Our results suggest a strong, graded association between increased level of depressive symptoms and poorer self-reported sleep quality in community-dwelling older women, a finding that is consistent with previously published reports (10
). We also found significant associations between greater levels of depressive symptoms and measures of sleep fragmentation (increased WASO and number of long wake episodes). Paudel and colleagues recently preformed a similar analysis of actigraphic sleep in a cohort of community-dwelling older men. The authors found a modest association between more depressive symptoms and increased sleep latency but no significant associations between level of depressive symptoms and objective measures of sleep fragmentation (17
). It is possible that this difference is related to the populations being tested since study population assessed by Paudel and colleagues was male and was younger (mean age 76.4 +/− 5.5 years) than the women in the SOF study.
After adjusting for multiple potential confounders, our objective assessment did not show any association between level of depressive symptoms and TST, SE, or SOL in this group. This is in contrast to other studies which have found objective evidence of decreased TST and SE as well as increased SOL in depressed patients by both subjective assessment (34
) and PSG (35
). This difference may be related to the method of sleep disturbance assessment. Further, those studies focused on individuals who either had a diagnosis of Major Depressive Disorder or who met DSM-IV criteria for a Major Depressive Episode whereas our analysis focuses on level of subjectively reported depressive symptoms. Alternatively, it may be that these specific sleep disturbances are less prominent in elderly depressed women compared to other groups of depressed individuals. A recent study comparing actigraphic sleep measures in elderly men and women found elderly men to have shorter and more fragmented sleep when compared to elderly women (37
). To our knowledge gender differences have not been examined in depressed older adults.
Objective sleep disturbances accounted for very little of the association between increased level of depressive symptoms and increased subjective sleep disturbance in this population. There are several possible explanations for this observation. Additional factors that were not assessed in this analysis may also play a role in the increased subjective perception of poor sleep quality. It is also possible that, in older community-dwelling women, women with and without depressive symptoms experience and report their sleep problems differently. Paudel and colleagues’ analysis in a cohort of community-dwelling older men found a similar discrepancy between the association of depressive symptoms with subjective as compared to objective sleep disturbances (17
). In particular, older men with depressive symptoms were much more likely to report poor sleep quality compared to non-depressed counterparts while depression was more modestly associated with objectively measured sleep disturbance. To our knowledge, there are no other studies evaluating the association between depressive symptoms and both subjectively and objectively assessed sleep disturbances specifically in older adults.
Several studies comparing subjective reports of sleep quality using the PSQI with objective sleep measurements by PSG in healthy older adults have found that, in that population, subjective reports of sleep quality were not as poor as might be expected considering the objective sleep disturbances that were observed by PSG. This suggests that healthy older adults may under-report their sleep disturbances or adapt their perception of acceptable sleep quality compared with younger adults (11
). The data presented here suggest that older women with depressive symptoms do not share this tendency to under-report their problems with sleep. Indeed, it has been previously shown that among depressed subjects, older adults are more likely to underestimate their total sleep when compared to younger adults when subjective assessments are compared to PSG (39
In this population, women with depressive symptoms were almost twice as likely to report excessive subjective daytime sleepiness compared with those in the normal category. Several studies in non-elderly populations have shown a similar relationship between depression and excessive daytime sleepiness (40
). Objective measurements also revealed an association between more depressive symptoms and daytime inactivity raising the possibility that there may be an increase in napping in women with more depressive symptoms. Although actigraphy is not able to distinguish definitively between quiet restfulness and sleep, previous studies have found similar associations between depressive symptoms and self-reported napping (42
). Further, Goldman and colleagues recently showed an association between objectively measured increased fragmentation of sleep and increased subjective daytime napping in older adults (43
). Studies employing PSG will be required to clarify this relationship.
The estimated point prevalence of major depressive disorder (MDD) in older adults has been reported to range from 3–10 % (44
) while “less than major” depressive syndromes (including dysthymia, minor depression and subsyndromal depression) are nearly twice as common with a reported prevalence of 9–24% (46
). The women in the SOF study did not undergo any formal psychiatric evaluation to determine diagnosis. However, the prevalence of differing levels of depressive symptoms based on GDS score in this cohort suggests a similar pattern. Given that subsyndromal depressive syndromes make up the majority of depressive syndromes in older adults, it is interesting that having “some depressive symptoms” conferred a similar increased risk for subjectively assessed poor sleep quality and excessive daytime sleepiness compared with being in the “depressed” group. Although there were significant trends suggesting graded associations between increasing levels of depressive symptoms and objective measures of sleep fragmentation in multivariable adjusted models, specific associations between having “some depressive symptoms” and objective measures of sleep fragmentation observed in age and site adjusted models were no longer significant in multivariable adjusted models. Future studies designed to better explore the relationship between subsyndromal depression and objective sleep disturbances would be useful to further investigate this relationship.
Our analyses add to the growing body of literature suggesting that both subjective sleep disturbance and excessive daytime sleepiness are common in older adults. More than half of the women in this study endorsed subjectively poor sleep (PSQI >5) and more than 10% of the women endorsed excessive daytime sleepiness. This is similar to what has been reported in other similar populations including the National Sleep Foundation’s Sleep in America survey in which more than 50% of respondents reported a sleep complaint occurring nearly every night. Similarly, in the Osteoporotic Fractures in Men study, the prevalence of subjective poor sleep (PSQI >5) in a large cohort of community dwelling older men was reported to be 44.2% while the prevalence of excessive daytime sleepiness (ESS >10) was12.7%. Subtle differences in the age and other covariates (e.g. burden of medical illness) of these different samples may explain small differences in the prevalence rates of these subjectively reported problems.
This study has several strengths. The sample size is large and is collected from four separate United States locations and the women were not selected based on depressive symptoms or on their report of sleep disturbances. Unlike most previous studies examining the relationship between depression and sleep disturbances in older adults, both subjective and objective measures of sleep disturbance and daytime sleep were assessed.
There are also several limitations to this analysis. The study is cross-sectional and it is therefore not possible to draw any conclusions about causality. The sample is made up only of community dwelling women, the vast majority of whom were white and over 80 years old, and the findings may not be applicable to other populations. Depressive symptoms were assessed by questionnaire only rather than a clinical diagnostic interview, so conclusions about psychiatric diagnosis cannot be made. Actigraphy relies on self-reported times in and out of bed which may be inaccurate and could introduce errors into those measurements such as sleep onset latency. However, these errors would not affect WASO and number of long wake episodes at night which we found to be associated with depressive symptoms. Additionally, actigraphy cannot definitively distinguish between daytime sleep and periods of quiet restfulness. Finally, the impact of adding multiple covariates to the many of the multivariable adjusted models underscores the complexity of the relationship between these covariates, sleep, and depression in this population. Our analyses were not designed to fully explore the role these covariates play in the relationship between depressive symptoms and sleep disturbances. Future studies designed specifically to address this issue would be useful, particularly since some of these covariates may be amenable to intervention.
In conclusion, these analyses add to the body of literature supporting an association between depressive symptoms and subjective sleep disturbance in older adults. Further, these data suggest an association between more depressive symptoms and fragmentation of sleep as well as increased daytime napping in older women. Longitudinal analyses will now be required to determine whether objectively measured sleep disturbances predict changes in depression over time.