Long-term surgical outcomes in congenital diaphragmatic
hernia: observations from a single institution
Tim Jancelewicza, Lan T. Vua, Roberta L. Kellerb, Barbara Brattona,
Hanmin Leea, Diana Farmera, Michael Harrisona, Doug Miniatia,
Tippi Mackenziea, Shinjiro Hirosea, Kerilyn Nobuharaa,⁎
aDivision of Pediatric Surgery, Department of Surgery, University of California, San Francisco, CA 94143, USA
bDepartment of Pediatrics, University of California, San Francisco, CA 94143, USA
Received 3 October 2009; accepted 6 October 2009
Background/Purpose: Surgical complications are common in survivors of congenital diaphragmatic
hernia (CDH), but little is known about long-term incidence patterns and associated predictors.
Methods: A cohort of 99 CDH survivors was prospectively followed at a single-institution
chest and spinal deformity, and operative small bowel obstruction (SBO), and then were retrospectively
analyzed in relation to perinatal and perioperative markers of disease severity to determine significant
predictors. Statistical methods used included univariate and multivariate regression analysis, hazard
modeling, and Kaplan-Meier analysis.
Results: At a median cohort age of 4.7 (range, 0.2-10.6) years, 46% of patients with patch repairs and 10%
of those with primary repairs had a hernia recurrence at a median time of 0.9 (range, 0.1-7.3) years after
repair. Chest deformity was detected in 47%. Small bowel obstruction and scoliosis occurred in 13%.
Recurrence and chest deformity were significantly more common with patch repair, liver herniation, age at
neonatal extubation greater than 16 days, oxygen requirement at discharge, and prematurity. The strongest
predictor of SBO was patch repair. Multivariate analysis showed that patch repair was independently
predictive of recurrence and early chest deformity (odds ratios of 5.0 and 4.8, confidence intervals of 1-24
Conclusions: For long-term survivors of CDH, specific perinatal and operative variables, particularly patch
repair, are associated with subsequent adverse surgical outcomes.
© 2010 Published by Elsevier Inc.
The survival of newborns with congenital diaphragmatic
hernia (CDH) has increased along with improvements in
pulmonary management [1,2]. A corresponding rise in the
number of survivors with significant chronic medical and
surgical morbidity must be anticipated. To improve short-
and long-term management of these infants, further studies
are necessary to clarify the prevalence of and risk factors for
specific surgical complications.
The most common adverse surgical outcomes in infants
with CDH include diaphragmatic hernia recurrence, chest
wall and spinal deformities, small bowel obstruction (SBO),
Presented at the 40th Annual Meeting of the American Pediatric Surgical
Association, Fajardo, Puerto Rico, May 28-June 1, 2009.
⁎Corresponding author. Tel.: +1 415 476 3070.
E-mail address: email@example.com (K. Nobuhara).
0022-3468/$ – see front matter © 2010 Published by Elsevier Inc.
Journal of Pediatric Surgery (2010) 45, 155–160
hernia incarceration, and midgut volvulus [3-5]. Prosthetic
patch repair is associated with subsequent hernia recurrence
in up to half of patients by 3 years of age, with many patients
having multiple recurrences . The prevalence of chest
deformity also approaches 50%; and the incidence of
scoliosis has been reported in up to 27% of CDH survivors,
although large studies are lacking [6-8]. Literature is also
sparse regarding the incidence and causes of intestinal
obstruction after CDH repair, but this has been reported to
occur in 4% to 21% of patients [4,5].
We prospectively followed a cohort of CDH survivors at a
single-center multidisciplinary clinic and documented surgi-
cal outcomes. The purpose of this study was to describe the
patterns and frequency of surgical complications in this
population and then retrospectively examine perinatal
clinical and operative data to identify significant risk factors
for specific adverse surgical outcomes.
Permission for this study was obtained from the
Committee on Human Research of the University of
California, San Francisco (UCSF), H11258-30536-01.
1.1. Study cohort
Ninety-nine CDH patients that survived to discharge and
were seen at the UCSF multidisciplinary clinic between
2000 and 2008 were prospectively followed. We documen-
ted perinatal and perioperative variables that could serve as
potential predictors of adverse outcomes, including prenatal
anatomical evaluations with determination of the presence
of liver herniation into the thorax by ultrasound  and the
lung-to-head ratio (LHR) [10,11]; fetal intervention, if any,
including tracheal occlusion; birth age and weight;
operative procedures and findings (type of hernia repair,
patch type, and presence of liver herniation); need for
fundoplication or gastrostomy tube; duration of intubation
and mode of ventilation; and oxygen requirements at
performed. Degree of chest deformity or scoliosis was noted.
Neurodevelopmental, audiometric, pulmonary, and growth
and nutritional characteristics were evaluated. Any surgical
procedures and intraoperative findings were documented,
operation for hernia recurrence or bowel obstruction.
1.2. Operative repair
All liveborn CDH patients were medically stabilized
before surgical closure of the hernia. Primary repair was
attempted if possible. For patients who required a patch
repair from 1998 to 2006, the type of patch used was either
polytetrafluoroethylene (Gore-Tex [GTX]; Gore Medical,
Flagstaff, AZ) or small intestinal submucosa (Surgisis [small
intestinal submucosal patch, SIS]; Cook Biotech, West
Lafayette, IN). Current practice at UCSF since 2006 has been
to use a custom-layered combination patch of Gore-Tex and
Surgisis (SIS + GTX).
1.3. Statistical analysis
Tabulated data were analyzed to determine which prenatal
and postnatal patient characteristics or operative variables
were significantly associated with the surgical complications
of hernia recurrence, chest deformity, scoliosis, and
operative SBO. Univariate analysis was performing using
the Pearson χ2and Fisher's Exact tests for categorical
variables, and the Student's t test and the Wilcoxon's rank
sum test for continuous variables. Odds ratios (ORs) and
positive and negative predictive values were generated to
evaluate the usefulness of variables as predictors of adverse
outcomes. Because hernia recurrence is a time-dependent
outcome, we calculated hazard ratios for different repair
types and used Kaplan-Meier survival curves using the log-
rank test for equality of survivor functions. To determine
predictor independence, multivariate regression analysis was
performed when appropriate with manual selection of
2.1. Cohort characteristics
General cohort characteristics are presented in Table 1.
There was 1 postdischarge death occurring in a 21-month-
old boy owing to delayed transfer to UCSF for manage-
ment of volvulus. For the purposes of statistical analysis,
perinatal predictors of adverse surgical outcomes were
given the following binary classifications: patch vs primary
repair, liver herniation vs no herniation, age at extubation
greater or less than 16 days (cohort median), neonatal
discharge with or without oxygen supplementation, prena-
tal intervention (tracheal occlusion, n = 8) or no prenatal
intervention, premature birth (b36 weeks) or term birth
(≥36 weeks), LHR less than or greater than 1.0, and use or
no use of extracorporeal membrane oxygenation (ECMO)
therapy (n = 8).
2.2. CDH recurrence
Thirty patients (31%) have had a hernia recurrence at a
median time of 0.9 (range, 0.1-7) years after initial repair. Of
patients who developed recurrence, 26 had an initial patch
repair and 4 had a primary repair. Thus, 46% of patch repairs
and 10% of primary repairs have recurred, although follow-
up times vary (Table 2). Of the 30 patients with recurrence,
156T. Jancelewicz et al.
11 (37%) have had more than 1 recurrence. Using Kaplan-
Meier analysis, the recurrence-free interval over 8 years for
the entire cohort is shown in Fig. 1A, which demonstrates
that, by 4 years postoperatively, 50% of patch repairs failed
(log-rank test for equality P b .001).
To estimate relative risk of recurrence over time based on
patch repair type, hazard ratios were calculated. Compared
with a primary hernia closure, an SIS repair had the highest
hazard ratio for recurrence at 8.6 (P b .001), whereas an
SIS + GTX repair had the lowest hazard ratio of 3.8 (P = .06,
Table 2). Fig. 1B shows the proportion of patients without
recurrence over 4 years stratified by patch type; the
recurrence-free interval was shortest in those with an SIS
repair vs a GTX or SIS + GTX repair (log-rank test for
equality, P b .001).
Because follow-up time is unequal between repair types,
we also examined the risk of early recurrence based on repair
type. Early recurrence was considered present when it
occurred less than 6 months after initial repair. Only an SIS
repair was significantly associated with early recurrence,
with an OR of 5.1 (confidence interval [CI], 2-17; P = .009).
Sevenof 23 SIS repairs (30%) failed in less than6 months,vs
2 GTX repairs (12%), 2 SIS + GTX repairs (13%), and 2
primary repairs (5%). Of the 11 patients with multiple
recurrences, 10 (91%) had an initial SIS repair and 1 had an
initial SIS + GTX repair.
Multiple perinatal markers of CDH severity were
significantly associated with hernia recurrence by univariate
analysis (Table 3). Multivariate regression analysis showed
Cohort characteristics (N = 99)
Variable Data point
(% or range)
Median current age
Mortality after discharge
Median gestational age at delivery
Median age at repaira
Initial nonprimary hernia repair
SIS + GTX composite
Minimally invasive repair
Median age at final extubation
Discharge with oxygen supplementation
4.7 (0.2-10.6) y
38 (28-41) wk
5 (0-540) d
16 (0-55) d
aThree patients were diagnosed with CDH well outside the neonatal
bTwo patients were referred for management after repair with an
unknown type of patch.
stratified by repair type. A, Proportion of cohort without a
recurrence measured over 8 years: primary vs patch repair. B,
Proportion of cohort without a recurrence measured over 4 years,
stratified by repair subtype: primary, GTX, SIS, and SIS + GTX
Kaplan-Meier analysis of recurrence-free intervals,
Estimates of recurrence risk
Repair type Number with
Median time to
recur (range), y
time (SD), y
ratio (95% CI)
SIS + GTX
157 Long-term surgical outcomes in CDH
that only patch repair was independently predictive of
subsequent recurrence, with an OR of 5.0 (CI, 1-24; P b .05).
Variables not listed in Table 3, including ECMO therapy,
low birth weight, CDH laterality, prenatal intervention, and
LHR less than 1.0, were not significant predictors of
recurrence or any other adverse surgical outcomes.
2.3. Chest deformity
Significant chest deformity, usually in the form of an
asymmetric pectus excavatum or chest wall hypoplasia, was
identified in 47 patients (47%). Nine patients with an initial
primary repair (21%) and 38 with a patch repair (67%)
developed chest deformity. By univariate analysis, patients
with a history of patch repair, liver herniation, oxygen
requirement at neonatal discharge, prolonged intubation,
and hernia recurrence had significantly increased odds of
chest deformity (P b .05, Table 3). On multivariate
analysis, no variable was an independent predictor of
overall chest deformity, although patch repair was inde-
pendently predictive of early chest deformity occurring
during the first year postrepair (OR, 4.8; CI, 1-21; P = .04).
An SIS repair was the only patch repair type significantly
associated with the development of early chest deformity
(OR, 3.0; CI, 1-8; P = .02).
A diagnosis of chest deformity occurred during the first 2
years postrepair in the vast majority of affected patients
(74%). Eleven patients with chest deformity (23%) demon-
strated improvement or resolution of the deformity without
intervention. Two patients of 5 (40%) who had a thoraco-
scopic primary repair developed mild chest deformity.
2.4. Bowel obstruction
Thirteen patients (13%) have undergone laparotomy for
SBO at a median interval of 1.2 (range, 0.1-3.6) years
postrepair. The cause of the obstruction was adhesions in 7
cases (54%), reherniation in 5 cases (39%), and midgut
volvulus in 1 case (8%). The patient with volvulus had a
contemporaneous hernia recurrence that was nonobstructing.
In most cases of adhesive SBO, interloop or incisional
adhesions were responsible for the obstruction, rather than
patch adherence. Significant univariate predictors of SBO
included a history of patch repair, liver herniation, prematu-
rity, and hernia recurrence (Table 3). Multivariate analysis
yielded no significant independent predictors of SBO.
Of the patients with operative SBO, 8 had an initial SIS
repair, 1 had a GTX repair, 3 had an SIS + GTX repair, and
1 had a primary repair. By univariate logistic regression and
Fisher's Exact test, the only repair subtype significantly
associated with SBO was SIS, with an OR of 8.1 (CI, 2-28;
P = .001).
Thirteen patients (13%) have been identified as having a
significant spinal curvature on physical examination (ap-
proximately N10°-20° curvature). Eleven of these patients
had a patch repair (85%), and 2 (15%) had an initial primary
repair. By univariate analysis, although patients with several
perinatal markers of disease severity were significantly more
likely to have scoliosis, the CI ranges obtained from this
analysis were too broad to reliably identify predictors of this
adverse outcome (data not shown).
Surgical complications are common among CDH survi-
vors, but data are limited regarding long-term incidence
patterns and associated risk factors [4,5,7,8,12-16]. As with
the adverse medical outcomes of neurodevelopmental delay,
hearing loss, pulmonary insufficiency, and growth and
nutritional failure, adverse surgical outcomes seem to be
most frequent in those patients with a large CDH defect
requiring patch repair. In keeping with existing reports, in
our cohort of 99 patients with a mean age of nearly 5 years,
we found that 31% had hernia recurrence, 47% developed
chest deformity, and 13% had SBO and scoliosis. Of all
clinical predictors, we found a history of patch repair to be
the most strongly and independently predictive of subse-
quent surgical complications, which correlates with the
observation in other reports that large hernia size—for which
patch repair is a surrogate marker—strongly reduces overall
survival as well as increases the risk of multiple adverse
Predictors of surgical complications
Variable Hernia recurrence,
30 affected (31%)
Chest deformity, 47 affected
13 affected (13%)
OR (95% CI)P OR (95% CI)P OR (95% CI)P
Oxygen at discharge
Extubation age N16 d
Premature birth (b36 wk)
158 T. Jancelewicz et al.
In contrast with our previous study , these longer-term
data have revealed that use of a bioabsorbable patch that
promotes native tissue ingrowth (SIS) without the presence
of a permanent material such as polytetrafluoroethylene
(GTX) predisposes patients to earlier hernia recurrence and
multiple recurrences. Existing reports supporting the use of
SIS for CDH repair do not examine results beyond 1 to 2
years of age [5,18,19]. Although the recurrence rate in our
cohort was high among all patch types as compared with
primary hernia closure, nearly a third of patients who had an
SIS repair had a recurrence within 6 months, which was
strikingly higher than the incidence for other repair types. In
addition, an SIS-only repair was significantly associated with
a higher frequency of early chest deformity and operative
bowel obstruction. The reasons for these latter increased
risks with SIS repair are unclear but may be related to disease
severity or hernia recurrence requiring repeat laparotomies
and patch repairs. Some reports have shown that the material
does not always provide durable defect repair in other
settings and may be proinflammatory [20-22]. In response to
these findings, our group no longer performs SIS-only
repairs. In addition, patients never received an SIS-only
patch as a repair for a recurrence; these were managed using
GTX or in some cases a muscular flap. Because no difference
was found with GTX + SIS repairs over GTX in terms of
adverse surgical outcomes, further study is required to
ascertain whether a composite repair is advantageous.
The risk of surgical complications was higher in patients
with severe disease as indicated by other perinatal clinical
markers besides patch repair, in particular liver herniation
oxygen supplementation required at neonatal discharge. We
found these same variables to be strongly predictive of
nonsurgical adverse outcomes in this cohort as well. Patients
without these clinical characteristics had a very low
likelihood of developing surgical complications; and thus,
these markers are perhaps most useful as negative predictors
of adverse outcomes. Based on predictive values data from
our cohort of children with a range of CDH severity, for
patients with a primary repair, no liver herniation, short
duration of intubation, and no oxygen dependence at
discharge, families may be counseled that the risk of hernia
recurrence is less than 10%, risk of chest deformity is less
than 25%, and risk of bowel obstruction is less than 5% over
the first 5 years of life.
Long-term skeletal deformity has been noted in CDH
survivors and tends to be most severe in those who had large
defects . Congenital diaphragmatic hernia–associated
chest deformity is particularly troublesome, as it tends to
be asymmetric and progressive. In our cohort, chest
deformity was most common after patch repair; but a
significant proportion of patients with an initial primary
repair still developed this problem (21%). Multisystem
disease severity, particularly pulmonary failure as assessed
by oxygen dependence at discharge, seems to increase the
risk of chest deformity, although all CDH survivors should
have close surveillance of their chest wall during childhood.
No patient in the cohort has had repair of a chest deformity.
Moderate scoliosis was also frequently noted in our
cohort; but onset tended to be later in childhood, and no
patients have yet undergone surgical correction. We could
not identify significant independent risk factors for this
condition, but patients with a patch repair tended to be more
likely to develop scoliosis.
Bowel obstruction after CDH repair may occur at any
time. In addition, the typically unfixed, nonrotated intestine
in CDH patients is a potential risk factor for volvulus; but we
observed only 1 case in our cohort. We found that SBO most
commonly occurred because of adhesive disease (54%), but
hernia recurrence was responsible for 5 obstructions (39%).
The most significant perinatal predictors of subsequent SBO
were not markers of severe pulmonary insufficiency but
rather the anatomical markers of CDH severity: patch repair,
liver herniation, and hernia recurrence. Although SIS repairs
were significantly associated with SBO, this observation
may be related to length of follow-up time or to the fact that
this was the most common type of patch repair in the cohort.
Our study is limited by missing data points for some
patients not seen consistently in clinic or who were lost to
follow-up. However, in clinic, we saw nearly 80% of all
CDH patients who survived to discharge; so the cohort data
are likely representative of the general CDH population. It
should be emphasized that the cohort consisted only of
patients who survived to discharge and were seen at clinic,
and there was a significant predischarge mortality (25%); this
partly explains the apparently low prevalence in the cohort of
patients with ECMO and with a low LHR, who had increased
perinatal mortality. The low prevalence of these conditions
may have reduced their significance as predictors.
In summary, we have confirmed a high frequency of
surgical complications in a large cohort of CDH survivors and
found that, although a history of patch repair is the most
significant predictor of adverse surgical outcomes, patients
with other perinatal markers of disease severity are also at risk
for hernia recurrence, chest deformity, and bowel obstruction.
 Chiu P, Hedrick HL. Postnatal management and long-term outcome
for survivors with congenital diaphragmatic hernia. Prenatal Diagn
 Weber TR, Kountzman B, Dillon PA, et al. Improved survival in
congenital diaphragmatic hernia with evolving therapeutic strategies.
Arch Surg 1998;133:498-502 [discussion 502-493].
 Moss RL, Chen CM, Harrison MR. Prosthetic patch durability in
congenital diaphragmatic hernia: a long-term follow-up study.
J Pediatr Surg 2001;36:152-4.
 St Peter SD, Valusek PA, Tsao K, et al. Abdominal complications
related to type of repair for congenital diaphragmatic hernia. J Surg
 Grethel EJ, Cortes RA, Wagner AJ, et al. Prosthetic patches for
congenital diaphragmatic hernia repair: Surgisis vs Gore-Tex. J Pediatr
Surgery 2006;41:29-33 [discussion 29-33].
159 Long-term surgical outcomes in CDH
 Lund DP, Mitchell J, Kharasch V, et al. Congenital diaphragmatic Download full-text
hernia: the hidden morbidity. J Pediatr Surg 1994;29:258-62
 Nobuhara KK, Lund DP, Mitchell J, et al. Long-term outlook for
survivors of congenital diaphragmatic hernia. Clin Perinatol 1996;23:
 Vanamo K, Peltonen J, Rintala R, et al. Chest wall and spinal
deformities in adults with congenital diaphragmatic defects. J Pediatr
 Metkus AP, Filly RA, Stringer MD, et al. Sonographic predictors of
survival in fetal diaphragmatic hernia. J Pediatr Surg 1996;31:148-51
 Laudy JA, Van Gucht M, Van Dooren MF, et al. Congenital
diaphragmatic hernia: an evaluation of the prognostic value of the
lung-to-head ratio and other prenatal parameters. Prenatal Diagn
 Yang SH, Nobuhara KK, Keller RL, et al. Reliability of the lung-to-
head ratio as a predictor of outcome in fetuses with isolated left
congenital diaphragmatic hernia at gestation outside 24-26 weeks.
Amer J Obstetr Gynecol 2007;197:30 e31-37.
 Bagolan P, Morini F. Long-term follow up of infants with congenital
diaphragmatic hernia. Semin Pediatr Surg 2007;16:134-44.
 Chiu PP, Sauer C, Mihailovic A, et al. The price of success in the
management of congenital diaphragmatic hernia: is improved survival
accompanied by an increase in long-term morbidity? J Pediatr Surg
 Lally KP, Engle W. Postdischarge follow-up of infants with congenital
diaphragmatic hernia. Pediatrics 2008;121:627-32.
 Ssemakula N, Stewart DL, Goldsmith LJ, et al. Survival of patients
with congenital diaphragmatic hernia during the ECMO era: an 11-
year experience. J Pediatr Surg 1997;32:1683-9.
 West SD, Wilson JM. Follow up of infants with congenital
diaphragmatic hernia. Semin Perinatol 2005;29:129-33.
 Lally KP, Lally PA, Lasky RE, et al. Defect size determines survival in
infants with congenital diaphragmatic hernia. Pediatrics 2007;120:
 Holcomb 3rd GW, Ostlie DJ, Miller KA. Laparoscopic patch repair of
diaphragmatic hernias with Surgisis. J Pediatr Surg 2005;40:E1-E5.
 Smith MJ, Paran TS, Quinn F, et al. The SIS extracellular matrix
scaffold-preliminary results of use in congenital diaphragmatic hernia
(CDH) repair. Pediatric surgery international 2004;20:859-62.
 Fumagalli U, Bona S, Caputo M, et al. Are Surgisis biomeshes
effective in reducing recurrences after laparoscopic repair of large
hiatal hernias? Surg Laparosc Endosc Percutan Tech 2008;18:
 Baroncello JB, Czeczko NG, Malafaia O, et al. The repair of
abdominal defects in rabbits with Parietex and Surgisis meshes
abdominal wall. Arq Gastroenterol 2008;45:323-9.
 Gupta A, Zahriya K, Mullens PL, et al. Ventral herniorrhaphy:
experience with two different biosynthetic mesh materials, Surgisis
and Alloderm. Hernia 2006;10:419-25.
Priscilla Chiu, MD (Toronto, Ontario): That is a wonderful
cohort, but you have left out the thoracoscopic or
laparoscopic repair patients. I want you to comment on
the recurrence rate there.
Tim Jancelewicz, MD (response): This cohort consists of 5
patients who had a thoracoscopic repair, and that
particular variable was not significantly associated with
an adverse outcome.
Charles Stolar, MD (New York City, NY): I congratulate
you for following this very important group of patients
because as more survive we are learning what they look
like as they grow up. I am particularly interested in those
patients who have an emerging incidence of scoliosis and
chest wall deformity, because it is not ordinarily seen in
boys but because it is becoming more common. I think
you recorded 14% at 4.7 years. I would say that the
relationship to the patch calls into question the causality
issue because diaphragmatic hernia is a field defect. The
whole side of the chest grows more slowly than the
contralateral side; and because the chest grows asymmet-
rically, they are going to get bent over that way. It is not
that they are tethered by the patch, at least that is how we
think about it.
My question—because you have an ongoing CDH
clinic, could you describe your protocol for following and
managing the axial skeletal abnormality?
Dr Jancelewicz (response): We think scoliosis is a late-
onset phenomenon; so we do not specifically formally
measure spinal curvature in clinic in about the first 5 years
of life, at least in the patients that I have seen. We suspect
that the incidence of scoliosis will become more common
beyond age 5 and a little bit older. There is actually at
present no formal measurement in clinic; but if we do see
something obvious, then we initiate referrals to appropri-
Unidentified speaker: Yesterday we heard some data that
a lower incidence of recurrence. Has the incidence of
recurrence with patch repairs caused you to change your
Dr Jancelewicz: There is definitely individual surgeon
preference in our practice. I think that flap repairs are
reserved in our practice, and I will defer to the surgeons to
comment on this on their own; but the flap repair is
reserved for multiply recurrent patients.
160 T. Jancelewicz et al.