METHODS

We conducted a prospective observational study based at a UK regional referral neonatal centre. Over eight months, the following information was collected whenever an echocardiogram was performed on the neonatal unit: basic patient details, indication for performing echocardiogram, whether the echocardiogram was performed by a neonatologist or paediatric cardiologist, findings on echocardiography, and any changes in clinical management following the scan.

Liverpool Women’s Hospital has a stand alone regional referral centre for neonatal intensive care without an on site paediatric cardiology service. A single paediatric cardiologist (DK) sees infants once a week for non-urgent review of suspected or proven cardiac disorders. At all other times, echocardiography is either performed by one of two neonatologists (CWY and NVS) or, in an urgent situation, by an on call paediatric cardiologist.

Indications for performing echocardiography were broadly categorised as follows: asymptomatic murmur; cardiorespiratory symptoms (cyanosis, shock, hypoxaemic respiratory failure, hypotension, and arrhythmia) with or without a murmur; antenatal diagnosis of cardiac defect; family history of cardiac defect; and non-cardiac congenital malformation. Two dimensional and colour Doppler echocardiographic examinations were performed using a Toshiba Corevision Pro system with a 7 MHz multifrequency imaging transducer combined with a 5 MHz Doppler transducer.

Infants found to have a completely normal echocardiogram, or a “trivial” structural abnormality such as patent foramen ovale, haemodynamically insignificant persistent ductus arteriosus (PDA), or mild “physiological” left pulmonary artery stenosis were classified as having no structural cardiac abnormality.¹¹ Persistent pulmonary hypertension of the newborn was diagnosed in the presence of normal cardiac anatomy with Doppler echocardiographic evidence of a right to left extrapulmonary shunt at atrial or ductal level.¹² Infants with poor left ventricular contractility (fractional shortening < 30%) or low left ventricular output (< 150 ml/kg/min) were classified as having left ventricular dysfunction.¹³

In infants who had paired scans (performed by both specialists at different times), the concordance between the two scans was evaluated. Two observers (DK and NVS) independently reviewed the echocardiography reports to grade the degree of concordance present. An arbitrary system was chosen: no concordance, moderate concordance, and complete concordance. In instances where no/moderate concordance was found, the reason for the discrepancy was described in further detail. Disagreement between the observers was resolved by discussion to achieve a consensus.

Infants who had one or more echocardiograms performed by a neonatologist alone (without confirmation of the diagnosis by the paediatric cardiologist) may have had a structural cardiac defect diagnosed after discharge from the neonatal unit. To pick up these false negative cases, we cross checked these infants’ records against admission and outpatient attendance data from the regional cardiac centre at Royal Liverpool Children’s Hospital, Alder Hey.

Statistical data were analysed using Arcus Quickstat 1.0 (Longman Software Publishing, Cambridge, UK). Groups were compared using either χ² or Fisher’s exact test for categorical data and Mann-Whitney U test for continuous data.

RESULTS

A total of 157 echocardiograms were performed in 82 infants during the study period. The median (interquartile range, IQR) gestation and birth weight of babies scanned was 32 (27–38) weeks and 1702 (1042–3105) g respectively. Their median (IQR) age at the time of the first echocardiogram was 2 (1–18) days. The median (IQR) number of scans/infant was 1 (1–2) with a maximum of nine scans performed in one baby.

The most common reasons for performing an echocardiogram were an asymptomatic murmur and cardiorespiratory symptoms with or without a murmur (table 1). Although in most babies the purpose of performing echocardiography was to diagnose or exclude structural heart disease, some babies were scanned to evaluate pulmonary haemodynamics and cardiac function. Approximately half of the infants scanned (44/82, 54% (95% confidence interval (CI) 42% to 65%)) had a definite structural cardiac abnormality identified on echocardiography (fig 1, table 2). In addition, 30 of the remaining 38 infants with essentially normal cardiac anatomy had evidence of either a functional cardiac abnormality (persistent pulmonary hypertension of the newborn or left ventricular dysfunction) or a trivial structural cardiac abnormality (patent foramen ovale, small PDA, or both).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Echocardiographic findings. PDA, Persistent ductus arteriosus; LV, left ventricular; PPHN, persistent pulmonary hypertension of the newborn; PFO, patent foramen ovale.

Initial echocardiographic assessment prompted a specific change in clinical management in 64 (78%, 95% CI 68% to 86%) babies. An urgent cardiology/cardiac surgery opinion was sought in 10 babies, medical treatment with routine cardiology follow up in nine babies, medical treatment without cardiology review in 16 babies, and routine cardiology follow up only in 29 babies. Echocardiography did not lead to an immediate change in clinical management in 18 infants, although a structural heart defect was excluded in five infants who had a suspected congenital heart defect on the basis of an antenatal ultrasound scan or a positive family history.

Echocardiography was performed by both neonatologist and paediatric cardiologist in 38 infants. These infants who had paired examinations were broadly similar to the whole group in terms of basic details and indications for echocardiography, but were significantly more likely to have an underlying diagnosis of structural cardiac abnormality or haemodynamically significant PDA, and had significantly more scans (table 1). In 31 of the 38 (82%, 95% CI 66% to 92%) infants who had a scan performed by both neonatologist and cardiologist, there was complete concordance between the two examinations. No infants had scans that were completely different. In seven infants, there were some differences between the recorded findings of the two observers (table 3).

DISCUSSION

Echocardiography is often performed on our neonatal unit usually to diagnose or exclude a structural heart defect, but also to evaluate pulmonary haemodynamics and cardiac function. The yield of abnormal findings is high, with abnormalities of structure and/or function being identified in about 70% of echocardiograms. Initial echocardiographic findings directly influenced clinical management in over three quarters of babies scanned.

A recent survey of UK practice suggested that babies are often transferred out of the neonatal unit for echocardiographic assessment or receive treatment inappropriately because of the lack of availability of 24 hour on site paediatric cardiology services in most tertiary neonatal units.¹ This study has again emphasised the need for ready access to echocardiographic skills on the neonatal unit. The most common reason for performing an echocardiogram was to determine whether the underlying cardiac anatomy was normal or abnormal. In addition to defining cardiac anatomy, echocardiography may also be useful in the assessment of neonatal cardiovascular physiology. In our study a considerable proportion of infants were diagnosed with a “physiological” abnormality, which led to a specific change in clinical management. Examples included identification of a haemodynamically significant PDA in a ventilated baby without a murmur, persistent pulmonary hypertension of the newborn in a baby with hypoxaemic respiratory failure, and poor left ventricular contractility in a hypotensive infant.

There was generally good agreement between the echocardiographic findings of the neonatologists and the subsequent scan performed by the cardiologist, with complete concordance in most cases. There was some disagreement in seven babies. Two infants with complex lesions were correctly identified, and, although the full details of the lesions were not defined, they were referred appropriately to a cardiac centre for further management. In two infants, a PDA was not seen on the initial scan, once by a neonatologist and once by the cardiologist. Initial failure to detect the PDA may have been because elevated pulmonary artery pressure prevented flow through the duct. Pulmonary stenosis was both overdiagnosed and underdiagnosed, and, although this did not appreciably contribute to a change in management, this experience should encourage neonatologists to evaluate the pulmonary valve particularly carefully.

In a retrospective review of referrals to a specialist paediatric cardiology centre, Ward and Purdie¹⁴ reported a misdiagnosis rate of 44% in scans performed by specialists other than trained paediatric cardiologists (mainly radiologists and adult cardiologists), with the highest proportion of errors occurring in the neonatal period. Our study is different in that we prospectively studied the performance of two neonatologists with some experience in neonatal echocardiography (but no formal training in paediatric cardiology) in an unselected infant population. In contrast with a previous suggestion that there should be a “ . . .restriction of individuals undertaking pediatric echocardiography . . .”,¹⁴ our findings support the widespread current practice in the United Kingdom of neonatologists performing echocardiograms, with close support from paediatric cardiologists. It has been proposed that telemedicine may also be useful in enabling prompt diagnosis and referral to a specialist cardiology centre, and this may be helpful especially in remote areas.¹⁵

The introduction of a formal, structured training programme and accreditation process in neonatal echocardiography (not yet available in the United Kingdom) is likely to further improve the performance of neonatologists in this area. The essential components of any training programme include opportunities for developing practical skills in echocardiography alongside specialists trained in neonatal echocardiography, reinforced by a sound understanding of the theory behind the use of ultrasound and the different ultrasound systems and modalities available. In addition to textbooks and other educational material, a number of practical courses in neonatal echocardiography are already available in the United Kingdom.^9,10 These are useful in that they provide an introduction to the subject, but cannot replace a structured training programme in neonatal echocardiography. Once acquired, maintaining echocardiographic skills will require a neonatologist to perform echocardiograms on a regular basis, with clinical sessions devoted to echocardiography in a neonatal unit able to provide a suitable caseload of patients both in terms of number and diversity of cases. We believe that this should be possible in most tertiary neonatal intensive care units in the United Kingdom. Close liaison with local colleagues in paediatric cardiology is essential to discuss echocardiographic findings, clinical diagnosis, and subsequent management. Neonatologists performing echocardiography should also audit their performance regularly to ensure that a high quality service is being delivered.

Some babies with structural heart disease may have been discharged, having been thought to have a normal heart or trivial structural abnormality from a neonatologist’s echocardiogram. However, of the 24 infants who had an echocardiogram performed by a neonatologist (but where the findings were not confirmed by the paediatric cardiologist) during the neonatal admission, none was subsequently diagnosed with a structural cardiac defect that had been missed on the initial echocardiogram. We acknowledge that we cannot completely exclude the possibility that an asymptomatic baby with a structural heart defect missed on the initial echocardiogram may not yet have presented for cardiological assessment. Although it is important to recognise that there is potential for structural heart disease to be missed (a problem not confined to neonatologists), our results do not suggest that this is a common occurrence when echocardiography is performed by neonatologists in this setting.^8,16

In summary, our findings suggest that echocardiography performed on the neonatal unit has a high yield for the diagnosis of structural and functional cardiac abnormalities, often results in a change in clinical management, and is a reliable tool in the hands of neonatologists with close support from paediatric cardiologists.