Background
Several quantification algorithms for measuring left ventricular (LV) size and function are used in clinical and research settings. The aims of this study were to investigate the effects of measurement algorithm and beat averaging on the reproducibility of measurements of the left ventricle and to assess the magnitude of agreement among the algorithms in children with dilated cardiomyopathy.
Methods
Echocardiograms were obtained in 169 children from eight clinical centers. Inter- and intrareader reproducibility was assessed on measurements of LV volumes using the biplane Simpson, modified Simpson, and 5/6 × area × length (5/6AL) algorithms. Percentage error was calculated as inter- or intrareader difference/mean × 100. Single-beat measurements and the three-beat average (3BA) were compared. Intraclass correlation coefficients were calculated to assess agreement.
Results
Single-beat interreader reproducibility was lowest (percentage error was highest) using biplane Simpson; 5/6AL and modified Simpson were similar but significantly better than biplane Simpson ( P < .05). Single-beat intrareader reproducibility was highest using 5/6AL ( P < .05). The 3BA improved reproducibility for almost all measures ( P < .05). Reproducibility in both single-beat and 3BA values fell with greater LV dilation and systolic dysfunction ( P < .05). Intraclass correlation coefficients were >0.95 across measures, although absolute volume and mass values were systematically lower for biplane Simpson compared with modified Simpson and 5/6AL.
Conclusions
The reproducibility of LV size and functional measurements in children with dilated cardiomyopathy is highest using the 5/6AL algorithm and can be further improved by using the 3BA. However, values derived from different algorithms are not interchangeable.
Echocardiographic measures of left ventricular (LV) size and systolic function are widely used as end points in clinical trials. However, the limited availability of data concerning the reproducibility of quantitative indices of ventricular function in pediatric dilated cardiomyopathy (DCM) is an impediment to controlled trials of therapy for ventricular dysfunction in children. Several algorithms for measuring LV volumes using two-dimensional echocardiographic methods are common in clinical and research use. These methods include the 5/6 × area × length (5/6AL), modified Simpson (MS), and biplane Simpson algorithms. Although the American Society of Echocardiography (ASE) has recommended the biplane Simpson (also known as the biplane method of disks) algorithm as the approach of choice for LV volume quantification in adults, the applicability of this recommendation to pediatric populations is unknown.
The objectives of this analysis were to (1) determine the impact of the method of calculating LV volumes on interreader and intrareader reproducibility of measured and calculated variables in children with DCM, (2) determine whether averaging multiple beats improves reproducibility, (3) assess the agreement among measurements by algorithm, and (4) determine whether the severity of cardiomyopathy affects the reproducibility of LV volumes.
Methods
The Ventricular Volume Variability (VVV) study was a multicenter observational study of pediatric subjects with stable DCM undertaken by the National Heart, Lung, and Blood Institute–sponsored Pediatric Heart Network. Enrolled subjects were followed for 18 months, and a study protocol echocardiogram was obtained at each clinical visit during this time. Inclusion and exclusion criteria are listed in the online Appendix . The study was conducted in accordance with the guidelines of the Pediatric Heart Network’s Data and Safety Monitoring Board and of each center’s institutional review board. Full details of the study design have been previously published.
The primary aim of the VVV study was to evaluate the longitudinal variance of echocardiographic indices of LV size and function. Subjects with histories of DCM by chart review were approached for consent for participation at the time of a clinical evaluation. Those subjects who met full inclusion criteria on the basis of the baseline study echocardiogram were eligible for follow-up echocardiography to determine longitudinal variability. The data from baseline echocardiograms obtained in patients who did not meet exclusion criteria but who also did not meet the dilation and/or dysfunction criteria (in other words, data obtained in those subjects whose echocardiograms had improved sufficiently to not meet entrance criteria for dilation and/or dysfunction since the diagnosis of DCM was made) were retained in the database and represent a normalized or near normalized population. Inclusion of these examinations permitted the analyses to be performed across a broader range of disease severity. For purposes of this report, only baseline evaluations are included.
Consented subjects underwent study echocardiography performed by sonographers at each site who were specifically trained on the standardized protocol for image acquisition. At least three cardiac cycles were recorded for each parameter. Height and weight were measured, and body surface area (BSA) was calculated using the Haycock formula. All baseline echocardiograms were submitted to the data-coordinating center and forwarded to the echocardiography core laboratory.
At the echocardiography core laboratory, two readers performed measurements on each echocardiogram to assess interreader reproducibility. The protocol specified 150 measured and derived parameters on each study. One ECL reader repeated all measurements 1 month later to assess intrareader reproducibility. All measurements were performed using custom Digital Imaging and Communications in Medicine software (EchoTrace; Marcus Laboratories, Boston, MA).
LV end-diastolic volume (LVEDV), LV end-systolic volume (LVESV), LV mass, and LV ejection fraction (LVEF) were calculated using three common algorithms. The ASE-recommended biplane Simpson method used areas from apical four-chamber and apical two-chamber views ( Figure 1 A). For the MS approach, we used an apical four-chamber area and a parasternal short-axis area ( Figure 1 B). The 5/6AL algorithm required the length of the left ventricle from the apical four-chamber and a parasternal short-axis area ( Figure 1 C).
Statistical Methods
Definitions
Inter- and Intrareader Reproducibility
In these analyses, the outcome measure is the percentage error of the mean. To evaluate interreader reproducibility, the absolute difference (“error”) between the measurements made by the primary and secondary readers was divided by the mean of those two measurements. To evaluate intrareader reproducibility, the difference (“error”) between the immediate and 1-month repeat measurements made by the primary reader was divided by the mean of those two measurements. Single-beat percentage error was based on the measurements obtained from the first beat, and three-beat average (3BA) percentage error was based on the average of all three measurements.
Reduction in Error
To quantify the reduction in error that was due to beat averaging, a “percentage reduction in percentage error” term (abbreviated to “error reduction”) was calculated using 3BA compared with single-beat averages as
( Single – beat percentage error − 3 BA percentage error ) / Single – beat percentage error × 100 % .
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
