Changes in liver volume from birth to adulthood: a meta-analysis.

A diversity of equations is available for the estimation of liver volume (LV), derived from studies in populations of ethnically homogeneous individuals and using a variety of methods of measurement. The aim of this study was to integrate all published pediatric data and to define a general equation for estimating LV from birth onward. Data were collated from 5,036 subjects (birth to 18 yr old). Equations were developed based on simple regression against body surface area (BSA) and multiple regression of LV with weight, height, BSA, age, gender, race, methodology, and year of publication as covariates. These equations, together with those reported in the literature, were compared for accuracy of prediction of LV from birth to 18 yr old. The most parsimonious equation to describe LV was selected according to the Akaike information criteria (AIC), precision and bias and following visual inspection of residual errors and observed vs. predicted plots: LV = 0.722 * BSA(1.176). The multiple regression models indicated that Japanese have up to 19% larger livers compared to Caucasians for a given body weight. Radiographic and ultrasonic measurements were associated with up to 8% lower estimates of liver size compared to measurements made at autopsy. There was no evidence that gender or the year in which a study was published (1933-1999) influenced the estimation of LV. The general equation was also applied to predict adult LV, and its precision and accuracy was found to be superior to those of 10/11 published adult models. In conclusion, we have developed a more general model to predict LV in pediatric populations and young adults, and have investigated a range of covariates.