Trends in sleep duration
Thirty-six-year secular trends in sleep duration and sleep satisfaction, and associations with mental stress and socioeconomic factors – results of the Population Study of Women in Gothenburg, Sweden
Cecilia Björkelund, Department of Primary Health Care, Box 454, SE-40530 Göteborg, Sweden. Tel.: +46 31 7866832; fax: +46 31 778 1 704; e-mail: email@example.com
Several European studies have reported sleeping problems in 20–40% of the population. We used data from the Population Study of Women in Gothenburg, based on medical examinations of three different representative cohorts of 38- and 50-year-old women in 1968–1969, 1980–1981 and 2004–2005 to study secular trends in sleep-related factors. The average reported sleep duration declined by about 15 min in the 38-year-old women during the 36 years of observation. No corresponding change in sleep duration was observed among 50-year-old women. During the same period, the proportion of women complaining of sleeping problems almost doubled in both age groups: from 17.7% in 1968 to 31.7% in 2004 in 38-year-old women, and from 21.6% to 41.8% in 50-year-old women. The prevalence of insomnia was higher in 50-year olds than in 38-year olds in all investigated cohorts. The use of sleeping medications remained unchanged since 1968. There were significant associations between perceived sleeping problems and reported lower satisfaction concerning economic, social and family situations, as well as with medical retirement and mental stress. There was, however, no association between alcohol consumption and sleeping problems. Regular leisure time physical activity was not, in most cases, associated with less perceived sleeping problems. Our study indicates that the physician should take socio-economic and family situations into consideration when examining female patients complaining of sleeping problems. Improvements on society level rather than on the individual level could be expected to be more efficient in improving women’s sleep.
Several surveys in western Europe have reported sleeping disorders in 20–40% of the general population. Women and the elderly are the most affected (Arber et al., 2009; Chen et al., 2005; Doi et al., 2000; Ohayon and Lemoine, 2004). Women have slightly longer sleep duration than men (Adams, 2006; Mallon and Hetta, 1997; Ohayon and Lemoine, 2004), but complaints of insomnia are more frequent in women (Chen et al., 2005; Groeger et al., 2004; Li et al., 2002; Mallon and Hetta, 1997; Ohayon and Lemoine, 2004; Walsh and Uestuen, 1999). The average sleep duration among US citizens has decreased for more than a generation. The median sleep duration in adults aged 40–79 years was 8 h per night in 1959, with 15% reporting less than 7 h per night (Hammond, 1964). In 2002 the median sleep duration had declined to 7 h, with more than 33% reporting less than 7 h of sleep. Several studies indicate that the mean sleep duration among adults is about 7 h (Groeger et al., 2004; National Sleep Foundation, 2003).
Quality of life is positively associated with longer sleep duration, up to 9 h; however, a negative association has been found with sleep duration exceeding 9 h (Groeger et al., 2004). Never-married individuals of both sexes reported less sleep disturbance than married and divorced individuals (Chen et al., 2005), while other studies report that unmarried, divorced, or widowed women and men tend to have shorter sleep and higher levels of sleeping problems than married people (Groeger et al., 2004; Stranges et al., 2008). The unemployed and medical retirees seem to have more sleeping problems than people who are working (Chen et al., 2005; Sekine et al., 2006). Sleeping complaints have been frequently associated with low socioeconomic status (Arber et al., 2009; Geroldi et al., 1996; Hunt et al., 1985; Hyyppäet al., 1997). Arber et al. (2009) found that gender differences concerning sleep problems were reduced following adjustment for socioeconomic characteristics.
Regular exercise improves sleep (Chen et al., 2005). Factors associated with poor sleep are smoking (Lexcen and Hicks, 1993) and high stress burden (Kim et al., 1999; Walsh and Uestuen, 1999). In the study by Lexcen as well the study by Kim, there was no association between alcohol intake and sleep disorders.
Use of sleeping medication increases with age; from 3.2% in individuals below age 45 years to 22% at ages 65–74 years, and to almost one-third of those aged 75 years or older in French population data (Ohayon and Lemoine, 2004). However, a study of Swedes aged 65–79 years reported that only 7.6% of women and 3.0% of men regularly used sleeping medications, with some geographic variation (Mallon and Hetta, 1997).
The purpose of this study was to describe secular trends in sleep duration, sleeping problems and sleep satisfaction in two age groups in a Swedish female population during a 36-year period, as well as to study associations between sleep and socioeconomic factors, lifestyle and quality of life factors.
Materials and methods
The Population Study of Women in Gothenburg, 1968–2004
In 1968–1969, a representative sample of 50- and 38-year-old women living in Gothenburg, Sweden, were invited to a free medical examination in the Prospective Population Study of Women in Gothenburg (Bengtsson et al., 1973). A total of 372 women aged 38 years and 398 women aged 50 years participated (91% participation rate in both cohorts). Analysis of survival rates among participants versus non-sampled women born in the same years revealed no major differences, providing additional evidence that participants were representative of the general population from which they were selected (Bengtsson et al., 1989).
In 1980–1981, a follow-up examination was conducted of the now 50-year-old women (n = 355; 82% of the women originally examined) (Bengtsson et al., 1989). A new age group, aged 38 years (n = 122; 85% participation rate) in 1980–1981 was recruited, based on date of birth. Women aged 38 years at the initial examination, who moved to Gothenburg between 1969 and 1980 and who had the same date of birth as the original cohort were invited to the 1980–1981 examination, which implied that also the 1980–1981 sample was representative of the female population in Gothenburg in 1980–1981.
In 2004–2005 a new study of 38- and 50-year-old women was conducted, and 343 women aged 38 years and 503 women aged 50 years, born in 1966 and 1954, living in Gothenburg were invited to the free medical examination. A total of 500 women (207 aged 38 years and 293 aged 50 years) accepted the invitation (60% and 58% respectively) (Björkelund et al., 2008).
Social and lifestyle-related variables
Participants were asked about past and/or current sleeping problems (‘Do you have or have had sleeping problems?’ and ‘Do you currently have sleeping problems?’). Beginning from the 1980–1981 examination, participants were also specifically asked about sleeping problems during the last 3 months (Have you had sleeping problems during the last 3 months?). Test of agreement between women’s reported sleeping problems during the last 3 months and the same women’s response to problems currently and/or in the past revealed 88% agreement (kappa test).
The women were asked for how many hours they slept during one 24-h period (including nap).
Self-evaluation of sleep quality
In the Gothenburg Quality of Life Instrument (GQOLI), introduced in the 1980 examinations, participants rated their overall sleep quality on a Likert scale. The GQOLI was developed in the early 1970s to measure symptoms and well-being in the population (Sullivan et al., 1993). The instrument was used in the 1980–1981 and 2004–2005 examinations. Subjects responded on a seven-point Likert-type scale, with 1 representing the best (‘excellent, couldn’t be better’, followed by 2 = ‘very good’, 3 = ‘good’, 4 = ‘neither good nor bad’, 5 = ‘not good really’, 6 = ‘poor’ and 7 = ‘very poor’). We defined scores 1–3 as good, 4–5 as average and 6–7 as poor. Respondents reported their current satisfaction with their situation (Indicate how you experience your situation now). The women were asked to complete the questionnaire at home before the examination.
Self-evaluation of economic situation
As part of the GQOLI questionnaire, participants were asked to self-evaluate their current economic situation (Indicate how you experience your situation now) on a seven-point Likert-type scale described above.
Self-evaluation of social and friendship situation
As part of the GQOLI questionnaire, participants were asked to self-evaluate their current social and friendship situation (Indicate how you experience your situation now) on a seven-point Likert scale described above.
Self-evaluation of home/family situation
As part of the GQOLI questionnaire, participants were asked to self-evaluate their current home/family situation (Indicate how you experience your situation now) on a seven-point Likert scale described above.
The women were asked whether they used sleeping medications or not (never or almost never/some times per month or more).
Subjects were asked whether they had retired for medical reasons.
Mental stress is defined as one or several, more than a month long, periods of anxiety, anxiousness, irritability, nervousness, tension or insomnia due to worries regarding work, own health, family or conflicts at home or at work during the last 5 years.
Subjects were asked if they had consumed beer, wine or spirits some days per week, some days per month or not at all during the last year.
Leisure time exercise
Subjects were classified as being physically active during their leisure time if they reported usually spending more than 4 h a week gardening, running, dancing, playing golf, tennis or being involved in similar activities during the last year.
Analyses of different variables were performed using parametric tests when normal distribution was assumed and non-parametric tests for non-normal distributions. Correlation was tested with Pearson’s correlation test. Logistic regression analyses were performed to compute odds ratios (OR) with 95% confidence intervals (CI). For continuous variables, a linear regression model was applied. Tests for trends were performed with linear-by-linear rank correlation tests. Differences were considered statistically significant at P < 0.05. The kappa test was used to test agreement: 0.81–1.0 was regarded as a very high degree of agreement.
Secular trends in sleep duration
The average sleep duration in women aged 38 years decreased over time from 7.32 h in 1968 to 7.24 h in 1980 and 7.08 h in 2004 (P = 0.008; Table 1). The proportion of women sleeping 5 h or less was almost unchanged, but the proportion sleeping 9 h or more decreased significantly from 11.3% in 1968 to 3.5% in 2004. Furthermore, the percentage sleeping 8 h was lower in 2004 (30.7%) than in 1968 (36.6%), while the percentage sleeping 7 h was higher in 2004 (42.6% versus 31.7%). The proportion sleeping 6 h was higher in 2004 than in 1968 (Table 1).
Table 1. Sleep duration, sleeping problems and use of sleeping pills in 38- and 50-year-old women in 1968–1969, 1980–1981 and 2004–2005 (%)
| ≥9 h||11.3||8.2||3.5|| ||7.5||9.0||4.6|| |
| 8 h||36.6||32.8||30.7|| ||25.6||30.4||26.6|| |
| 7 h||31.7||39.3||42.6|| ||40.5||36.3||43.6|| |
| 6 h||15.3||14.8||18.8|| ||17.1||14.9||16.3|| |
| ≤5 h||5.1||4.9||4.5|| ||9.3||9.3||8.8|| |
| ||100%||100%||100%|| ||100%||100%||100%|| |
| Mean (h)||7.32||7.24||7.08||0.008||7.02||7.12||7.00||0.90|
|Sleeping problems Have or|
| have had before||17.7||21.5||31.7||0.0003||21.6||31.8||41.8||<0.000|
| Still have||11.6||13.2||15.8||0.33||16.1||20.3||23.7||0.11|
| Last 3 months|| ||17.2||36.5||0.005|| ||34.7||44.8||0.049|
| Use of sleeping pills||8.1||6.6||5.4||0.27||10.8||15.3||11.4||0.72|
In contrast to the 38-year-old women, there was no significant change in average sleep duration in the 50-year-old women; i.e. 7.02 h in 1968, 7.12 h in 1980 and 7.00 h in 2004 (P = 0.90).
The proportion of women reporting sleeping problems last 3 months was significantly higher in both age groups in 2004–2005, compared with that in 1980–1981. Among the 38-year-old women, 31.7% reported having or have had sleeping problems in 2004, compared with 17.7% in 1968 (Table 1). In the case of 50-year-old women, the proportion of having or have had sleeping problems was 41.8% versus 21.6% in 2004 and 1968 respectively (Table 1).
Overall, there was no significant change in the use of sleeping medications between 1968 and 2004 (Table 1). The use of sleeping pills was higher in women aged 50 years than in those aged 38 years at all examinations. The OR for total use of sleeping pills among 50-year-old women, compared with women aged 38 years, was 1.73 (CI 1.25–2.37, P = 0.0008).
Sleep satisfaction according to GQOLI
There was no statistically significant difference between 1980 and 2004 concerning the proportion of 50-year-old women reporting satisfactory sleep (sleep satisfaction scores 1–3; GQOLI), while the proportion of 38-year-old women who reported higher sleep satisfaction scores in 2004 was significantly lower than in 1980, also when sleep duration (<7, 7–8 and >8 h per 24 h) was included in the model (Table 2). Only 38.1% of 38-year-old women sleeping <7 h per day reported sleep satisfaction scores 1–3 in 2004, compared with 66.7% in 1980 (Table 2). Of the group sleeping 7–8 h, 71.9% reported sleep satisfaction scores 1–3 in 2004, more corresponding to the proportion in 1980 (84.1%). However, separate comparisons for each age group and within each group of women sleeping <7, 7–8 and >8 h per 24 h, respectively, showed no statistically significant differences between 1980–1981 and 2004–2005, although there was a close to significant difference in 38-year olds sleeping <7 h between 1980–1981 and 2004–2005 as well as those sleeping 7–8 h. Table 2 also shows that the proportion of all 38-year-old women who stated average sleep quality (scores 4–5) almost doubled in 2004 compared with that in 1968, and the major shift was thus from good to average sleep quality.
Table 2. Evaluation of sleep quality (GQOLI, 1–3 good, 4–5 average, 6–7 poor) in women sleeping <7, 7–8 and >8 h per 24 h in 1980–1981 and 2004–2005 respectively
|Evaluation of sleep (GQOLI)|
|<7 h|| 24 (19)|| || 47 (23)|| || || 86 (24)|| || 71 (25)|| || |
| Good|| ||66.7|| ||38.1|| || ||25.6|| ||34.3|| |
| Average|| ||16.7|| ||33.3|| || ||44.2|| ||42.9|| |
| Poor|| ||16.7|| ||28.6||0.058|| ||30.2|| ||22.9||0.18|
|7–8 h|| 88 (72)|| ||148 (73)|| || ||237 (67)|| ||198 (70)|| || |
| Good|| ||84.1|| ||71.9|| || ||75.1|| ||67.2|| |
| Average|| ||13.6|| ||24.4|| || ||21.0|| ||29.6|| |
| Poor|| ||2.3|| ||3.7||0.053|| ||3.9|| ||3.2||0.17|
|>8 h|| 10 (8)||(8.2)|| 7 (4)|| || || 32 (9)|| || 13 (5)|| || |
| Good|| ||70.0|| ||42.9|| || ||71.9|| ||61.5|| |
| Average|| ||30.0|| ||57.1|| || ||18.8|| ||23.1|| |
| Poor|| 0|| || ||0||0.18|| ||9.4|| ||15.4||0.47|
|Total||122|| ||202|| || ||355|| ||282|| || |
| Good|| ||79.5|| ||62.7|| || ||62.7|| ||57.8|| |
| Average|| ||15.6|| ||28.1|| || ||26.5|| ||32.7|| |
| Poor|| ||4.9|| ||9.2||0.004*|| ||10.8|| ||9.5||0.521†|
Associations between reported sleeping problems and quality of life variables
Potential associations between sleeping problems and quality-of-life variables were analysed by testing the association between GQOLI variables ‘economic situation’, ‘social situation and friendship’ and ‘home/family situation’ (good, average and poor), and ‘sleeping problems last 3 months’ (confirmation or denial). There was a statistically significant association between perceived sleep problems during the last 3 months and GQOLI scores for ‘economic situation’ in both age groups. This was also seen between perceived sleep problems and ‘social situation – friendship’ and ‘home/family situation’ in almost all of both age groups (1980–1981 and 2004–2005 samples; Table 3). There was a pattern of increasing percentage confirming sleeping problems across worsening categories of economic, social and family situation in all observed age groups and examinations, indicating a general, strong association between socio-economic status and women’s sleep (Table 3).
Table 3. Associations between sleeping problems in the last 3 months and economic situation, social situation–friendship and home/family situation (GQOLI, 1–3: good; 4–5: average; 6–7: poor) respectively
|Sleeping problems in last 3 months by economic situation GQOLI|
| Good situation||11 (13.8)||1.0||41 (32.3)||1.0||85 (30.5)||1.0||85 (41.3)||1.0|
| Average situation||6 (17.6)||1.3 (0.5–4.0)||27 (43.5)||1.6 (0.9–3.0)||29 (50.9)||2.4 (1.3–4.2)||30 (48.4)||1.3 (0.8–2.4)|
| Poor situation||4 (50.0)||6.3 (1.4–29)||6 (50.0)||2.1 (0.6–6.9)||5 (55.6)||2.9 (0.7–11)||12 (63.2)||2.4 (0.9–6.5)|
|Sleeping problems in last 3 months by social situation–friendship GQOLI|
| Good situation||14 (13.9)||1.0||54 (33.8)||1.0||87 (29.9)||1.0||104 (41.8)||1.0|
| Average situation||6 (37.5)||3.7 (1.2–12)||15 (42.9)||1.5 (0.7–3.1)||25 (56.8)||3.1 (1.6–5.9)||23 (62.1)||2.3 (1.1–4.5)|
| Poor situation||1 (20.0)||1.6 (0.2–15)||5 (83.3)||9.8 (1.1–86)||7 (70.0)||5.5 (1.4–21.7)||3 (75.0)||4.2 (0.4–40.8)|
|Sleeping problems in last 3 months by home/family situation GQOLI|
| Good situation||14 (13.7)||1.0||52 (31.9)||1.0||93 (31.9)||1.0||95 (41.3)||1.0|
| Average situation||4 (25.0)||2.1 (0.6–7.4)||16 (53.3)||2.4 (1.1–5.4)||21 (48.8)||2.0 (1.1–3.9)||30 (58.8)||2.0 (1.1–3.8)|
| Poor situation||3 (75.0)||18.9 (1.8–194)||5 (83.3)||10.7 (1.2–93.7)||4 (50.0)||2.1 (0.5–8.7)||4 (57.1)||1.9 (0.4–8.7)|
Association between reported sleeping problems, medical retirement and lifestyle
In the case of medical retirees, both 38- and 50-year-old women had a significantly higher frequency of sleeping problems than not medically retired women in 2004, as did 50-year-old women in 1980 (Table 4).
Table 4. Associations between sleeping problems and medical retirement, mental stress, physical activity and consumption of beer, wine and spirits in the 1968–1969, 1980–1981 and 2004–2005 samples
|Sleeping problems by medical retirement|
| Medical retirement||0 (0)||1 (100)||6 (75.0)||4 (40.0)||16 (60.0)||23 (71.9)|
| Working||43 (12.2)||20 (20.8)||68 (29.9)||60 (20.9)||104 (29.6)||107 (38.0)|
| OR (CI) for sleeping problems associated with medical retirement||_||14.9 (0.58–377.6)||5.6 (1.1–28.5)||3.64 (0.99–13.3)||3.7 (1.6–8.6)||3.6 (1.6–8.1)|
|Sleeping problems by mental stress|
| No/limited stress last 5 years||12 (6.2)||4 (5.6)||4 (9.3)||23 (11.3)||51 (22.8)||9 (17.3)|
| One or more month-long periods of stress last 5 years||53 (33.5)||22 (44.0)||55 (36.4)||61 (34.4)||60 (48.0)||107 (46.5)|
| OR (CI) for sleeping problems associated with stress||7.7 (3.9–15.1)||13.2 (4.2–42.7)||5.6 (1.9–16.5)||4.1 (2.4–7.0)||3.13 (1.9–5.2)||4.16 (1.85–9.6)|
|Sleeping problems by beer drinking|
| Never||20 (20.0)||8 (33.3)||19 (34.6)||28 (19.9)||29 (30.9)||24 (32.9)|
| Not in last year||1 (100.0)||4 (28.6)||6 (50.0)||0 (0)||25 (41.0)||10 (71.4)|
| Some days per month||27 (16.3)||11 (18.3)||34 (27.4)||34 (23.0)||46 (29.1)||80 (44.9)|
| Some days per week||18 (17.1)||3 (13.0)||1 (16.7)||24 (22.2)||11 (30.6)||2 (13.3)|
| OR (CI) for sleeping problems associated with more frequent beer drinking||0.92 (0.73–1.16)||0.66 (0.43–1.02)||0.81 (0.59–1.12)||1.06 (0.87–1.29)||0.94 (0.75–1.19)||1.07 (0.83–1.38)|
|Sleeping problems by wine drinking|
| Never||35 (20.6)||5 (50.0)||6 (23.1)||38 (20.1)||10 (23.8)||10 (47.6)|
| Not in last year||1 (100.0)||7 (43.8)||3 (60.0)||0 (0)||22 (31.4)||4 (50.0)|
| Some days per month||25 (13.7)||12 (13.6)||44 (32.1)||40 (22.0)||71 (33.2)||79 (43.2)|
| Some days per week||5 (29.4)||2 (28.6)||8 (26.7)||7 (28.0)||8 (34.8)||24 (34.3)|
| OR (CI) for sleeping problems associated with more frequent wine drinking||0.87 (0.68–1.12)||0.44 (0.24–0.79)||1.05 (0.73–1.51)||1.09 (0.87–1.36)||1.19 (0.88–1.60)||0.81 (0.59–1.11)|
|Sleeping problems by spirits drinking|
| Never||44 (15.9)||5 (26.3)||27 (37.5)||58 (20.6)||23 (32.9)||41 (37.6)|
| Not in last year||1 (100.0)||5 (21.7)||7 (36.8)||0 (0)||26 (30.2)||15 (51.7)|
| Some days per month||20 (22.2)||16 (20.5)||27 (26.0)||27 (24.1)||62 (32.6)||59 (42.8)|
| Some days per week||1 (33.3)||0 (0)||0 (0)||0 (0)||0 (0)||2 (66.7)|
| OR (CI) for sleeping problems associated with more frequent sprits drinking||1.25 (0.94–1.66)||0.84 (0.49–1.46)||0.75 (0.54–1.03)||1.08 (0.84–1.40)||0.98 (0.74–1.29)||0.35 (0.88–1.45)|
|Sleeping problems by physical activity last year|
| None/low activity||41 (12.9)||17 (18.3)||45 (39.0)||50 (14.9)||98 (36.7)||88 (50.9)|
| Regular activity||2 (5.1)||4 (14.3)||26 (31.4)||14 (23.7)||22 (28.6)||40 (37.2)|
| OR (CI) for sleeping problems associated with regular physical activity||0.54 (0.30–0.99)||0.83 (0.45–1.54)||0.80 (0.57–1.19)||1.22 (0.77–1.94)||0.78 (0.58–1.05)||0.62 (0.45–0.85)|
There was a significant association between perceived mental stress and perceived poor sleep. Women reporting one or more month-long periods of mental stress during the last 5 years reported sleeping problems three times as often; these associations were significant in both 38- and 50-year-old women and in all samples (Table 4).
There was no association between sleeping problems and consumption of beer, wine or spirits (Table 4).
Regular physical activity was associated with less sleeping problems in only two of six analyses (Table 4).
Our study shows that the prevalence of sleeping problems among women was almost twice as high in 2004 as in 1968, and that sleeping problems were reported more often in the 50-year-old cohort than in the 38-year-old age group at all examinations. The mean sleep duration was unchanged in 50-year-old women between 1968 and 2004, while the 38-year-old women slept 0.24 h less in 2004 than in 1968. In both age groups, women with a sleep duration of 7 h or more were generally satisfied with their sleep, while women sleeping less than 7 h generally expressed dissatisfaction. Furthermore, we found strong associations between several socioeconomic factors, as well as mental stress and sleeping problems.
In the original 1968–1969 study, participation rates among women who were invited to the free medical examination were over 90%, resulting in a sample of 843 individuals, 38 and 50 years of age in the present study who can be considered representative of middle-aged women in Gothenburg at that time. In 1980–1981, the participation rate was over 80%, while in 2004–2005 participation was lower than at earlier examinations (60% in 38-year olds and 58% in 50-year olds). Analysis of non-participants in the different examinations showed that participants were somewhat healthier than non-participants (Bengtsson et al., 1973; Björkelund et al., 2008), implying that differences between cohorts probably would have been even more pronounced if the participation rate in the 2004–2005 examination had been higher.
Comparisons with other studies
This study is one of a few in the world that have examined secular trends in sleeping problems in different age cohorts over such a long time span. In contrast to a British study, which found that sleep duration had not significantly changed between reported 1967 and 2003 (Groeger et al., 2004; Tune, 1969), our study showed that sleep duration in 38-year-old women has declined since 1968. During the same time period, the proportion of women reporting sleeping problems has increased. However, it is interesting that the proportion of 50-year-old women reporting sleeping problems also increased over time despite no change in mean duration of sleep. Notwithstanding, an increase in the use of sleeping medication could not be observed between 1968–1969 and 2004–2005 in either age group.
The use of sleeping pills seems to differ between, as well as within, countries; reported use of sleeping medications in Gothenburg exceeded that in the city of Uppsala, Sweden (Mallon and Hetta, 1997). The use of sleeping medications increases significantly with age. Our study registered a 5% versus 11% usage of sleeping pills for ages 38 and 50 years, respectively, in 2004. The study from the Uppsala region showed 8% usage among women aged 65–79 years. The reason for this difference might be that the Uppsala study was not population based.
In contrast to geographic differences concerning the use of sleeping pills, the average sleep duration in our study (7.1 versus 7.0 h) is in line with multiple European and American surveys indicating an average of 7 h per night (Groeger et al., 2004; Hyyppäet al., 1997; Kripke et al., 2002; Li et al., 2002; Walsh and Uestuen, 1999).
An interesting finding in our study was sleep satisfaction related to different sleep durations. Among those sleeping 7–8 h, a current general recommendation, the majority was satisfied. No significant changes in sleep satisfaction could be distinguished in the groups of different sleep duration, from 1980 to 2004, possibly depending on small group numbers, but a significantly higher rate of women aged 38 years reported having better sleep quality, regardless of sleep duration, in 1980 compared with 2004.
An important finding was how economic status, mental stress and lifestyle are related to women’s reports of sleeping problems. In our study, the association between sleeping problems and perceived poor economic, family and social situation as well as mental stress was robust and across all examinations and both age groups, with increasing share of sleeping problems with perceived poorer situation. According to Arber et al. (2009), socioeconomic inequalities explain a major part of gender-related differences in sleep problems. Improved socioeconomic conditions in women have been proposed to be one factor that could improve women’s sleep. Studies have shown that divorce and separation led to deterioration in sleep in married women (Arber et al., 2009; Li et al., 2002; National Sleep Foundation, 2003), but only slighter or no difference concerning sleeping problems and economic situation (National Sleep Foundation, 2003).
In our study as well as in others, medical retirees had poorer sleep quality (National Sleep Foundation, 2003; Heslop et al., 2002). One explanation might be that people not working generally have a more burdensome social situation and a higher prevalence of depression.
Like other studies (Hyyppäet al., 1997; Kim et al., 1999), our study found no association between consuming alcohol and sleep. Regular exercise was at least partly associated with good sleep, which is also in line with other studies (Chen et al., 2005).
Strengths and limitations of the study
This is one of few cohort comparison studies on women’s sleep, with a relatively large number of participants at each examination. The questions were, to a large extent, similar at each examination in order to increase accuracy as much as possible. Initial participation rates were high, and documentation in the Population Study of Women in Gothenburg regarding sleep habits and lifestyle factors has been thorough. Thus, our study has revealed several statistically significant associations not only between sleep and lifestyle, but also in changes in sleep patterns over time. However, this study is limited to women, and patterns and associations are not necessarily applicable to men. In addition, data concerning sleep were based on the women’s own perceptions stated in questionnaires and doctor’s interview. Types of sleep medication were not fully documented directly in connection with the sleep interview. Finally, as mentioned earlier, participation rates differed over time in the different cohorts studied. However, decreasing participation in later examinations would probably tend to result in underestimated secular increases in sleep problems, rather than the reverse.
Despite these limitations, several conclusions may be drawn about trends over time, based on this cohort comparison. We could reliably conclude that middle-aged women well before menopause have shorter average sleeping time and perceive more sleeping problems than in earlier generations. In our study, the average sleep duration in 38-year-old women declined by about 15 min during the last 36 years. During the same period, the proportion of women complaining of sleeping problems almost doubled, but the use of sleeping medications remained unchanged since 1968. Middle-aged women of today generally seem to perceive more sleeping problems than earlier generations, which could for instance be explained by increased participation in working life and at the same time still taking major responsibilities for family and household work. In the Population Study of Women, working outside the home in these age groups increased from 55% in 1968 to 90% in 2004 (Björkelund et al., 2008).
The association between perceived sleeping problems and reported poor perceived economic, social and family situations, as well as with medical retirement across all examinations and both age groups, with increasing share of sleeping problems with perceived poorer situation is a confirmation of findings from several other population studies and should be an important issue in the context of improvements of women’s health. Short sleep duration and perceived sleep problems are both symptoms and causal factors of ill health. Changes in societal and work-related matters that improve women’s socioeconomic and family situations could be expected to be very health promotive for middle-aged women. As there was no apparent association between alcohol consumption and sleeping problems or regular leisure time physical activity, improvements on society level rather than on the individual level could be expected to be more efficient in improving women’s sleep.
This work was supported by grants from the Swedish Research Council and the Swedish Council for Working Life and Social Research (EpiLife, WISH), the Bank of Sweden Tercentary Foundation, and the Sahlgrenska Academy, University of Gothenburg. We thank the staff at the Department of Primary Health Care, Sahlgrenska Academy, the University of Gothenburg, for helping us by contacting our subjects and scheduling examinations.