Our aim was to study the prevalence and the socio-economic differences in self-reported insomnia and stress over the years 1979–2002 in Finland. Compared to the first study period, 1978–1982, there was increase in the prevalence of stress until the period 1993–1997 among men and women. There was also an increase in the prevalence of self-reported insomnia among both men and women during the last study period, 1998–2002. Consistently more insomnia and stress was among the unemployed and retired (early retirees in this data). Lowest education was associated with more insomnia especially among men, and less stress among both sexes. Those in the intermediate levels of household income had least stress. Income level differences in insomnia were less consistent. Socio-economic differences slightly fluctuated over the total period 1979–2002; however, there were no substantial changes in socio-economic differences in insomnia and stress.
This study provided new information about trends in self-reported insomnia and stress by socio-economic factors utilising a repeated cross-sectional study design on a 24-year time scale. It benefited from the use of nationally representative population survey data, which were supplemented with reliable educational level and household income data from Statistics Finland. As a limitation of the survey, the data included only non-specific single-item measures of insomnia and stress, which may cover a wide range of psychological symptoms from transient to severe, chronic symptoms. We have pointed out in this article that these measures are not used for any diagnostic purposes but instead to represent the subjective dimension of mental well-being which can be easily used to monitor differences in population subgroups . However, as was demonstrated in our prior study, single-item measures of insomnia and extremely high stress had significant associations with cause-specific mortality, such as coronary heart disease mortality and so-called unnatural mortality (including alcohol-related mortality, accidents, violence and suicide) .
Some other methodological issues need to be addressed concerning the variables used. In this study we used ‘more stress than in people general’ and ‘my life is nearly unbearable’ categories combined as indicator for stress. Educational differences were reversed in stress; those in the highest education were more stressed compared to the less educated. This finding is in line with many other previous research, however, the phenomenon is not clearly explained in the literature [24, 35]. It has been proposed, especially related to work stress, that those with higher education have for example gained occupational position with greater responsibilities, higher expectations and higher stress . One explanation is relating to the complexity of the stress construct and measurement. In general, our stress measure is exploring only perception of stress, with no further information about sources of stress, or possible coping resources or outcomes. In additional analyses we used ‘my life is nearly unbearable’ category alone as an indicator for extremely high stress (2.5% in the exposed group). Extremely high stress was more common in the lower levels of education compared to the highest education, although the associations were not independent of other socio-economic factors. Our results indicate that even though stress seems to be more common among the highest educated, some of the most extreme stressful situations may be experienced among those in the lowest education. Specific measures of stress would be needed in order to examine different sources and exposures of stress, duration, as well as coping resources and responses. Stress, and less consistently insomnia, also produced an u-shaped distribution with household income level, suggesting intermediate levels being protective for insomnia and stress. More stress and insomnia in the lowest income group may relate, for example, to low social participation and material resources, whereas more symptoms among the highest income group might relate to the factors associated with the higher social position and occupational status, as was discussed with education.
Psychological symptoms are known to be associated with each other with complex interrelations. In our study, Pearson’s correlations (p < 0.001) between insomnia and depression were r = 0.37 in males and r = 0.34 in females; and between stress and depression r = 0.40 in males and r = 0.38 in females. We made adjustment for self-reported depression in order to control the possible effect of depression on the associations for insomnia and stress. Most significant effect on following this adjustment was for retired and unemployed respondents, which had no longer statistically significantly higher stress among neither of sexes. Therefore, some of the employment status differences in stress may be explained by depression. However, if stress and insomnia symptoms are preceding and predictors of depression, as some of the evidence is demonstrating [36–38], then adding depression in the analyses may have caused over adjustment in the models.
Moreover, insomnia and stress are commonly thought to relate to each other ; for example, work stress is linked to insomnia [39–41]. In our data, 39% of those having stress reported insomnia, and 41% of those who reported insomnia also had stress; however, the cross-sectional study design of our data does not allow us to make conclusions about the causality of the stress-insomnia relationship, or any other studied associations. Trends in the prevalence of insomnia and stress were not identical to either each other or to what we previously found using the same data with self-reported depression, which for example showed a decreasing trend contrary to the increasing trend in insomnia and stress . Even though insomnia and stress are known to be related to each other and other indicators for mental health problems, they may also produce an independent risk factor for health and well-being.
Studies of the effects of retirement on mental health, including sleep outcomes, has produced inconsistent findings showing both improvement and increase in symptoms [42, 43]. In a longitudinal follow-up study, sleep disturbances have been found to improve after retirement, which were explained by removal of work-related risk factor exposures. Retirement on health grounds was, however, associated with increase in sleep disturbances following retirement . In the Finnish register-based follow-up study sleeping problems were found to be associated with subsequent disability retirement . In our study retired respondents, which were early retirees, had more insomnia and stress. Early retirement is known to be associated with lower mental and physical health [45–47], and mental and musculoskeletal disorders are the most common reasons for granting disability pension in Finland [48–50].
Over the years, the overall response rate of the “Health Behaviour and Health among the Finnish Adult Population” survey has decreased from 84% down to 65%. Similar rates have been found in other population surveys . We have conducted several non-respondent analyses on our survey data [52, 53] in which we found that the non-respondents were more likely to be male, young and lower educated. Gender and educational differences in the response rate have further widened over time. Additionally, we found higher total and cause-specific excess mortality (such as due to alcohol, external causes, suicide) among survey non-respondents, which was partly explained by educational and income differences between respondents and non-respondents. The results of those non-respondent analyses indicate that the non-respondents may have, for example, unhealthy lifestyles, more severe illnesses and mental health problems, and may also differ from the respondents in terms of self-reported psychological symptoms. In the present study, respondents with missing data on insomnia and stress variables were also more likely to be in the lower SES groups.
After the economic boom of the 80s, Finland experienced its most severe financial crisis to date at the beginning of the 90s. The economic recession caused unemployment and loss of income for a wide population regardless of socio-economic background. Several studies have examined the implications of the recessionary period for psychological health, which is hypothesised to have deteriorated during that time [54–56]. This was only partially supported in a Finnish longitudinal cohort study, which showed no drastic decline in sleep quality during 1991–1995 except among the prospectively unemployed, most of whom were blue-collar workers . We found no statistically significant increase in insomnia during the period of high unemployment in 1993–1997. The prevalence of stress had a linearly increasing trend, which during recession period strengthened especially among women. However, general increase in stress was not attributable to any specific socio-economic levels. During the recession, changes in socio-economic differences were somewhat inconsistent, but differences especially between the employed and unemployed respondents rather narrowed in both insomnia and stress. This indicates that when national unemployment rates were relatively high, being unemployed was obviously not as selective as usual. Valkonen et al.  found that economic recession slowed down rather than sped up the growth of relative inequalities in mental health related mortality (such as alcohol-related causes, accidents and suicide) in Finland.
The prevalence of self-reported insomnia increased over the last time period, 1998–2002. Various studies have found similar evidence for an increase in insomnia. In a review and re-analysis conducted in Finland it was found that insomnia-related symptoms increased during 1995–2005 among the general population and especially among the employed working-aged population . It was concluded that the increase in insomnia-related symptoms may be due to changes in working life that have raised the employees’ stress level. In the Swedish population study of women  the proportion of respondents with sleeping problems almost doubled from 1968 to 2004. Perceived poor economic status, poor family and social situation as well as mental stress were related to sleeping problems in women . In our data the increase in insomnia was evenly distributed (excluding housewives) over different socio-economic groups, making it difficult to account for socio-economic inequality.
Although socio-economic differences slightly fluctuated over the total period 1979–2002, significant changes in socio-economic differences were rare. It is noteworthy, that some of the differences in insomnia and stress were reversed and curvilinear. Future studies are needed to explore the complexity and significance of socio-economic differences, especially in stress, as well as the growing prevalence of psychological symptoms.