Some investigators have separately analyzed alcohol’s effects during the first and second half of the nighttime sleep period. These studies found that particularly at higher alcohol doses, increased wake periods or light stage 1 sleep periods occurred during the second half of the sleep period (Williams et al. 1983; Roehrs et al. 1991). This second-half disruption of sleep continuity is generally interpreted as a “rebound effect” once alcohol has been completely metabolized and eliminated from the body. The term “rebound effect” means that certain physiological variables (e.g., sleep variables, such as the amount of REM sleep) change in the opposite direction to the changes induced by alcohol and even exceed normal levels once alcohol is eliminated from the body. This effect results from the body’s adjustment to the presence of alcohol during the first half of the sleep period in an effort to maintain a normal sleep pattern. Once alcohol is eliminated from the body, however, these adjustments result in sleep disruption. This hypothesis is supported by the known rate of alcohol metabolism, which leads to a decrease in BrAC of 0.01 to 0.02 per-cent per hour. Given that in such experiments, the typical peak BrACs measured shortly before sleep are 0.06 to 0.08 percent, alcohol metabolism at this rate would be completed within 4 to 5 hours of sleep onset; thus, the sleep disruption during the second half of the night would coincide with the clearance of alcohol from the body.

In addition to these effects on sleep initiation and sleep maintenance, researchers have found that alcohol consistently affects the proportions of the various sleep stages. Thus, most studies have reported a dose-dependent suppression of REM sleep at least during the first half of the sleep period (Williams and Salamy 1972). As noted earlier, the amount of REM sleep time is lower during the first half of the night relative to the second half of the night; consequently, the full REM-suppressive effect of alcohol is probably underestimated in most studies. To determine alcohol’s full effect on REM sleep, investigators would need to administer an additional alcohol dose in the middle of the night, thereby causing alcohol’s peak concentrations to coincide with the majority of REM sleep time. No such studies have been conducted, however.

Those studies that have demonstrated alcohol-induced REM suppression during the first half of the sleep period also have frequently found an REM rebound (i.e., longer-than-normal REM periods) during the second half of the night (Williams and Salamy 1972). As a result, the overall amount of REM sleep in subjects receiving alcohol before sleeping did not differ from that in subjects receiving a nonalcoholic drink (i.e., a placebo). As with the increased periods of wakefulness or light sleep, the REM rebound during the second half of the night is associated with the completed alcohol metabolism and elimination from the body. The neurobiological mechanisms responsible for the rebound of either wakefulness or REM sleep are still unknown.

Some studies also found an alcohol-related increase in the amount of SWS (i.e., stages 3 and 4 NREM sleep) in the first half of the sleep period (Williams and Salamy 1972). In addition to the alcohol dose consumed, the basal (i.e., normal) level of SWS in the study population appeared to be the most likely factor determining whether SWS was increased. For example, in a study of insomniacs who had lower amounts of SWS than did healthy people when taking a placebo--a typical finding in insomniacs--SWS increased when they consumed alcohol (Roehrs et al. 1999). Conversely, alcohol did not affect SWS in a group of age-matched healthy control subjects.

Another population that typically shows lower levels of SWS compared with healthy young adults are the elderly, but no studies have assessed alcohol’s effects on the sleep of healthy elderly people. In sleep deprivation studies, however, elderly participants show increases in SWS on the recovery night after the sleep-deprivation period; possibly alcohol could similarly promote SWS in elderly people. This finding does not imply, however, that alcohol should be considered a potential sleep therapy in elderly people, because tolerance to the SWS enhancement develops rapidly (Prinz et al. 1980).

Several studies have assessed the effects of alcohol administration over several nights. Such studies clearly demonstrated that tolerance to alcohol’s sedative and sleep-stage effects develops within 3 nights (Williams and Salamy 1972) and that the percentages of SWS and REM sleep return to basal levels after that time. Furthermore, in some studies, the discontinuation of nightly alcohol administration resulted in a REM sleep rebound--that is, an increase in REM sleep beyond basal levels (Williams and Salamy 1972). However, not all studies found such a rebound effect. This variability in results may be related to several factors specific for each study, including the basal level of REM sleep in the participants, the degree of alcohol-related REM suppression, the extent of prior tolerance to REM suppression, and the dose and duration of alcohol administration.