Indirect evidence for biological mortality bias in growth from two temporo-spatially distant samples of children.

Biological mortality bias in growth is a challenge to the analysis and understanding of past populations. In this analysis, we address two interrelated aspects of the bias: its potential magnitude in terms of linear growth and the association between height and survivorship. A contemporary sample of 292 children, whose recumbent length was measured at autopsy in Cuyahoga County, USA, was used to quantify the magnitude of mortality bias. Differences between survivors and non-survivors were quantified using t-tests and Cohen's d for effect size. While survivors were consistently taller than non-survivors, the difference did not become significant until after 7 years of age. A historical sample of 656 girls, whose height and weight were measured at admission to a tuberculosis sanitarium, was used to examine the relationship between height, weight, and survivorship. The survivors and non-survivors were compared using t-tests and Cohen's d, and odds of survival were modeled with logistic regression. Surviving girls were consistently taller and heavier than non-surviving girls. However, while taller girls were more somewhat more likely to survive, survivorship was more strongly associated with heavier weight at admission. Taken together, these results suggest that while mortality bias in growth may exist, it may not be large enough to impact interpretations of past population growth patterns. It should be noted that this is the case only if mortality bias does not vary significantly between different populations and if it does not significantly affect dental development.