Plasma melatonin, core body temperature, and sleep efficiency before and during transdermal melatonin administration during daytime. Data in (a) represent hourly means ± 1 SEM (N=8 for melatonin and body temperature; N=7 for sleep efficiency). Application and removal of the dermal patch are indicated by dashed vertical lines. Note that elevated melatonin levels prior to patch application are due to endogenous secretion. Horizontal black bars on top and bottom indicate scheduled sleep opportunities. Gray bars indicate scheduled wake episodes in ∼1.8 lux. Subjects maintained a semi-recumbent position (upper part of bed at 45°) for the 6 h before and for the hours after the 8-h daytime sleep opportunity to allow for measurements of body temperature undisturbed by changes in body position. During the sleep episodes subjects were supine. Note that at 20:00 h and 21:00 h in the placebo condition blood samples could only be obtained from 6 subjects. Mixed-model ANOVA on log-transformed hourly melatonin values following patch application was significant for Condition, Time, and Condition × Time (P<0.0001 in all cases). ANOVA on temperature data was significant for Condition and Time (P<0.0001), but not for their interaction (P=0.99). Black diamonds indicate significant differences from placebo [P<0.05, t-tests (melatonin, body temperature) or Wilcoxon signed rank test (sleep efficiency) for planned comparisons]. (b) Individual melatonin profiles in males (top) and females (bottom). Note that one male subject with very low exogenous plasma melatonin levels also had very low endogenous melatonin levels.

Effect of transdermal melatonin administration on EEG power density during NREM sleep during daytime. Top: for both melatonin and placebo, power densities in the last 6 h of time in bed were expressed as percentage of first 2 h. Data represent means ± 1 SEM (N=7). Filled (melatonin) and open (placebo) diamonds above abscissa indicate frequency bins for which difference from the first 2 h was significant (P<0.05, paired t-tests on log-transformed values). Power density values and significance are plotted at upper limit of 0.5-Hz frequency bins. Two-way mixed-model ANOVA on log-transformed power densities was significant for Condition, and EEG Frequency bin (P<0.0001), but not for their interaction (P=0.23). Bottom: power densities in last 6 h of melatonin condition were expressed relative to corresponding interval of placebo. Filled triangles refer to significant differences from placebo (P<0.05, t-tests), and gray triangle to a statistical tendency (P=0.058).