Materials and Methods
A CDH database was established in France in 2007 by the National Center for Rare Diseases. At an annual national meeting, participating perinatal centers are encouraged to enter CDH cases in the database. Diagnosis, management, and delivery in CDH cases were standardized in the network. Pregnancies were managed in a fetal medicine unit, and deliveries occurred in a tertiary care center. In all cases amniocentesis and ultrasound was offered to rule out chromosomal abnormality and associated malformations. Microarrays were not performed on a routine basis at this time. Confirmation of diagnosis and LHR measurement were done by a referring physician.
Isolated left CDH was defined as CDH with a normal karyotype and no other malformation detected at ultrasound. The newborns were followed up until 6 months. All babies were treated following a standardized protocol established by the members of the Center for Rare Diseases. Neonates were intubated in the delivery room and admitted to a neonatal intensive care unit. Ventilatory support was initiated in all cases, with high-frequency oscillation ventilation when necessary.
Extracorporeal membrane oxygenation (ECMO) was not used because its usefulness is still debated because of the risk of long-term complications. Indeed, the use of ECMO is associated with a higher risk of neurological impairment. Whether the increased incidence of adverse outcome associated with ECMO indicates a more severe form of CDH or reflects ECMO-associated complications is still a matter of debate. In general, ECMO is reserved for the sickest newborns.
Surgical repair (with or without a prosthetic patch) was done in all cases as soon as the neonate was hemodynamically stable, usually within 24–48 hours.
Data were recorded through a web-based questionnaire accessible by password. They concerned gestational age established at diagnosis, the side of the hernia, the position of the liver, LHR, associated anomalies, and postnatal follow-up data such as mortality, persistent pulmonary hypertension of the newborn (PPHN), and oxygen therapy dependence. Gestational age was established in all cases by ultrasound measurement of crown-rump length between 11 and 14 WG.
The LHR was computed by multiplying the 2 longest perpendicular diameters of the contralateral lung at the level of the 4-chamber view of the heart and dividing this by the head circumference. LHR was routinely measured at the second-trimester ultrasound scan between 22 and 24 weeks. However, it could be measured later in pregnancy if CDH was diagnosed later. In practice, ratio of the observed over the expected lung-to-head circumference ratio (LHR o/e ) was used because it is independent of gestational age. It is computed by dividing the observed LHR by the expected value for gestational age on 2-dimensional ultrasound.
The study was limited to isolated CDH (ie, normal fetal karyotype, absence of other structural anomalies), and terminations of pregnancy and intrauterine fetal deaths were excluded.
The primary outcome was neonatal mortality, which was evaluated at 28 days and at 6 months. The secondary outcome was postnatal morbidity, defined by respiratory and gastrointestinal morbidity. Respiratory morbidity was characterized by PPHN at 48 hours and oxygen therapy dependence at 28 days. PPHN (assessed by echocardiography at 48 hours of life) was defined by pulmonary artery systolic pressure above 40 mm Hg. It was assessed among children who were alive at 48 hours.
Oxygen therapy dependence at 28 days was assessed among children who were alive at 28 days. Gastrointestinal morbidity was characterized by oral disorders (defined by gastroesophageal reflux disease and/or swallowing disorders) and enteral feeding. It was studied among children who were alive at 6 months.
We also studied prosthetic patch repair during surgery as a marker of a more severe defect and worse prognosis.
To take into account a potential effect of the center, we adjusted for the size of the center, defined by the number of CDH cases per year: fewer than 5 cases per year; between 6 and 15 cases per year; more than 16 cases per year.
This study received the approval of the Scientific Committee of the National Center for Rare Diseases and the Institutional Review Board of the southern Paris area (referent Comité de Protection des Personnes 14–044).
Statistical analysis
Descriptive results are displayed as a percentage or median and interquartile range. The relationships between gestational age at diagnosis and outcome (mortality or morbidity) were modeled with a logistic regression using fractional polynomials that give an optimal data fit. This method provides curves and their 95% confidence intervals and allows calculation of odds ratios for mortality and morbidity after having divided gestational age into the usual 3 trimesters (≤14 WG, 15–28 WG, and ≥29 WG). Univariate analyses were first conducted, and then adjustments were made for the factors associated with CDH severity: side of the hernia, thoracic herniation of the liver, gestational age at birth, LHR o/e , and treatment by fetal tracheal occlusion.
The statistical analyses were performed with Stata 12 (StataCorp, College Station, TX).
Results
Between January 2008 and November 2013, 567 pregnancies with CDH were recorded in 30 fetal medicine units. Of these, 506 (89.2%) were diagnosed antenatally and 449 with isolated CDH ( Figure 1 ). Of these 449 pregnancies, 420 ended in a live birth after exclusion of 26 terminations of pregnancy, 2 intrauterine fetal deaths, and 1 of unknown outcome. Among those 420 live births, gestational age at diagnosis was unknown for 26 and follow-up at 28 days after birth was not complete in the database for 17, leaving 377 cases for the sample analyzed.
In this sample, the median birthweight was 2950 g (interquartile range, 2500–3300). The median gestational age at diagnosis was 23 + 5 WG (interquartile range, 22–29); 18 cases were diagnosed in the first trimester, 263 in the second, and 96 in the third. Table 1 shows the results of the prenatal ultrasound examination. In these 377 cases, 364 children (96.6%) were followed up until 6 months.
| n (%) or median [IQR] | Total | First trimester (n = 18) | Second trimester (n = 263) | Third trimester (n = 96) |
|---|---|---|---|---|
| Left-sided CDH | 319 (0.85) | 17 (0.94) | 223 (0.85) | 79 (0.82) |
| Thoracic herniation of the liver | 152 (0.41) | 7 (0.42) | 125 (0.48) | 20 (0.22) |
| LHR o/e | 0.42 [0.33–0.54] | 0.41 [0.34–0.44] | 0.40 [0.33–0.52] | 0.50 [0.37–0.68] |
The mortality rate was 32.9% at 28 days and 39.0% at 6 months ( Table 2 ). Among neonates who died before 28 days of life, 18% died immediately on the first day of life in the delivery room.
| Outcome | Total (n = 377) | Trimester of CDH diagnosis | P value a | ||
|---|---|---|---|---|---|
| First (n = 18) | Second (n = 263) | Third (n = 96) | |||
| Mortality | |||||
| At 28 d | 32.9% (28.2–37.9) | 61.1% (35.7–82.7) | 39.2% (33.2–45.3) | 10.4% (0.05–18.3) | < .001 |
| At 6 mo b | 39.0% (34.0–44.2) | 82.3% (56.6–96.2) | 44.5% (38.3–50.8) | 16.1% (0.09–25.2) | < .001 |
| Respiratory morbidity | |||||
| PPHN at 48 h c | 61.6% (56.0–66.9) | 100% (78.2–100) | 70.0% (63.3–76.0) | 35.9% (26.1–46.5) | < .001 |
| Oxygen dependence at 28 d d | 37.7% (31.7–44.0) | 100% (59.0–100) | 46.5% (38.6–54.6) | 16.3% (9.2–25.8) | < .001 |
| Gastrointestinal morbidity e | |||||
| Oral disorders | 19.3% (14.3–25.1) | 33.3% (0.84–90.6) | 24.6% (17.7–32.7) | 9.1% (3.7–17.8) | .006 |
| Enteral feeding | 15.0% (10.6–20.4) | 0.0% (0–70.8) | 20.1% (13.8–27.8) | 6.4% (2.1–14.3) | .011 |
| Surgery c | |||||
| Prosthetic patch repair | 27.1% (22.0–32.8) | 55.6% (21.2–86.3) | 32.6% (25.9–39.9) | 12.6% (6.5–21.5) | < .001 |
a P value for the association between outcome and gestational age at diagnosis was modeled by fractional polynomials
b Among newborns not lost to follow-up at 6 months (n = 364)
c Among children who were alive at 48 hours of life (n = 323)
d Among children who were alive at 28 days of life (n = 253)
There was a statistically significant relationship between gestational age at diagnosis and mortality at 28 days, as shown by Figure 2 , in which the mortality rate decreases steadily with the date of diagnosis. Mortality rates at 28 days according to the trimester of diagnosis were 61.1%, 39.2%, and 10.4% in the first, second, and third trimesters, respectively ( P < .001) ( Table 2 ). Similarly, mortality rate at 6 months decreased significantly with the trimester of diagnosis ( P < .001): 82.3%, 45.5%, and 16.1% for cases diagnosed in the first, second, and third trimesters, respectively.
After adjustment for the main factors of severity of CDH (see Statistical analysis section), the relationship between gestational age at diagnosis and mortality at 28 days remained significant ( P < .001, Table 3 ). Adjusted odds ratios and 95% confidence intervals were 3.10 (1.85–5.21) and 0.35 (0.18–0.66) for the first and third trimesters, respectively, compared with the second trimester.
| Variable | Trimester of CDH diagnosis | P value a | ||
|---|---|---|---|---|
| First (n = 18) | Second (n = 263) (reference) | Third (n = 96) | ||
| Mortality | ||||
| At 28 d | 3.12 [1.86–5.25] | 1 | 0.35 [0.18–0.66] | < .001 |
| At 6 mo b | 3.09 [1.86–5.11] | 1 | 0.38 [0.21–0.70] | < .001 |
| Respiratory morbidity | ||||
| PPHN at 48 h c | 2.85 [1.76–4.64] | 1 | 0.27 [0.15–0.49] | < .001 |
| Oxygen dependence at 28 d d | 3.22 [1.52–6.81] | 1 | 0.18 [0.08–0.39] | < .001 |
| Gastrointestinal morbidity e | ||||
| Oral disorders | 1.65 [0.73–3.75] | 1 | 0.66 [0.28–1.52] | .014 |
| Enteral feeding | 4.56 [1.58–13.2] | 1 | 0.58 [0.21–1.59] | .002 |
| Surgery c | ||||
| Prosthetic patch repair | 1.47 [0.85–2.53] | 1 | 0.57 [0.29–1.10] | < .001 |
a P value for the association between outcome and gestational age at diagnosis modeled by fractional polynomials
b Among newborns not lost to follow-up at 6 months (n = 364)
c Among children who were alive at 48 hours (n = 323)
d Among children who were alive at 28 days (n = 253)
Results were similar for respiratory morbidity characteristics. For instance, the rate of PPHN at 48 hours was significantly greater when diagnosis was made earlier in the pregnancy ( Figure 3 and Table 2 ). This relationship remained significant after adjustment for the main factors of CDH severity ( Table 3 ).
