Blood pressure and capillary refill time are poor predictors of volume of blood flow in the transitional circulation.1 ,2 Echocardiography is increasingly used in the assessment of the critically ill newborn,3 and can be performed without appreciably disturbing cardiorespiratory status.4 Doppler echocardiographic estimations of left ventricular output5 and right ventricular output6 are reliable in the newborn period, but persistent shunting through fetal pathways means that neither represents true systemic perfusion.7 ,8

Doppler echocardiographic measurement of blood flow in the superior vena cava (SVC) has been suggested as a repeatable marker of upper body systemic perfusion.9 Low SVC flow in the neonatal period has been linked to periventricular haemorrhage10 and adverse long-term neurodevelopmental outcome.11 However, this technique has been evaluated by only one group of researchers.9^–11 Volume of blood flow in the post-ductal descending aorta (DAo) provides a potential marker of lower body systemic perfusion which has been assessed in older children.12 Although the pattern of DAo flow is well-studied in preterm neonates,13 ,14 volume of flow has only been assessed in small numbers of infants.15

This study aimed to assess the repeatability of measurement of SVC and DAo flow volume in the first 48 h of postnatal life in preterm infants. In addition, normative values for the two measurements were estimated.

METHODS

Infants of less than 31 weeks completed gestation were recruited to the study at the National Women’s Hospital, Auckland, between 1 December 2002 and 1 May 2004. Among the total cohort, infants who did not require circulatory support and required less than 48 h respiratory support were prospectively deemed “healthy”.16 ,17 A cohort of healthy term infants, who did not require admission to neonatal intensive care unit or have evidence of respiratory or circulatory compromise were also examined.

Echocardiography was performed in preterm infants as close as possible to 5, 12, 24 and 48 h postnatal age. Term infants were examined at similar ages, but we did not perform scans after 12 h if the infants had been discharged. Scans were performed by one of two investigators (AMG and CAK), both of whom had more than 2 years’ experience in neonatal echocardiography at the start of the recruitment period. We standardised the technique for assessment of both SVC and DAo flow on a small number of infants prior to enrolment of the cohort described. All scans were performed with an ATL 3000 ultrasound scanner (Advanced Technological Laboratories, Bothell, Washington, USA) equipped with a 7 MHz probe. Images were recorded on videotape and measurements performed by a single investigator (AMG) away from the cotside.

SVC flow velocity time integral (VTI) was assessed using pulsed wave Doppler from a low subcostal view as originally described.9 Any reversal of flow in the SVC was quantified and deducted from the measured forward flow to give an integrated VTI. SVC diameter was assessed from a high parasternal long axis view, rotated towards the true sagittal plane. The transducer head was placed as close to the midline as possible to acquire directly anteroposterior views of the SVC. Maximum and minimum SVC diameters were assessed for each cardiac cycle, and the mean of these used to quantify volume of flow. Although previous studies have used two-dimensional echocardiography to assess SVC diameter,9 we elected to use M mode trailing edge-leading edge echocardiography as it improves vessel wall definition and repeatability.18

DAo flow VTI was assessed twice during each scan, from a low subcostal sagittal view flow and from a high parasternal view. In both cases the DAo was viewed with minimum angle of insonation between the ultrasound beam and the direction of blood flow. Pulsed wave Doppler was used with appropriate angle correction. Wall thump filters were set to their minimum level to aid detection of low-velocity diastolic DAo flow. Any reversal of diastolic flow in the DAo was quantified and deducted from the measured forward flow to give an integrated VTI. DAo diameter was assessed as close as possible to the plane of the aortic valve when seen in the parasternal short axis view. Systolic and diastolic DAo diameter was assessed by M mode echocardiography using the trailing edge-leading edge technique.

All scans were analysed using in-built cardiac analysis software on the ultrasound machine. All measures were averaged over five consecutive cardiac cycles19 except in the case of SVC flow velocity, for which measures were averaged over 10 consecutive cycles to minimise the impact of variation in flow related to respiratory movements. Volume of flow (expressed as ml/kg/min) was calculated from:

where VTI  =  velocity time integral in cm, π = 3.14, vessel diameter in cm, birth weight in kg

To analyse the within-observer and between-observer repeatability of the ultrasound techniques a subgroup of infants was re-examined within 10 min of the initial scan by either the same (AMG) or a different examiner. Images were recorded onto a separate videotape, and these images were analysed after completion of analysis of the primary images of the entire cohort. The repeatability coefficient (the difference between measurements one would have to witness to have a 95% probability that it had not occurred due to chance or measurement inaccuracy) was calculated as described by Bland and Altman.20 The repeatability index was calculated as the repeatability coefficient divided by the mean of all the measures.21 Median and range of variability between pairs of scans were also calculated. When assessing temporal changes in the normative data, comparisons between pairs of time points were made using the Wilcoxon signed rank test. Tendency of flow volume to change consistently over time was examined by repeated measures analysis of variance. All data were analysed using Statview software (SAS Institute, Cary, NC, USA), and statistical significance was taken where p<0.05.

RESULTS

Within-observer repeatability

Within-observer repeatability was assessed from 18 scans of 13 preterm infants in the first week of life, performed by a single operator (AMG). The infants had a median (range) gestation of 29 weeks (27–30) and median birth weight of 1235 g (850–1900 g). Four infants were included in the healthy cohort described below. Three infants were intubated, 12 were on continuous positive airway pressure (CPAP) and three were not receiving respiratory support. The infants were not given muscle relaxants or sedated during the study period.

Within-observer repeatability of assessment of SVC flow

The repeatability coefficient of within-observer measurement of SVC flow was 30 ml/kg/min (95% CI 17–43 ml/kg/min) (fig 1, table 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Bland–Altman plot of intraobserver difference between two measures of SVC flow for 18 scans in 13 preterm infants.

View this table:

• View inline
• View popup

Table 1 Within-observer repeatability for diameter, velocity time integral, heart rate and flow volume for the superior vena cava (SVC) and descending aorta (DAo) in 18 scans in 13 preterm infants

Within-observer repeatability of assessment of DAo flow

The repeatability coefficient of within-observer measurement of DAo flow from the subcostal view was 60 ml/kg/min (95% CI 34–86 ml/kg/min) (fig 2, table 1). There was minimal (median 4°) variability in angle correction used. In our cohort, assessment of DAo flow volume from the suprasternal view showed poorer repeatability than when assessed from the subcostal view (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Bland–Altman plot of intraobserver difference between two measures of DAo flow for 18 scans in 13 preterm infants.

DISCUSSION

Assessment of circulatory status is an integral component of care in the neonatal unit.22 Episodes of low systemic perfusion are associated with adverse outcomes in preterm23 and term24 infants. Although adequacy of the circulation in most clinical settings has previously been assessed using surrogate markers which are “imperfect predictors” of flow in the transitional circulation,1 reviews on circulatory support in the neonate are increasingly emphasising the importance of supporting blood flow as well as blood pressure.25 ,26

Fully assessing the error inherent in any new measurement technique is critical in determining the technique’s applicability. The analysis suggested by Bland and Altman20 ,27 is now accepted as the most robust technique for assessing repeatability.28 We found the repeatability coefficient for SVC flow in preterm infants, when assessed by a single observer, to be 30 ml/kg/min. This is the difference in SVC flow volume one would have to witness to have a 95% probability that it had not occurred due to chance alone. This equates to a repeatability index (the percentage difference required to suggest genuine change) of 31%. Although these values are high, the wide range of flow volumes seen in preterm infants (40–193 ml/kg/min) means that quantification of SVC flow volume may be a relatively sensitive technique for detecting haemodynamic change in the clinical setting.

The repeatability coefficient and the repeatability index were not quoted in the only previous study of repeatability of SVC flow in preterm infants (Kluckow and Evans9), but the range of variability seen in both cohorts was 1–33%. Given that the maximum variability in the 35 scans in Kluckow and Evans’ cohort was 33%, the repeatability index was probably similar to that seen in our study. Disappointingly, the repeatability coefficient for DAo flow volume, when assessed by a single observer from the subcostal view in preterm infants was 60 ml/kg/min. Further work is clearly required to improve repeatability of the measurement.

Inconsistencies in DAo diameter measurement made a greater contribution to the variability in DAo flow volume estimation than inconsistencies in flow velocity. As flow depends on area (π × diameter2/4) any discrepancies in diameter measurement are compounded in the calculation of flow. The repeatability index for VTI measurement in the descending aorta was 24%, which is comparable with published repeatability for ascending aortic29 and pulmonary arterial21 VTI in neonates. The descending aortic diameter can be assumed to be constant over short periods when arterial blood pressure is constant. Therefore the reasonable within-observer repeatability of DAo stroke distance measurement provides some scope for the technique to be used to monitor trends in flow volume over time. In contrast, given the reasonable repeatability of assessment of volume of SVC flow and the distensibility of the SVC, monitoring of SVC flow volume will possibly have greater clinical utility than monitoring of SVC stroke distance alone.

Between-observer repeatability for measurement of both SVC and DAo flow volume in our cohort was very poor. This highlights the importance of a careful and consistent approach to the measurement of flow volume. The lack of a significant difference in mean SVC or DAo flow volume assessed by the two observers in our cohort suggests that there was no systematic difference in the diameter measures. The 95% confidence limits of the between-observer repeatability coefficients were wide and small numbers of infants were studied, so it is possible that we have overestimated the repeatability coefficient. However, even if one accepts the lower limit of the 95% confidence limits, the repeatability coefficients for both measures are still high. This may be due in part to the operators gaining proficiency in what are novel techniques. Further work in the area is likely to improve imaging technique, as training in techniques may be associated with improved repeatability.19 Nevertheless it is clear that the techniques are relatively difficult to learn, and that standardisation is critical. Until between-observer repeatability is substantially improved, measurements of SVC or DAo flow volume obtained by one observer should not be presumed to be comparable with those obtained by another.

Repeatability is probably influenced to an extent by the degree of spontaneous fluctuation of the measured variables. Our previously published data on spontaneous fluctuation in invasively monitored mean blood pressure demonstrate a repeatability index of up to 20%.4 While changes in blood pressure are not directly related to changes in volume of blood flow in the transitional circulation, similar spontaneous fluctuations may also be seen in volumes of blood flow.

Although our cohort included only small numbers of relatively healthy term and preterm infants, data on these infants can provide some estimate of normative data for comparison with sick infants. Defining what constitutes “normal” blood flow in preterm infants is difficult. However, preterm infants who require minimal respiratory support in the first postnatal days have been shown to have a near-normal postnatal circulatory adaptation.16 ,17

The fifth percentile for SVC flow volume in healthy preterm infants at any time in the first 48 h of postnatal life in our cohort was 55 ml/kg/min. The median SVC flow in our cohort of healthy preterm infants was approximately 30% higher than that previously reported by Kluckow and Evans.9 As the SVC VTI in the present study was similar to that reported by these authors, the discrepancy between these preterm cohorts comes entirely from differences in assessment of SVC diameter and is most likely due to the oval cross-sectional shape of the SVC, where imaging the vessel other than in the true anteroposterior plane will lead to higher estimates of diameter.9 We tried to ensure imaging of the vessel in the anteroposterior plane in our study, but a relatively lateral imaging window is required in some infants to visualise the vessel beyond overlying lung.9 The imaging window used in our study was probably more lateral than that used by Kluckow and Evans. This may also account for the lack of discrepancy in SVC diameter measurement in term infants where hyperinflation of lung tissue is less of a hindrance to imaging.9 However, since the SVC is oval in cross-section, neither study produces “true” SVC flow volume. Any measurement need only be reproducible and correlated closely to true flow to have clinical utility. The fifth percentiles for DAo flow volume and VTI in healthy term infants at any time in the first 48 h were 116 ml/kg/min and 9.1 cm, respectively. The fifth percentiles for DAo flow volume and VTI in healthy preterm infants at any time in the first 48 h were 90 ml/kg/min and 4.5 cm, respectively. Estimates of DAo flow volume in our cohort are similar to those previously reported in term infants30 ,31 and using an intra-aortic Doppler probe in a small cohort of preterm infants.30

An important limitation of the present study is the small number of infants studied. However, the repeatability of assessment of SVC flow volume was similar to that reported previously9 and repeatability of DAo stroke distance was similar to that previously reported for stroke distance in the ascending aorta29 and main pulmonary artery.21

In conclusion, assessment of volume of SVC flow has repeatability similar to that of many other neonatal haemodynamic measurements. A change in SVC flow volume in a preterm infant of more than 30 ml/kg/min, when assessed by a single observer, probably represents a true alteration in upper body systemic venous return. Although the repeatability of assessment of volume of DAo flow by a single observer was poorer, assessment of stroke distance (that is, VTI) of DAo flow has good repeatability. A change in DAo stroke distance in a preterm infant of more than 2.2 cm, when assessed by a single observer, is likely to represent a true alteration in lower body blood flow. Serial measurements of SVC flow and DAo stroke distance by a single observer may therefore have utility in assessing response to clinical intervention in newborn infants.

Using the techniques studied in this cohort, SVC flow volumes below 55 ml/kg/min and DAo flow volumes below 90 ml/kg/min in the first 48 h of postnatal life in preterm infants may represent low or borderline systemic perfusion.