Visual Guesswork Falls Short in Reading Newborn Hearts, Study Finds

When a critically ill newborn needs an urgent assessment of heart function, the clinician at the bedside often has only seconds to decide whether the heart is pumping well. A new exploratory study suggests that this rapid, eyeball-based judgement, while valuable in extreme cases, is far less reliable than many clinicians might hope when the problem is mild or moderate rather than severe.

The research, published in Echo Research & Practice, compared how well visual estimation of ventricular systolic function in newborn infants matched up against standard quantitative echocardiographic measurements. Neonatologist-performed echocardiography, known as NPE, is increasingly practiced in neonatal units, and its operators typically judge systolic function by eye first, confirming with quantitative markers only when indicated. The study set out to test how concordant those two approaches actually are, both for experienced NPE operators and for clinicians with little or no echocardiography experience.

The prospective, observational, dual-centre cohort study took place at Homerton University Hospital in London and Aberdeen Maternity Hospital in Scotland between April and October 2024. Sixty newborn infants who underwent NPE as part of routine care were enrolled. Their mean gestational age was 34 weeks, with a standard deviation of 6.2 weeks, and their mean birth weight was 2311 grams, with a standard deviation of 1319 grams. Infants with congenital heart disease, apart from patent ductus arteriosus, and those with major congenital abnormalities or missing data were excluded.

Blinded NPE and non-NPE operators independently assessed left and right ventricular systolic function visually, using apical four-chamber and parasternal views. For the left ventricle, operators evaluated the approximation of the free wall and the septum and the reduction in the cavity during systole; for the right ventricle, they watched the vertical movement of the free wall and similar cues. Each operator then classified contractility as either normal or impaired. Crucially, NPE operators assessed scans from infants not cared for at their own site to ensure blinding, while non-NPE operators, who received no prior training or feedback during the study, viewed anonymised scans with no clinical information at all.

These visual judgements were then compared against established quantitative markers: shortening fraction, Simpson’s ejection fraction and global longitudinal strain for the left ventricle, and tricuspid annular plane systolic excursion, or TAPSE, along with fractional area change, for the right ventricle. Impaired left ventricular function was defined pragmatically as a shortening fraction below 26 percent, an ejection fraction below 50 percent, or global longitudinal strain less negative than minus 20 percent, consistent with commonly used neonatal echocardiographic criteria. For the right ventricle, TAPSE and fractional area change more than two standard deviations below age-related values were considered impaired.

The headline numbers were sobering. Sensitivity testing across all echocardiographic parameters ranged from 25 to 80 percent, while specificity ranged from 38 to 78 percent. For left ventricular parameters specifically, sensitivity ranged from 25 to 44 percent for the NPE operator and the non-NPE operator, with specificity between 56 and 76 percent. Overall positive and negative likelihood ratios did not demonstrate a strong association with the presence or absence of disease. Concordance between NPE and non-NPE operators was generally limited across most echocardiographic parameters, although test-retest reliability was acceptable for several quantitative measures, including Simpson’s ejection fraction, global longitudinal strain and right ventricular fractional area change.

One pattern stood out clearly: sensitivity increased with worsening grades of left ventricular dysfunction defined by shortening fraction. For the NPE operator, sensitivity rose from 20 percent for mild dysfunction to 100 percent for severe dysfunction. In other words, clinicians were most accurate at the extremes, correctly identifying hearts that were either very normal or severely impaired, while borderline cases risked misclassification. This mirrors findings from other settings. In animal studies, practitioners of varying experience correctly identified severe shortening fraction reduction 92 to 96 percent of the time but managed only 50 to 54 percent success for moderate reduction, and in adult patients, agreement between operators visually grading left ventricular function was just 50 percent.

The authors point to several reasons why visual estimation struggles in this population. The human eye can reliably recognise very normal or very abnormal heart function, but infants with borderline dysfunction fall into a grey zone. Right ventricular assessment carries additional challenges owing to the ventricle’s complex structural geometry, muscle fibre orientation and response to afterload. Atrioventricular valvar regurgitation can create a false impression of good systolic function, and suboptimal images, operator experience and artefacts all contribute to poor concordance. The neonatal circulation adds a further layer of complexity, because quantitative markers themselves can be imperfect when exposed to the dynamic physiology of intracardiac shunts and altered loading conditions in transitional circulation. The study also did not account for the haemodynamic significance of patent ductus arteriosus, which may alter loading conditions and influence quantitative measurements.

The study’s limitations are substantial and the authors are candid about them. The sample was small, reflecting the exploratory design, and the prevalence of significant cardiac dysfunction was low, particularly for left ventricular parameters. Convenience sampling limited control over the prevalence and severity of pathology. The two cohorts were heterogeneous: infants at one centre were significantly more premature, needed more respiratory support, and included more invasively ventilated infants, 80 percent versus 46.7 percent at the other centre, although inotrope use, which was modest overall, differed without statistical significance. Test-retest analysis was available for only one NPE operator, and the binary classification of left ventricular function as simply normal or impaired may have masked differences that severity grading would have revealed. The authors note that it is difficult to disentangle the effects of acquisition expertise from interpretative experience, since NPE operators acquired the images that non-NPE operators interpreted.

Despite these caveats, the study carries a practical message for neonatal intensive care. Visual estimation may serve as a useful bedside screening tool for detecting severe ventricular dysfunction, for example in a neonate with sepsis where rapid evaluation may facilitate timely optimisation of therapy, or in hypoxic ischaemic encephalopathy with pulmonary hypertension, where ruling out severe left ventricular dysfunction matters because it is a key contraindication to pulmonary vasodilator therapy. But visual assessment alone is likely insufficient for detecting mild or borderline impairment, and reliance on qualitative evaluation in such cases risks under-recognition of clinically relevant dysfunction. Where feasible, the authors conclude, quantitative echocardiographic measures should be incorporated alongside visual assessment, with visual estimation treated as a screening rather than a definitive tool that prompts further evaluation when clinical concerns persist. Larger studies enriched for infants with ventricular dysfunction are needed to confirm these findings and to refine the evaluation of ventricular function in newborn infants.

Subject of Research: Concordance between visual estimation and quantitative echocardiographic markers of ventricular systolic function in newborn infants

Article Title: Comparison of visual estimation of ventricular systolic function with quantitative echocardiographic markers in newborn infants

Article References: Pereira, S. S., Yeldo, B., Fleming, R., & Satas, S. (2026). Comparison of visual estimation of ventricular systolic function with quantitative echocardiographic markers in newborn infants. Echo Research & Practice, 13(1), Article 40. https://doi.org/10.1186/s44156-026-00142-2

Image Credits: AI Generated

DOI: 10.1186/s44156-026-00142-2

Keywords: neonatology, echocardiography, ventricular systolic function, visual estimation, shortening fraction, ejection fraction, TAPSE, global longitudinal strain, neonatologist-performed echocardiography, point-of-care ultrasound, diagnostic accuracy, newborn infants