observation ensures that the catheter is intraluminal and not intraperitoneal or partially formed in the wall of the stomach. In addition,
Downloaded from www.ajronline.org by UCSF LIB & CKM/RSCS MGMT on 12/19/14 from IP address 169.230.243.252. Copyright ARRS. For personal use only; all rights reserved
728
HICKS
the two complications requiring surgery may have been preventable if Cope catheters had been used instead of Foley catheters. As we found no difference in feeding tolerance or requirementfor tube change within 30 days (between Cope catheters and gastrojejunostomy catheters), we have not altered our practice of placing a gastrostomy tube rather than gastrojejunostomy tube for routine feeding access. We reserve gastrojejunostomy tubes for patients with known gastroesophageal reflux or documented aspiration, gastric atony, or partial gastric obstruction. The orientations of the muscular layers in the stomach wall are advantageous to percutaneous gastrostomy placement. Three distinct layers run in different directions: an inner oblique layer, a middle circular layer, and an outer longitudinal layer. This arrangement is probably the reason we were able to puncture and dilate the stomach wall without peritoneal leakage even if the access route was lost at the time of initial tube placement. This occurred in 1 0 cases; an additional puncture proved adequate for access in all cases. In only one case was additional puncture and tube placement associated with morbidity; as mentioned previously, this was not a statistically significant factor in morbidity. The muscular coat of the stomach is also probably responsible for the fact that it has proved unnecessary to appose the gastric wall and the anterior abdominal wall. Our series serves as further evidence that routine use of T fasteners for this purpose is not required. We have used T fasteners only in selected instances. We have found them useful for uncooperative patients when it is feared that the tube might be removed inadvertently. They are also helpful for conversion of a gastrostomy tube to a gastrojejunostomy tube within 7 days after placement. Because gastrocutaneous tracts are usually mature at 7 days, we do not use T tacks for conversion of gastrostomy tubes to gastrojejunostomy tubes more than 1 week after placement [1 8, 20]. Finally, we have used them in patients with ascites or those on peritoneal dialysis. The indications for gastrostomy or gastrojejunostomy and the classification of complications vary among institutions and make comparisons between reported series difficult. In addition, there are differences in patient populations and referral patterns. Our definition of major morbidity included some complications that would have been classified as minor complications in other studies, such as tube migration or infection requiring antibiotics or tube removal [1 2, 1 7, 21 , 22]. However, we believe that our classification reflects a definite distinction between levels of care necessary for patient support in the morbidity phase of the complication. The method of classification of complications must be a consideration when comparing results of different series of gastrostomy placement, whatever the means of insertion. In summary, our results compare favorably with surgical and percutaneous endoscopic gastrostomy as a safe and effective means of gastroenteric feeding or decompression.
ET AL.
AJR:154,
April 1990
We compared Foley catheters, Cope catheters, and gastrojejunostomy catheters for differences in morbidity, mortality, feeding tolerance, and tubal patency. Foley catheters had a higher frequency of morbidity requiring surgery and required more tube changes in the 30-day follow-up period. Therefore, we prefer the Cope type of gastrostomy tube for initial gastrostomy tube placement unless a gastrojejunostomy tube is indicated.
REFERENCES 1 . Preshaw RM. A percutaneous tube. Surg Gynecol
2. Gauderer
Obstet
method for inserting a feeding gastrostomy
1981;1 52:658-660
MW, Ponsky JL, Izant RJ. Gastrostomy
percutaneous
endoscopic
technique.
J Pediatr
without laparotomy:
Surg
3. Ponsky JL, Gauderer MW, Stellato TA. Percutaneous tomy:
review
of 150 cases.
Arch
Surg
4. Tao HH, Gillies RR. Percutaneous
a
1980; 1 5 :872-875
endoscopic
gastros-
1983;1 18:913-914
feeding gastrostomy.
1983;141:
AJR
793-794
5. Wills JS, Oglesby JT. Percutaneous
gastrostomy.
Radiology
1983;149:
449-453
6. Wills JS, Oglesby JT. Percutaneous Radiology 1985;154:71-74 7. Ho CS, Gray
RR, Goldfinger
M, Rosen
gastrostomy:
further
lE, McPherson
experience.
R. Percutaneous
gastrostomy for enteral feeding. Radiology 1985;156:349-351 8. vanSonnenberg E, Wittich GR, Cabrera OA, et al. Percutaneous tomy
and
gastrojejunostomy.
2. Clinical
experience.
AJR
581 -586 9. Alzate GD, Coons HG, Elliott J, Carey PH. Percutaneous
gastros1986;1 46:
gastrostomy
for
jejunal feeding: a new technique. AJR 1986;147:822-825 1 0. Gray RR, St. Louis EL, Grosman H. Percutaneous gastrostomy and gastrojejunostomy. Br J Radio! 1987;60: 1067-1 070 1 1 . Ho CS, Yee ACN, McPherson R. Complications of surgical and percutaneous nonendoscopic gastrostomy: review of 233 patients. Gastroentero!ogy 1988;95: 1206-1 210
12. Halkier BK, Ho CS, Yee ACN. Percutaneous Seldinger
technique:
review
of
252
feeding gastrostomy
patients.
Radiology
with the 1989;171:
359-362 13.
Brown AS, Mueller PR, Ferucci nylon T-fastener for fixation 1986;158:543-545
JT. Controlled percutaneous gastrostomy: of the anterior gastric wall. Radiology
gastrostomy. Radiology 1988:167: 41-43 Ruge J, Vazquez RM. An analysis of the advantages of Stamm and percutaneous endoscopic gastrostomy. Surg Gynecol Obstet 1986;
14. Wills JS, Oglesby JT. Percutaneous 15.
162: 13-16 16. Wasiljew BK, Vjiki GT, Beal JM. Feeding gastrostomy: mortality.
Am J Surg
complications
and
EP. Percutaneous
en-
1982;143:194-195
17. Larson DE, Burton DD, Schroeder
KW, DiMagno
doscopic gastrostomy: indications, success, in 314 consecutive patients. Gastroenterology
complications,
and mortality
1987:93:48-52 TA, Aszodi A. Percutaneous approaches to enteral alimentation. Am J Surg 1985:149: 102-1 05 19. Stiegmann G. Goff J, VanWay C, Perino L, Peariman N, Norton L. Operative versus endoscopic gastrostomy. Preliminary results of a prospective randomized trial. Am J Surg 1988:155:88-92 18. Ponsky
20.
21
.
JL, Gauderer
MW, Stellato
vanSonnenberg E, Wittich GR, Brown LK, et al. Percutaneous gastrostomy and gastroenterostomy. 1 . Techniques derived from laboratory evaluation. AIR 1986;146:577-580 Grant JP. Comparison of percutaneous endoscopic gastrostomy with Stamm gastrostomy. Ann Surg 1988:207:598-603
22. Sangster W, Cuddington gastrostomy.
Am J Surg
GD, Bachulis 1988:155:677-679
BL. Percutaneous
endoscopic
