Sleep in Older Adults and Mortality

By Dr. Gregg D. Jacobs

A recent study in the journal Psychosomatic Medicine (65:63-73, 2003) examined the relationship between sleep and mortality in older adults. This study was conducted by Dr. Amanda Dew and her colleagues at the University of Pittsburgh School of Medicine. The authors examined the sleep EEG records of 185 adults, ages 60-80, who had previously been enrolled in one of eight research protocols in which all-night EEG sleep studies had been conducted. These older adults had no history of mental illness, sleep complaints (including insomnia), or current cognitive impairment. The researchers conducted follow-up assessment about 13 years after the original sleep studies had been conducted to determine whether patients were still alive or had died. They found that 118 individuals were still alive at follow-up while 66 were positively ascertained as deceased.

After controlling for age, gender, and baseline medical problems, the researchers found that certain baseline sleep parameters positively predicted increased risk of dying during the follow-up period. These included prolonged sleep latencies (time to fall asleep) of greater than 30 minutes, sleep efficiencies (time asleep divided by time in bed) less than 80%, and high or low percentages of REM (dream) sleep (low percentages of REM sleep may suggest excessive in light sleep while high REM percentages may indicate depression).

The authors concluded that these specific sleep EEG alterations are associated with increased risk of dying beyond those associated with age, gender, and medical problems. They also suggest that interventions to prevent these alterations in the sleep of older adults may be warranted.

When interpreting the results of this study and the authors' conclusions, it is important to note that sleep durations of either greater or less than six hours per night were not associated with mortality. Additionally, these were adults who did not complain of sleep problems in daily life (e.g., they were considered normal sleepers in daily life despite the fact that they showed sleep alterations in the sleep lab).

As the authors point out, these findings may suggest that the sleep EEG alterations that were associated with increased mortality may represent early signs of brain alterations or organic brain damage such as Alzheimer's disease that ultimately predict death. Another possibility is that the individuals who exhibited alterations such as prolonged sleep latencies (despite being normal sleepers outside the sleep lab) may be individuals who were the greatest "stress reactors" in the laboratory environment since stress consistently prolongs sleep latencies. Because stress has been repeatedly associated with increased risk of morbidity and mortality, it is possible that abnormal stress responses in the laboratory, and not sleep alteration per se, may be associated with increased risk of death.

Before recommendations should be made about interventions to prevent sleep alterations in normal older sleepers who only show evidence of certain abnormal sleep parameters in a sleep lab, future research must address the question of whether it is truly specific sleep alteration per se, or other possible mechanisms such as those discussed above that truly predict increased risk of death. The fact that the individuals in this study did not complain of sleep problems, and that there was no association between sleep duration and mortality, casts significant doubt on the likelihood that sleep disturbance itself truly predicts mortality.