A wide variety of cardiopulmonary disease processes can lead to the elevation of right heart pressure. In broad categories, elevation of right ventricular (RV) pressure can result from RV outflow tract obstruction, branch pulmonary artery obstruction, pulmonary vascular obstructive disease, pulmonary venous obstruction, or left ventricular inflow obstruction. Right heart catheterization (RHC) remains the gold standard for evaluating right heart hemodynamics in these patients. The risk for exposure to radiation and the invasive nature of RHC prompted investigators in the 1980s to introduce noninvasive methods of assessing right heart hemodynamics, specifically Doppler echocardiography (DE). These studies showed excellent correlation between DE-estimated RV pressure using tricuspid regurgitation (TR) and direct RHC-measured RV pressure. Because of its widespread availability and noninvasive nature, DE is now a widely used surrogate for RHC to determine and monitor right heart pressure in a wide variety of disease processes.

Recently, the accuracy of DE in estimating right heart hemodynamics has come into question. The initial studies comparing DE and RHC to assess right heart hemodynamics used correlation and regression analysis. This type of statistical analysis has its shortcomings, as highlighted by Bland and Altman in their seminal report in 1986. Bland and Altman argued that a high degree of correlation between two methods of measurement should not be interpreted as indicating accuracy and interchangeability of the two methods. Instead, they advocated using bias and precision statistics to compare different techniques of measuring the same physiologic variable. Bland-Altman analysis is now the accepted statistical method of comparing two methods of measuring the same physiologic variable, superseding correlation and regression. Recent studies in adults comparing DE and RHC using Bland-Altman analyses have suggested that DE is not as accurate as previously described for the assessment of right heart hemodynamics. A prospective, simultaneous study comparing DE and RHC using Bland-Altman analyses has not been performed in children.

In a prospective, simultaneous DE and RHC study in children, we sought to analyze the accuracy of DE in estimating right heart pressure. We further sought to determine if the degree of RV hypertension affects the accuracy of DE in assessing right heart pressure.

Methods

Patient Population

All children aged 0 to 17.9 years who presented to the St. Louis Children’s Hospital cardiac catheterization laboratory for clinically indicated RHC between November 2011 and November 2012 were considered for enrollment. Inclusion criteria were two-ventricle physiology, adequate TR envelope on precatheterization echocardiography, and sinus rhythm. Exclusion criteria were single-ventricle physiology, the presence of a subpulmonary morphologic left ventricle, and the presence of a mechanical tricuspid valve. Informed, written parental consent was obtained for all subjects. Informed, written assent was obtained for all subjects >12 years of age. The study was approved by the institutional review board for human studies at Washington University School of Medicine.

Catheterization Data

All patients underwent simultaneous measurement of the peak RV–to–right atrial (RA) pressure gradient (ΔP) by RHC and by DE. All data collection was done with the patient in the supine position and under the usual sterile conditions of the catheterization laboratory. RV and RA pressure tracings were recorded using a standard fluid-filled, end-hole catheter in pullback fashion. RV and RA pressure tracings were recorded simultaneous to the performance of DE. RHC-measured ΔP (ΔP _C ) was calculated by subtracting the RA V-wave pressure from the peak RV systolic pressure (RVSP) at end-expiration. Cardiac output was calculated using the Fick method and indexed to body surface area. Simultaneous systemic blood pressure was obtained by arterial catheterization in patients in whom arterial access was clinically indicated and by upper extremity noninvasive cuff measurement in those in whom arterial access was not clinically indicated. Interpretation of RHC data was done by a single, experienced interventional cardiologist (J.J.M.) to exclude interobserver variability. The interventional cardiologist was blinded to the patients’ diagnoses and Doppler echocardiographic data.

Echocardiographic Data

Two-dimensional and color DE was performed using commercially available ultrasound equipment (Vivid 9; GE Medical Systems, Milwaukee, WI). With the patient in the supine position, continuous-wave Doppler recordings of the maximal TR jet velocity (V) were obtained in standard apical or parasternal windows (whichever produced the highest velocity signal). The Doppler beam was kept in parallel alignment with the direction of the TR jet as visualized by color Doppler. The majority (80%) of echocardiographic studies were performed by a single cardiac sonographer. The remainder of the studies was performed by two additional cardiac sonographers. No saline or contrast injection was used to enhance weak signals. Patients were excluded from the study if there was no measurable TR by color Doppler at the time of data acquisition. DE-estimated ΔP (ΔP _E ) was calculated using the modified Bernoulli equation, ΔP _E = 4V ² . An average of three consecutive cardiac cycles was used to account for respiratory variation. Interpretation of Doppler echocardiographic data was done by a single, experienced echocardiographer (G.K.S.) to exclude interobserver variability. The echocardiographer was blinded to the patients’ diagnoses and RHC data.

Assessment of TR Envelope Quality

To minimize the effect of TR spectral envelope quality on the accuracy of DE, the data were analyzed using only patients with optimal TR envelopes. A TR envelope was deemed “optimal” by the interpreting cardiologist (G.K.S.) when the spectral Doppler envelope edge was clearly identifiable in at least the first three quarters of systole. Optimal envelopes had sharp, rounded edges clearly demonstrating the peak of the TR velocity. The remainder were deemed “suboptimal” and were not included in data analysis. Examples of TR envelope quality are shown in Figure 1 . An independent assessment of TR envelope quality was also done in 20 randomly selected patients by a separate, blinded observer (G.K.G.), and interobserver reliability was assessed using Cohen’s κ method.

TR envelope quality. The TR envelope was deemed “optimal” if the spectral envelope was filled in at least the initial three quarters of systole, with a sharp, rounded edge clearly demonstrating the peak of the TR velocity. (A) Half filled (patient excluded); (B) two-thirds filled (patient excluded); (C) three quarters filled (patient included); (D) full (patient included).

Stratification of Patients on the Basis of RV Pressure

Patients were categorized on the basis of the ratio between RHC-measured RVSP and simultaneously measured systemic systolic blood pressure (SBP). In patients with left heart obstruction, a simultaneous catheter-measured left ventricular systolic pressure was used as the SBP. Two clinically relevant thresholds were used to define RV hypertension and subsequently categorize patients: (1) RVSP ≥ 1/2 SBP and (2) RVSP ≥ 2/3 SBP. Group 1 included patients with RVSP < 1/2 SBP, and group 2 included those with RVSP ≥ 1/2 SBP. Group 3 included patients with RVSP < 2/3 SBP, and group 4 included those with RVSP ≥ 2/3 SBP.

Statistical Analysis

Values for descriptive statistics are reported as mean ± SD for normally distributed data and as median (interquartile range) for data not normally distributed. Correlation coefficients between RHC-derived and DE-derived pressures were calculated using Pearson’s correlation method, and the relationship of the methods was assessed using linear regression. Agreement between methods was assessed using Bland-Altman analysis. Accuracy of Doppler echocardiographic measurements was predefined as 95% limits of agreement within ±10 mm Hg. Fisher’s exact test was used to compare categorical data between groups. Receiver operating characteristic analysis was used to assess the associations between high RVSP (the reference standard) and ΔP _E . The previously used clinically relevant thresholds were used to define RV hypertension: (1) RVSP ≥ 1/2 SBP and (2) RVSP ≥ 2/3 SBP. Binomial exact confidence intervals were calculated for the area under the curve. Youden’s index was used to determine on the receiver operating characteristic curve the point maximizing the difference between the true-positive ratio and the false-positive ratio. Bootstrapped (1,000 replications) confidence intervals were calculated for this point (criterion value). On the basis of the criterion value, sensitivity, specificity, positive predictive value, negative predictive value, and their 95% confidence intervals were calculated. For the calculation of positive and negative predictive values, the ratio of cases in the positive and negative groups was used to represent disease prevalence. P values < .05 were considered statistically significant. Statistical analyses were performed using SPSS version 19 (SPSS, Inc, Chicago, IL), JMP Statistical Software release 10.0.0 (SAS Institute Inc, Cary, NC), and MedCalc Statistics for Biomedical Research version 12.3.0.0 (MedCalc Software, Mariakerke, Belgium).