The different sleep problems of men and women – a comprehensive study of objective sleep, stress, immune activity, sickness absence, morbidity and mortality
Sleep disturbances are associated with a number of diseases and mortality, and women complain more of disturbed sleep than men, and seek treatment more often. The reason for the differences is not clear, but stress is one proposed factor. However, apart from sleep complaints we know very little about the other aspects of sleep and gender, but we would expect to see sleep complaints reflected in objective sleep, as well as in sleep related: immune system activation, morbidity & sickness absence, fatigue, and mortality. Information on these influences is necessary for understanding the pattern of gender differences in sleep and to improve interventions. The project includes an objective sleep study (polysomnography, the “gold standard” of sleep measurement) in 800 men and women (half each), C-reactive protein (indicator of inflammation) from blood samples in the morning after the recording, fatigue ratings at 5 points during the day, a cognitive test and ratings of stress. It also includes the covariation of sleep, stress and sickness absence across 8 years (5000 participants) within genders, and a large (N=43000) epidemiological study of the prospective association between reported sleep problems and later morbidity and mortality –in relation to gender. This is a comprehensive and unique effort to understand gender related differences in sleep. An additional strength is that the same self-report sleep questionnaire is used in all three sub-studies.
Final report
Background and purpose
Previous work has found that women in survey studies complain more of sleep problems than men, despite some objective sleep measures indicating better sleep. This paradoxical observation has been much discussed over the years, but no reasonable explanation has been found. The purpose of the present project was to understand gender differences in sleep and their consequences. In particular we were interested in if physiological sleep measures (polysomnography, measures of EEG, EMG, EOG, labeled “PSG” for short) would be related to reported sleep variables for the same recorded sleep. We were also interested in whether women’s more frequent sleep complaints would link to outcomes like sickness absence over time or to long-term mortality.
The work was divided into three approaches
1) Comparing in a large sample, objective sleep in men and women in relation to self-reported sleep complaints. We also included some newly developed measures of objective sleep, hypothesized to better reflect objective sleep problems.
2) Comparing the association across years between sleep complaints and sickness absence in men and women. Do sleep complaints link to sickness absence across repeated measurements across years, and is the association different in men and women?
3) Comparing the association between sleep duration and mortality in men and women during a 16 year follow-up time. Does mortality differ between men and women for long or short sleep.
Genomförande/Methods
The three research approaches were organized into three indipendent studies.
In “1”, Gender and objective/subjective sleep, 620 men and women participated with one sleep recorded in the home, complemented with ratings of sleep quality in the morning. Men and women were compared (analysis of variance) for each collected variable, with a subdivision into age groups (since sleep tends to differ considerably depending on age). We also analyzed the association between objective sleep and sleep quality ratings and its links to gender. Several new measures were introduced to attempt to identify causes of differences.
In “2”, Gender-related association between reported sleep and sickness absence across years, we analyzed a dataset of c:a 5000 individuals. This focused on 8 years with measurements every two years (5 measurements). It was attempted to predict changes in sickness absence from changes in sleep problems and of stress, depression and other possible conflunders. Mixed model regresson was used for analyses.
In“3”, gender, sleep duration and mortality, a large dataset of 34.000 individuals was used to predict mortality (from the cause of death registry) from a baseline data collection 18 years before the end of follow-up time. Cox regression analysis was applied to the data, and the analysis was controlled for a number of background variables.
The three most important results
1, Gender and objective/subjective sleep: Reported sleep quality was lower in women, particularly in younger age groups. For objective sleep, women had a longer TST (sleep duration) higher sleep efficiency, fewer awakenings, less Stage N1 sleep (superficial sleep), more stage N3% (deep sleep) han men. All indicate better objective sleep in women. The differences were more marked in higher age groups, and to some extent in lower ones. Particularly Stage N3 (deep sleep) showed a dramatic fall with increasing age in men, while normal levels were maintained in women. The only variables with poorer objective sleep in women was sleep latency, waketime per awakening, and subjective/abjective sleep latency (time to sleep onset). Thus, women have a longer time to perceive each awakening, and this is linked to women’s report of more sleep problems than men. We also found that women had few objective awakenings per subjective awakening (2) compared to men (5). Thus, women had a more accurate perception of their objective awakenings than men (or men had a more inaccurate perception). In addition, we analyzed the association between objective and reported sleep quality and found that it was relatively good in both genders, but that women was on a higher level of sleep complaints compared to men. for most variables reported lower quality for each level of objective sleep. That is, women rated sleep with, for example, 5 awakenings as having a much lower quality than did men for the same number of awkakenings. This was true for all variables except for waketime/awakening, for which both gender gave similar ratings. And, importantly, men with high rated sleep quality had a much higher underestimation of their number of awakenings. It was concluded that women has a higher perceptive sensitivity to objective awakenings than men, and that this may be contributing to women’s higher complaints of poor sleep.
2. Gender-related association between reported sleep and sickness absence across years. Both women and men showed a significant association between change in reported sleep quality, sleep duration during days off, and changes in sickness absence, but the effect was stronger in women. The results mean that increased sleep problems and increased sleep duration during days off are associated with increased sickness absence. Possibly, long sleep during days off may be a reflection of poor sleep during working days, or an indication of ongoing disease. Sleep durations at 9h and above, seem particularly critical. Women had a higher probability of sickness absence with increasing sleep duration during days off. Women were also overall more prone to sickness absence than men. It was concluded that women seem more sensitive to poor sleep in relation to increased sickness absence. Probably, interventions to improve women’s sleep may reduce sickness absence.
3, Gender, sleep duration and mortality, Mortality was high for both men and women for long (>8h) sleep durations, it was low for both groups at the most common sleep duration (7h), but women had a significantly higher mortality for short (=5h) sleep duration than did men. The effect of short sleep increased with increasing follow-up time, suggesting an accumulation across time. It was suggested that short sleep seemed a particular problem for women in terms of mortality.
Conclusions
In conclusion, in the present project we have identified possible causes of women’s higher level of sleep complaints – a previously unknown sensitivity of women to awakenings, dependent on a longer time awake per awakening (or that men has a remarkable insensitivity to awakenings). The data does not indicate the reason for this discrepancy, but hormonal differences is one tentative possibility. We also found that increased sleep problems and increased sleep duration during days off are are linked to increased sickness absence over time, particularly in women. Furthermore, short sleep was linked to a higher mortality in women than for men. The results open up for new research approaches to gender differences in sleep.
New research questions
One major new research question is the reason for women’s higher mortality risk with short sleep. When does this occur in time, and what other factors are associated with it?
Another new major research issue is what other factors beyond sleep and stress link to women’s higher sickness absence, for example, family situation.
A third major new research question is follow-up if the waketime/awakening may be the missing link between objective sleep and more sleep complaints in women. This also suggests that men are less able to perceive their sleep quality. The latter may constitute a health problem. New studies of experimental awakenings and detailed monitoring of waketime in relation to awakenings are needed, and links to, for example, hormonal changes should be included. There is also need for studies of time awake/ awakening in relation to ratings of sleep “in general”, as in survey studies or in a clinician’s office. A related question is the implications of the dramatic drop in N3 sleep in men with increasing age. Is this possibly related to health and longevity problems, and are there biological factors that link to this drop, for example, testosterone or IGF-1 (an indicator of secretion of growth hormone during sleep - important for bodily repair processes)?
Spridning av forskningsresultaten
Results are being disseminated in scientific journals, at scientific congresses, and in media. We also plan dissemination via lectures and popular science journals, for exampoe, “Medicinsk Vetenskap” och newspaters (we have discussed the results with a major Swedish newspaper, which will write about the results in about two months’ time / when the last paper has been accepted).
Previous work has found that women in survey studies complain more of sleep problems than men, despite some objective sleep measures indicating better sleep. This paradoxical observation has been much discussed over the years, but no reasonable explanation has been found. The purpose of the present project was to understand gender differences in sleep and their consequences. In particular we were interested in if physiological sleep measures (polysomnography, measures of EEG, EMG, EOG, labeled “PSG” for short) would be related to reported sleep variables for the same recorded sleep. We were also interested in whether women’s more frequent sleep complaints would link to outcomes like sickness absence over time or to long-term mortality.
The work was divided into three approaches
1) Comparing in a large sample, objective sleep in men and women in relation to self-reported sleep complaints. We also included some newly developed measures of objective sleep, hypothesized to better reflect objective sleep problems.
2) Comparing the association across years between sleep complaints and sickness absence in men and women. Do sleep complaints link to sickness absence across repeated measurements across years, and is the association different in men and women?
3) Comparing the association between sleep duration and mortality in men and women during a 16 year follow-up time. Does mortality differ between men and women for long or short sleep.
Genomförande/Methods
The three research approaches were organized into three indipendent studies.
In “1”, Gender and objective/subjective sleep, 620 men and women participated with one sleep recorded in the home, complemented with ratings of sleep quality in the morning. Men and women were compared (analysis of variance) for each collected variable, with a subdivision into age groups (since sleep tends to differ considerably depending on age). We also analyzed the association between objective sleep and sleep quality ratings and its links to gender. Several new measures were introduced to attempt to identify causes of differences.
In “2”, Gender-related association between reported sleep and sickness absence across years, we analyzed a dataset of c:a 5000 individuals. This focused on 8 years with measurements every two years (5 measurements). It was attempted to predict changes in sickness absence from changes in sleep problems and of stress, depression and other possible conflunders. Mixed model regresson was used for analyses.
In“3”, gender, sleep duration and mortality, a large dataset of 34.000 individuals was used to predict mortality (from the cause of death registry) from a baseline data collection 18 years before the end of follow-up time. Cox regression analysis was applied to the data, and the analysis was controlled for a number of background variables.
The three most important results
1, Gender and objective/subjective sleep: Reported sleep quality was lower in women, particularly in younger age groups. For objective sleep, women had a longer TST (sleep duration) higher sleep efficiency, fewer awakenings, less Stage N1 sleep (superficial sleep), more stage N3% (deep sleep) han men. All indicate better objective sleep in women. The differences were more marked in higher age groups, and to some extent in lower ones. Particularly Stage N3 (deep sleep) showed a dramatic fall with increasing age in men, while normal levels were maintained in women. The only variables with poorer objective sleep in women was sleep latency, waketime per awakening, and subjective/abjective sleep latency (time to sleep onset). Thus, women have a longer time to perceive each awakening, and this is linked to women’s report of more sleep problems than men. We also found that women had few objective awakenings per subjective awakening (2) compared to men (5). Thus, women had a more accurate perception of their objective awakenings than men (or men had a more inaccurate perception). In addition, we analyzed the association between objective and reported sleep quality and found that it was relatively good in both genders, but that women was on a higher level of sleep complaints compared to men. for most variables reported lower quality for each level of objective sleep. That is, women rated sleep with, for example, 5 awakenings as having a much lower quality than did men for the same number of awkakenings. This was true for all variables except for waketime/awakening, for which both gender gave similar ratings. And, importantly, men with high rated sleep quality had a much higher underestimation of their number of awakenings. It was concluded that women has a higher perceptive sensitivity to objective awakenings than men, and that this may be contributing to women’s higher complaints of poor sleep.
2. Gender-related association between reported sleep and sickness absence across years. Both women and men showed a significant association between change in reported sleep quality, sleep duration during days off, and changes in sickness absence, but the effect was stronger in women. The results mean that increased sleep problems and increased sleep duration during days off are associated with increased sickness absence. Possibly, long sleep during days off may be a reflection of poor sleep during working days, or an indication of ongoing disease. Sleep durations at 9h and above, seem particularly critical. Women had a higher probability of sickness absence with increasing sleep duration during days off. Women were also overall more prone to sickness absence than men. It was concluded that women seem more sensitive to poor sleep in relation to increased sickness absence. Probably, interventions to improve women’s sleep may reduce sickness absence.
3, Gender, sleep duration and mortality, Mortality was high for both men and women for long (>8h) sleep durations, it was low for both groups at the most common sleep duration (7h), but women had a significantly higher mortality for short (=5h) sleep duration than did men. The effect of short sleep increased with increasing follow-up time, suggesting an accumulation across time. It was suggested that short sleep seemed a particular problem for women in terms of mortality.
Conclusions
In conclusion, in the present project we have identified possible causes of women’s higher level of sleep complaints – a previously unknown sensitivity of women to awakenings, dependent on a longer time awake per awakening (or that men has a remarkable insensitivity to awakenings). The data does not indicate the reason for this discrepancy, but hormonal differences is one tentative possibility. We also found that increased sleep problems and increased sleep duration during days off are are linked to increased sickness absence over time, particularly in women. Furthermore, short sleep was linked to a higher mortality in women than for men. The results open up for new research approaches to gender differences in sleep.
New research questions
One major new research question is the reason for women’s higher mortality risk with short sleep. When does this occur in time, and what other factors are associated with it?
Another new major research issue is what other factors beyond sleep and stress link to women’s higher sickness absence, for example, family situation.
A third major new research question is follow-up if the waketime/awakening may be the missing link between objective sleep and more sleep complaints in women. This also suggests that men are less able to perceive their sleep quality. The latter may constitute a health problem. New studies of experimental awakenings and detailed monitoring of waketime in relation to awakenings are needed, and links to, for example, hormonal changes should be included. There is also need for studies of time awake/ awakening in relation to ratings of sleep “in general”, as in survey studies or in a clinician’s office. A related question is the implications of the dramatic drop in N3 sleep in men with increasing age. Is this possibly related to health and longevity problems, and are there biological factors that link to this drop, for example, testosterone or IGF-1 (an indicator of secretion of growth hormone during sleep - important for bodily repair processes)?
Spridning av forskningsresultaten
Results are being disseminated in scientific journals, at scientific congresses, and in media. We also plan dissemination via lectures and popular science journals, for exampoe, “Medicinsk Vetenskap” och newspaters (we have discussed the results with a major Swedish newspaper, which will write about the results in about two months’ time / when the last paper has been accepted).