Postnatal chest X‐ray in children with asymptomatic congenital lung malformations

Abstract

Objective

The clinical implications of a postnatal chest X‐ray (CXR) in asymptomatic children with a prenatally diagnosed congenital lung malformation (CLM) are uncertain. We assessed the justification for the postnatal use of CXR in these children.

Methods

We included patients with CLM confirmed through chest computed tomography angiography or histopathological analysis who were asymptomatic at birth, underwent routine postnatal CXR, and participated in our standard of care prospective structured longitudinal follow‐up program. Children with major associated morbidities were excluded. Primary outcomes were the positive and negative predictive values (PPV and NPV) of CXR findings for symptom development at 4 weeks and 6 months of age. Secondarily, we sought to establish whether CXR findings were associated with undergoing additional diagnostics during the initial observational hospital stay or prolonged postnatal hospital admission.

Results

Among 121 included patients, CXR showed no abnormalities in 35 (29%), nonspecific abnormalities in 23 (19%), and probable CLM in 63 (52%). The PPV of CXR in relation to symptom development at 4 weeks and 6 months was 0.05 and 0.25, respectively. Corresponding NPVs were 0.96 and 0.91. An association was identified between CXR findings and undergoing further diagnostics during the initial observational hospital stay (p = .047). Additional diagnostic findings did not influence clinical management. CXR findings were not associated with prolonged initial hospital stay (p = .40).

Conclusion

The routine practice of postnatal CXR in asymptomatic patients with prenatally diagnosed CLM can be omitted, as CXR findings do not influence subsequent clinical management.

1. INTRODUCTION

Congenital lung malformations (CLM) encompass various anatomical abnormalities affecting the respiratory tract. These include congenital pulmonary airway malformation (CPAM), bronchopulmonary sequestration (BPS), bronchogenic cyst (BC), congenital lobar overinflation (CLO), bronchial atresia, and hybrid lesions, the latter usually referring to combinations of CPAM and BPS. ¹ Children with CLM may experience symptoms and complications such as respiratory distress, respiratory tract infections, pneumothorax, wheezing, and chronic cough. ² , ³ , ⁴ While most babies with CLM are asymptomatic at birth and remain symptom‐free throughout childhood, ⁵ literature suggests that 3%–64% of initially asymptomatic babies with CLM may develop symptoms, but precise numerical evidence is lacking. ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ The reported incidence of CLM is 4/10,000 births and has risen over the past decades. This increase can be attributed in part to the introduction of routine prenatal ultrasound screening—implemented in the Netherlands in 2007—and advancements in ultrasound resolution quality. ¹² , ¹³

The standard of care for children with CLM varies worldwide and is primarily influenced by the preferences of local clinicians. ¹⁴ , ¹⁵ , ¹⁶ As per international consensus, surgery is the standard treatment for symptomatic patients with CLM. However, there is ongoing controversy about the optimal management of asymptomatic lesions—including postnatal imaging—as evidence is lacking. ¹⁷ , ¹⁸ , ¹⁹ Presently, chest computed tomography angiography (CTA) is considered the gold standard for confirming the prenatal diagnosis of CLM, with MR imaging also gaining relevance. ²⁰ A chest X‐ray (CXR) is often the first postnatal imaging modality during the postnatal hospital stay of children with CLM. ¹⁴ , ²¹ , ²² Performing postnatal CXR is advocated in several surgical textbooks as it can indicate the extent of the CLM. ²³ , ²⁴ In symptomatic patients with CLM, CXR is performed to detect abnormalities such as mediastinal shift or pneumothorax. Some centers still perform postnatal CXR for all patients with prenatally diagnosed CLM to assess whether a lesion can be detected. ²⁵ However, CXR is not the optimal imaging modality for this purpose, as evidenced by its sensitivity of 61% for detecting CPAM, meaning that a normal CXR does not rule out CLM. ²⁶ Furthermore, postnatal CXR does not predict the need for respiratory support during the postnatal hospital stay or the development of symptoms within the first 30 days following discharge. Consequently, its clinical relevance is questionable. ²² To the best of our knowledge, no studies have investigated the short and longer‐term predictive value of a postnatal CXR in relation to early‐life symptom development in asymptomatic children with prenatally diagnosed CLM.

We hypothesized that performing a postnatal CXR is superfluous in asymptomatic patients with prenatally diagnosed CLM. The primary aim of this study was to determine if a postnatal CXR in children with an asymptomatic CLM has clinical value. We sought to achieve this by assessing the predictive value of CXR for early‐life symptom development and exploring whether CXR findings influenced postnatal clinical management.

2. MATERIALS AND METHODS

2.1. Population and data collection

Children with CLM born in the Erasmus MC Sophia Children's Hospital between January 1999 and August 2022 were eligible for inclusion if they were asymptomatic at birth, defined as the absence of respiratory or other symptoms after the first 24 h of life. Need for short‐term supportive oxygen—defined as oxygen supplied via nasal cannula or mask—provided and stopped within the first 24 h of life in the absence of other symptoms was considered to reflect an asymptomatic condition, as this could be related to the transition from the intra‐ to extrauterine environment. CXR performed within the first 24 h of the postnatal observational hospital stay and chest CTA performed within 12 months of age were other inclusion criteria.

Children with major associated morbidities, defined as any medical condition requiring surgical intervention, affecting normal respiratory function, or leading to failure to thrive in early life, were excluded. Since 1999, all children with CLM at our hospital have been prospectively enrolled in a structured longitudinal follow‐up program integrated into our standard of care management. This program aims to monitor these children's overall health and neurodevelopment through scheduled outpatient visits and includes several assessments of lung function and exercise capacity. ²⁷ , ²⁸ Data for this study was retrieved from electronic patient files and included patient characteristics, CXR findings, CLM type, postnatal management, and symptom development at 4 weeks and 6 months. Additionally, operative characteristics of children who underwent CLM‐related surgery were retrieved. The Medical Ethical Review Board of the Erasmus University Medical Center Rotterdam approved this retrospective study design and waived the need for informed consent (MEC‐2022‐0721). All parents and children were informed that data collected in the context of the longitudinal follow‐up program could be used for research purposes.

2.2. Patient characteristics

We collected the following perinatal patient characteristics: sex, gestational age at birth, preterm births (<37 weeks), birth weight, being small for gestational age (birth weight <P10 for gestation according to Perined ²⁹ ), delivery method, and the department where the child was initially admitted based on institutional protocol or postnatal clinical status (ICU or maternity ward). Postnatal CXR findings were categorized into three groups: normal CXR, nonspecific abnormalities (e.g., consolidation, hyperlucency, atelectasis), or probable CLM. The CXR images were labeled as “probable CLM” when the radiologist specifically reported that the characteristics of the abnormalities found (including localization, shape, and density) likely indicated the presence of a CLM. Conversely, CXR images were labeled as “nonspecific abnormalities” when aberrant findings were too subtle for the radiologist to suggest a probable relation with a CLM. We presented CLM type based on histopathological diagnosis when available, and on the CTA performed in the first year of life when no histopathological analysis was available. The CLM diagnosis was categorized into seven groups: CPAM, BPS, BC, CLO, bronchial atresia, hybrid lesions, and other/atypical findings.

CLM‐related symptoms, assessed at 4 weeks and 6 months of age, were defined as parent‐reported feeding difficulties, respiratory problems, failure to thrive, or doctor‐diagnosed respiratory tract infections.

2.3. Primary outcomes

The primary outcomes were the positive predictive value (PPV) and negative predictive value (NPV) of postnatal CXR findings in relation to the development of symptoms at 4 weeks and 6 months of age. To calculate these outcomes, we categorized the CXR findings into “probable CLM” and “no clear CLM visible” as binary classification is required for PPV and NPV calculation. The latter category encompasses normal CXR and nonspecific abnormalities.

2.4. Secondary outcomes

Secondary outcomes included possible associations between the CXR findings and the performance of further diagnostic assessments during the initial observational hospital stay, as well as the length of the initial observational hospital stay. Additionally, we provided supplementary clinical information for patients whose CXR findings prompted further diagnostics or prolonged the initial hospital stay.

2.5. Statistical analysis

Results were presented as mean ± standard deviation (SD) for normally distributed variables and median (interquartile range) for continuous variables that were not normally distributed. The normal distribution of continuous variables was assessed using the Shapiro–Wilk test. The Chi‐square test was used for the analysis of categorical data, for example, for comparing how often additional diagnostics were administered in children with a normal postnatal CXR, nonspecific abnormalities on CXR, and probable CLM on CXR. The Kruskal–Wallis test was used for the comparison of continuous data that were not normally distributed, for example, to assess if there were differences in hospital stay duration between children with a normal postnatal CXR, nonspecific abnormalities on CXR, and probable CLM on CXR.

The PPV was calculated as $\frac{\mathrm{CLM}\mathrm{on}\mathrm{CXR}\mathrm{and}\mathrm{symptom}\mathrm{development}}{\mathrm{CLM}\mathrm{on}\mathrm{CXR}\mathrm{and}\mathrm{symptom}\mathrm{development}+\mathrm{CLM}\mathrm{on}\mathrm{CXR}\mathrm{and}\mathrm{no}\mathrm{symptom}\mathrm{development}}$, and the NPV as $\frac{\mathrm{no}\mathrm{CLM}\mathrm{on}\mathrm{CXR}\mathrm{and}\mathrm{no}\mathrm{symptom}\mathrm{development}}{\mathrm{no}\mathrm{CLM}\mathrm{on}\mathrm{CXR}\mathrm{and}\mathrm{no}\mathrm{symptom}\mathrm{development}+\mathrm{no}\mathrm{CLM}\mathrm{on}\mathrm{CXR}\mathrm{and}\mathrm{symptom}\mathrm{development}}$. We assumed a significance level of 5% in all statistical tests. The statistical package IBM SPSS (version 28.01) was used for data analysis.

3. RESULTS

In our database, we identified data from 266 patients with CLM at the time of this study, of whom 127 candidates fulfilled the inclusion criteria. The reasons for exclusion are explained in Figure 1. Follow‐up data were missing for six children, resulting in a final study population of 121 patients.

Figure 1.

Flowchart of patient inclusion. CLM, congenital lung malformation, CTA, computed tomography angiography. ᵃMajor associated morbidities: cardiac (N = 3), chromosomal (N = 3), nonchromosomal syndromes (N = 2), anorectal malformation (N = 1), esophageal atresia (N = 1), congenital diaphragmatic hernia (N = 1), hepatic arteriovenous malformation (N = 1), filamin A deficiency (N = 1).

3.1. Patient characteristics

In our cohort, proportions of males and females were balanced, and gestational age, as well as birth weight, were within normal range, with most deliveries occurring vaginally (Table 1). Among all patients, CPAM was the most common CLM type (N = 48; 40%), followed by hybrid lesions (N = 26; 21%) and BPS (N = 20; 17%). Fifteen patients (12%) underwent surgery at any point during follow‐up, primarily due to respiratory tract infections (N = 4; 27%) or cardiac overload (N = 4; 27%) (Table 2). Lobectomy was the most frequently performed procedure (N = 11; 73%) (Table 2).

Table 1.

Demographic and general patient characteristics.

	Study population (N = 121)
Sex
Male	60 (50%)
Female	61 (50%)
Gestational age (weeks)	39.1 [38.6; 40.3]
Preterm births	9 (7%)
Birth weight (grams)	3345 ± 478
Small for gestational agea	11 (9%)
Delivery method
Vaginal	97 (80%)
C‐section	24 (20%)
Initial admission
ICU	74 (62%)
Maternity ward	45 (38%)
Missing	2
CXR findings
Normal	35 (29%)
Nonspecific abnormalities	23 (19%)
Probable CLM	63 (52%)
CLM typeb
CPAM	48 (40%)
BPS	20 (17%)
CLO	7 (6%)
BC	3 (2%)
Bronchial atresia	10 (8%)
Hybrid	26 (21%)
Other/unclearc	7 (6%)

Note: Data is presented as: N (%), median [interquartile range], or mean ± standard deviation.

Abbreviations: BC, bronchogenic cyst; BPS, bronchopulmonary sequestration; CLM, congenital lung malformation; CLO, congenital lobar overinflation; CPAM, congenital pulmonary airway malformation; CTA, computed tomography angiography; CXR, chest X‐ray.

^a Defined as birth weight <P10 for gestation according to Perined (Hoftiezer L. Perined Hoftiezer geboortegewichtcurven. https://www.perined.nl/geboortegewichtcurven: Perined; 2014).

^b Based on the CTA performed within 12 months, or on the histopathological analysis when available (N = 15).

^c Other: duplication cyst (N = 2), lung agenesis (N = 1). Unclear diagnosis: N = 4.

Table 2.

Surgical characteristics.

	Patients who underwent surgery (N = 15)
Timing of surgery
0–6 months	3 (20%)
6–12 months	4 (27%)
1–2 years	4 (27%)
2–5 years	3 (20%)
5–8 years	1 (7%)
Surgical indication
LRTIa	4 (27%)
Cardiac overload	4 (27%)
Failure to thrive	1 (7%)
Mass effect	2 (13%)
Respiratory distress	1 (7%)
Unclear/missing	3 (20%)
Surgical procedure
Lobectomy	11 (73%)
Sequestrectomy of BPSb	2 (13%)
Cystectomy	2 (13%)
Histopathological diagnosis
CPAMc	6 (40%)
BPS	6 (40%)
BCd	1 (7%)
Hybrid (CPAM + BPS)	1 (7%)
Duplication cyst	1 (7%)
Diagnosis CTAe = histopathological diagnosis?
Yes	12 (80%)
Nof	3 (20%)

Note: Data is presented as: N (%).

^a LRTI = lower respiratory tract infections.

^b BPS = bronchopulmonary sequestration.

^c CPAM = congenital pulmonary airway malformation.

^d BC = bronchogenic cyst.

^e CTA = computed tomography angiography.

^f One suspected CPAM turned out to be a BPS, one duplication cyst turned out to be a BC and vice versa.

3.2. Primary outcomes

In 35 children (29%), postnatal CXR was normal, while nonspecific abnormalities—further described in Supporting Information File S1—were reported in 23 children (19%). Probable signs of CLM were found in 63 children (52%). The PPV of CXR in relation to symptom development at 4 weeks and 6 months was 0.05 and 0.25, respectively (Table 3). Corresponding NPVs were 0.96 and 0.91, respectively (Table 3).

Table 3.

Chest X‐ray predictive value.

	No clear CLM on CXR	Probable CLM on CXR	PPV	NPV
Symptoms at 4 weeks	N  = 52	N  = 57	0.05	0.96
No	50	54
Yes	2	3
Missing	6	6
Symptoms at 6 months	N  = 56	N  = 59	0.25	0.91
No	51	44
Yes	5	15
Missing	2	4

Abbreviations: CLM, congenital lung malformation; CXR, chest X‐ray; NPV, negative predictive value; PPV, positive predictive value.

3.3. Secondary outcomes

CXR findings were associated with the performance of additional diagnostics during the initial postnatal hospital stay (p = .047), as presented in Table 4.

Table 4.

Chest X‐ray findings and initial hospital stay.

	Normal CXR	Nonspecific abnormalities	Probable CLM on CXR	p value
Study population (N = 121)	35 (29%)	23 (19%)	63 (52%)
Additional diagnostics during initial hospital staya	N = 35	N = 23	N = 63	p = .047
No	35 (100%)	20 (87%)	61 (97%)
Yes	0	3 (13%)	2 (3%)
Length of initial hospital stay (days)b	2 [1; 2]	2 [2; 3]	2 [1; 3]	p = .40

Abbreviations: CLM, congenital lung malformation; CXR, chest X‐ray.

^a Presented as N (%), analysis through Chi‐square test.

^b Presented as median [interquartile range], analysis through Kruskal–Wallis test.

Five patients underwent additional diagnostics during the initial hospitalization due to CXR findings, among whom three showed nonspecific abnormalities, and two had probable CLM based on the initial CXR. Three of these patients underwent repeat CXR during the initial admission, which provided no new insights. One patient with nonspecific abnormalities on the postnatal CXR underwent a chest CTA on the second day of life without an immediate clinical indication. One patient underwent both a repeated CXR and a chest CTA within the first 2 days of life due to the lack of clarity of the postnatal CXR images. In none of these cases did the additional radiographic studies influence clinical management. All five patients were discharged within 1 day after the additional assessments.

The five patients who underwent additional radiographic studies were initially admitted to the ICU. In four cases, this was due to the local protocol in place before 2019, which recommended a more lenient approach with observational ICU admission compared to the current protocol. The fifth patient was initially admitted to the ICU due to respiratory distress upon birth, which was briefly treated with positive pressure ventilation via a nasal mask.

In the total study population, no association was identified between CXR findings and the length of the initial hospital stay (p = .40) (Table 4). None of the children's initial observational hospital stays were extended because of CXR findings.

4. DISCUSSION

Despite the common practice of performing a CXR within the first postnatal day in asymptomatic patients with prenatally diagnosed CLM, the added clinical value of this approach is uncertain. We calculated the predictive values—both PPV and NPV—of CXR in relation to symptom development at 4 weeks and 6 months. The PPV of CXR for symptom development—which was our focus—was low, while the NPV was high. Abnormal CXR findings were associated with undergoing additional diagnostics during the initial hospital stay, but these additional diagnostics did not lead to changes in clinical management.

Previous research indicated that chest radiography neither predicts the need for respiratory support during the initial postnatal hospital stay nor symptom development in the neonatal period. ²² Consequently, routine postnatal CXR in patients with prenatally diagnosed CLM was deemed a “low‐value thoracic imaging” entity. ³⁰ Our study supports this conclusion and offers added value by presenting data from a larger group of patients enrolled in a structured follow‐up program, with a follow‐up duration extending beyond the initial month after discharge. Based on the existing literature and our findings, we advocate against the routine use of postnatal CXR and postnatal ICU admission in asymptomatic patients with prenatally diagnosed CLM.

Several potential limitations of this study are worth addressing. First, radiology reports were not consistently documented by a standard protocol, leading to variability in the amount of information they contained. However, each report contained sufficient information to categorize the images reliably into one of the three established categories (normal CXR, nonspecific abnormalities, probable CLM).

A second potential limitation was reporting bias. Parents aware of CLM being visible on the postnatal CXR might have been more vigilant and may have reported symptoms earlier compared to parents of children with normal CXR results. This potential bias might have influenced our results by increasing the probability of a significant difference in reported symptom rates between children with normal CXR findings and those with probable CLM. To minimize this bias, we took detailed medical histories during outpatient follow‐up visits and evaluated radiological assessments from other clinics.

Third, our standard care for patients with congenital anomalies has evolved significantly over the past decades. Under our previous protocol on CLM—which was in use until 2019—clinicians tended to admit children with prenatal suspicion of CLM to the ICU more readily than under our current protocol. As alluded to in the presentation of our secondary outcomes, these ICU admissions may have led to more accessible and potentially unnecessary additional diagnostics compared to when children were admitted to the maternity ward. This may have been a confounder partially explaining the reported association between CXR findings and additional diagnostics. The current protocol prescribes observational ICU admission for asymptomatic patients only if they have a CPAM volume ratio greater than 0.5, polyhydramnios or mediastinal shift on prenatal ultrasound, or other congenital conditions.

Last, we acknowledge that we did not assess the cost‐effectiveness of postnatal CXR. However, given its demonstrated subpar clinical predictive value, conducting a cost‐effectiveness analysis might not significantly impact clinical decision‐making or patient outcomes, thus yielding limited meaningful insights.

In conclusion, routine postnatal CXR in asymptomatic patients with prenatally diagnosed CLM offers minimal clinical value and can be refrained from, mainly due to its limited PPV in relation to early‐life symptom development. For cases of unexplained neonatal clinical deterioration, CXR remains useful in exploring underlying causes. In the absence of clinical manifestations, a chest CTA or chest MRI after the neonatal period should be the initial imaging tool employed. Future research efforts could focus on establishing predictive models for symptom development and the need for surgical intervention in children with CLM.

AUTHOR CONTRIBUTIONS

Louis W. J. Dossche, Casper M. Kersten, Hanneke IJsselstijn, Rene M. H. Wijnen, and J. Marco Schnater contributed to the conceptualization and design of the study. Louis W. J. Dossche, Charlotte S. van den Aardwegh, and Casper M. Kersten acquired and interpreted the data. Louis W. J. Dossche, Charlotte S. van den Aardwegh, and Joost van Rosmalen carried out the statistical analyzes. Louis W. J. Dossche, Charlotte S. van den Aardwegh, Casper M. Kersten, Hanneke IJsselstijn, and J. Marco Schnater drafted the work, which Joost van Rosmalen and Rene M. H. Wijnen reviewed. All authors have read and approved the final version of the manuscript for publication. Moreover, they agree to be accountable for all aspects of the work. They will ensure that questions about the accuracy and integrity of any part of the work are appropriately investigated and resolved.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHICS STATEMENT

The Medical Ethical Review Board of Erasmus University Medical Center approved this retrospective study design and waived the need for informed consent (MEC‐2022‐0721). All parents and children were informed that outcome data were used for research purposes.

Supporting information

Supporting information.

Dossche L. W. J, van den Aardwegh CS, Kersten CM, et al. Postnatal chest X‐ray in children with asymptomatic congenital lung malformations. Pediatr Pulmonol. 2024;59:3333‐3339. 10.1002/ppul.27201

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.