Abstract

Quantifying melatonin, specifically dim-light melatonin onset (DLMO), in pediatric populations allows us to better understand the implications of circadian rhythm on health. However, current melatonin metrics were developed for adult populations and melatonin collection is typically time and resource intensive. The first goal of this thesis was to identify melatonin metrics traditionally used in pediatric literature and to examine alternative metrics. Data extracted from twenty-two articles identified via scoping review revealed that the most common DLMO metrics focused on a single melatonin dynamic: timing (i.e., the time melatonin concentration surpasses a predetermined threshold). Thus, alternative metrics to quantify timing, concentration, and rate dynamics, such as total melatonin output and rate derived by linear regression, were proposed, and applied to a pediatric secondary dataset. Traditional and alternative timing (ravg=.86) and concentration (r=.93) metrics where highly correlated, while rate metrics were moderately correlated (r=.38). The second goal was to explore potential proxy variables to estimate DLMO. This was achieved by augmenting the secondary dataset using multiple imputation, then by conducting a Classification and Regression Tree analysis to estimate DLMO dynamics. Sleep midpoint and bedtime measured by actigraphy were some of the best proxies for the timing dynamic (R2=.91) but not for concentration (R2=.50) and rate (R2=.41). Thus, other proxies to estimate concentration and rate dynamics, such as temperature, diet, or race, should be explored. Results of this thesis highlight the lack of melatonin metrics for youth populations in the literature and the need to better quantify melatonin and circadian rhythms.