Superior Vena Flow Quantification in the Newborn: Reply to a Letter by Kluckow and Evans

We thank Drs Kluckow and Evans for their comments and acknowledge that they have pioneered the use of echocardiography to drive improvements in neonatal care. We share their enthusiasm for neonatal echocardiography and aim not to prevent its use but rather to make it more robust and hence yet more valuable. We do not, however, accept their premise that it can be “premature and unjustified” to cast doubt on the utility of a relatively novel measure; rather, we argue that skepticism should be the starting point for assessing all novel techniques and that robust validation should occur before clinical confidence replaces doubt.

We are not the first to urge caution in the use of superior vena caval (SVC) flow. In its consensus statement, the writing group of the American Society of Echocardiography highlighted that measurement of SVC flow in neonates was subject to significant variability and that “use of SVC flow requires further validation before being incorporated into standard clinical practice.” Dr Evans has himself acknowledged that assessment of SVC flow is “a more difficult technique to master” and that “its role is probably more as a research tool rather than a routine clinical tool.”

Incorporation into clinical practice requires robust validation, yet this is severely limited in premature neonates by the lack of a gold standard. Instead, only repeatability and indirect validation can be performed. In their original report, Kluckow and Evans reported intraobserver scan-rescan variability for SVC flow volume of up to 32.8%. An indirect validation (of SVC flow against left ventricular output [LVO] in infants without a ductus arteriosus) was also performed on the assumption that both measures would represent systemic blood flow. However the association between the two was relatively weak ( R 2 = 0.40), such that 60% of the variability in SVC flow was not accounted for by variability in LVO and must therefore be due to other factors, such as variability in the proportion of LVO supplying the upper body (a reasonable hypothesis) or to measurement inaccuracy.

In 2008, we reported scan-rescan repeatability for SVC flow in neonates using the widely accepted Bland-Altman methodology of limits of agreement (95% confidence limits of repeated measures) and repeatability index (limits of agreement/mean of population) and showed an intraobserver repeatability index of 31%, very similar to the likely 95% confidence limits of Kluckow and Evans, given their range of 32.8%. To our knowledge, the only other group with repeatability data reported a much higher intraobserver repeatability index (53%), though these investigators used a lower frequency probe, which is suboptimal in newborns. A particular concern in this latter study is that this repeatability was assessed merely from the analysis and reanalysis of a single acquisition of SVC flow. It is likely that a true scan-rescan analysis would have produced even poorer repeatability. Despite, or perhaps because of, Kluckow and Evans’s assertion that SVC flow is so widely used in clinical studies, we felt that further validation of SVC flow was warranted.

We have been applying phase-contrast magnetic resonance imaging (PCMRI) to the neonatal population since 2006. We have previously shown robust validation of our PCMRI sequences against a gold-standard ex vivo flow phantom ( R 2 = 0.995). We have validated PCMRI against magnetic resonance imaging volumes of cardiac output generated from end-diastolic and end-systolic left ventricular volumes in preterm infants (repeatability index = 16.6%) and have internally validated PCMRI sequences of systemic perfusion by comparing LVO with the sum of SVC and descending aortic flow in neonates without a ductus arteriosus (repeatability index = 13.2%). Last, we have demonstrated that the scan-rescan repeatability of SVC flow by PCMRI is superior to that by echocardiography (repeatability index = 12.8%).

Kluckow and Evans raised a number of specific concerns about our report that we feel merit a reply. First, they state that we are proposing PCMRI as a gold standard. We went to great lengths to avoid such a conclusion, indeed stating that PCMRI “is not a flawless gold standard”; we do, however, feel that the technique shows significant promise. Second, we acknowledge that the time delay between PCMRI and echocardiographic assessments in our cohort was longer than desirable (median, 3.67 hours; range, 0.25-33.0 hours), but this issue alone does not explain the disparity in performance between the assessment of LVO (repeatability index = 28.2%) and SVC flow (repeatability index = 68.0%). Moreover, choosing to study a more stable cohort of infants outside the transitional period allowed us to tolerate a certain delay between magnetic resonance imaging and echocardiography, because these infants are less likely to have abrupt changes in cardiovascular parameters.

Third, Kluckow and Evans questioned whether PCMRI reliably measures stroke distance. On the basis of our published external and internal validations (above), we feel relatively confident that the PCMRI techniques are robust, though we would welcome third-party repetition of our findings. Our current hypothesis for why echocardiography may overestimate SVC stroke distance is related to the nonlaminar flow within the vessel, but at present, this is purely speculative.

Last, Kluckow and Evans stated that they and others have found SVC flow to be less reliable after 48 hours of age. This is certainly a possibility, given that echocardiographic windows often become compromised in preterm infants as lung disease progresses, and our study could not address that hypothesis. We are, however, unaware of data confirming this pattern, but we would welcome its publication.

There are many points on which we agree with Kluckow and Evans. Targeted echocardiography has the scope to improve neonatal care, all novel techniques should be subject to validation, multiple measures should be included in any clinical assessment to minimize the impact of variability, and cardiac magnetic resonance imaging is unlikely to be applicable to unstable newborn infants in the foreseeable future. Our hope, however, is that in time, PCMRI will become a true gold standard capable of driving improvements in echocardiographic techniques. We have already demonstrated that assessment of SVC area from the axial rather than the sagittal view improves performance, and we are currently assessing the quantification of stroke distance from the suprasternal view, as described by Harabor and Fruitman.

In the meantime, we argue that clinicians should be aware of the limitations of all clinical assessment techniques, particularly those that are operator and equipment dependent. We would encourage any practitioners routinely using SVC flow (or any quantification technique) in neonates to perform audits of 10 to 20 scans with repeated measurements and define their own levels of repeatability; only then will individuals know what confidence to place in their measures.

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

May 31, 2018 | Posted by in CARDIOLOGY | Comments Off on Superior Vena Flow Quantification in the Newborn: Reply to a Letter by Kluckow and Evans

Full access? Get Clinical Tree

Get Clinical Tree app for offline access