Predicting heart failure hospitalization and mortality by quantitative echocardiography: is body surface area the indexing method of choice? The Heart and Soul Study

Abstract

Background: Echocardiographic measurements of left ventricular (LV) mass, left atrial (LA) volume, and LV end-systolic volume (ESV) predict heart failure (HF) hospitalization and mortality. Indexing measurements by body size is thought to establish limits of normality among individuals varying in body habitus. The American Society of Echocardiography recommends dividing measurements by body surface area (BSA), but others have advocated alternative indexing methods.

Methods: Echocardiographic measurements were collected in 1024 ambulatory adults with coronary artery disease. LV mass, LA volume, and LV ESV were calculated using truncated ellipse method and biplane method of disk formulae. Comparison between raw measurements and measurements divided by indexing parameters was made by hazard ratios per standard deviation increase in variable and c-statistics for BSA, BSA(0.43), BSA(1.5), height, height(0.25), height(2), height(2.7), body weight (BW), BW(0.26), body mass index (BMI), and BMI(0.27).

Results: Mean LV mass was 192 +/- 57 g, mean LA volume was 65 +/- 24 mL, and mean LV ESV was 41 +/- 26 mL. Average height was 171 +/- 9 cm, average BSA was 1.94 +/- 0.22 m(2), and average BMI was 28.4 +/- 5.3 kg/m(2). At an average follow-up of 5.6 +/- 1.8 years, there were 148 HF hospitalizations, 71 cardiovascular (CV) deaths, and 269 all-cause deaths. There was excellent correlation between raw measurements and those indexed by height (r = 0.98-0.99), and moderate correlation between raw measurements and those indexed by BW (r = 0.73-0.94). C-statistics and hazard ratios per standard deviation increase in indexed variables were similar for HF hospitalization, CV mortality, and all-cause mortality. There were no significant differences among indexing methods in ability to predict outcomes.

Conclusion: The choice of indexing method by parameters of BSA, height, BW, and BMI does not affect the clinical usefulness of LV mass, LA volume, and LV ESV in predicting HF hospitalization, CV mortality, or all-cause mortality among ambulatory adults with coronary artery disease. Continued use of BSA to index measurements of LV mass, LA volume, and LV ESV is acceptable.

Figure 1

Radial plot of c-statistics, or area under the receiver operating characteristic curve, for standard echocardiographic variables in predicting HF hospitalization. Further distance from the center of the plot indicates higher predictive ability. LVEF, Left ventricular ejection fraction (%); LV ESV, left ventricular end-systolic volume (mL); EDV, end-diastolic volume (mL); LAV, left atrial volume (mL); RAV, right atrial volume (mL); EDPRv, end-diastolic pulmonary regurgitation velocity (m/s); TRv, TR velocity (m/s); MRv, mitral regurgitation velocity (m/s); PV s/d VTI, pulmonary vein systolic to diastolic flow ratio by velocity time integral (dimensionless ratio); IVRT, isovolumic relaxation time (ms).

Figure 2

HRs per standard deviation (as shown in Table 3) increase in variables indexed to different parameters of body size in the entire sample (n = 1,024). A circular configuration of the radial plot indicates equivalence among indexing methods. LV, Left ventricular; BSA, body surface area; h, height; bw, body weight; BMI, body mass index. The numeric values used to create these plots are listed in Table 4.

Figure 3

C-statistics, or area under the receiver operating characteristic curve, for HF hospitalization and mortality by different methods of indexing for LV mass LA volume, and LV ESV in the entire sample (n = 1024). A circular configuration of the radial plot indicates equivalence among indexing methods. LV, Left ventricular; BSA, body surface area; h, height; bw, body weight; BMI, body mass index. The numeric values used to create these plots are listed in Table 4.

Figure 4

Percentage of HF hospitalizations, CV deaths, or all-cause deaths correctly classified by the highest quartile of LV mass indexed to BSA or height^2.7 with results limited to men. The cutoff at the highest quartile indexing was 114 g/m² for BSA and 52 g/m^2.7 for height^2.7.