Standards for Predicting Basal Metabolism

Abstract

The compiling of a statistically significant number of metabolism measurements on girls from 17 through 20 was undertaken at the University of Wisconsin for the purpose of deciding upon a valid standard of normality for such subjects. This small but important sex and age group represents a gap relatively unexplored by the older data upon which prediction standards still in general use were based. Diverse methods used for tentatively bridging this gap gave rise to predictions which are contradictory and confusing in practice. Basal heat production was determined in 163 tests on 97 girls whose ages were approximately uniformly distributed over the 4 years from 17 through 20. These were University students who were classed as Grade A in their medical and physical examinations and who were judged by further special examination and questioning at the time of their tests to be free from indisposition or defects that should disqualify them from serving as physiological normals. The results of the tests were subjected to various types of mathematical analysis and the first accepted tests on the 97 subjects were made the basis of a prediction standard expressing the equation: Basal heat production in calories per 24 hours = 10.63×weight in kg.+3.23×stature in cm.+184.61. The equation was arrived at by standard methods of multiple correlation. Comparisons by various criteria were made of the “fit” of this standard and that of the Aub-DuBois, the Harris-Benedict prediction for adult women, extrapolated for these ages, and of Benedict's special prediction for girls from 12 through 20, when each was applied 1.—to the data of the Wisconsin series; and 2.—to a comparably large series of normal controls of similar ages collected from different observers and widely different localities. The latter included data on 77 young women studied in nongoiter-belt regions of the south. The fit of the different standards was similar for the two series of controls. As they stand the Aub-DuBois prediction averaged from 8 to 15 per cent and the Harris-Benedict from 6 to 10 per cent too high, while the Benedict averaged from 1 to 8.5 per cent too low to describe the measured metabolism of the various sub-groups of controls tested. Tests of fitness more fundamental than mere success in predicting average absolute level of heat-production for groups—i.e., standard errors of estimating the rates, percentiles falling within ±10 per cent and ±15 per cent of the respective predictions, and extreme ranges of calculated rates, were applied after the older standards had been put on a comparable basis with the Wisconsin standard by applying constant percentage corrections to center them for the Wisconsin data. Finally a special set of comparisons was made for the 38 individuals of the two test series who were atypical in proportion of weight to height. The results of the various comparisons indicated little choice in accuracy of predicting for either group of controls by the Wisconsin prediction, the corrected Harris-Benedict, or the corrected Aub-DuBois, though by all the criteria taken together, they arranged themselves in the order named. The Benedict prediction compared with the other three showed excessive scatter of individual rates about their mean. According to this standard, but not judged by the others, the underweight girls of these series showed very different types of metabolic rates than the overweight girls. The older standards should be modified only after many groups of data, covering all ages, can be compared to show the ultimate corrections needed to make them fit the largest possible number of cases. It is hoped that the present study can contribute toward an eventual agreement in this respect. Meanwhile, since the Wisconsin prediction has proved to be valid for these considerable numbers of normal girls examined with comparable technic in both the north and south of America, it is offered for its immediate interest in either clinical or physiological comparisons.