Chapter 15
Newborns, infants and children
Francesco Raimondi1, Fiorella Migliaro1, Antonietta Giannattasio1, Letizia Capasso1, Claudia Lucia Piccolo2, Margherita Trinci3, Vittorio Miele4 and Stefania Ianniello3
1Division of Neonatology, Dept of Translational Medical Sciences, Università “Federico II”, Naples, Italy. 2Dept of Medicine and Health Sciences, University of Molise, Campobasso, Italy. 3Emergency Radiology Unit, Dept of Diagnostic Imaging, S.Camillo-Forlanini Hospital, Rome, Italy. 4Dept of Emergency Radiology, Azienda Ospedaliero Universitaria Careggi, Florence, Italy.
Correspondence: Francesco Raimondi, Dept of Translational Medical Sciences, Università “Federico II” di Napoli, via Sergio Pansini 5, I-80131 Naples, Italy. E-mail: raimondi@unina.it
Chest US is a rapidly growing field in the imaging of the developing human, and is based on the same images and artefacts employed by adult emergency physicians. Solid evidence has recently been published about the US features of respiratory distress syndrome, transient tachypnea of the neonate and meconium aspiration syndrome. Tension pneumothorax can be accurately diagnosed with US and rapidly treated. Research into endotracheal tube positioning and chronic lung disease is ongoing. Semiquantification of the US signal has generated an intriguing comparison with the conventional radiogram with regard to the role of the first-line technique in several respiratory diagnoses. In the infant and the child, LUS can be applied in the work up of lung consolidations and several groups have investigated its accuracy in diagnosing pneumonia and bronchiolitis. Further paediatric applications include the study of chest wall masses and pleural effusions; focused assessment with sonography for trauma protocol remains an important area of current investigation.
Cite as: Raimondi F, Migliaro F, Giannattasio A, et al. Newborns, infants and children. In: Laursen CB, Rahman NM, Volpicelli G, eds. Thoracic Ultrasound (ERS Monograph). Sheffield, European Respiratory Society, 2018; pp. 206–225 [https://doi.org/10.1183/2312508X.10007217].
Current LUS applications in neonatology
Adult emergency medicine physicians first applied LUS to the clinical field. As LUS semiology is almost the same in patients of any age, neonatologists and paediatricians have quickly learned how to apply it and are now using it with remarkable success. It is a novel point-of-care, easy-to-learn imaging technique with the significant advantage of sparing the developing human from exposure to ionising radiation. ESCOURROU and DE LUCA [1] provided a convincing demonstration of this in an elegant report. By instituting a LUS protocol in their French neonatal intensive care unit (NICU), the number of CXR per baby in 2014 was roughly half that in 2012 and the mean radiation dose per baby decreased from 180 to 60 EGy.
This chapter provides a concise review of current applications in neonatal and paediatric respiratory medicine, with emphasis on the strength of the evidence and areas of future investigation. The fact that around half of the studies discussed were published in the last 4–5 years shows how rapidly interest is growing around the diagnostic potential of LUS in the newborn infant.
At birth and in the nursery
The normal lung transitions at birth from fluid to air content. A recent study by BLANK et al. [2] used LUS to document the rapid liquid clearance from the lungs in a cohort of mostly term infants. All infants had an established pleural line at the first examination (median 2 (range 1–4) min). Only 14% of infants had substantial liquid retention 10 min after birth. 49%, 78% and 100% of infants had completed airway liquid clearance at 2, 4 and 24 h, respectively. In a smaller, single-centre study, MARTELIUS et al. [3] showed a significant difference in postnatal liquid absorption attributable to the mode of delivery; in particular, there was a significantly higher fluid retention at 3 h in term infants born by caesarean section compared with vaginal delivery. The potential of LUS as a noninvasive technique to anticipate the need of respiratory support after birth was assessed by RAIMONDI et al. [4] in a cohort of 154 late preterm and term babies who underwent sequential LUS. Images were classified as type 1 (white lung image), type 2 (prevalence of comet-tail artefacts or B-lines) and type 3 (prevalence of horizontal or A lines) profiles. Shortly after birth, 14 neonates were assigned to the type 1, 46 to the type 2 and 94 to the type 3 profiles. Within 36 h there was a gradual shift from types 1 and 2 to type 3. LUS had a high accuracy in predicting NICU admission, with a sensitivity of 77.7% and a specificity of 100%. This information may be particularly valuable to healthcare workers at level 1 birth centres with difficult access to intensive care.
In the NICU
Transient tachypnoea of the newborn
In a seminal paper, COPETTI and CATTAROSSI [5] demonstrated that neonates with transient tachypnoea of the newborn (TTN) had a white lung image and a regular pleural image at LUS. Moreover, a sharp increase in echogenicity was described in the lower lung fields of all infants with TTN. The authors named the latter US feature the “double lung” sign. In subsequent publications, VERGINE et al. [6] and LIU et al. [7] showed that the double lung sign is not pathognomonic of TTN, and that its US appearance may be quite variable. Current discrepancies between the available single-centre investigations may be due to the fact that their original designs lack any correlation between US and clinical features. A multicentre study conducted with stringent methodology might help solve the issue [5–7].
Respiratory distress syndrome
A typical LUS scan in a baby with respiratory distress syndrome (RDS) shows a white lung image with no spared areas and an irregular pleural surface. Small subpleural consolidations can also be detected [8]. These results have been confirmed by several authors with minor modifications [9, 10].
LUS can also be useful in the diagnosis of RDS complications. A prospective study conducted by LOVRENSKI [11] included 120 preterm infants with clinical and radiographic signs of RDS. Within the first 24 h of life, LUS was performed using both a transthoracic and a transabdominal approach; sequential scans were then performed. In the 47 detected pulmonary complications of RDS (including haemorrhage, pneumothorax (PTX), pneumonia, atelectasis, bronchopulmonary dysplasia), comparisons between LUS and CXR were made. In 45 instances, the same complication was diagnosed with LUS as with CXR, indicating that this method has a high reliability in premature infants with RDS.
CATTAROSSI et al. [12] made the interesting observation that, unlike conventional CXR, the LUS scan is not modified shortly after surfactant administration. However, in a subsequent small study of 12 preterm infants, LOVRENSKI et al. [13] proposed a grading system for RDS evolution after surfactant replacement.
Investigating LUS as a monitoring tool for RDS course is an ambitious yet achievable goal and there is a considerable amount of work still to be done. For instance, no solid longitudinal data is currently available showing the long-term evolution of neonatal RDS, and we do not know whether stratification by gestational age of LUS scan might yield useful information.
Meconium aspiration syndrome
PIASTRA et al. [14] provided the original description of six neonates with meconium aspiration syndrome (MAS). All patients showed interstitial involvement, either coalescent or sparse, consolidations, atelectasis and bronchograms. No pattern was observed for the distribution of signs in lung areas, although the signs varied with time, probably due to the changing localisation of meconium in the lungs. There was a good correlation between LUS and CXR findings. Similar findings were also reported by LIU et al. [15] in a large series of 117 neonates with MAS.
Tension pneumothorax
Although individual case reports and single-centre series were already available [16, 17], in 2016, RAIMONDI et al. [18] led the first international, prospective study on tension PTX in the neonate. Taking the conventional chest radiogram as a reference standard in a series of 42 infants presenting with significant sudden desaturation and bradycardia, RAIMONDI et al. [18] showed that LUS has an absolute diagnostic accuracy for tension PTX, outperforming the clinical diagnosis. After sudden decompensation, a LUS scan was performed in a mean±SD time of 5.3±5.6 min, in comparison with the 19±11.7 min required for a chest radiography. Emergency drainage was performed after US but before radiography in nine cases.
Neonatal pneumonia
In infectious pneumonia, LUS findings include large areas of lung consolidation with irregular margins and air bronchograms, pleural line abnormalities, and interstitial syndrome. These features assured absolute diagnostic accuracy for LUS in the study by LIU et al. [19]. Investigating 40 neonates with symptomatic pneumonia and 40 healthy controls, the authors failed to provide precise masking details. A different approach has been adopted in studies investigating the aetiology of neonatal respiratory distress with LUS. This was the case in a study of 63 symptomatic infants performed by RACHURI et al. [10], which found six pneumonia cases. In a large cohort of 3405 Chinese neonates (2658 of whom were symptomatic) screened with LUS, 1016 received an US diagnosis of pneumonia [20]. In addition, 1692 cases were examined using LUS and chest radiography for the first time within 48 h of admission. Among 81 cases who underwent CXR and who did not receive a diagnosis of lung disease, 32 had clinical and US evidence of pneumonia. The fact that LUS outperforms the conventional radiogram in diagnosing pneumonia is well established in adult medicine where chest tomography has been used as a reference standard [21].
Chronic lung disease
Predicting which infant with respiratory distress will eventually develop chronic lung disease (CLD) is an ambitious goal of a few LUS studies. In 1996, AVNI et al. [22] investigated a cohort of 105 premature babies with weekly transabdominal LUS scans. In all patients with RDS, diffuse retrodiaphragmatic hyperechogenicity was observed. Hyperechogenicity resolved completely in patients with an uncomplicated clinical evolution. In contrast, in patients with CLD (defined as oxygen dependency at day 28 of life), hyperechogenicity only partially resolved, resulting in a less severe and less extensive picture. Day 18 was the earliest time at which the persistence of abnormal retrodiaphragmatic hyperechogenicity was observed in 100% of the patients presenting CLD at day 28. At that time, 95.2% of the patients without abnormal hyperechogenicity showed uncomplicated evolution and no CLD. Similar results were then published by PIEPER et al. [23] in 2004, in a study of 36 infants examined 7 years earlier. Both studies lack significant clinical data and are outdated in terms of choice of LUS technique and disease definition. Convincing, properly powered and well conducted studies are needed to investigate this fundamental issue.
Functional LUS
As previously mentioned, LUS has been used to provide detailed nosographic description. However, a functional approach may be adopted by linking a specific LUS profile to a clinical decision.
RAIMONDI et al. [24] investigated a cohort of 54 preterm neonates admitted to the NICU with moderate respiratory distress who were stabilised on nasal continuous positive airway pressure for 120 min. LUS was performed and blind classification to types 1 (white lung), 2 (prevalence of B-lines) or 3 (prevalence of A-lines) was carried out. An experienced radiologist blinded to the infant’s clinical condition examined and graded the chest radiograph. The study’s primary outcome was the accuracy of bilateral type 1 in predicting intubation within 24 h of scanning. A secondary outcome was the performance of the highest radiographic grade within the same time interval. Type 1 lung profile showed a high diagnostic accuracy with a sensitivity of 88.9% and a specificity of 100%, while CXR had a sensitivity of 38.9% and a specificity of 77.8%
The advantages of LUS in predicting the failure of noninvasive respiratory support were later confirmed by two other clinical research teams.
BRAT et al. [25] correlated a LUS score with several oxygenation parameters (transcutaneous partial pressure of oxygen to fraction of inspired oxygen ratio, alveolar–arterial gradient, oxygenation index, arterial to alveolar ratio) and with the administration of exogenous porcine surfactant. Among the 130 neonates included in this study, the LUS score was significantly correlated with all indices of oxygenation, independent of gestational age. The LUS score was a better predictor of the need for surfactant in preterm babies with a gestational age of <34 weeks (area under the curve (AUC) 0.93; 95% CI 0.86–0.99; p<0.001) than in term and late preterm neonates with a gestational age of ≥34 weeks (AUC 0.71; 95% CI 0.54–0.90; p=0.02); the AUC for these two gestational age subgroups are significantly different (p=0.02). In babies with a gestational age of <34 weeks, a LUS score cut-off of 4 predicted surfactant administration with 100% sensitivity and 61% specificity, yielding a post-test probability of 72%.
Similarly, RODRÍGUEZ-FANJUL et al. [26] investigated a cohort of neonates who were >32 weeks of age. Findings from the LUS helped the ultrasonographer classify the infants into two groups according to the potential risk of a bad respiratory outcome: low risk (normal or TTN) or high risk (RDS, MAS, PTX or pneumonia). A second investigator made the same classification after reading the CXR. 105 neonates (64.8% in the low-risk group and 35.2% in the high-risk group) were recruited in total; of those, 20% needed intubation, and this was more frequent in the high-risk group (relative risk 17.5; 95% CI 4.3–70.9; p<0.01). LUS and CXR showed a high index of agreement in predicting the risk of respiratory failure.
The research from these three groups of investigators demonstrates LUS to be a rapid, repeatable, radiation-free tool that supports the neonatologist in the difficult decision of whether to intubate a patient.
Endotracheal tube positioning
US has been shown to be effective in the verification of endotracheal tube (ETT) position in adults and older children; however, it use has been less frequently studied in the neonates and infants. A recent study by SHETH et al. [27] reviewed nine relevant papers on the topic. US interpretation of the ETT position correlated with the radiography position in 73–100% of cases when the ETT tip was visible. There were variations in technique, sonographer and sonographer training between studies. In addition, one needs to bear in mind the limited information on detecting endotracheal versus oesophageal positioning with LUS in this population. Unless convincing evidence is published soon, it is unlikely that US will supplant capnography for this purpose any time in the near future.
Miscellaneous
Preliminary reports on possible LUS applications range from rare conditions, such as localised interstitial emphysema [28], to common diseases, such as bronchiolitis [29, 30] and atelectasis [31, 32].
Some adult cardiologists and emergency medicine physicians have integrated LUS data into their cardiac and haemodynamic US evaluation [33]. Neonatologists are following in this direction. RODRÍGUEZ-FANJUL et al. [34] investigated the accuracy of LUS in assessing pulmonary overflow in 51 infants with congenital heart disease. Neonates were classified into two groups depending on their predisposition to developing pulmonary overflow as evaluated by the abundance of B-lines. The results were compared with physical examination, CXR and echocardiography. No differences in the abundance of B-lines were present between the groups; however, those with a cardiac defect with a tendency to develop pulmonary overflow had a higher B-line score after 72 h (p<0.05), and showed a good correlation with echocardiography findings and a better sensibility than physical examination and CXR.
Research perspectives
Standardisation and quality of the scientific evidence
The previous section has summarised the rapidly cumulating evidence in favour of LUS as a useful extension of physical examination in the critical neonate. Now that groundbreaking evidence has been published, the search for a common language is mandatory. Large clinical studies from China have shown that no US sign is typical of a neonatal respiratory disease [7, 20]. It is also difficult to trace a combination of US features that univocally depict a respiratory condition across all published studies. Part of the problem stems from the different tools (e.g. the types of probes and machines) and examination techniques (transthoracic versus transabdominal approach; number and type of US views) in use. Most of the available evidence comes from single-centre studies, often with poor consideration for masking and sample power procedures. There is great need for sufficient methodological stringency to overcome the still widespread clinical and forensic concerns about LUS.
To provide a better understanding about the influence of these variables and to promote collaborative studies, we have founded NeoLUS (Neonatal LUS for the neonate and the small infant; https://www.facebook.com/groups/1493243264284547/), a dedicated research network currently counting 140 members around the world.
Quantification of the US signal: fact or fiction?
A semiquantitative approach has long been in use in neonatology, correlating the neurological outcome of a preterm baby with the extension of his/her intraventricular haemorrhage assessed by a widespread head US scoring system. A similar strategy has already been adopted by adult lung ultrasonographers, correlating a LUS score with the patient’s clinical conditions. Close correlation with clinical data is what is often lacking in the available neonatal LUS literature [35]. The aforementioned papers on functional LUS are a remarkable exception to this rule. Standardising a semiquantitative approach to LUS may lead to a new way to manage everyday NICU operations, such as alveolar recruiting, individualised surfactant administration, follow-up of respiratory failure and early prediction of bronchopulmonary dysplasia.
In the near future, both standardisation and accurate quantification may become possible with the help of computer-assisted image analysis techniques supporting LUS. This has already been attempted in adults using the grayscale analysis techniques [36].
In a study of 48 mechanically ventilated cardiac surgery patients, CORRADI et al. [36] compared the performance of a semiquantitative LUS visual score (visual LUS (V-LUS)) based on B-lines versus the computer-aided quantitative LUS (Q-LUS) analysis in assessing the degree of pulmonary oedema [36]. Both V-LUS and Q-LUS were acceptable indicators of pulmonary oedema in mechanically ventilated patients. However, at high positive end-expiratory pressure (PEEP), only Q-LUS provided data that was significantly correlated with postcapillary wedge pressure and extravascular lung water. Computer-aided Q-LUS had the advantage of being not only independent of operator perception but also of PEEP.
A different and more refined strategy of computer-aided analysis has recently been proposed by VEERAMANI and MUTHUSAMY [37] using machine learning image classification. This sophisticated technology has already been shown to be very effective when analysing foetal lung to predict neonatal respiratory morbidity [38].
Conclusion
LUS is a novel, radiation-free, point-of-care imaging technique that is under intensive investigation in all fields of neonatal respiratory medicine. Current evidence is solid and the potential offered by computer-assisted image analysis is very intriguing. Further standardised, reproducible results will strongly encourage neonatologists to adopt LUS as an extension of their clinical examination in their daily practice.
Current LUS applications in infants and children
For a long time, US has been performed solely to recognise pleural effusion in the thorax. This is because the physical limitations of the normal aerated lung and thoracic cage strongly affected the application of US in the diagnosis of thoracic disease [39]. In recent years, lung evaluation has become feasible owing to advances in technology; the use of multifrequency probes, together with the application of tissue-harmonic imaging, lead to improved spatial resolution and tissue penetration, as well as a specific focus on decreasing ionising radiation [40–42]. Children are at least four times more sensitive to ionising radiation than adults because of a longer life expectancy and a faster cell rate division, thus increasing the risk of the development of radiation-induced cancer. LUS, which lacks ionising radiation, plays an essential role in the implementation of the ALARA (as low as reasonably achievable) principle, which aims to reduce exposure to ionising radiation using the imaging technique with little to no radiation [43–45].
In infants and children, TUS becomes useful in a variety of settings, especially in situations of clinical and radiographic concern, as well as superficial lumps and bumps, the characterisation of structures such as the thymus, and diaphragmatic motion, because of the superior acoustic window and the relative paucity of adipose tissue compared with adults [46, 47]. Its application in the paediatric lung assists children with a small thoracic width and lung mass, allowing easier detection of lung anomalies with LUS because of a more rapid involvement of the pleura [48].
Technique
In the typical clinical setting, LUS is usually preceded by a chest radiograph that may raise several clinical questions relating to lung opacification, driving the examination to the appropriate anatomical region.
The transducer is chosen according to patient size and the depth of penetration; infants are usually studied with high-frequency probes, whereas older children are studied with lower-frequency probes. The classic acoustic windows are the supraclavicular, suprasternal, parasternal, intercostal, subxiphoid and transdiaphragmatic windows in longitudinal and transversal sections. The young patient is usually studied in the supine, lateral decubitus or sitting position, depending on their clinical condition and the clinical question [49].
The anterior chest wall and mediastinum are well depicted through sternal and costochondral cartilages because of the low ossification of the thoracic cage and the relatively large thymus [41, 42].
As regards LUS physics and semeiotics, we refer to what is explained elsewhere in this text.
Lung evaluation
White hemithorax on CXR
White hemithorax (a homogeneous radiopaque hemithorax at chest radiograph) are commonly found in children and are caused by a variety of disorders, such as lung agenesis, severe hypoplasia and lung collapse [50, 51]. The latter is responsible for severe respiratory symptoms. In infants and children, the differential diagnosis of lung collapse includes bronchial mucous plugs, foreign body aspiration, endobronchial tumours and bronchial compression. In children with endobronchial foreign bodies, tumours or bronchial mucous plugs, it is common to observe associated findings, such as pneumomediastinum, subcutaneous emphysema and an interrupted bronchus [52, 53]. When bronchial aspiration is suspected, immediate bronchoscopy is required. All these conditions are associated with ipsilateral mediastinal displacement.
The white hemithorax associated with contralateral mediastinal displacement could be caused by congenital malformations, pneumonia with pleural effusions (figure 1), cancer, isolated pleural effusions or vascular lesions [54–56].