Cranial ultrasounds: Are they reliable?

Abstract

Most newborn intensive care unit centers perform serial cranial ultrasound evaluations early in the course of hospitalization for premature infants and often a follow–up examination is done at 4-6 weeks of age. These evaluations are done to document the presence of intracranial hemorrhage, to guide choice of therapies that may exacerbate risk of further hemorrhage, and to counsel families about neurodevelopmental outcomes. These data are also often used for benchmark comparisons. There have been few studies that investigated the reliability or accuracy of these findings, and yet randomized controlled trials in neonatal patients and interventional studies frequently include presence and degree of intraventricular hemorrhage (IVH) in their outcome data. Hintz et al describe the interobserver reliability and accuracy of the interpretation of cranial ultrasounds in premature infants. The authors took advantage of a large collection of ultrasound examinations performed as part of the NICHD Neonatal Network’s trial on the safety of inhaled nitric oxide. This was an important outcome variable because early case reports had suggested a potential risk of IVH; however, the trial found no differences in the rates of severe IVH or periventricular leukomalacia (PVL) between treated infants and controls. Scans were read by central readers who were board-certified radiologists and employed a structured data instrument. The study demonstrated a high degree of agreement between “local readers” and the central “experts.” This agreement was best for Grades 3 and 4 IVH, less so for periventricular leukomalacia, and poor for milder degrees of IVH. These observations are reassuring but also raise a note of caution. There is a strong association between severe degrees of IVH and poor neurodevelopmental outcome. The strength of those associations is less for milder degrees of bleeding. Given the differences in interpretation between the experts and the home-based readers for milder degrees of IVH, care should be exercised by clinicians in counseling families regarding the association between cranial ultrasound findings and neurodevelopmental prognosis. The study also raises a note of caution for investigators who use cranial ultrasound findings as a study outcome. Future designs should employ strategies to confirm the reliability of the ultrasound findings. Likewise, additional data of this kind on inter-observer reliability of intracranial pathology from computerized tomography or MRI evaluations is needed. All that said, our understanding of the reliability of cranial ultrasound examinations is advanced by this report. Interobserver Reliability and Accuracy of Cranial Ultrasound Scanning Interpretation in Premature InfantsThe Journal of PediatricsVol. 150Issue 6PreviewTo assess interobserver reliability between 2 central readers of cranial ultrasound scanning (CUS) and accuracy of local, compared with central, interpretations. Full-Text PDF