Melatonin concentration and core body temperature (CBT) follow endogenous circadian biological rhythms. In the evening, melatonin level increases and CBT decreases. These changes are involved in the regulation of the sleep-wake cycle. Therefore, the authors hypothesized that age-related changes in these rhythms affect sleep quality in older people. In a cross-sectional study design, 11 older poor-sleeping women (aged 62-72 yrs) and 9 older good-sleeping women (60-82 yrs) were compared with 10 younger good-sleeping women (23-28 yrs). The older groups were matched by age and body mass index. Sleep quality was assessed by the Pittsburgh Sleep Quality Index questionnaire. As an indicator of CBT, oral temperature was measured at 1-h intervals from 17:00 to 24:00 h. At the same time points, saliva samples were collected for determining melatonin levels by enzyme-linked immunosorbent assay (ELISA). The dim light melatonin onset (DLMO), characterizing the onset of melatonin production, was calculated. Evening changes in melatonin and CBT levels were tested by the Friedman test. Group comparisons were performed with independent samples tests. Predictors of sleep-onset latency (SOL) were assessed by regression analysis. Results show that the mean CBT decreased in the evening from 17:00 to 24:00 h in both young women (from 36.57°C to 36.25°C, p < .001) and older women (from 36.58°C to 35.88°C, p < .001), being lowest in the older poor sleepers (p < .05). During the same time period, mean melatonin levels increased in young women (from 16.2 to 54.1 pg/mL, p < .001) and older women (from 10.0 to 23.5 pg/mL, p < .001), being lowest among the older poor sleepers (from 20:00 to 24:00 h, p < .05 vs. young women). Older poor sleepers also showed a smaller increase in melatonin level from 17:00 to 24:00 h than older good sleepers (mean ± SD: 7.0 ± 9.63 pg/mL vs. 15.6 ± 24.1 pg/mL, p = .013). Accordingly, the DLMO occurred at similar times in young (20:10 h) and older (19:57 h) good-sleeping women, but was delayed ∼50 min in older poor-sleeping women (20:47 h). Older poor sleepers showed a shorter phase angle between DLMO and sleep onset, but a longer phase angle between CBT peak and sleep onset than young good sleepers, whereas older good sleepers had intermediate phase angles (insignificant). Regression analysis showed that the DLMO was a significant predictor of SOL in the older women (R(2) = 0.64, p < .001), but not in the younger women. This indicates that melatonin production started later in those older women who needed more time to fall asleep. In conclusion, changes in melatonin level and CBT were intact in older poor sleepers in that evening melatonin increased and CBT decreased. However, poor sleepers showed a weaker evening increase in melatonin level, and their DLMO was delayed compared with good sleepers, suggesting that it is not primarily the absolute level of endogenous melatonin, but rather the timing of the circadian rhythm in evening melatonin secretion that might be related to disturbances in the sleep-wake cycle in older people.
