A comparison of four methods to estimate dim light melatonin onset: a repeatability and agreement study.

Dim light melatonin onset (DLMO) is considered the most reliable circadian phase marker in humans. However, the methods to calculate it are diverse, which limits the comparability between studies. Given the key role of DLMO to diagnose circadian rhythm sleep-wake disorders and determine the optimal timing of chronotherapies, the establishment of clear and validated guidelines on the methodology to assess DLMO is very important. We performed a repeatability study (n = 31) and an agreement study (n = 62) in healthy young adults with hourly blood samples collected under dim light conditions (<8 lux) during a chronobiological protocol. We assessed the repeatability of DLMO with three different methods (fixed threshold, dynamic threshold and hockey stick) across two nights and assessed agreement of each method with the mean visual estimation made by four chronobiologists. Analyses included Bland-Altman diagrams, intraclass correlation coefficients and equivalence tests. The repeatability of the four methods across two nights ranged from good to perfect. The agreement study highlighted that the hockey stick showed equivalent or superior performance (ICC: 0.95, mean difference with visual estimation: 5 min) in healthy subjects compared to the dynamic and fixed thresholds. Thanks to its objective nature, the hockey stick method may provide better estimates than the mean of the visual estimations of several raters. These findings suggest that the hockey stick method provides the most reliable estimate of DLMO within the tested methods and should be considered for use in future studies.