Interobserver agreement and image quality of functional cardiac ultrasound measures used in a randomised trial of delayed cord clamping in preterm infants

Objective Functional cardiac ultrasound measures are used clinically and in trials for assessing the haemodynamic status of newborn infants. Superior vena cava (SVC) flow and right ventricular output (RVO) are established measures of systemic blood flow on the first postnatal day. The objective was to assess image quality and interobserver agreement of these measures in preterm infants enrolled in a randomised trial of immediate versus delayed cord clamping. Design and setting Image quality and interobserver agreement for SVC flow, RVO and ductus arteriosus (DA) size were assessed on measurements taken at 3-6, 6-12 and 20-28hours for the first 10 infants enrolled at each of four sites (total 40). Bland-Altman plots were constructed; mean difference (bias) and limits of agreement (LOA) were calculated. Potential sources of variation were explored. Results Quality was judged satisfactory for >97% of images. The mean difference and LOA between the observers were 5.4mL/kg/min and -49.0 to 59.8mL/kg/min for SVC flow, -26.6mL/kg/min and -131.4 to 78.2mL/kg/min for RVO, and 0mm and -0.8 to 0.8mm for DA diameter, respectively. The principal source of measurement error for SVC flow was diameter, and for RVO, diameter and velocity time integral. The difference between observers for both SVC and RVO was significantly associated with site. Conclusion Interobserver variability for SVC flow is consistent with that previously reported, but higher for RVO. The findings should be incorporated into clinical practice, training, accreditation and trial design.