We read with interest the article by Ficial et al . published as a validation study of magnetic resonance imaging (MRI) versus echocardiographic measurements of systemic blood flow. This is the first published validation study of phase-contrast MRI, in a group of babies with no demographic similarities to participants in the original studies of low blood flow. The authors concluded that echocardiographic assessment of superior vena cava (SVC) flow is of limited accuracy in their studied population and further claim that this casts doubt on the utility of this measurement generally for diagnostic decision making. We believe that these conclusions are premature and unjustified.

There are two types of validation studies: clinical validation and measurement validation against a gold standard. There are no clinical validation studies of cardiac MRI in preterm neonates, and Ficial et al .’s study does not provide any. Contrast this with other measures of systemic blood flow, including SVC flow, for which there are well over 100 publications from many groups around the world correlating it with important clinical outcomes ranging from intraventricular hemorrhage and 3-year neurodevelopment to improved cardiac output after delayed cord clamping and umbilical cord milking. Clearly, much more data about venous MRI studies in preterm infants are required before it can be regarded as superior to other measures of systemic blood flow from a clinical point of view.

Robust repeated measurement validation needs to fulfill several criteria, few of which are satisfied here.

First, a strong gold standard is required. Ficial et al . propose that MRI assessment is the “gold” standard for the validation of systemic blood flow (including venous blood flow) in newborn infants on the basis of accuracy and repeatability studies, mainly of arterial blood flow in adults and older children. There is little detail about the use of MRI to measure venous blood flows, making it hard for the reader to decide if this is a robust technique.

Second, the compared measurements must be closely related in time in order to minimize physiologic variation as opposed to measurement error. In Ficial et al .’s study, the measures are separated by up to 33 hours, which severely compromises the validity of the study. Preterm infants are clinically labile, and blood pressure and other cardiovascular measures change frequently, meaning that even a 1-hour difference between studies is problematic.

Third, the two comparators must measure the same thing. We have concerns about whether the techniques used to measure systemic blood flow were correct, as the difference in velocity-time integral or stroke distance is large and counterintuitive given that a likely source of error in Doppler measurement (imperfect angle of insonation) will tend to underestimate rather than overestimate VTI, as found in this study.

Fourth, the two measures must be compared in the population of clinical interest. The population studied in this correlation study bears little resemblance to the infants included in our original studies or in whom we use these measures currently. Our original publication on SVC flow included a population of 126 infants (vs 49 infants in Ficial et al .’s study), with a mean gestational age of 27 weeks (vs 34 weeks), all studied at 5, 12, 24, and 48 hours (vs 11 days) of life, many of whom were mechanically ventilated. We (and others) have found that measurement of SVC flow by ultrasound in infants >48 hours old and of older gestation is less accurate and repeatable. We are therefore not surprised that, leaving aside the major issues about timing of the two measures, Ficial et al . found a lack of correlation in the much older group of infants studied here.

Cardiac MRI may well have some complementary role in the future of cardiovascular assessment of neonates but is unlikely to be applied to those who most need flow measures. These are the extremely preterm infants born at <28 weeks’ gestation, in the first 48 hours of life, often requiring significant respiratory support, including mechanical ventilation. The investigators themselves acknowledged this in their original report describing cardiac MRI. In this group, we recommend bedside Doppler ultrasound measurement of both SVC flow and right ventricular output, as a cross-check against each other, to minimize risks of measurement error. This provides useful cardiovascular information, which aids diagnostic decision making and has been clearly linked to both short- and long-term clinical outcomes. Concluding that echocardiographic measures of flow have limited clinical utility in identifying circulatory failure on the basis of this single validation study is premature and unjustified.