Midlife Insomnia and Subsequent Mortality
Midlife Insomnia and Subsequent Mortality
Baseline demographic and clinical characteristics are presented in Table 1. The prevalence of insomnia was 5.6%. Insomnia was more prevalent among women, current smokers, persons with low education, and among those with low physical activity. Insomnia was also positively associated with BMI, symptoms of anxiety, depression and OSA, self-reported angina and number of somatic symptoms and musculoskeletal pain, as well as use of sleep medication and short sleep duration. Insomnia was unrelated to alcohol consumption, blood pressure, and to reported diabetes, myocardial infarction and stroke.
During the follow-up period from 1997–1999 through 2012, 160/6233 (2.6%) persons died, of which 93/3858 (2.4%) were female and 67/2377 (2.8%) were male. In the crude analyses, insomnia was associated with an almost three-fold increase in mortality (hazards ratio (HR) = 2.74; 95% CI: 1.75–4.30; Figure 1 and Table 2). Including the confounders as adjustment variables, one at a time, only slightly changed the effect estimates. In the fully adjusted model, insomnia was even more strongly associated with mortality than in the crude analysis (HR = 3.34; 95% CI: 1.67–6.69).
(Enlarge Image)
Figure 1.
Kaplan-Meier survival curves by insomnia status (fully adjusted analyses) in the Hordaland Health Study (1997–1999).
When stratifying the analyses by gender we found that male insomniacs had almost 4 times higher risk of mortality compared to male good sleepers (crude HR = 4.72; 95% CI: 2.48–9.03; Figure 2), whereas the corresponding effect for women was HR = 1.96 (95% CI: 1.04–3.67). The interaction between insomnia and gender was statistically significant with a stronger effect of insomnia for males compared to females (P = 0.050).
(Enlarge Image)
Figure 2.
Unadjusted hazard-ratios of mortality risk associated with insomnia, stratified by sex and sleep duration during 13–15 years follow-up of the Hordaland Health Study (1997–1999). Error bars represent 95% confidence intervals. Note that y-axis is on a logarithmic scale.
The mortality risk was higher in insomniacs sleeping less than 6.5 hours (crude HR = 2.79; 95% CI: 1.26–6.17; Figure 2) compared those sleeping 6.5 hours or more. The association between insomnia with normal or greater sleep duration and mortality was not significant (crude HR = 1.78; 95% CI: 0.78–4.06). However the interaction between insomnia and sleep duration was not statistically significant (P = 0.44).
As a sensitivity analyses, we repeated the analyses after excluding deaths occurring in the first 2 years of follow-up. This resulted in 8 deaths being omitted, still the associations remained practically identical: crude HR = 2.76 (95%CI: 1.74–4.38) and adj. HR = 3.29 (95% CI: 1.60–6.78).
We also repeated the analyses using the continuous insomnia score as exposure variable. These analyses showed that the crude HR was 1.08 (95% CI: 1.02–1.15), equal to an 8.3% increase in mortality rate for every increasing value of the insomnia score (range 0–16). The fully adjusted HR was 1.10 (95% CI: 1.01–1.20), indicating a 10.1% increase in mortality rate.
Although the low numbers of deaths in the current study prevented detailed analyses of cause of death, crude analyses showed no clear pattern of specific causes being more prevalent in the insomnia vs. the non-insomnia group. The most common cause of death among insomniacs was cancer (48%) followed by CVD (9.5%), as was also the case in the non-insomnia group.
Results
Sample Characteristics
Baseline demographic and clinical characteristics are presented in Table 1. The prevalence of insomnia was 5.6%. Insomnia was more prevalent among women, current smokers, persons with low education, and among those with low physical activity. Insomnia was also positively associated with BMI, symptoms of anxiety, depression and OSA, self-reported angina and number of somatic symptoms and musculoskeletal pain, as well as use of sleep medication and short sleep duration. Insomnia was unrelated to alcohol consumption, blood pressure, and to reported diabetes, myocardial infarction and stroke.
The Effect of Insomnia on All-cause Mortality
During the follow-up period from 1997–1999 through 2012, 160/6233 (2.6%) persons died, of which 93/3858 (2.4%) were female and 67/2377 (2.8%) were male. In the crude analyses, insomnia was associated with an almost three-fold increase in mortality (hazards ratio (HR) = 2.74; 95% CI: 1.75–4.30; Figure 1 and Table 2). Including the confounders as adjustment variables, one at a time, only slightly changed the effect estimates. In the fully adjusted model, insomnia was even more strongly associated with mortality than in the crude analysis (HR = 3.34; 95% CI: 1.67–6.69).
(Enlarge Image)
Figure 1.
Kaplan-Meier survival curves by insomnia status (fully adjusted analyses) in the Hordaland Health Study (1997–1999).
When stratifying the analyses by gender we found that male insomniacs had almost 4 times higher risk of mortality compared to male good sleepers (crude HR = 4.72; 95% CI: 2.48–9.03; Figure 2), whereas the corresponding effect for women was HR = 1.96 (95% CI: 1.04–3.67). The interaction between insomnia and gender was statistically significant with a stronger effect of insomnia for males compared to females (P = 0.050).
(Enlarge Image)
Figure 2.
Unadjusted hazard-ratios of mortality risk associated with insomnia, stratified by sex and sleep duration during 13–15 years follow-up of the Hordaland Health Study (1997–1999). Error bars represent 95% confidence intervals. Note that y-axis is on a logarithmic scale.
The mortality risk was higher in insomniacs sleeping less than 6.5 hours (crude HR = 2.79; 95% CI: 1.26–6.17; Figure 2) compared those sleeping 6.5 hours or more. The association between insomnia with normal or greater sleep duration and mortality was not significant (crude HR = 1.78; 95% CI: 0.78–4.06). However the interaction between insomnia and sleep duration was not statistically significant (P = 0.44).
As a sensitivity analyses, we repeated the analyses after excluding deaths occurring in the first 2 years of follow-up. This resulted in 8 deaths being omitted, still the associations remained practically identical: crude HR = 2.76 (95%CI: 1.74–4.38) and adj. HR = 3.29 (95% CI: 1.60–6.78).
We also repeated the analyses using the continuous insomnia score as exposure variable. These analyses showed that the crude HR was 1.08 (95% CI: 1.02–1.15), equal to an 8.3% increase in mortality rate for every increasing value of the insomnia score (range 0–16). The fully adjusted HR was 1.10 (95% CI: 1.01–1.20), indicating a 10.1% increase in mortality rate.
Cause of Death
Although the low numbers of deaths in the current study prevented detailed analyses of cause of death, crude analyses showed no clear pattern of specific causes being more prevalent in the insomnia vs. the non-insomnia group. The most common cause of death among insomniacs was cancer (48%) followed by CVD (9.5%), as was also the case in the non-insomnia group.