Keywords
congenital diaphragmatic hernia - neonatal - pulmonary hypertension - prematurity
- NICU - pulmonary hypoplasia - respiratory failure
Congenital diaphragmatic hernia (CDH) is a life-threatening anomaly occurring in 1
of every 3,000 to 5,000 live births. It is associated with respiratory failure, pulmonary
hypoplasia, pulmonary hypertension, and mortality ranging from 24 to 40%.[1]
[2]
[3]
[4] International groups have studied these infants from both a medical and surgical
perspective,[5]
[6] and many short-term outcomes have been previously defined.[7] Although some risk factors for mortality have been described,[8] early prediction of mortality and other morbidities remains a challenge.[9]
[10]
[11]
Most neonatal intensive care units (NICUs) have few patients with CDH.[12]
[13] Even in the regional NICU that cares for infants referred with CDH, patient accrual
is insufficient to adequately describe patient characteristics, the distribution of
applied therapies, or short-term outcomes over a contemporary period.[14] Longitudinal single-center reports are difficult to interpret as institutional practice
and technologic advances vary over time, making the use of historic controls fraught
with bias. Detailed description of infants with CDH in a contemporary, multi-institutional
cohort is a necessary step to associate risk factors with patient outcomes. Moreover,
a traditional focus on the surgical procedures has left the exploration of the risks
and benefits of the applied medical therapies less well defined.
Using an extensive clinical data set of infants referred to participating regional
NICUs, we focused this report on the characterization of infants with CDH and their
in-hospital outcomes for a large, multicenter contemporary cohort in the United States.
Because infants born less than 34 weeks' gestation have a higher risk or mortality
and are typically not considered candidates for therapies such as extracorporeal membrane
oxygenation (ECMO), we stratified the analysis of medical and surgical inventions
and outcomes by gestational age (GA) and birth weight (BW).
Patients and Methods
The Children's Hospital Neonatal Database (CHND) captures clinical data on all infants
admitted to 27 participating regional NICUs.[13] For this study, all infants with a diagnosis of CDH over a 4-year period (2010–2013)
were identified. Infants who had surgical repair before referral or had previously
been discharged home were not included in this analysis. The cohort included patients
from 22 of 27 hospitals who contributed expanded data focused on CDH patients; thus,
patients from the remaining five hospitals that did not participate in the CDH module
were prospectively excluded (n = 49). Chart abstractors at each site completed prospective training including review
of clinical definitions, participation in web-based seminar tutorials, and case-based
practice.[13] The Stanley Manne Children's Research Institute's Institutional Review Board (Chicago,
Illinois) approved this analysis (#2011–14673), and each participating site has ongoing
local institutional review board approval for participation in the data registry.
Eligible infants were described using maternal and infant demographic, birth, and
clinical characteristics including selected variables that occurred before referral
to the regional NICU. These were reported for the whole cohort and after stratification:
those born < 34 weeks' gestation or < 2 kg (cohort 1) and those born ≥ 34 weeks' gestation
and ≥ 2 kg (cohort 2). These groups were chosen to isolate the larger, more mature
infants who may have been eligible candidates to receive ECMO.[15]
The characteristics of the type of defect, type of repair, surgical approach, and
distribution of selected perioperative sequelae were reported. Clinical interventions
provided were described, including respiratory and vasopressor support, and receipt
of ECMO or pulmonary vasodilator medications. For these variables, the cohort was
stratified by mortality to demonstrate the unadjusted differences in the receipt of
therapies.
The outcomes described are those achieved before hospital discharge from the regional
NICU. Some infants were transferred to another institution, and data on the final
disposition of these infants were not measured systematically. The prevalence of other
selected morbidities including the prevalence of central line associated blood stream
infections, duration of mechanical ventilation, and hospital length of stay (LOS)
are reported for the duration of stay within the participating CHND hospitals.
Data description and analyses were performed with SAS v9.3 (Cary, NC). Student t-test, chi-squared, and bivariable analyses were applied, as appropriate. Nonparametric
testing was applied when the distribution of selected measures did not conform to
a normal distribution.
Results
Of the 49,403 records in CHND, 642 infants (1.4%) had a diagnosis of CDH for their
first admission to these participating regional NICUs. After excluding infants with
incomplete data (n = 49) and those who were repaired (n = 3) or previously home before referral (n = 18), 572 infants with CDH were eligible for analysis. Patient volume varied by
center (median, 23 patients per center; [inter-quartile range, IQR, 10, 34]), and
six centers had more than 36 CDH patients included.
Prenatal and Perinatal Characteristics
Prenatal diagnosis was made in 64% of CDH patients, and most infants (89%) were born ≥ 34
weeks' gestation and ≥ 2 kg (cohort 2). Small for GA,[16] multiple gestation, cesarean delivery, antenatal steroid administration, APGAR scores ≤ 5
at 1, 5, or 10 minutes, and exogenous surfactant therapy before referral were more
frequently observed in cohort 1 compared with those in cohort 2 ([Table 1]). Infants were referred early after birth (median, 3.6 hours; IQR, 1.5, 6.3 hours)
and 89% were referred before 24 hours of age. Only one infant at any gestation had
confirmed sepsis present at the time of referral.
Table 1
Demographic variables associated with CDH patients referred to CHND NICUs
|
Variable
|
All
|
GA/BW stratification
|
p Value
|
|
< 34 wks or < 2 kg
|
≥ 34 wks and > 2 kg
|
|
Number of CDH patients
|
572
|
64 (11)
|
508 (89)
|
|
|
Median GA in weeks (IQR)
|
38 (36, 39)
|
32 (31, 33)
|
38 (37, 39)
|
< 0.001
|
|
Median BW in grams (IQR)
|
3,000 (2,580, 3,345)
|
1,777 (1,438, 1,970)
|
3,068 (2,745, 3,400)
|
< 0.001
|
|
SGA (< 10%ile)
|
68 (12)
|
14 (22)
|
54 (11)
|
0.014
|
|
Multiple gestations
|
23 (4)
|
8 (13)
|
15 (3)
|
0.002
|
|
Cesarean delivery
|
256 (45)
|
42 (66)
|
214 (42)
|
0.001
|
|
Antenatal steroids
|
65 (11)
|
26 (41)
|
39 (8)
|
< 0.001
|
|
APGAR @1 minute ≤ 5
|
333 (58)
|
47 (73)
|
286 (56)
|
0.004
|
|
APGAR @5 minute ≤ 5
|
142 (25)
|
26 (41)
|
116 (23)
|
0.007
|
|
APGAR @10 minute ≤ 5
|
55 (10)
|
11 (17)
|
44 (9)
|
0.005
|
|
Surfactant before referral
|
90 (16)
|
33 (52)
|
57 (11)
|
< 0.001
|
Abbreviations: BW, birth weight; CDH, congenital diaphragmatic hernia; GA, gestational
age; IQR, interquartile range; SGA, small for gestational age.
Note: Using published gender-specific and gestational age-specific normative values.[8] Data presented are N (%) or median (IQR).
Congenital Diaphragmatic Hernia Defect and Operative Repair
Characteristics of the diaphragmatic hernia itself were recorded among those who were
surgically repaired (n = 450; 78.6% of CDH patients). Right-sided defects were more common in cohort 1 (28.2
vs. 14.8%, p = 0.028) as was liver herniation into the thorax (61.5 vs. 47.5%, p = 0.002). In contrast, thoracostomy tubes were more frequently placed in the larger,
more mature infants (45.5 vs. 28.2%; p = 0.043), although most were on the side ipsilateral to the defect and placed after
CDH repair (84%). In both the strata, the majority of infants received surgical repair
in the first (49%) or second (26%) weeks of life, with a median age at repair of 7
days (IQR = 3, 13). Most operative repairs occurred in the operating room (62%) rather
than in the NICU (38%). Patch (51%) and primary (47%) repairs were divided equally
among infants, and were similar in frequency in both strata (p = 0.66). Patch repair was associated with an increased risk for mortality relative
to primary repair (12 vs. 1.3%, p < 0.01) as was need for thoracostomy tube on either the ipsilateral or contralateral
side before surgical repair (p < 0.0001). Of those infants who were surgically repaired, 11% (n = 52) received an operative repair while receiving ECMO. Complications after operative
repair were infrequent with postoperative chylothorax (6.4%), liver/splenic laceration
(2.9%), and recurrent hernia before discharge (2.4%) noted as the three most common
problems. Infections, postoperative hemorrhage, intestinal ischemia/injury, and abdominal
compartment syndrome were reported as singular events in the entire cohort. Complications
were similar by GA and BW strata.
Medical Therapies Provided during Hospitalization
CDH infants received complex medical support during their NICU hospitalization ([Table 2]). Infants were typically intubated on their day of birth (94%), and both mechanical
ventilation and noninvasive respiratory supports were nearly universally applied with
few differences between the GA/BW strata. Infants in cohort 1 were treated less commonly
with conventional mechanical ventilation than those in cohort 2, although use of high-frequency
oscillatory ventilation (HFOV) was similar between the groups. For those infants receiving
mechanical ventilation, the total duration of respiratory support was longer in the
preterm cohort. Not surprisingly, preterm infants were less likely to receive ECMO,
but for those treated, the median duration of ECMO therapy was significantly longer
(cohort 1: 19 days vs. cohort 2: 11 days; p = 0.04), and survival after ECMO trended lower (cohort 1 = 17 vs. cohort 2 = 52%;
p = 0.11). Vasopressors and pulmonary vasodilator medications were frequently used;
of note, inhaled nitric oxide was used in the majority (62%) of infants and 22% received
sildenafil.
Table 2
Respiratory support during CDH hospitalization, stratified by gestational age and
birth weight
|
VARIABLE
|
ALL, (N = 572)
|
< 34 wks or < 2 kg (n = 64)
|
≥34 wks and > 2 kg (n = 508)
|
p Value
|
|
Advanced respiratory support
|
|
Median days on respiratory support (median, IQR)
|
18 (10, 38)
|
46 (13, 74)
|
18 (10, 36)
|
0.039
|
|
Median days on mechanical ventilation
|
15 (6, 29)
|
17 (5, 38)
|
15 (6, 28)
|
0.735
|
|
High-frequency oscillatory ventilation, N (%)
|
225 (39)
|
23 (36)
|
202 (40)
|
0.590
|
|
Median duration in days
|
5 (2, 13)
|
7 (2, 19)
|
5 (2, 13)
|
0.214
|
|
Conventional ventilation
|
510 (89)
|
50 (78)
|
460 (91)
|
0.005
|
|
Median duration in days
|
12 (5, 23)
|
14 (3, 47)
|
12 (5, 22)
|
0.366
|
|
Vasopressors
|
|
Dopamine
|
419 (73)
|
50 (78)
|
369 (73)
|
0.454
|
|
Epinephrine
|
192 (34)
|
23 (36)
|
169 (33)
|
0.675
|
|
Milrinone
|
172 (30)
|
17 (27)
|
155 (31)
|
0.565
|
|
Dobutamine
|
76 (13)
|
12 (19)
|
64 (13)
|
0.173
|
|
Vasopressin
|
30 (5)
|
4 (6)
|
26 (5)
|
0.764
|
|
Norepinephrine
|
15 (3)
|
1 (1.5)
|
14 (3)
|
1.000
|
|
Vasodilators
|
|
Inhaled nitric oxide
|
353 (62)
|
39 (61)
|
314 (62)
|
0.892
|
|
Sildenafil
|
126 (22)
|
12 (19)
|
114 (22)
|
0.631
|
|
Prostaglandin E1
|
37 (6)
|
4 (6)
|
33 (7)
|
1.000
|
|
Prostacyclin
|
19 (3)
|
1 (1.5)
|
18 (4)
|
0.711
|
|
Bosentan
|
12 (2)
|
0 (0)
|
12 (2)
|
0.378
|
|
ECMO
|
186 (33)
|
6 (9)
|
180 (35)
|
< 0.001
|
|
Total days on ECMO
|
11 (7, 18)
|
19 (15, 37)
|
11 (7, 17)
|
0.037
|
Abbreviations: CDH, congenital diaphragmatic hernia; ECMO, extracorporeal membrane
oxygenation.
Note: Data presented are n (%) or median, (interquartile range).
Mortality
Overall mortality was 29%; preterm infants had a mortality rate of 50%, compared with
27% for infants in cohort 2 (p < 0.001) ([Table 3]). Prenatal diagnosis of CDH was more frequently made in nonsurviving infants (77
vs. 59%; p < 0.001). Mortality associated with ECMO use was 49% compared with 20% in those who
did not receive ECMO (p < 0.001). Nonsurviving infants were less likely to receive conventional mechanical
ventilation, and more likely to be treated with high-frequency mechanical ventilation
([Table 4]). Only seven infants in the entire cohort did not receive mechanical ventilation
outside the immediate perioperative period, and one was a nonsurvivor. Vasopressor
and prostanoid medications as well as ECMO were used more frequently among nonsurvivors,
and the median duration of ECMO was longer than in survivors (14.3 vs. 8.6 days, p < 0.001). Associated major congenital anomalies (cardiac, genetic syndrome, and other
major anomalies) were more frequent in nonsurviving infants (41 vs. 27%; p = 0.008). The recorded causes of death were related to pulmonary hypoplasia, pulmonary
hypertension, and/or cardiopulmonary failure in all but 10 patients. For these 10
patients, their cause of death was related to hemorrhage (n = 2), brain injury (n = 6), vascular dissection (n = 1), and thrombosis (n = 1).
Table 3
Predischarge outcomes in infants with CDH
|
Variable
|
All (N = 572)
|
< 34 wk or < 2 kg (n = 64)
|
≥ 34 wk and ≥ 2 kg
(n = 508)
|
p Value
|
|
Mortality, n (%)
|
167 (29)
|
32 (50)
|
135 (27)
|
< 0.001
|
|
Median hospital length of stay (IQR)
|
32 (16, 66)
|
40 (5, 76)
|
32 (17, 63)
|
0.890
|
|
CDH patients discharged home
|
356 (66)
|
21 (33)
|
352 (69)
|
|
|
Route of feed at discharge
|
|
Breast
|
139 (40)
|
6 (29)
|
133 (41)
|
0.259
|
|
Bottle
|
255 (74)
|
15 (71)
|
240 (74)
|
0.789
|
|
Tube[a]
|
142 (41)
|
15 (71)
|
127 (39)
|
0.004
|
|
Bottle and tube[a]
|
73 (21)
|
9 (43)
|
64 (20)
|
0.012
|
|
Discharge medications
|
|
Supplemental oxygen
|
83 (24)
|
10 (48)
|
73 (23)
|
0.009
|
|
Sildenafil
|
47 (14)
|
5 (24)
|
42 (13)
|
0.158
|
|
Furosemide
|
24 (7)
|
5 (24)
|
19 (6)
|
0.002
|
|
Thiazide
|
22 (6)
|
4 (19)
|
18 (6)
|
0.036
|
Abbreviation: CDH, congenital diaphragmatic hernia.
a Includes gastrostomy, nasogastric, or transpyloric administration of feedings. Data
presented are n (%) or median (interquartile range).
Table 4
Therapies utilized by CDH patients during initial hospitalization
|
Variable
|
All (N = 572)
|
Survivors (n = 405)
|
Nonsurvivors (n = 167)
|
p Value
|
|
Advanced respiratory support
|
|
Median days on mechanical ventilation (median, IQR)
|
15 (6, 29)
|
15 (7, 29)
|
15.5 (2, 27)
|
0.013
|
|
High frequency oscillatory ventilation, N (%)
|
347 (61)
|
199 (49)
|
148 (89)
|
< 0.001
|
|
Median duration in days
|
5 (2, 13)
|
6 (2, 13)
|
3.5 (1, 14)
|
0.039
|
|
High-frequency jet ventilation
|
11 (2)
|
6 (1.5)
|
5 (3)
|
0.31
|
|
Median duration in days
|
4 (1, 9)
|
2 (1, 4)
|
9 (6, 15)
|
0.016
|
|
Conventional ventilation
|
510 (89)
|
385 (95)
|
125 (75)
|
< 0.001
|
|
Median duration in days
|
12 (5, 23)
|
12 (6, 23)
|
10 (3, 22)
|
0.009
|
|
Vasopressors
|
|
Dopamine
|
419 (73)
|
271 (67)
|
148 (89)
|
< 0.001
|
|
Epinephrine
|
192 (34)
|
103 (25)
|
89 (53)
|
< 0.001
|
|
Milrinone
|
172 (30)
|
91 (22)
|
81 (49)
|
< 0.001
|
|
Dobutamine
|
76 (13)
|
36 (9)
|
40 (24)
|
< 0.001
|
|
Vasopressin
|
30 (5)
|
18 (4)
|
12 (7)
|
0.215
|
|
Norepinephrine
|
15 (3)
|
11 (3)
|
4 (2)
|
1.000
|
|
Vasodilators
|
|
Sildenafil
|
126 (22)
|
81 (20)
|
45 (27)
|
0.076
|
|
Prostaglandin E1
|
37 (6)
|
19 (5)
|
18 (11)
|
0.014
|
|
Inhaled Prostacyclin
|
19 (3)
|
9 (2)
|
10 (6)
|
0.036
|
|
IV Prostacyclin
|
19 (3)
|
7 (2)
|
12 (7)
|
0.003
|
|
Bosentan
|
12 (2)
|
7 (2)
|
5 (3)
|
0.346
|
|
ECMO
|
186 (33)
|
95 (23)
|
91 (54)
|
< 0.001
|
|
Total days on ECMO
|
11 (7, 18)
|
9 (6, 14)
|
14 (9, 21)
|
< 0.001
|
|
VV
|
42 (7)
|
22 (5)
|
20 (12)
|
0.012
|
|
VA
|
140 (24)
|
72 (18)
|
68 (41)
|
< 0.001
|
|
VV-VA
|
7 (1.2)
|
2 (0.5)
|
5 (3.0)
|
0.025
|
Abbreviations: CDH, congenital diaphragmatic hernia; ECMO, extracorporeal membrane
oxygenation; VA, venoarterial; VV, venovenous; VV-VA, venovenous converted to venoarterial.
Note: Data presented are n (%) or median (interquartile range).
Additional Short-Term Outcome Metrics
Median hospital LOS was 32 days (IQR 16, 66 days), and correspondingly, 11% of CDH
infants had an LOS in excess of 90 days. The use and duration of central venous lines
were both frequent (mean of 2 central lines per patient) and prolonged (mean of 20
days per patient), and 9% developed a blood stream infection during their hospitalization.
A brain injury was documented on imaging studies in 1.6% of patients, and 19 CDH patients
were treated with therapeutic hypothermia for hypoxic–ischemic encephalopathy, of
which 10 patients survived. Seizures were common (7.3%) and 3.7% were diagnosed with
hearing deficits before discharge. The pharmacologic treatment of gastroesophageal
reflux was nearly universal (99.3% of CDH patients), although surgical fundoplication
was infrequent (8.1%). Preterm infants were discharged home with supplemental oxygen,
diuretic medications, and tube feedings more frequently than infants born near or
full term ([Table 3]).
Discussion
Despite recent advances in medical technology and care, infants with CDH experience
significant morbidity and mortality.[1]
[2]
[3]
[4] Several groups have examined the management of CDH as well as for identification
of factors that predict outcome.[5]
[6]
[7]
[8]
[9]
[10]
[11] We describe a large multicenter cohort of CDH infants referred to regional NICUs
for care with an emphasis on the medical therapy provided because these therapies
in a large contemporary cohort have not been well described. Better understanding
of the variability in practice and center-specific practices as they relate to outcome
may ultimately help to identify best practices that may improve outcomes for affected
infants.
Although prematurity has been identified as an independent predictor of adverse outcomes
in CDH,[17]
[18] relatively few studies have examined this group of CDH patients in detail. Our cohort
shows that infants born < 34 weeks' gestation or < 2 kg in weight are as likely to
receive aggressive care as more mature infants, with the exception of ECMO. Overall,
the preterm, smaller infants received respiratory support and ECMO therapy for a longer
duration. We also found that CDH infants born prematurely more commonly had a right-sided
CDH, and previous reports have shown that these infants are more likely to have associated
anomalies[17]; both risk factors associated with mortality.[7]
[19]
[20] Clearly, preterm infants with CDH represent an extremely high-risk group of patients,
with a higher morbidity and mortality than more mature infants.
In our cohort, the survival of preterm infants was 50%, which included one infant
treated with ECMO. For those preterm infants who survived, more were discharged on
oxygen or with feeding assistance than mature infants, but with comparable rates of
documented brain injury, seizures, and hearing loss. Although increased risk of long-term
morbidity must be considered in this population, counseling and overall approach to
care of preterm infants with CDH should balance this risk with the likelihood of short-term
survival—and the burdens of lengthy hospitalizations—when considering aggressive treatment.
Overall survival in this group of patients referred for care of CDH was 71%, similar
to that in previous reports.[1]
[4]
[10]
[12] As expected, the majority of nonsurvivors died as a result of pulmonary hypoplasia,
pulmonary hypertension, and cardiopulmonary failure. Nonsurvivors were treated with
more medical therapies including HFOV, vasopressors, pulmonary vasodilator medications,
and ECMO despite uncertain efficacy to increase the likelihood of survival. Interestingly,
we did find that after referral to regional NICUs, only a single nonsurvivor was not
mechanically ventilated, suggesting that these infants are nearly universally offered
aggressive intervention. Despite this, mortality remains high for this condition.
Continued efforts to define those infants least likely to respond to aggressive medical
and surgical intervention are warranted.
Despite a high risk of morbidity related to treatment of CDH, we found that in-hospital
outcomes for survivors were relatively favorable. Specifically, surgical, infectious,
and neurologic complications existed, but they were infrequent, and the majority of
patients were discharged home without supplemental oxygen or medications to treat
pulmonary hypertension. This cohort of infants was treated with many medical therapies,
for which evidence is limited or incomplete, including inhaled nitric oxide, sildenafil,
diuretic medications, and acid suppression medications for gastroesophageal reflux
during their initial hospital course.[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26] Use of this data in conjunction with additional prospective studies may help to
determine most effective practices for medical management of CDH.
There are limitations of this study that surround the issue of referral bias. This
cohort referred to regional NICUs may not be representative of the entire population
of infants with CDH; it is plausible that the infants who were most likely to experience
complications or poor outcomes associated with CDH were referred to these NICUs. Alternatively,
infants deemed as nonsurvivable, whether term or preterm, may have remained at the
referral hospital and not transferred for aggressive care. Nevertheless, this report
summarizes the therapies received and the short-term outcomes of a large, contemporary,
US-born cohort of infants with CDH referred to participating children's hospitals'
NICUs to provide data for counseling families of affected infants.
In summary, these results facilitate a greater understanding of the burden CDH infants
face, both for clinicians and for families of affected infants. Although individual
hypotheses were not tested, these data serve to generate hypotheses on which future
prediction and prospective studies will be based. Moreover, these results demonstrate
the successful collaboration between 27 hospitals in the United States focused on
high-risk infants with complex diseases and rare conditions. Specifically, these data
provide the framework for future studies to predict short-term outcomes using clinical
risk factors as well as to quantify the intercenter variation in care delivered and
outcomes achieved between these regional NICUs.
