0241 Evaluation of Four Computational Methods to Estimate Dim Light Melatonin Onset in Pediatric Populations

Abstract Introduction Dim light melatonin onset (DLMO) is the gold standard circadian phase marker in humans. There are several methods that can be employed to estimate DLMO from salivary or plasma melatonin levels. These methods include utilizing fixed thresholds (3 pg/ml, 4 pg/ml), individual thresholds (2 SD above baseline), and the hockey stick method. The extant literature suggests that DLMO estimates in adults and adolescents are influenced by DLMO quantification methods. To date, no studies have compared these methods for calculating DLMO in pediatric populations. Given that measuring DLMO in children presents unique methodological challenges, research is needed to compare agreement across calculation methods. Methods Salivary melatonin was collected from 49 healthy children aged 3.3 – 6.0 years and retrospectively analyzed. Children maintained a consistent sleep/wake schedule for 7 days. On the evening of the 8th day, children provided saliva samples every 30 min until 1 h past habitual bedtime under dim light conditions (< 10 photopic lux). Saliva was collected by having each child mouth/chew on absorbent cotton swabs. DLMO was compared across four methods of calculation: 4 pg/ml fixed threshold, 3 pg/ml fixed threshold, individual thresholds calculated as 2 SD above baseline levels, and the hockey stick method. Results Intraclass correlations among the four methods were all significant (p < 0.001), with the strongest agreement observed between the 3 pg/ml, 4 pg/ml, and hockey stick methods. DLMO values from all methods were positively correlated with bedtime, sleep onset, midsleep, and sleep offset (p < 0.001). Although the 2 SD method was more discrepant from the other three estimation methods on average, it did yield a more plausible DLMO estimate in one individual identified as a high secretor. Conclusion The 3 pg/ml, 4 pg/ml, and hockey stick methods provide reliable estimates of DLMO in young children, whereas the 2 SD method may yield a more accurate DLMO estimation in edge cases of high or low melatonin secretors. More research is needed to generalize these overall findings to other pediatric populations, including those with known variability in melatonin output. Support (if any) Support for data collection came from F32-HD103390; R01-HD087707; T32-HL149646.