Predictors of body surface area

To determine the accuracy of body surface area (BSA) calculations proposed in the literature and their applicability to patient populations such as neonates and parturients.

Using 395 actual BSA measurements in the literature, 15 prediction formulas were assessed for accuracy using the root mean squared error (RMSE) method of prediction. Height/weight distributions of infants and parturients, and the relationship of DuBois and DuBois predictions to actual BSA, were compared using scatter plots. Percentage errors across different body sizes were determined.

Obstetrics clinic and labor and delivery rooms at the University of Chicago Medical Center.

Sixty women (gestational week 34 to 40) and 148 neonates.

We measured the height and weight of the neonates and the women. We also used the height, weight, and BSA of 395 subjects reported in the literature. Although the commonly used DuBois and DuBois formula was derived from only 10 subjects, our statistical analysis demonstrates that it can be used over a wide range of measured BSAs (RMSE = 6.3%) and patients, including both infants and pregnant women. As BSA increases, so does the absolute prediction error, but the percentage error is greatest for infants, for whom the formula tends to underestimate BSA. There were no other significant differences in predictive accuracy for gender, age, or body habitus.

Several BSA formulas, including the DuBois and DuBois formulas adequately predict measured BSA over a wide range of patient populations. Although the original subjects studied by Dubois and DuBois excluded extremes of height and weight, their formula appears to be a valid predictor.