Outcomes of congenital diaphragmatic hernia among preterm infants: inverse probability of treatment weighting analysis

To evaluate the survival and intact-survival rates among preterm infants with congenital diaphragm hernia (CDH). Multicenter retrospective cohort study of 849 infants born between 2006 and 2020 at 15 Japanese CDH study group facilities. Multivariate logistic regression analysis adjusted using inverse probability treatment weighting (IPTW) method was used. We also compare trends of intact-survival rates among term and preterm infants with CDH. After adjusting using the IPTW method for CDH severity, sex, APGAR score at 5 min, and cesarean delivery, gestational age and survival rates have a significantly positive correlation [coefficient of determination (COEF) 3.40, 95% confidence interval (CI), 1.58–5.21, p value <0.001] and higher intact-survival rate [COEF 2.39, 95% CI, 1.73–4.06, p value 0.005]. Trends of intact-survival rates for both preterm and term infants had significantly changed, but improvement in preterm infants was much smaller than in term infants. Prematurity was a significant risk factor for survival and intact-survival among infants with CDH, regardless of adjustment for CDH severity.


INTRODUCTION
Congenital diaphragmatic hernia (CDH) is a fetal abnormality that occurs in 2.7-4.9 per 10,000 live births [1]. The pathophysiology of CDH is a diaphragmatic defect accompanied by complications, such as lung hypoplasia, pulmonary hypertension, and cardiac dysfunction which may occur and significantly affect a child's life [2]. Although the survival rate has improved dramatically over the years, owing to prenatal diagnosis and advances in various treatments [3][4][5], there are still several problems to be solved, including severe cases and complications.
Prematurity is an essential factor among the various factors that affect the prognosis of CDH. Preterm births have gotten worse for survival rates and newborn developmental outcomes. Survival rates for preterm births are improving worldwide, and as the survival rate increases, the proportion of children who require long-term medical care increases [6][7][8][9]. Although data in the U.S. showed that the prognosis for preterm CDH naturally improves with increasing weeks of gestation [10]; however, this is not the case in Japan.
It is unknown how many preterm infants with CDH will require medical care. Many cases have reported life expectancy; however, complications have not yet been reported. Preterm birth outcomes have improved over the years, but there are no data on changes over time. Additionally, these data do not account for disease severity, and the evidence is insufficient.
In this study, we hypothesized that preterm infants with CDH would have a lower survival rate and require additional medical care than term infants with CDH. The survival rate of preterm infants with CDH and extent of medical care needed at discharge were examined using the Inverse Probability treatment of Weighting (IPTW) method according to the severity of CDH.

MATERIALS AND METHODS Study design and population
This multicenter, retrospective cohort study was conducted among infants with CDH born between 2006 and 2020 at 15 Japanese CDH study group facilities. The study enrolled 1037 infants with CDH registered in the Japanese CDH group database. Outborn infants, those receiving palliative care, and those with an unknown gestational age (GA) were excluded. We first excluded those who were outborn because of lack of markers to assess CDH severity, such as lung area to head circumference ratio (LHR) and liver-up. We also excluded patients who underwent fetal endoscopic tracheal occlusion (FETO) because it is not yet widely used and can only be performed at some of the 15 participating facilities in Japan. Infants were categorized, as shown in the flowchart in Fig. 1. The institutional ethics committee approved the study protocol of Osaka Women's and Children's Hospital (No. 11017), and the institutional review board of all the participating institutions. The authors have no conflicts of interest to declare. This work was supported by a grant from the Ministry of Health, Labor and Welfare of Japan (Grant No.20FC1017).
We also adopted an opt-out system and posted a study summary on each institutional website. Individuals had the right to decline participation. After the birth of CDH infants, clinician verbally informed to the parents that the opt-out format was used in this study. The opt-out rate was 0% in this study. The study was performed in accordance with the principles of the Declaration of Helsinki and ethical guidelines for medical and health research involving human participants.

Outcomes
Primary outcomes were survival and intact-survival rates among preterm infants with CDH at discharge. In this study, Preterm infants included 33-36 weeks GA and term infants included 37 weeks GA over. We excluded ≤32 weeks preterm because of the small number of cases and much variation in perinatal background and treatment. Intact-survival patients did not require medical care at discharge, such as supplementary oxygen, mechanical ventilator, tracheostomy, tube feeding, central venous feeding, or pulmonary vasodilators. Baseline characteristics included perinatal factors, such as GA, birth weight (BW), sex, APGAR score at 5 min, mode of delivery, chromosomal abnormalities, congenital heart disease, small for GA (≤−1.5 SD) ( Table 1).
Clinical coarse as treatment and complications showed Table 2 as antenatal steroids, polyhydramnios, hydrops, surfactant administration, shunt direction at the ductus within 24 h after birth, use of high-frequency oscillation ventilation (HFO), inhaled nitric oxide (iNO), and extracorporeal membrane oxygenation (ECMO) and central nervous system (CNS) abnormalities. We defined CNS abnormality as intraventricular hemorrhage (IVH), periventricular leukomalacia (PVL), hydrocephalus, and hypoxicischemic encephalopathy. We examined outcomes related to CDH on surgery day, pneumothorax, chylothorax, pulmonary hemorrhage, surgical wound infection, observed/expected LHR (o/e LHR), lung-to-thorax transverse area ratio (L/T ratio), and liver-up presence into the thorax. As not all cases of prenatal diagnosis had been performed, we first investigated whether prenatal diagnosis had been completed, and then examined these factors.
These therapies and surgery were initiated according to the clinical decisions of each team; indication criteria were not defined prospectively. Our data were reported by each institution according to a standardize protocol. These measurements as previously reported in the literature [4,11,12]. Liver-up was defined as prenatally detected liver herniation occupying more than one-third of the thoracic space [12]. As for o/eLHR, the contralateral area (in square millimeters) and the thorax area (in square millimeters) were measured by manual tracking of the limit of the lung and thorax at the transverse section containing the four-chamber view of the heart in ultrasonography. The head circumstance (in millimeters) was measured in the standard biparietal view of ultrasonography. The observed LHR, which was the ratio of the contralateral lung to the head circumference, we divided by the appropriate normal mean of gestational age and multiplied by 100 to derive the o/e LHR and expressed as a percentage. The expected LHRs were determined by the published formulas, which are freely available to all by the official calculator in the Tracheal Occlusion To Accelerate Lung Growth (TOTAL) trial website (access http://www.totaltrial.eu/) [13]. A cut-off point of 25% for o/e LHR was set based on the results of a previous report [14]. We defined severe CDH as one or both liver-up and o/eLHR ≤25%.
We examined perinatal factors, CDH-related factors, and the proportion of infants who needed medical care at discharge for each GA group (33-36 weeks, and ≥37 weeks). We investigated the trend in the percentage of intact survival of preterm infants with CDH over three years. In the multivariable analysis, we compared the survival and intactsurvival rates between preterm (33-36 weeks GA) and term (≥37 weeks)

Statistical analysis
Categorial variables were described as frequency and percentage. We described distributions of continuous variables as median or minimum and maximum values. We compared the baseline characteristics and outcomes of preterm and term infants using logistic regression analysis to calculate Odds ratio (OR), 95% confidence intervals (CI) and P value. P value <0.05 was recognize significant. We used the Jonckheere-Terpstra test to analyze between three years epoch and intact-survival rates. A sample size of 138 patients for each study group was originally calculated to be sufficient for detecting an assumed reduction in survival and intact-survival from 90% in the term group to 70% in the preterm group with 80% power by a twosided test with a 5% significance level.
To account for selection bias and potential confounding factors between groups in comparisons of outcome, we performed weighted propensity scores (PSs) analysis control for differences in perinatal factors and CDH severity. A PSs for each patient were calculated as the predicted probability of preterm from multivariable logistic regression that included all possible prenatal confounders likely to have affected neonatal mortality and morbidity, including sex, APGAR score at 5 min, cesarean delivery, and CDH severity. The propensity model was generated using inverse probability of treatment weighting method. Each patient was weighted by the inverse probability of being preterm and term group, with the goal balancing observable characteristics. The weights of patients in the preterm CDH group were the inverse of the PS, and those of patients in the term group were the inverse of (1-PS) [15]. Multiple imputations were performed before multiple logistic regression analysis because some data were missing from our dataset. Prenatal diagnosis was not performed in all the patients, and some missing values were found. After weighted, we used linear regression analysis to indicate the relationship between survival/ intact-survival rates and gestational age.
Data management and statistical analyses were performed using the R statistical software (version 3.4.1(The R Foundation for Statistical Computing, Vienna, Australia). Inverse probability of treatment weighting was conducted using Survey (V.3.37). All reported p values were two-sided, and values of p < 0.05 were considered statistically significant. Data are presented as medians (interquartile range) or numbers (%).

RESULTS
In our study, 849 infants with CDH were enrolled. Outborn infants (n = 136), infants with unknown GA (n = 3), infants receiving palliative care (n = 35), and infants undergoing FETO (n = 14) were excluded, and one infant was received palliative care and FETO. We studied 167 (19.7%) preterm infants and 682 term infants with CDH (Fig. 1). Table 1 presents detailed infant characteristics, showing that late-preterm infants constitute a large percentage.
The survival and intact-survival rates among 33-36 weeks GA had a significantly lower compare with 37-41 weeks GA (Table 2).
Using the IPTW method based on CDH severity and perinatal factors, gestational age and survival rates have a significantly positive correlation [coefficient of determination (COEF) 3.40, 95% confidence interval (CI), 1.58-5.21, p value <0.001] and higher intactsurvival rate [COEF 2.39, 95% CI, 1.73-4.06, p value 0.005] ( Table 3). Trends of intact-survival rates for both preterm and term infants had significantly changed, but improvement in preterm infants was much smaller than in term infants (Fig. 2).

DISCUSSION/CONCLUSION
Using data from a multicenter study in Japan, our study evaluated survival and intact-survival rates at discharge among preterm and term infants with CDH matched for CDH severity. The results showed that preterm infants with CDH had significantly lower survival and intact-survival rates at discharge than did full-term infants. The trends showed that intact survival rates increased among term and preterm infants, but the degree of preterm have smaller compared with term infants. We attribute that the trend has improved over the epoch is due to the development of fetal diagnostic techniques, advances in ventilators and drugs, and improvements in treatment strategies, including surgical treatment and postoperative management.
Prematurity decreased the survival rate of infants with CDH, regardless of the severity of CDH, which is consistent with previous reports. We obtained similar results, although there were fewer categories of GA. The Neonatal Research Network of Japan, a Japanese neonatal database that registers babies born in Japan at 32 weeks or less, also found that the survival rate of 30,973 babies over a ten-year period from 2003 to 2012 increased as the number of weeks of gestation increased [16]. The trend for infants with CDH appears to be similar to that for preterm infants. Peluso et al. [10] reported a survival rate of 49% for 2,356 CDH preterm infants, while our figures were 55.5% for 32 weeks or less and 67.1% for 33-36 weeks, slightly higher than previously reported. In our study, it is difficult to determine whether these rates were higher because of the small number of patients enrolled at 32 weeks or less. We excluded only 8% of patients with CDH severity at 32 weeks or less when matching for CDH severity. The comparison population comprised almost exclusively late-preterm patients between 33 and 36 weeks of age, with poor outcomes of survival and development compared with term infants [17]. This finding is consistent with the results of our study.
In this study, we found that preterm infants with CDH had an increased need for medical care at discharge from the hospital, even if the severity of CDH was matched. To the best of our knowledge, this has not been reported yet. Neurodevelopmental impairment is an important complication in preterm infants with CDH [18][19][20][21][22]. Although the detailed mechanism is unknown, there are reports of central nervous system (CNS) abnormalities in CDH [23][24][25]. Congenital diaphragmatic hernia is a risk factor for hypoxic acidosis, which is a risk factor for CNS abnormalities during the disease. But, in our study there was no significant difference in CNS abnormality between preterm and term groups. We would need further cases and studies to confirm this hypothesis.
There are various reports on the medical care needed at discharge for CDH [26], but very few are available. There have been several studies on each medical care category, and our investigation is consistent with these studies. Baroudi et al. [27] reported that lower GA was a risk factor for tracheostomy in univariate analysis, but not in multivariate analysis. These findings require additional cases and further investigations.
We defined CDH severity based on two criteria: liver elevation and o/e LHR. The importance of liver-up has been shown in a meta-analysis [28] to help predict survival rates and respiratory disease in infants with isolated left-sided CDH. Observed/expected LHR was also an attractive risk predictor [29], and low o/e LHR was a sensitive predictor of lower survival and higher incidence of ECMO [30]. Terui et al. reported that a combination of factors, including o/e LHR < 25%, liver-up, thoracic stomach, right-side CDH, and severe malformations were sensitive in predicting risk [31]. Liver-up and o/e LHR were combined to determine severity, which is reasonable, considering the evidence obtained in previous studies. In our multicenter study, we used the L/T ratio to estimate illness severity in actual practice. Masahata et al. [32] found that a combination of L/T ratio and liver-up was significantly associated with postnatal severity in a single-center study, and Usui et al. [33] reported an linear relationship between the L/T ratio and o/e LHR in early pregnancy. Although our group used L/T ratio for diagnosis of CDH severity, because several international reports used LHR to assess CDH severity, and because we considered preterm as exposure, we adopted o/e LHR could be corrected by gestational weeks.
This study has several limitations. First, this was a multicenter retrospective study. Therefore, the therapeutic strategies for preterm infants with CDH depended on attending physicians in each facility. Consequently, there is a possible bias in survival and intact survival among CDH preterm infants. Second, this study lacked information after hospital discharge and did not allow long-term follow-up. Our database has many categories for longterm outcomes; therefore, further research is needed to enforce evidence about preterm infants with CDH. Third, we analyzed the effects, including chromosomal and cardiac abnormalities. Nonisolated CDH seems worse than isolated CDH, but we intended to assess outcomes among all preterm infants with CDH. In addition, this study did not evaluate neonatal morbidity under admission for bronchopulmonary dysplasia or intraventricular hemorrhage. Our database lacked this information, and we could not assess the association between neonatal morbidity and preterm infants with CDH. Finally, in our study, preterm infants with CDH showed poor improvement in intact-survival rates at 5-year intervals compared to term infants. The cause of this difference could not be clearly shown, although the influence of CDH was more pronounced in preterm infants. Our study covered only 15 year period, and the number of preterm births per 5-year period was small, and the selection bias was significant. It is necessary to further expand the sample size and conduct more studies in the future.
In conclusion, survival and intact survival rates at discharge for preterm infants with CDH were lower than those for term infants with CDH after matching for CDH severity. Further evaluation of the impact of immaturity on CDH will require more studies on long-term neurodevelopmental outcomes.  Fig. 2 Trends in intact-survival rates among term and preterm infants with CDH.