Repeatability and Reproducibility of Neonatal Echocardiography: The Copenhagen Baby Heart Study

Abstract

Background: The Copenhagen Baby Heart Study (CBHS) is a population-based cohort study of neonates (N = 25,000), including echocardiography. Echocardiography in neonates is mainly focused on congenital heart disease (CHD), whereas general aspects of cardiac dimensions and function in neonates without CHD remain to be further addressed.

Purpose: This study was conducted to assess the reliability of neonatal echocardiography and validity of echocardiographic methods used in the CBHS.

Methods: Reliability and agreement were tested for two-dimensional (2D), M-mode, spectral Doppler, and tissue velocity echocardiography for the following. (1) Measurements: seven sonographers independently performed two measurement rounds: (a) measurement of the same 50 echocardiograms (n = 350 echocardiograms measured) and (b) repeated measurement of 25 of the 50 echocardiograms (n = 175 echocardiograms measured). (2) Acquisition: four sonographers independently performed two rounds of echocardiographic acquisition and subsequent measurement of the same 22 neonates (n = 176 acquisitions and measures). Intra- and interobserver variabilities were assessed by determinations of coefficient of variation (CV), intraclass correlation coefficient (ICC), Bland-Altman plot, and 95% limits of agreement.

Results: (1) Measurements: we found intra- and interobserver ICC ≥ 0.67 for 2D parameters, except for left ventricular (LV) wall thicknesses and LV diameter (interobserver); ICC ≥ 0.84 for tricuspid annular plane systolic excursion (TAPSE); ICC ≥ 0.93 for pulsed-wave Doppler (PW); ICC ≥ 0.84 for continuous-wave Doppler; and ICC ≥ 0.87 for tissue velocity parameters. We found CV < 15% for all parameters except LV wall thicknesses. (2) Acquisition: we found intra- and interobserver ICC ≥ 0.69 for 2D parameters, except for LV wall thicknesses, aortic valve annulus (interobserver), and LV end-systolic diameter (interobserver); ICC = 0.45-0.49 for TAPSE; ICC = 0.48-0.64 for PW; and ICC ≥ 0.70 for continuous wave. We found CV < 15% for all parameters.

Conclusions: Reliability of echocardiographic measurements and acquisition of cardiac dimensions and function were good for most parameters but lower for TAPSE (acquisition) and PW Doppler (acquisition) and poor for LV wall thicknesses. In general, echocardiography of cardiac dimensions and function in the neonate is reliable.

Trial registration: ClinicalTrials.gov NCT02753348.

Keywords: Acquisition; Agreement; Echocardiography; Measurements; Neonate; Reliability.