Clinical characteristics and outcomes of neonates with congenital diaphragmatic hernia at the Vietnam National Children’s Hospital: a retrospective observational study

Abstract

Background

Congenital diaphragmatic hernia (CDH) has a high mortality rate, particularly in low- and middle-income countries. This study aimed to investigate mortality rates and associated factors in CDH neonates in Vietnam.

Methods

This retrospective observational study included CDH neonates admitted to a central children’s hospital in Vietnam between November 2021 and September 2023. We collected data on neonates’ characteristics, management, complications, and outcomes, comparing these data between survivors and non-survivors. We also employed logistic regression analysis to identify factors associated with hospital mortality.

Results

Of 74 neonates with CDH, 64.9% (48/74) were male. The hospital mortality rate was 50.0% (37/74). The median gestational age at birth was 38 weeks (interquartile range [IQR]: 38–39), and the median age at admission was 5.2 h (IQR: 3.0–15.8). All neonates were referred from various prior hospitals, with 83.8% (62/74) requiring immediate postnatal intubation. The neonates presented in critical condition, as reflected by a median pre- and post-ductal SpO₂ difference of 2.5% (IQR: 1.0–10.0), a mean pulmonary artery systolic pressure (PASP) of 51.7 mmHg (standard deviation: 18.3) on admission, and a median Oxygenation Index at 6 h of life of 13.3 (IQR: 6.8–28.2). The median peak Vasoactive Inotropic Score (VIS) during surgical intensive care unit stay was 35.0 (IQR: 15.0–80.0). High-frequency oscillatory ventilation was used as the initial ventilatory mode in 52.7% (39/74) of cases. Supportive therapies included vasopressors (84.9%; 62/73), inotropic agents (29.6%; 21/71), inhaled nitric oxide (13.5%; 10/74), Ilomedin (29.7%; 19/64), and extracorporeal membrane oxygenation (6.8%; 5/74). Surgical repair was performed in 70.3% (52/74) of neonates. Multivariable logistic regression analysis identified higher peak VIS (adjusted odds ratio [AOR]: 1.061; 95% confidence interval [CI]: 1.011–1.113; p = 0.017) and elevated admission PASP (AOR: 1.140; 95% CI: 1.041–1.247; p = 0.005) as independent predictors of hospital mortality.

Conclusions

In this selected cohort of CDH neonates admitted to a central children’s hospital in Vietnam, a high hospital mortality rate was observed. The findings suggest that limited access to neonatal intensive care and surgical repair may have influenced outcomes, warranting further evaluation in similar resource-limited settings to improve care strategies.

Trial registration

Not applicable.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12887-025-06108-3.

Introduction

Congenital diaphragmatic hernia (CDH) is a life-threatening developmental defect of the diaphragm that allows abdominal organs to herniate into the thoracic cavity, leading to severe respiratory distress in neonates. Despite improvements in neonatal intensive care, mortality rates remain significant [1–9], ranging from 17.0% (15/88) [8] to 30.5% (634/2,078) [2] in high-income countries (HICs) and as high as 43.8% (7/16) [1] to 78.8% (965/1,225) [5] in low- and middle-income countries (LMICs). Over the past two decades, studies from Asian countries have reported varying neonatal mortality rates [1, 6]10– [16]. These studies, which often involved relatively small sample sizes and diverse selection criteria, found mortality rates of 41.9% (62/148) in India [10], 29.2% (35/120) in Malaysia [6], 20.8% (5/24) in Taiwan [12], and 21.1% (4/19) in Singapore [14]. These countries represent a range of economic statuses, from low-middle to upper-middle and high-income. The significant disparities in mortality rates may be attributed to variations in the characteristics of neonates, as well as differences in the clinical capacity to manage CDH between HICs and LMICs. Postnatal mortality predictors include prematurity or low birth weight [17, 18], a large defect size [19, 20], associated cardiac anomalies [21, 22], severe pulmonary hypertension (PH) [23, 24], the need for extracorporeal membrane oxygenation (ECMO) [25, 26], birth at a non-tertiary centre [26–29], and poor gas exchange in the early postnatal period [30–33]. Addressing these factors through targeted interventions can significantly reduce mortality rates. Essential measures include improving prenatal diagnosis and monitoring, optimizing neonatal intensive care, expanding access to advanced treatments, strengthening healthcare infrastructure, providing robust support for families and communities, and driving policy reforms. Each of these strategies plays a pivotal role in enhancing survival outcomes.

Vietnam is an LMIC, ranking 15th globally and 3rd in Southeast Asia by population, with 96.462 million people [34]. Over the years, economic and political reforms have driven rapid growth; [35] however, healthcare providers continue to face significant challenges in managing neonates with CDH [36]. Limited resources and a lack of advanced diagnostic and treatment options in local settings hinder the delivery of effective care [36, 37]. Although national health insurance, introduced in 1992, aimed to improve healthcare access and alleviate financial burdens associated with user fees established in 1989, coverage for the latest medical advancements remains inadequate. Furthermore, medical personnel may lack sufficient training and experience to accurately diagnose and manage CDH and other critical neonatal conditions, delaying essential interventions [36]. In Vietnam’s healthcare system, central hospitals are responsible for handling complex cases beyond the capacity of local facilities [38]. As a result, neonates with CDH often experience delays in receiving timely treatment and supportive care [36]. Understanding the country-specific causes, risk factors, and prognoses for CDH is crucial to improving outcomes and reducing neonatal mortality. Therefore, this study aimed to investigate the mortality rate and associated factors in neonates with CDH at a central children’s hospital in Vietnam.

Methods

Study design and setting

We conducted a retrospective observational study on neonates with CDH admitted to the National Children’s Hospital in Hanoi, Vietnam, between November 22, 2021, and September 24, 2023. This study received approval from the Scientific and Ethics Committees of the National Children’s Hospital on August 31, 2023. The National Children’s Hospital is designated as a central hospital with 2,000 beds in northern Vietnam by the Ministry of Health (MOH). In the Vietnamese healthcare system, central hospitals (level I) are responsible for training medical staff and providing care for patients who cannot receive adequate treatment in local hospitals, including provincial and district hospitals (levels II and III, according to the Ministry of Health of Vietnam) [38].

Participants

This study included all neonates diagnosed with CDH at the National Children’s Hospital. Expert paediatricians at the study site established the diagnosis according to the Canadian CDH Collaborative Clinical Practice Guidelines for diagnosis and management [39, 40]. The criteria for diagnosis include: (i) clinical presentation, characterized by respiratory distress within the first few minutes after birth, a barrel-shaped chest, a scaphoid abdomen, and diminished breath sounds on the ipsilateral side; and (ii) chest radiography that revealed herniation of abdominal contents. In cases of diagnostic uncertainty, cross-sectional imaging techniques, such as ultrasound or computed tomography (CT), were employed. At the study site, neonates with CDH were managed according to the Canadian CDH Collaborative Clinical Practice Guidelines for diagnosis and management [39, 40]. Surgical repairs were conducted based on the criteria established by the CDH EURO Consortium Consensus [41].

Data collection

We utilized a unified case record form (CRF) to collect data on study variables from hospital medical records. For data entry, we used EpiData Entry software (EpiData Association, Denmark, Europe), which allowed for streamlined and programmed data input, thereby reducing the likelihood of entry errors. We did not include patient identifiers in the database to ensure patient confidentiality.

Variables

The CRF included five sections with variables primarily derived from the Canadian CDH Collaborative Clinical Practice Guidelines for diagnosis and management and the CDH EURO Consortium Consensus [39–41]. This information was collected by fully trained paediatricians and comprised various aspects such as:

• (i)The first section focused on perinatal characteristics, including whether the neonates had a prenatal diagnosis of CDH. It also covered gestational age at birth, gender of the neonate, delivery method, birth weight at delivery, whether immediate postnatal intubation was performed, and details of admission. This study defined immediate postnatal intubation as the intervention administered to neonates immediately after birth at the delivery hospital.
• (ii)The second section described the admission characteristics of neonates, including age and weight, heart rate (HR), arterial blood pressure (BP), pre-ductal and post-ductal peripheral oxygen saturation (SpO₂), pre-ductal and post-ductal SpO₂ difference, and various laboratory parameters. These parameters comprised white blood cell counts, haemoglobin, platelet count, C-reactive protein, urea, creatinine, and albumin levels. Additionally, imaging findings were noted, specifically the presence of herniated organs (i.e., stomach, spleen, kidney, liver, small intestine, or large intestine) and the location of the hernia (i.e., left side, right side, bilateral sides, or in other areas) observed on chest X-rays, which were confirmed by ultrasound or CT scans when the diagnosis remains uncertain. The presence of PH was identified through echocardiography. Arterial blood gases were also evaluated, which included measurements of blood potential hydrogen (pH), arterial oxygen partial pressure (PaO₂), and arterial carbon dioxide partial pressure (PaCO₂). The section further addressed any associated congenital abnormalities, including congenital heart diseases (e.g., patent ductus arteriosus, ventricular septal defect, atrial septal defect, or patent foramen ovale) and non-cardiac anomalies.
• (iii)The third section described the overall severity of the illness, assessed using the Vasoactive Inotropic Score (VIS) at its peak, which quantifies the dosage and variety of vasoactive drugs administered during the neonate’s stay in the surgical intensive care unit (SICU) [42–44]. Additionally, we evaluated the severity of PH through two primary measurements: (i) pulmonary artery systolic pressure (PASP), measured via echocardiography upon admission [45, 46], and (ii) the Oxygenation Index (OI), monitored at each calendar hour after birth [47–49].
• (iv)The fourth section captured resource utilization in the SICU, highlighting various surgical interventions. It detailed techniques such as mechanical ventilation (MV) and its parameters, including continuous mandatory ventilation (CMV) and high-frequency oscillatory ventilation (HFOV) modes. Additionally, it covered hemodynamic support by administering vasopressors (e.g., adrenaline, noradrenaline, dopamine), and inotropes (e.g., dobutamine, milrinone). In cases where iNO and ECMO were unavailable, we additionally utilized milrinone as a pulmonary vasodilator therapy for managing persistent PH in neonates with CDH. The management of PH was also emphasized, including the administration of iNO and ilomedin. Furthermore, this section focused on ECMO management and surgical repairs. The types of surgical repairs included suture-only surgical repairs, which utilized sutures alone, and patch-enhanced surgical repairs, which incorporated patches.
• (v)The fifth section focused on complications, such as bleeding, chylothorax, nosocomial infections, and acute kidney injury, pneumothorax, and clinical outcomes, such as hospital mortality. Nosocomial infections were identified as infections that neonates acquire while receiving healthcare. These infections, which can include surgical wound infections, bloodstream infections, and urinary tract infections, are not present or in the incubation stage at the time of admission. Instead, they typically develop 48 h or more after the patient has been admitted [50, 51].

Outcome measures

The primary outcome was hospital mortality, defined as death from any cause during hospitalization. We also examined secondary outcomes, including complications and hospital length of stay (LOS).

Sample size

In this retrospective observational study, we aimed to investigate the hospital mortality rate as the primary outcome. To determine the minimum sample size required for estimating a population proportion with an 85% confidence level and a margin of error of ± 8.26%, we relied on an assumed population proportion of 41.9%, derived from a previous study on hospital mortality rates [10]. As a result, our sample size should be at least 74 neonates, which should be sufficient to represent a normal distribution.

where:

z is the z score (z score for a 85% confidence level is 1.44)

ε is the margin of error (ε for a confidence interval of±8.26% is 0.0826)

p is the population proportion (p for a population proportion of 41.9% is 0.419)

n is the sample size

Statistical analyses

We utilized Stata 18 (StataCorp LLC, Texas, USA) for data analysis. We reported the data as numbers (no.) and percentages (%) for categorical variables and medians with interquartile ranges (IQRs) or means with standard deviations (SDs) for continuous variables. Furthermore, comparisons were made between survival and death in the hospital and between surgical repair and no surgical repair for each variable using the Chi-squared or Fisher’s exact test for categorical variables and the Mann–Whitney U test, Kruskal–Wallis test, or one-way analysis of variance for continuous variables.

We assessed the factors associated with hospital mortality using logistic regression analysis. We included all variables related to perinatal characteristics, neonatal admission characteristics, overall severity of illness, resource utilization in the SICU, surgical interventions, and complications in the univariable logistic regression model. To mitigate the risk of overfitting and ensure a balance between statistical robustness and clinical applicability, we employed a systematic approach in selecting variables for the multivariable logistic regression model. Specifically, we selected variables with a P-value < 0.10 in the univariable analysis comparing hospital mortality and survival (e.g., admission pre-ductal and post-ductal SpO₂ difference, admission PaO₂, admission PaCO₂, admission PASP, peak VIS, and HFOV as the initial ventilation mode), as well as variables deemed clinically essential (e.g., birth weight at delivery [52], gender of the neonate [53], age at hospital admission [52], admission serum lactate level [54], congenital heart diseases [21, 22], and immediate postnatal intubation [41, 55]) based on prior literature and domain expertise to include in the multivariable logistic regression model. Conversely, we excluded variables with a P-value < 0.10 in the univariable analysis if they had a missing Data rate exceeding 10% (e.g., OI measured 6 h after birth). The final set of selected variables included perinatal characteristics (i.e., birth weight at delivery, gender of the neonate, and immediate postnatal intubation), admission characteristics of neonates (i.e., age at hospital admission, pre-ductal and post-ductal SpO₂ difference, serum lactate level, PaO₂, PaCO₂, and congenital heart diseases), overall severity of illness (i.e., admission PASP and peak VIS), and resource utilization in the SICU (i.e., HFOV as the initial ventilation mode). Using a stepwise backward elimination method, we began with the complete multivariable logistic regression model, employing the forced entry approach. We presented the odds ratios (ORs) and 95% confidence intervals (CIs) from the univariable logistic regression model, as well as the adjusted odds ratios (AORs) and 95% CIs from the multivariable logistic regression model. We evaluated the goodness-of-fit of the multivariable model using several criteria, including the log-likelihood, a pseudo-R-squared value, and the Hosmer-Lemeshow goodness-of-fit test.

The significance levels were two-tailed for all analyses, and we considered the P < 0.05 as a statistically significant value.

Results

In our study, we initially entered Data from 97 paediatric patients with CDH into our Database. We then excluded 18 patients who were older than 30 days, as they fell outside the neonatal age range. Additionally, we removed 5 neonates (6.3%; 5/79) due to extensive missing data across multiple variables, including those critical to our analysis. This absence was primarily due to clinical circumstances, such as the issuance of a Do-not-resuscitate order either at birth or upon admission to the hospital. Consequently, our analyses were conducted with 74 eligible neonates (Fig. 1).

Fig. 1.

Flowchart Illustrating the Study Design

The primary and secondary outcomes

Of the 74 eligible neonates, 50.0% (37/74) died in the hospital (Fig. 1). The perioperative complications observed included chylothorax (10.8%; 8/74), nosocomial infections (8.1%; 6/74), acute kidney injury (18.9%; 14/74), and pneumothorax (8.2%; 6/73) (Table 1). The median LOS was 13.5 days with an IQR of 5.2 to 26.0 days (Table 1). Table 1 provides detailed comparisons of complications and LOS between neonates who survived and those who died in the hospital.

Table 1.

Perioperative complications and outcomes in neonates with congenital diaphragmatic hernia based on hospital survivability

Variables	All cases n = 74	Survived n = 37	Died n = 37	P value
Perioperative complications
Chylothorax, no. (%)	8 (10.8)	5 (13.5)	3 (8.1)	0.454
Nosocomial infection, no. (%)	6 (8.1)	5 (13.5)	1 (2.7)	0.088
Acute kidney injury, no. (%)	14 (18.9)	6 (16.2)	8 (21.6)	0.553
Pneumothorax, no. (%), n = 73	6 (8.2)	0 (0.0)	6 (16.2)	0.012
Clinical time course and outcomes
Length of MV (days), median (IQR)	7.4 (5.0–13.0)	7.9 (6.9–11.9)	5.0 (1.9–13.0)	0.008
Length of SICU stay (days), median (IQR)	12.6 (5.2–22.4)	17.1 (11.4–30.4)	5.2 (1.2–12.8)	< 0.001
Length of hospital stay (days), median (IQR)	13.5 (5.2–26.0)	17.7 (14.0–33.1)	5.2 (1.2–12.9)	< 0.001
Outcome status				< 0.001
Died in SICU stay, n (%)	37 (50.0)	0 (0.0)	37 (50.0)	NA
Alive upon discharge from SICU stay, but died in hospital stay, n (%)	0 (00)	0 (00)	0 (00)	NA
Alive upon hospital stay discharge, n (%)	37 (50.0)	37 (50.0)	0 (0.0)	NA

^a To indicate comparisons between neonates who survived and those who died in the hospital.

Perinatal characteristics

In this study, all neonates were transferred to the Vietnam National Children’s Hospital from various prior hospitals, including the central obstetrics hospital (17.6%; 13/74), provincial obstetrics and paediatrics hospitals (70.3%; 52/74), and central or provincial general hospitals (12.1%; 9/74) (Table 2). Among these neonates, 64.9% (48/74) were male (Table 2). The median gestational age at birth was 38 weeks (IQR: 38–39) (Table 2), with a preterm (< 37 weeks) rate of 18.9% (14/74) (Table S1 in Additional file 1). Additionally, 98.7% (73/74) had a prenatal diagnosis of CDH (Table 2). The delivery methods included vaginal delivery (28.4%; 21/74) and caesarean Sect. (71.6%; 53/74) (Table 2). The median birth weight was 3.0 kg (IQR: 2.6–3.2) (Table 2), with 79.7% (59/74) classified as normal birth weight (2.5–4.0 kg) and 20.3% (15/74) as low birth weight (< 2.5 kg) (Table S1 in Additional file 1). Immediate postnatal intubation was performed in 83.8% (62/74) of neonates with CDH (Table 2). Table 2 and Table S1 (Additional file 1) present a comparison of perinatal characteristics between neonates who survived and those who died in the hospital.

Table 2.

Perinatal Characteristics of Neonates with Congenital Diaphragmatic Hernia Based on Hospital Survivability

Variables	All cases n = 74	Survived n = 37	Died n = 37	P valuea
Prior hospitalization				0.644
Central obstetrics hospitals, no. (%)	13 (17.6)	5 (13.5)	8 (21.6)
Provincial obstetrics and paediatrics hospitals, no. (%)	52 (70.3)	27 (73.0)	25 (67.6)
Central or provincial general hospitals, no. (%)	9 (12.1)	5 (13.5)	4 (10.8)
Prenatal diagnosis of CDH, no. (%)	73 (98.7)	36 (97.3)	37 (100.0)	0.314
Gestational age at birth (weeks), median (IQR)	38 (38–39)	38 (38–39)	38 (37–39)	0.081
Gender of the neonate (male), no. (%)	48 (64.9)	21 (56.8)	27 (73.0)	0.144
Delivery method				0.797
Vaginal delivery, no. (%)	21 (28.4)	10 (27.0)	11 (29.7)
Caesarean section, no. (%)	53 (71.6)	27 (73.0)	26 (70.3)
Birth weight at delivery (kg), median (IQR)	3.0 (2.6–3.2)	3.0 (2.7–3.2)	2.9 (2.6–3.1)	0.175
Immediate postnatal intubation, no. (%)	62 (83.8)	27 (73.0)	35 (94.6)	0.012

^a To indicate comparisons between neonates who survived and those who died in the hospital

Clinical characteristics and management

In our study cohort of neonates, the median age at hospital admission was 5.2 h (IQR: 3.0–15.8) (Table 3). The admission clinical examination revealed a median HR of 127 beats/min (IQR: 114–144), along with median systolic and diastolic BPs of 68.0 mmHg (IQR: 62.0–75.0) and 40.0 mmHg (IQR: 34.0–46.0), respectively (Table 3). Additionally, the median pre-ductal and post-ductal SpO₂ levels were 97.5% (IQR: 92.5–99.0) and 95.0% (IQR: 81.5–97.0), respectively, which resulted in a median pre- and post-ductal SpO₂ difference of 2.5% (IQR: 1.0–10.0) (Table 3). In the majority of cases of CDH, herniation predominantly occurred on the left side (82.4%; 61/74), followed by the right side (14.9%; 11/74), with a small rate in other areas (2.7%; 2/74) (Table 4). Notably, no instances of bilateral herniation were observed (Table 4). The herniated abdominal organs included the stomach (25.7%; 19/74), spleen (47.3%; 35/74), liver (20.3%; 15/74), small intestine (58.1%; 43/74), and large intestine (56.8%; 42/74) (Table 4). Additionally, the most commonly associated congenital abnormalities were congenital heart defects (90.1%; 64/71), including patent ductus arteriosus (87.1%; 61/70), ventricular septal defect (5.6%; 4/71), atrial septal defect (2.8%; 2/71), and patent foramen ovale (49.3%; 35/71) (Table 4). Among all patients, the mean admission PASP was 51.7 mmHg (SD: 18.3) (Table 4). The median OI, measured six hours after birth, was 13.3 (IQR: 6.8–28.2) (Table 4). Additionally, the median peak VIS recorded during the neonate’s stay in the SICU was 35.0 (IQR: 15.0–80.0) (Table 4). Tables 2 and 3, along with Tables S2 and S3 (Additional file 1), present a comparison of clinical characteristics and laboratory findings between neonates who survived and those who died in the hospital.

Table 3.

Clinical Characteristics and Gas Exchange of Neonates with Congenital Diaphragmatic Hernia at Admission, Based on Hospital Survivability

Variables	All cases n = 74	Survived n = 37	Died n = 37	P valuea
Clinical characteristics
Age at hospital admission (hours), median (IQR)	5.2 (3.0–15.8)	9.0 (3.5–23.8)	3.5 (2.3–7.3)	0.012
Heart rate (beats/min), median (IQR), n = 51	127 (114–144)	127 (113–144)	139 (124–143)	0.463
Blood pressure
Systolic blood pressure (mmHg), mean (SD), n = 51	68.0 (62.0–75.0)	65.5 (61.0–75.0)	71.0 (62.0–75.0)	0.336
Diastolic blood pressure (mmHg), median (IQR), n = 51	40.0 (34.0–46.0)	39.5 (34.0–45.0)	44.0 (38.0–48.0)	0.185
Oxygen saturation (SpO2)
Pre-ductal SpO2 (%), median (IQR), n = 68	97.5 (92.5–99.0)	99.0 (98.0–99.0)	92.5 (87.0–97.0)	< 0.001
Post-ductal SpO2 (%), median (IQR), n = 68	95.0 (81.5–97.0)	97.0 (95.0–98.0)	81.5 (70.0–93.0)	< 0.001
Pre-/post-ductal SpO2 difference (%), median (IQR), n = 68	2.5 (1.0–10.0)	1.0 (1.0–3.0)	10.0 (2.0–17.0)	< 0.001
Gas exchange
pH, median (IQR)	7.3 (7.2–7.3)	7.3 (7.3–7.4)	7.2 (7.1–7.3)	< 0.001
PaCO2 (mmHg), median (IQR)	58.5 (45.0–76.0)	50.0 (42.0–60.0)	76.0 (56.0–87.0)	< 0.001
PaO2 (mmHg), median (IQR)	47.7 (34.0–83.0)	69.0 (44.0–133.0)	42.6 (30.5–50.0)	< 0.001
HCO3− (mmol/L), mean (SD), n = 73	25.8 (4.4)	25.9 (3.5)	25.6 (5.2)	0.812

^a To indicate comparisons between neonates who survived and those who died in the hospital.

Table 4.

Associated Congenital Abnormalities, Herniated Organs, and Severity of Illness in Neonates with Congenital Diaphragmatic Hernia at Admission, Based on Hospital Survivability

Variables	All cases n = 74	Survived n = 37	Died n = 37	P valuea
Associated congenital abnormalities
Congenital heart diseases, no. (%), n = 71	64 (90.1)	30 (83.3)	34 (97.1)	0.051
Patent ductus arteriosus, no. (%), n = 70	61 (87.1)	29 (80.6)	32 (94.1)	0.090
Ventricular septal defect, no. (%), n = 71	4 (5.6)	0 (0.0)	4 (11.4)	0.037
Atrial septal defect, no. (%), n = 71	2 (2.8)	1 (2.8)	1 (2.9)	0.984
Patent foramen ovale, no. (%), n = 71	35 (49.3)	19 (52.8)	16 (45.7)	0.552
Diagnostic imaging of herniated organs
Herniated organs
Stomach, no. (%)	19 (25.7)	11 (29.7)	8 (21.6)	0.425
Spleen, no. (%)	35 (47.3)	24 (64.9)	11 (29.7)	0.002
Liver, no. (%)	15 (20.3)	8 (21.6)	7 (18.9)	0.772
Small intestine, no. (%)	43 (58.1)	29 (78.4)	14 (37.8)	< 0.001
Large intestine, no. (%)	42 (56.8)	28 (75.7)	14 (37.8)	0.001
Location of hernia				0.110
Left side, no. (%)	61 (82.4)	32 (86.5)	29 (78.4)
Right side, no. (%)	11 (14.9)	3 (8.1)	8 (21.6)
Other areas, no. (%)	2 (2.7)	2 (5.4)	0 (0.0)
The severity of illness
Severity of cardiac dysfunction
Peak VIS, median (IQR), n = 73	35.0 (15.0–80.0)	20.0 (10.0–35.0)	70.0 (42.5–107.5)	< 0.001
Severity of pulmonary hypertension
Admission PASP (mmHg), mean (SD)	51.7 (18.3)	43.0 (17.8)	60.4 (14.3)	< 0.001
OI measured 6 h after birth, median (IQR), n = 42	13.3 (6.8–28.2)	8.6 (5.2–9.8)	20.1 (11.5–34.0)	0.002

^a To indicate comparisons between neonates who survived and those who died in the hospital

In our study, only 70.3% (52/74) of neonates underwent surgical repairs, with 36.5% (19/52) receiving suture-only and 63.5% (33/52) undergoing patch-enhanced repairs (Table 5). The remaining 22 neonates (29.7%; 22/74) did not undergo surgery due to death before surgical intervention (Table S4 in Additional file 1). All neonates received invasive MV, with HFOV as the initial ventilation mode in 52.7% (39/74) of cases (Table 5). Upon admission, our neonates received hemodynamic support, which included vasopressors for 84.9% (62/73) and inotropic agents for 29.6% (21/71), as detailed in Table 5. Additionally, pulmonary vasodilator therapy was provided, with iNO administered to 13.5% (10/74) of the neonates and Ilomedin given to 29.7% (19/64) (Table 5). Notably, only 6.8% (5/74) of the neonates required ECMO (Table 5). Table 5 presents the management strategies and compares these variables between neonates who survived and those who died in the hospital.

Table 5.

Resource Utilization in the Surgical Intensive Care Unit and Surgical Interventions for Neonates with Congenital Diaphragmatic Hernia, Based on Hospital Survivability

Variables	All cases n = 74	Survived n = 37	Died n = 37	P valuea
Respiratory support
The initial MV modes				< 0.001
CMV, no. (%)	35 (47.3)	29 (78.4)	6 (16.2)
HFOV, no. (%)	39 (52.7)	8 (21.6)	31 (83.8)
Haemodynamic support
Vasopressor administration, no. (%), n = 73	62 (84.9)	26 (78.3)	36 (100.0)	< 0.001
Maximum vasopressor rate (µg/kg/min)
Adrenaline, median (IQR), n = 71	0.25 (0.10–0.46)	0.15 (0.00–0.24)	0.40 (0.25–0.60)	< 0.001
Noradrenaline, median (IQR), n = 71	0.00 (0.00–0.30)	0.00 (0.00–0.06)	0.28 (0.00–0.40)	0.020
Dopamine, mean (SD), n = 71	1.1 (3.6)	1.4 (4.2)	0.9 (3.0)	0.615
Inotrope administration, no. (%), n = 71	21 (29.6)	5 (13.5)	16 (47.1)	0.002
Maximum inotrope rate (µg/kg/min)
Dobutamine, mean (SD), n = 71	2.4 (5.1)	0.8 (2.2)	4.1 (6.6)	0.005
Milrinone, median (IQR), n = 70	0.03 (0.11)	0.00 (0.00)	0.06 (0.15)	0.024
Adjunctive therapies
iNO administration, no. (%)	10 (13.5)	0 (0.0)	10 (27.0)	0.001
Ilomedin administration, no. (%), n = 64	19 (29.7)	4 (11.1)	15 (53.6)	< 0.001
Ilomedin rate (ng/kg/min), mean (SD), n = 64	0.33 (0.62)	0.09 (0.30)	0.64 (0.77)	< 0.001
ECMO management, no. (%)	5 (6.8)	0 (0.0)	5 (13.5)	0.021
Surgical repair
Surgical repairs, no. (%)	52 (70.3)	37 (100.0)	15 (40.5)	< 0.001
Type of surgical repairs, n = 52				0.004
Suture-only surgical repairs, no. (%)	19 (36.5)	18 (48.7)	1 (6.8)
Patch-enhanced surgical repairs, no. (%)	33 (63.5)	19 (51.4)	14 (93.3)

^a To indicate comparisons between neonates who survived and those who died in the hospital

Factors associated with hospital mortality

In univariable logistic regression analyses, we found that immediate postnatal intubation was significantly associated with a higher risk of hospital mortality, with an OR of 6.481 (95% CI: 1.310–32.072; p = 0.022) (Table 6). Additionally, a greater pre-ductal and post-ductal SpO₂ difference (OR: 1.367; 95% CI: 1.145–1.632; p = 0.001), an elevated PaCO₂ (OR: 1.066; 95% CI: 1.031–1.101; p < 0.001), a higher peak VIS (OR: 1.053; 95% CI: 1.027–1.080; p < 0.001), an increased PASP at admission (OR: 1.068; 95% CI: 1.032–1.106; p < 0.001), and the use of HFOV as the initial ventilation mode (OR: 18.729; 95% CI: 5.793–60.549; p < 0.001) were also significantly associated with an increased risk of hospital mortality (Table 6). Conversely, a higher PaO₂ was significantly associated with a decreased risk of hospital mortality (OR: 0.969; 95% CI: 0.950–0.987; p = 0.001) (Table 7). In a subsequent multivariable logistic regression analysis, a higher peak VIS (AOR: 1.061; 95% CI: 1.011–1.113; p = 0.017) and a higher admission PASP (AOR: 1.140; 95% CI: 1.041–1.247; p = 0.005) remained independent predictors of hospital mortality. Furthermore, we identified lower birth weight at delivery (AOR: 0.112; 95% CI: 0.017–0.733; p = 0.022) as a new independent predictor of hospital mortality (Table 6).

Table 6.

Factors Related to Hospital Mortality in Neonates with Congenital Diaphragmatic Hernia: Logistic Regression Analyses

Factors	Univariable logistic regression analysesa				Multivariable logistic regression analysisb
	OR	95% CI for OR		p-value	AOR	95% CI for AOR		p-value
		Lower	Upper			Lower	Upper
Perinatal characteristics
Birth weight at delivery	0.574	0.252	1.310	0.187	0.112	0.017	0.733	0.022
Gender (male)	2.057	0.776	5.451	0.147	0.608	0.089	4.159	0.612
Immediate postnatal intubation	6.481	1.310	32.072	0.022	NA	NA	NA	NA
Clinical characteristics and laboratory investigations
Age at hospital admission (hours)	0.975	0.946	1.004	0.090	NA	NA	NA	NA
Pre-/post-ductal SpO2 difference	1.367	1.145	1.632	0.001	1.332	0.997	1.780	0.053
Lactate (mmol/L)	1.253	0.970	1.619	0.084	NA	NA	NA	NA
PaO2 (mmHg)	0.969	0.950	0.987	0.001	NA	NA	NA	NA
PaCO2 (mmHg)	1.066	1.031	1.101	< 0.001	1.076	0.996	1.163	0.064
Associated congenital abnormalities
Congenital heart diseases	6.800	0.774	59.746	0.084	NA	NA	NA	NA
Severity of illness
Peak VIS	1.053	1.027	1.080	< 0.001	1.061	1.011	1.113	0.017
Admission PASP	1.068	1.032	1.106	< 0.001	1.140	1.041	1.247	0.005
Respiratory support
HFOV as the initial ventilation mode	18.729	5.793	60.549	< 0.001	NA	NA	NA	NA
Constant					0.000			0.034

^a)Each variable of perinatal characteristics, neonatal admission characteristics, overall severity of illness, resource utilization in the SICU, surgical interventions, and complications was initially analyzed using the univariable logistic regression model. Variables with a P-value < 0.10 in the univariable analysis, alongside clinically significant factors, were considered for inclusion in the multivariable logistic regression model.

 ^b)All selected variables were incorporated into the multivariable model, which was then refined using a stepwise backward elimination process, starting with the forced entry method. In addition, the final model demonstrated a good model fit, with a log-likelihood of −14.43 and a pseudo R² of 0.689, indicating strong explanatory power. The Hosmer-Lemeshow goodness-of-fit test yielded a χ²(8) = 4.12, p-value = 0.84, suggesting excellent calibration and no evidence of poor fit.

Discussion

In this study, we observed that all neonates with CDH were transferred from various prior hospitals to our study site, with a notably low rate of immediate postnatal intubation (Table 2). Half of these neonates died in the hospital (Fig. 1). The majority of neonates presented with severe conditions, characterized by significant differences in pre- and post-ductal SpO₂ levels, elevated PASP at the time of admission, high OI measured 6 h after birth, and increased peak VIS during SICU stay (Table 4). Despite more than half of the neonates with CDH receiving HFOV as the initial ventilation mode, only a low rate received management with iNO, Ilomedin, or ECMO during SICU stay (Table 5). The rate of surgical repairs performed was also relatively low (Table 5). Multivariable analysis identified lower birth weight at delivery as an independent predictor of hospital mortality, along with higher peak VIS and elevated admission PASP as additional independent predictors (Table 6).

In the present study, our mortality rate is significantly higher than the rate reported in a previous study (10.1%; 14/139) conducted in Hanoi, Vietnam [16]. The discrepancy in mortality rates may be attributed to differences in inclusion criteria between the studies. For instance, our study included neonates with CDH who received neonatal intensive care, regardless of whether they underwent surgical repair. In LMICs, our mortality rate aligns with the rate reported in India (41.9%; 62/148) [10], while it is much lower than the rate reported in a population-based study (78.8%; 965/1,225) conducted in São Paulo State, Brazil, between 2004 and 2015 [5]. Population-based studies highlight the concept of ‘hidden mortality’ [5, 56, 57], first described by Harrison in 1978 [56]. Advances in prenatal diagnosis have led to an increasing number of prenatally diagnosed cases. Up to 60% of all neonates with CDH are prenatally diagnosed [58], but this has also led to an increased number of terminations of pregnancies due to the diagnosis. These findings explain why our mortality rate is lower than the rates reported in these population-based studies. However, our observed mortality rate is higher than those reported in several other studies in HICs, such as Canada (17.0%; 15/88) [8], United States (28.3%; 206/728) [27], and a report from the Congenital Diaphragmatic Hernia Study Group (CDHSG) (30.5%; 634/2,078) [2]. These disparities may arise from various factors, as highlighted below:

Firstly, all neonates with CDH included in our study were referred from various prior hospitals to the participating central hospital (Table 2). This pattern of referral may have introduced selection bias, as patients with more severe clinical presentations were more likely to be transferred [38, 59]. As a result, our cohort may disproportionately represent critically ill cases, which could limit the generalizability of our findings to the broader CDH population, particularly those with milder conditions treated at local facilities. A distinctive characteristic of neonates with CDH in Vietnam is that many are initially diagnosed with CDH or other thoracic lesions, including congenital cystic adenomatoid malformation, bronchopulmonary sequestration, bronchogenic cysts, bronchial atresia, or teratomas, at local hospitals. These cases are only transferred to the central hospital when their conditions become severe [38]. Since 1961, the Ministry of Health in Vietnam has managed healthcare provision through a system known as the Direction of Healthcare Activities, which mandates collaboration between different levels of health facilities [38]. Higher-tier hospitals are responsible for supporting lower-level facilities to improve access to primary care for local communities while simultaneously handling complex cases that exceed the capacity of local hospitals [38]. However, this referral system often results in delayed diagnosis, treatment initiation, and surgical intervention for CDH, potentially contributing to higher mortality rates [59]. A retrospective study found that hospitals with a higher volume of cases and surgical procedures achieved better outcomes, including lower mortality rates and shorter durations of MV [27]. Thus, to improve survival rates for neonates with CDH in Vietnam, it is crucial to enhance human resources, medical infrastructure, and sociological support systems at the local level.

Secondly, the Transfer of neonates with CDH from local hospitals to specialized centres can further compromise their already critical condition. A study conducted in the United States found that only 47.7% (1020/2140) of infants with CDH were transported to specialized facilities [28]. In contrast, our study revealed that all neonates with CDH were transferred from various prior hospitals to our study site (Table 2). However, a significant proportion of these neonates did not receive immediate postnatal intubation and ventilation, essential interventions for stabilizing their condition during transfer (Table 2). In Vietnam, patient Transfers between medical facilities are regulated by Circular No. 14/2014/TT-BYT, issued by the MOH of Vietnam on April 14, 2014 [60]. This directive specifies the procedures, conditions, and medical protocols for patient transfers. In emergencies, medical facilities must ensure the following measures are in place: (i) the use of ambulances or other appropriate vehicles; (ii) the availability of essential medical equipment and emergency medications during transport; and (iii) the assignment of qualified medical personnel, including doctors, nurses, and midwives, to monitor the patient’s condition and apply suitable transport techniques based on their specific needs. Despite these regulations, pre- and inter-hospital transportation services in Vietnam remain underdeveloped [61–65]. The availability of ambulances, properly trained medical personnel, and essential life-saving equipment is limited, while medical oversight and systematic quality monitoring are rare [65]. Consequently, both basic life support and advanced life support interventions are frequently delayed for critically ill patients until they reach a hospital, increasing the risk of adverse outcomes [61–64]. Previous studies have indicated that neonatal transport of infants with CDH is associated with poorer survival rates compared to those born at tertiary centres that specialize in the management of CDH [28, 29]. This decline in survival can be attributed to the stress of transport, particularly affecting the CDH population, as they are highly susceptible to developing PH. Additionally, transported infants may experience periods of hypoxia and/or hemodynamic instability during transport [28], especially in LMICs where the quality of pre-hospital or inter-hospital care can significantly impact patient outcomes [61–64, 66]. Therefore, improvements are needed in Vietnam’s neonatal transfer services, including critical interventions during transportation (such as intubation and ventilation) and the optimization of ambulance and pre- and inter-hospital care.

Lastly, the observed variation may stem from differences in the neonates and clinical ability to care for critically ill neonates between LMIC and HIC settings. Congenital heart disease is the most common congenital abnormality associated with CDH in infants, and those with severe cardiac defects tend to have poorer outcomes [21, 22]. According to a report from the CDHSG covering the period from 2001 to 2006, only 16% (322/2,077) of infants with CDH were found to have congenital heart diseases [2]. In contrast, our study revealed that nearly all neonates with CDH had congenital heart diseases (Table 4), substantially contributing to the high mortality rate observed (Table 4). The present study found that immediate postnatal intubation which is crucial for care and gentle ventilation was often limited (Table 2). While small retrospective studies at single centres investigated the use of non-invasive respiratory support (e.g., continuous positive airway pressure or oxygenation via nasal cannula) in neonates with mild prenatally diagnosed CDH as an alternative to immediate intubation and reported favourable outcomes [67, 68], the CDH EURO Consortium and the American Heart Association/American Academy of Paediatrics recommend immediate intubation for infants with prenatally diagnosed CDH as the standard of care [41, 55]. This approach avoids the use of oronasal positive pressure (e.g., bag-mask ventilation), which can distend the stomach and compress the lungs, leading to further lung hypoplasia, more severe PH, and an increased risk of mortality [41, 69]. In our study, over half of the neonates with CDH received HFOV as the initial ventilation mode (Table 5). This rate is significantly higher than the rate (34.8%; 674/1,936) reported by the CDHSG [2]. Currently, limited data is available on comparing CMV and HFOV in neonates with CDH [70, 71]. In a randomized trial that included 171 infants with prenatally diagnosed CDH (87% of whom had left-sided CDH), the rates of bronchopulmonary dysplasia or death were lower in those assigned to CMV compared to those receiving HFOV (45% versus 54%); however, this difference was not statistically significant, with an OR of 0.62 (95% CI: 0.25–1.55) [70]. Our univariable logistic regression analysis indicates that the use of HFOV as the initial ventilation mode was significantly associated with a higher risk of hospital mortality (Table 6). This association likely reflects the severity of illness, as patients who required HFOV were in a more critical condition compared to those who did not receive this intervention during their SICU stay (Table S5 in Additional file 1). However, in the subsequent multivariable logistic regression analysis, HFOV as the initial ventilation mode did not emerge as an independent predictor of hospital mortality (Table 6), suggesting that illness severity, rather than the ventilation strategy itself, was the primary determinant of patient outcomes. The present study identified significant differences between pre- and post-ductal SpO₂ levels, elevated PASP upon admission, a high OI at 6 h post-birth and increased peak VIS during the SICU stay (Table 4). Despite these critical indicators, only a small rate of neonates with CDH received ECMO support during their stay in the SICU (Table 5), a markedly lower rate than the 28.6% (595/2,078) reported by the CDHSG. Observational studies support the efficacy of ECMO for neonates with severe PH due to CDH, suggesting that survival rates for these critically ill infants have improved since the introduction of ECMO [25, 72, 73]. According to a report from the Extracorporeal Life Support Organization registry, which examined ECMO use in infants with CDH over 25 years, mortality rates declined from 62% in the earlier era (1986–1990) to 49% in the later era (2006–2010) [25]. ECMO can be life-saving for neonates with CDH because it temporarily takes over the function of the heart and lungs, allowing these underdeveloped or compromised organs to rest and recover. In CDH, the herniation of abdominal organs into the chest cavity severely impairs lung development and function, often leading to persistent PH and respiratory failure. In our cohort, the limited use of ECMO may have contributed to the persistence of elevated PASP and increased peak VIS, both of which emerged as independent predictors of hospital mortality (Table 6). Our study also revealed that surgical repairs for neonates with CDH (Table 5 and Table S4 in Additional file 1) are less frequently observed than those reported by the CDHSG, which Documented a rate of 83% (1,729/2,078) [2]. Surgical repair for neonates with CDH is generally deferred until the infant achieves physiologic stability rather than being performed immediately after birth [39–41]. This strategy, commonly known as delayed repair, has become the standard of care in many specialized centres. In CDH, underdeveloped lungs and the frequent occurrence of persistent PH complicate early intervention. Immediate surgery in this fragile state may worsen PH and precipitate hemodynamic instability. Delaying surgical repair allows time for (i) pulmonary and cardiovascular stabilization using gentle ventilation strategies, pulmonary vasodilators, and ECMO when indicated; (ii) reduction in right-to-left shunting, thereby enhancing oxygenation; and (iii) improved surgical tolerance with reduced intraoperative and postoperative complications [39–41]. However, findings from the present study reveal that although we initially managed a high rate of neonates with HFOV, there was a low utilization of pulmonary vasodilators and ECMO (Table 5). These limited therapies may have hindered the optimization of preoperative stabilization, thereby restricting the feasibility of timely surgical repair. Overall, these findings discussed earlier highlight the importance of enhancing neonatal intensive care and increasing the availability of surgical repairs at the Vietnam National Children’s Hospital to reduce mortality rates in neonates with CDH.

The present study has certain limitations. First, the retrospective nature of this study limited data availability for several variables (Table S6, Additional file 1). For example, OI at 6 h post-birth was available for only 42 neonates. Incomplete data may introduce variable selection bias in the multivariable logistic regression analysis, potentially leading to the omission of clinically relevant predictors and compromising the model’s robustness. Additionally, some excluded neonates had missing data for most variables, as these might have received a do-not-resuscitate order at birth or upon admission, making their inclusion in the analysis impossible. Excluding neonates with incomplete data might also introduce selection bias and an underestimation of fatalities. Secondly, the study was conducted at a single centre in Hanoi, Vietnam, focusing on a highly selected population of cases transferred from local hospitals to the highest-level public sector hospitals in Vietnam. As a result, the number of neonates with CDH is likely to be significantly higher. In Addition, not all neonates with CDH from local hospitals were referred to the participating central children’s hospital; only those with established diagnoses were transferred. This implicit selection bias and incomplete patient inclusion in the study database could potentially skew the mortality rate, resulting in an underestimation of fatalities. Finally, although the sample size was large enough, the CI was wide (± 8.26%), which might influence the normal distribution of the collected sample. Future studies should focus on multicentre prospective designs with comprehensive data collection to minimize selection bias and improve the generalizability of findings across diverse populations of neonates with CDH.

Conclusion

This study investigated a selected cohort of neonates with CDH admitted to a central children’s hospital in Vietnam, revealing a high mortality rate. These neonates presented with severe conditions and received limited intensive neonatal care, with a low rate of surgical repairs. Lower birth weight at delivery, higher peak VIS, and elevated admission PASP were independent predictors of hospital mortality. These findings highlight the potential value of enhancing neonatal intensive care services and improving access to surgical repairs, which may lead to improved outcomes for neonates with CDH in similar resource-limited settings.

Abbreviations

AORs

Adjusted odds ratios

BP

Blood pressure

CDH

Congenital diaphragmatic hernia

CDHSG

Congenital Diaphragmatic Hernia Study Group

CIs

Confidence intervals

CMV

Continuous mandatory ventilation

CRF

Case report form

CT

Computed tomography

ECMO

Extracorporeal membrane oxygenation

HFOV

High-frequency oscillatory ventilation

HICs

High-income countries

HR

Heart rate

iNO

Inhaled nitric oxide

IQRs

Interquartile ranges

LMICs

Low- and middle-income countries

LOS

Hospital length of stay

MOH

Ministry of Health

MV

Mechanical ventilation

OI

Oxygenation Index

ORs

Odds ratios

PaCO₂

Arterial carbon dioxide partial pressure

PaO₂

Arterial oxygen partial pressure

PASP

Pulmonary artery systolic pressure

pH

Blood potential hydrogen

PH

Pulmonary hypertension

SDs

Standard deviations

SICU

Surgical intensive care unit

SpO₂

Peripheral oxygen saturation

VIS

Vasoactive Inotropic Score

Authors’ contributions

DAD contributed to the conception, design of the work, data acquisition, and interpretation of data for the work, wrote the first draft of the work, and revised the draft critically for important intellectual content; TVN and DTTN contributed to the design of the work, data acquisition, and interpretation of data for the work, and revised the draft critically for important intellectual content; DTP and PTHN contributed to the design of the work, data analysis, and interpretation of data for the work; TDN, TAP, DTTL, and MTN contributed to the data acquisition, and interpretation of data for the work; MHN contributed to the data analysis and interpretation of data for the work; HHD, SHP, MTTN, HTTT, and SND contributed to the interpretation of data for the work and revised the draft critically for important intellectual content; NHN and CQL contributed to the design of the work, interpretation of data for the work, wrote the first draft of the work, and revised the draft critically for important intellectual content. All authors reviewed and edited the work and approved its final version. DAD and CQL are responsible, as guarantors, for the overall content.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability

All data generated or analysed during this study are included in this published article [and its supplementary information files].

Declarations

Ethics approval and consent to participate

The Scientific and Ethics Committees of Vietnam National Children’s Hospital approved this study (Approval Number: VN01037/IRB00011976/FWA00028418) on August 31, 2023. The study was conducted according to the principles outlined in the Declaration of Helsinki. The Committees waived the need for informed consent for this retrospective observational study. Public notification of the study was provided through published postings, following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines, which include an explanation and elaboration of the STROBE Statement and a checklist of items that should be included in reports of cohort studies. The authors responsible for data analysis stored the dataset in password-protected systems and only presented anonymized data.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.