The ability to determine body size-independent data for linear left ventricular dimensions is important to facilitate group comparisons and construct appropriate normative data. Recently, the adoption of non-linear scaling has been supported both theoretically and empirically (George et al. 2001). However, potential inaccuracies in echocardiography and anthropometry have not been addressed. Therefore this study employed MRI as this has become the ‘gold-standard’ technique for both cardiac and body composition measurements. It is more accurate and reliable than standard echocardiography and has been validated against dissection and chemical analysis (Fowler et al. 1992).

After local ethical approval data were collected in 172 young adult male military recruits (age range 17Ð28 years) who were free of known cardiovascular disease. Multiple ECG-gated transverse slices through the heart were obtained. Maximal LV septal (LVSWT) and posterior wall thickness (LVPWT) as well as LV internal dimension at end-diastole (LVIDd) were obtained at the level of the mitral valve by digitisation. Forty transaxial spin echo image slices (TE40, TR500, field of view 45 X 45 cm, slice thickness 10 mm) were obtained of the whole body. Adipose tissue and thus fat mass was measured by an automated technique and was then subdivided into various fat compartments (e.g. subcutaneous fat mass). This allowed the calculation of fat-free mass (FFM). Height (HT), body mass (BM) and body surface area (BSA) were determined anthropometrically. LogÐlog least-squares linear regression analyses were performed to determine the slope exponents (b) for all relationships between linear dimensions and body size/composition variables. All data are means ± 95 % confidence intervals.

Geometric consistency was confirmed for LVSWT-FFM (b = 0.29 ± 0.07), LVPWT-FFM (b = 0.27 ± 0.09) and LVSWT-BM (b = 0.27 ± 0.07), whereas LVIDd-FFM (b = 0.19 ± 0.09), LVPWT-BM (b = 0.24 ± 0.08) and LVIDd-BM (b = 0.22 ± 0.07) did not quite reach the geometrically consistent value of 0.33. Exponents for HT were not geometrically consistent (1.0) and demonstrated greater variability (b = 0.03 ± 0.03 to 0.34 ± 0.24). Consequently, exponents for BSA were not geometrically consistent. Exponents for all indices of adiposity were small but positive (b = 0.03 ± 0.02 to 0.07 ± 0.02).

This study lends support to previous echocardiographic data that FFM^0.33 can be used in scaling procedures to determine a body size-independent index for LV wall thickness. The use of HT^1.0 and BSA^0.5 for scaling of LV linear dimensions is not supported by the current data. These data should be substantiated by further research.

S. Myerson was supported by a British Heart Foundation Fellowship.

View Table 1 in new window

All procedures accord with current local guidelines.