Objectives: This study was designed to determine the most appropriate method to normalize left ventricular mass for body size.

Background: Left ventricular mass has been normalized for body weight, surface area or height in experimental and clinical studies, but it is uncertain which of these approaches is most appropriate.

Methods: Three normotensive population samples--in New York City (127 adults), Naples, Italy (114 adults) and Cincinnati, Ohio (444 infants to young adults)--were studied by echocardiography. Relations of left ventricular mass to body size were similar in all normal weight groups, as assessed by linear and nonlinear regression analysis, and results were pooled (n = 611).

Results: Left ventricular mass was related to body weight to the first power (r = 0.88), to body surface area to the 1.5 power (r = 0.88) and to height to the 2.7 power (r = 0.84), consistent with expected allometric (growth) relations between variables with linear (height), second-power (body surface area) and volumetric (left ventricular mass and body weight) dimensions. Strong residual relations of left ventricular mass/body surface area to body surface area (r = 0.54) and of ventricular mass/height to height (r = 0.72) were markedly reduced by normalization of ventricular mass for height2.7 and body surface area1.5. The variability among subjects of ventricular mass was also reduced (p < 0.01 to p < 0.002) by normalization for body weight, body surface area, body surface area1.5 or height2.7 but not for height. In 20% of adults who were overweight, ventricular mass was 14% higher (p < 0.001) than ideal mass predicted from observed height and ideal weight; this increase was identified as 14% by left ventricular mass/height2.7 and 9% by ventricular mass/height, whereas indexation for body surface area, body surface area1.5 and body weight erroneously identified left ventricular mass as reduced in overweight adults.

Conclusions: Normalizations of left ventricular mass for height or body surface area introduce artifactual relations of indexed ventricular mass to body size and errors in estimating the impact of overweight. These problems are avoided and variability among normal subjects is reduced by using left ventricular mass/height2.7. Simple nomograms of the normal relation between height and left ventricular mass allow detection of ventricular hypertrophy in children and adults.