2. Children's Hospital/University of Michigan Medical Center
found 13 children who were treated for a recurrent TEF. Four
of these were initially repaired at outside institutions. The
other 13 children included in the study were operated on in 4
hospitals throughout the United States and across the world.
The senior author (ACG) was involved in all the cases.
Gestational age and birth weight were recorded. Details of
the original repair as well as complications were identified.
The symptoms leading to diagnosis and the method of
diagnosis were documented. Data are reported as mean
(range). Descriptive statistical analysis was performed using
the SPSS program (SPSS, Inc, Chicago, IL). Missing data
were excluded from the statistical analysis.
2. Results
Twenty-six children with a recurrent TEF had a mean
gestational age of 35.3 (24-40) weeks and a mean birth weight
of 2100 (795-3500) g. Of those developing a recurrence; 21
initially had an EA and distal TEF, 1 had an EA with both a
proximal and distal TEF, and 4 had an H-type TEF. Their first
operation was a primary repair (n = 21), gastrostomy tube
placement and delayed repair (n = 3), and ligation of the
fistula followed by delayed repair (n = 2). A leak was noted in
18 (68%) of 26 infants on their postoperative esophagram
after original repair.
Most children who develop a recurrent fistula have
respiratory and/or feeding problems leading to the diagnosis.
These symptoms are often difficult to distinguish from those
of tracheomalacia and gastroesophageal reflux (GER) that
occur frequently in infants born with EA. Our group
included 11 with tracheomalacia, 1 requiring an aortopexy,
and 19 with GER, of which 14 required a fundoplication.
Symptoms leading to the diagnosis of recurrence were
respiratory symptoms only (n = 7), a combination of
respiratory symptoms and feeding issues (n = 15), and
feeding issues only (n = 1). Three recurrences were
asymptomatic and discovered on routine contrast studies.
These symptoms usually occurred relatively soon after the
original repair. There were 2 patients who were diagnosed a
long time after their original diagnosis, one 5 years after the
original repair and one 18 years after her first repair.
Excluding these 2 outliers, the mean time to recurrence was
2.1 months. Diagnosis was made in 24 children with a
contrast study and then confirmed with bronchoscopy. When
bronchoscopy was used, a fistula was seen directly in one
half of the recurrences, whereas in the others, a dye study
was required to confirm the presence of the TEF. In both
recurrent fistulas diagnosed only with bronchoscopy, a dye
study was required.
Three children in this series underwent at least 2 unsuc-
cessful attempts at endoscopic closure before definitive repair.
At the time of exploration, the recurrent fistulas were
located almost exclusively at the site of the original TEF. On
2 occasions, the fistula was found at the site of an injury to
the trachea that occurred during the original operation. The
repairs entailed locating the fistula, separating the trachea
from the esophagus sharply, and then placing some type of
tissue between the suture lines to prevent future recurrences.
Multiple types of tissue were used for this purpose
including pleura (n = 11), pericardium (n = 8), a muscle flap
(n = 3), and lymph nodes (n = 2). One recurrent fistula was
repaired with no tissue between the suture lines, and one
eventually resulted in a gastric pull-up. Postoperatively, the
majority of the children did well; however, there were some
complications. Contrast studies performed postoperatively in
25 children revealed 7 leaks, 5 being minor and resolving by
the next contrast study. Strictures developed in 4, and 3
repairs resulted in a second recurrent fistula.
In the first of these re-recurrences, the original repair had
been attempted in a 920-g, 24-week-premature infant that
resulted in a tracheal injury—the site of recurrence. At the
subsequent operation, no tissue was placed between the
suture lines; and a second recurrence resulted. This was
repaired through the neck, and a portion of the sternoclei-
domastoid muscle was secured between the suture lines. The
second patient had a pleural flap interposed at the time of first
repair and a pericardial flap used for the second successful
repair. The third patient was a 1.43-kg, 32-week-premature
infant who underwent repair at day 2 and then had a “revision
of the TEF with tracheal repair,” both at an outside hospital.
This infant had an esophageal stricture and a fistula
extending from the stricture to the carina. At the repair of
the first recurrence, a catheter was not placed across the
fistula; and there was difficulty identifying the esophageal
side of the fistula. Routine postoperative esophogram
showed a recurrence, and he went on to have a cervical
esophagostomy and eventually a gastric pull-up.
Long-term outcome was reviewed with a median follow-
up of 84 (6-287) months looking at method of feeds and
growth. Almost all children were eating exclusively by
mouth, 3 took all nutrition by tube, and 2 required a
combination of oral and tube feeding. Growth chart data
were available in 23 of the children and showed the
following distribution: 0 to 25th percentile (n = 16), 26th
to 50th percentile (n = 4), and 51st to 75th percentile (n = 3).
3. Discussion
Recurrent TEF continues to occur at a rate of up to 10%,
making this a complication that most pediatric surgeons will
have to deal with at some time [1-3]. In our series, the infants
who developed a recurrent fistula tended to be born more
prematurely and at a lower birth weight than expected, had a
tracheal suture line, and tended to have a leak on contrast
study after their initial repair. Purisch et al [4] in a review of
congenital malformations in preterm infants found a mean
gestational age of 37.6 weeks and a mean birth weight of
338 S.W. Bruch et al.
3. 2742 g in 93 infants born with a tracheoesophageal anomaly.
Our study group was born at a mean of 35.4 weeks, with a
mean birth weight of 2100 g. This birth weight falls between
the fifth and 10th percentile of the expected birth weight of
infants born at 35.4 weeks' gestation [5]. All of the
recurrences had a tracheal suture line after their initial repair.
The recurrences entered the trachea at the original fistula site
or at the site of a tracheal injury repaired at the initial
procedure. No fistulae occurred in infants after a pure EA
repair. An anastomotic leak was noted in 68% of the initial
repairs that eventually developed a recurrence. This leak rate
is more than triple the expected 15% to 20% leak rate after
TEF repair as quoted in a review by Spitz in 2007 [1].
Most infants who develop a recurrent TEF usually have
symptoms. These symptoms are often difficult to distinguish
from symptoms of tracheomalacia or GER commonly found
in infants with repaired EA. These symptoms were evaluated
with a contrast study looking for reflux and ended up
showing the recurrence (Fig. 1). Most of our recurrences
were detected in this way. This probably has more to do with
the frequency with which the study is ordered and not
necessarily with its sensitivity. Recurrent fistulas can be
missed quite easily on routine contrast studies, especially if a
prone pull-back method is not used. The best way to
demonstrate a fistula is to place a tube down the esophagus to
the gastroesophageal junction, place the infant prone, and
inject contrast into the esophagus as the tube is pulled back
into the proximal esophagus [6]. Bronchoscopy is then used
to help make or confirm the diagnosis of a recurrence. At the
time of bronchoscopy, the pit from the original fistula that
was repaired is easily seen usually near the carina or in the
upper trachea in the case of an H-type fistula. A small
Fogarty catheter may be passed into the pit and then retrieved
in the esophagus with esophagoscopy. This is ideally placed
just before definitive repair, as the catheter can be left in
place to assist in finding the exact location of the fistula
intraoperatively (Fig. 2). However, as most have a small
diameter fistula, it may be difficult to place a tube through it
to make the diagnosis. In those cases, a methylene blue test
can be used to make the diagnosis of a recurrence. The
bronchoscope is placed in the trachea, and the pit is
identified. The bronchoscope is then removed, and an
Fig. 1 Esophagram showing a recurrent TEF with contrast
running off into the distal bronchus.
Fig. 2 Depiction of a Fogarty catheter placed through the TEF
via bronchoscopy. Having this catheter in place makes locating the
fistula at the time of repair much easier.
Fig. 3 Schematic depicting the pericardial flap that is raised after
dissecting away the phrenic nerve with its accompanying blood
supply. The pericardial flap is rolled into the grove between the
trachea and the esophagus and sutured in place to provide 2 layers
of viable pericardium between the 2 suture lines.
339Management of recurrent tracheoesophageal fistulas
4. endotracheal tube one-half size smaller than would fit snug is
introduced and placed above the area of the pit. An
endoscope is then placed in the esophagus. Methylene blue
(about 5 mL) is then placed in the endotracheal tube, and the
baby is ventilated to spread the methylene blue around the
trachea. Inspection of the esophagus will reveal blue dye in
the esophagus at the location of the fistula.
There are several options for repair. There have been a
number of reports describing endoscopic techniques placing
adhesives in the recurrent fistula tract. The quoted rates of
success vary from 60% to 90% overall, with about 2 attempts
on average being required for closure [7-10]. This method
works best when the fistula tract is first de-epithelialized by
mechanical means and then fibrin glue or a similar agent is
placed into the fistula tract. There was more than the usual
amount of inflammation at the site of recurrence in those
children who had failed endoscopic closures.
We believe there are 3 key points that should be adhered
to at the time of repair to minimize the chances of a second
recurrence: the placement of a catheter through the fistula
before opening the chest, complete separation of the trachea
from the esophagus before dividing the fistula, and
placement of viable tissue between the suture lines. If the
fistula cannot be cannulated, the dye study described earlier
can be done to localize the fistula opening in the esophagus.
A small flexible endoscope can then be placed in the
esophagus; and during the thoracotomy, its light will help
identify the site of the fistula. With the fistula identified, the
esophagus and trachea are completely separated before
dividing the fistula. The 2 ends can then be sutured closed,
and viable tissue is placed between the suture lines. Our
tissue preference is pericardium, if available [11]. This flap
should be fashioned avoiding the phrenic nerve and inter-
posed between the suture lines (Fig. 3) to provide a well-
vascularized piece of tissue that has enough substance to
prevent recurrence.
A recurrent TEF presents a difficult problem, both with
diagnosis and treatment. Hopefully, this series, the largest so
far reported in the literature, will provide a framework to
assist in its management.
References
[1] Spitz L. Oesophageal atresia. Orphanet J Rare Dis 2007;2:24.
[2] Ghandour KE. Recurrent tracheo-oesophageal fistula: experience with
24 patients. J Paediatr Child Health 1990;26:89-91.
[3] Kovesi T, Rubin S. Long-term complications of congenital esophageal
atresia and /or tracheoesophageal fistula. Chest 2004;126:915-25.
[4] Purisch SE, DeFranco EA, Mugllia LJ, et al. Preterm birth in
pregnancies complicated by major congenital malformations: a
population-based study. Am J Obstet Gynecol 2008;199:287.e1-8.
[5] Kramer MS, Platt RW, Wen SW, et al. A new and improved
population-based Canadian reference for birth weight for gestational
age. Pediatrics 2001;e35:108.
[6] Ein SH, Stringer DA, Stephens CA, et al. Recurrent tracheoesophageal
fistulas seventeen-year review. J Pediatr Surg 1983;18:436-41.
[7] Richter GT, Ryckman F, Brown RL, et al. Endoscopic management of
recurrent tracheoesophageal fistula. J Pediatr Surg 2008;43:238-45.
[8] Meier JD, Sulman CG, Almond PS, et al. Endoscopic management of
recurrent congenital tracheoesophageal fistula: a review of techniques
and results. Int J Pediatr Otorhino 2007;71:691-7.
[9] Tzifa KT, Maxwell EL, Chait P, et al. Endoscopic treatment of
congenital H-type and recurrent tracheoesophageal fistula with
electrocautery and histoacryl glue. Int J Pediatr Otorhino 2006;70:
925-30.
[10] San Roman CG, Barrios JE, Lluna J, et al. Long-term assessment of the
treatment of recurrent tracheoesophageal fistula with fibrin glue
associated with diathermy. J Pediatr Surg 2006;41:1870-3.
[11] Wheatley MJ, Coran AG. Pericardial flap interposition for the
definitive management of recurrent tracheoesophageal fistula. J Pediatr
Surg 1992;27:1122-6.
340 S.W. Bruch et al.