This was a retrospective observational cohort study. This retrospective study was part of the larger, on-going CODE-HEART (Congenital Diaphragmatic Hernia Evaluation of the Heart) study, which includes a prospective study of cardiac function in newborns with CDH compared with that of a control group. In this study the retrospective examination was performed on prior echocardiograms that had been saved digitally for clinical purposes. Medical data were retrieved from medical records at Karolinska University Hospital. The systolic and diastolic function were assessed with LV strain (LVS) and Left atrial reservoir strain (LASr) and were tested for correlation with LOS in the PICU. The outcome LOS was divided into two groups, more (>) or less (≤) than 8 days in the PICU, according to Larsen et al. (31).

Since 2018, the care of children born with CDH in Sweden has been centralized, and Karolinska University Hospital is the larger of two pediatric surgical centers in Sweden offering this specialized care. This study focused on children born with CDH between 2018 and 2020 who were treated at Karolinska University Hospital, Stockholm, Sweden. The families were asked to participate via telephone, email, or postal mail, and participants were included after oral and written consent. Exclusion criteria were not being born in Sweden, no available presurgical echocardiogram, poor-quality echocardiograms, echocardiograms performed after 72 h of age, and the presence of complex congenital heart disease or complex genetic or chromosomal anomalies. The hernia severity was assessed according to prenatal measurements of the lung size, observed-to-expected lung-to-head ratio (O/E LHR), and postnatal defect size, measured as the need for a patch (32).

All echocardiographic exams were reviewed retrospectively. The exams were performed between 2018 and 2020 by different pediatric cardiologists and stored digitally in SyngoDynamics version VA40D (Siemens Healthineers, Erlangen, Germany). Echocardiographic exams performed before the surgical intervention and before 72 h of age were reviewed by a senior consultant pediatric cardiologist (KÖT) who also performed all measurements and calculations and was blinded to the severity and LOS. Echocardiograms with poor quality and that were inadequate for strain measurements were excluded, according to the guidelines of the EACVI/ASE/Industry Task Force to standardize deformation imaging (33). Intraobserver variability tests were performed to reduce subjective data, ensure reproducibility, and validate the observations. Strain analyses were performed using the TomTec version TTA2 (TOMTEC Imaging Systems GmbH, Unterschleissheim, Germany). The echocardiographic parameters assessed were the left ventricular end-diastolic dimension (LVED) measured in the apical four-chamber view, z-score according to the Detroit data, LVS, LASr, direction of blood flow through the PFO, and direction of blood flow through the arterial duct (DA). For all echocardiographic parameters the mean of three measurements was used for analysis.

For strain analyses, 2D speckle tracking strain analyses were used. The endocardial contour was initially tracked automatically by the software defining a region of interest and adjusted manually, as needed. Both LA and LV strain were measured in one plane from 2D apical four-chamber view (34, 35). LVS was measured as peak left ventricular global longitudinal strain and presented in negative percentage the more negative values indicate better LV function. Left atrial reservoir strain (LASr) is a marker of LV filling and LV diastolic function. Other LA strain measurements as conductive and contractile strain have not shown as strong correlations to LV dysfunction as LASr and were accordingly not analyzed in this study (27, 36). When measuring LASr, the pulmonary veins and/or LA appendage orifice were excluded according to the guidelines of the EACVI/ASE/Industry Task force to standardize deformation imaging (33) (Figure 1). To our knowledge, no normal values for the LASr in the neonatal population have been established. Therefore, the chosen cutoff value for a normal LASr was set to >33% in our study, as suggested by Ficial et al. (34). The cutoff value for LVS was −16%, where more negative values than −16% were considered normal according to Patel et al. (11, 22).

Categorical data are expressed as frequencies or proportions. Continuous data are expressed as mean ± standard deviation (SD) or median (interquartile range). Non-normally distributed data were analyzed with the Mann-Whitney U-test to investigate differences between groups. To compare more than two groups, the Kruskal Wallis test was used. For nominal categorical data, the Chi-square or Fisher's exact test was used. Spearman's non-parametric correlation coefficient was performed to assess linear correlations between variables. A logistic regression model was used to calculate the odds ratio for sex, birth weight, LVED, patch/no patch, direction of blood flow through the DA, and LASr <33%. Uni- and multivariate models were used and compared using forward selection. Intraclass correlation coefficients were used for reliability testing using the two-way mixed effect model. A p-value <0.05 was considered statistically significant, and 95% confidence intervals were used. Statistical analyses were performed using SPSS software (version 27; IBM).

Ten randomly selected echocardiograms, blinded to the observer, were measured twice by the same observer. The interval between measurements was >8 weeks. An intraobserver variability test was then performed using SPSS.

This study was approved by the Swedish Ethical Board (Dnr 2020-03543), according to Swedish ethical regulations.

From 2018 to 2020, 55 Swedish newborns with CDH were treated at Karolinska University Hospital. Overall, 46 caregivers provided consent to participate, and 35 cases were included in the study. Reasons for exclusion are displayed in Figure 2.

Demographic characteristics of participants are shown in Table 1. Among the included cases, no children had complex congenital heart disease (CHD); six children were diagnosed with minor CHD, including atrial septal defects (ASD), ventricular septal defects (VSD), and left superior vena cava (SVC); and one child was born with omphalocele. No children were diagnosed with associated genetic disorders or chromosomal anomalies. Five children received VA (venoarterial) ECMO treatment, 3 echocardiograms were performed on ECMO. Four of the included children required dialysis during their PICU stay or ECMO treatment, all echocardiograms included were performed before dialysis was initiated the dialysis was a part of the intensive care treatment. None of the cases had a primary nephrological diagnosis. At the time of the echocardiograms during the first 72 h, five neonates had on-going Milrinone infusion and 16 had on- going support with inotropes. The inotropes used were noradrenalin (n = 10), adrenaline (n = 1, vasopressin (n = 3), dopamine (n = 3), and dobutamine (n = 1). A combination of inotrope drugs was used in three cases. In one case it was not possible to find out which inotrope drug that was used. A total of 34 neonates were intubated and on mechanical ventilation when the echocardiograms were performed. Ten neonates had on-going iNO treatment while the echocardiograms were performed.

The mean LASr for the whole group of CDH children was 27.4% (SD 10.5). The mean LVS for the whole group of CDH children was −16.9% (SD 4.2). The LASr was correlated with the LOS in the PICU (correlation coefficient: −0.378; p = 0.025) but the LVS and LOS were not correlated (correlation coefficient: 0.115; p = 0.510). When echocardiograms performed on ECMO support were excluded (n = 3) from the analysis, the correlation remained significant (correlation coefficient: −0.428; p = 0.015).

The LVED was correlated with the LOS (correlation coefficient: −0.546; p = 0.01), but the LASr and LVED were not correlated (correlation coefficient: 0.109; p = 0.532) (Figure 3).

The LASr measurements were divided into two groups: ≥33% (n = 8) or <33% (n = 27). In our study, measurements >33% were considered normal. These two groups were compared regarding the LOS in the PICU. The group with LASr <33% had a significantly longer PICU stay than the group with LASr ≥33% (20.8 vs. 8.6 days; p < 0.002) (Figure 4).

The LVS measurements were divided into two groups, > or ≤−16%, based on the cutoff value suggested by Patel et al. (11) The LOS in the PICU were not significantly different between the two groups (18.1 vs. 17.92 days; p = 0.961) (Figure 5).

Signs of pulmonary hypertension were assessed based on the direction of the blood flow through the DA. Of the cases, 29 (82%) had an open DA, 8 (28%) had a strict right-to-left shunt, and 21 (72%) had a bidirectional shunt. The groups showed no differences regarding the LOS (11.7 vs. 16.2 days; p = 0.200).

The use of inhaled nitric oxide (iNO) treatment (n = 16) was correlated with increased LOS in the PICU (11.0 vs. 26.4 days; p < 0.001).

Of the cases, 33 (88.5%) had PFO, 2 (5.7%) had a right-to-left shunt, 25 (71.4%) had a left-to-right shunt, and 4 (11.4%) had a bidirectional shunt. The groups had no significant differences when compared to the LOS in the PICU (p = 0.587) or LASr (p = 0.838).

Logistic regression analysis was used to analyze the association between the LOS in the PICU and LASr, after adjusting for LVED, sex, birth weight, and patch/no patch.

After adjusting for possible confounders, the association between the LASr and LOS in PICU remained. [OR = 0.9 (0.817–0.995); p = 0.039].

The intraobserver variability test showed an intraclass correlation coefficient for the LASr of 0.94, which was considered excellent. For the LVS, the correlation coefficient was 0.73, which was considered acceptable.

This study has several limitations, the study was retrospective; thus, the echocardiographic exams were not performed according to a pre-defined research protocol, and not all required echocardiographic assessments were eligible. The assessment of pulmonary hypertension was limited by the retrospective analysis, with few cases displaying the typical signs of pulmonary hypertension as tricuspid regurgitation, flattening of the septum or right to left shunt in arterial duct. In addition, the exams were performed by different pediatric cardiologists. The frame rate was not considered, which might have affected the strain measurements. The age of the children when the echocardiograms were performed also differed. Several exams were excluded, due to poor quality and being performed after 72 h of age, resulting in a small sample size, which may have impacted the results. Another limiting factor was the lack of a control group. Furthermore, the absence of normal values for LA strain in the neonatal period was a limitation. Comorbidities could also affect the LOS in the PICU. Minor congenital heart disease is unlikely to affect the LOS in the PICU and should therefore not have influenced the results. One child was diagnosed with a minor omphalocele, and we cannot rule out the possibility that the omphalocele impacted the LOS (42). In Sweden, the termination of pregnancy (TOP) rate for prenatal CDH is approximately 30% (2), which may have resulted in a selection bias for less severe CDH cases among the newborns with CDH.

This study also had strengths. The same experienced cardiologist reviewed all echocardiograms and performed all measurements in a standardized setting. The experienced cardiologist reviewing the examinations had no personal involvement in the initial care of the included study participants and was blinded to the outcome during analysis. The echocardiograms were stored on a server that was separate from the medical records, and echocardiograms and medical journals were reviewed separately. The intraobserver variability was excellent in the LASr correlation test and acceptable in the LVS correlation test. LASr is a user-friendly echocardiographic measurement that can be used in clinical practice, even when conditions to perform echocardiography are not optimal. Furthermore, the software used was not vendor dependent (TomTec), and all echocardiographic exams were performed by trained cardiologists in the same pediatric cardiology unit and stored on the same server system. All children were cared for and managed at Karolinska University hospital; therefore, few cases were missing medical data. The standardization of treatment at Karolinska University Hospital of children born with CDH before, during, and after birth enhances the robustness of the study findings.

In our study population, we identified a correlation between a reduced LA strain and a longer stay in the PICU, where the echocardiographic measurement of LASr reflects the diastolic cardiac function of the left ventricle. LASr strain is a feasible echocardiographic parameter and may be used as a marker of severity in children born with CDH.