ABSTRACT
Determination of circadian phase is required to diagnose and treat circadian abnormalities, but the measurement of dim light melatonin onset (DLMO), the most common phase marker, is laborious. As sleep timing reflects circadian phase, measurement of sleep markers (e.g., sleep onset, sleep midpoint, sleep offset) provides a simple way to estimate DLMO. The study aim was to compare methods to estimate DLMO from markers derived from the Pittsburgh Sleep Quality Index (PSQI), Munich Chronotype Questionnaire (MCTQ), sleep diaries, and actigraphy. PSQI, MCTQ, and 1 week of diary and actigraphy data were collected from 72 (36 f, 36 m) healthy adults aged 23.1 (± 3.6) y prior to a laboratory sleep study. Saliva samples were collected hourly in dim light during the second evening of the study. The sleep markers most strongly associated with DLMO from each source were PSQI onset, MCTQ average midpoint, 7-d diary midpoint, and 7-d actigraphy midpoint. Estimates of DLMO as a fixed interval before the sleep marker exhibited proportional bias. DLMO estimated from regression models based on sleep midpoint from 7 d of diary or 7 d of actigraphy showed the narrowest limits of agreement with measured DLMO without proportional bias (±1.8 h and ±1.9 h, respectively). Our findings indicate none of the methods provided precise estimates of DLMO from sleep markers. The best estimates were from linear regressions on sleep midpoints from 7 d of diary or actigraphy, and these estimates of DLMO may be suitable for limited research purposes.