SEARCH

SEARCH BY CITATION

Keywords:

  • adolescents;
  • epidemiology;
  • risk-taking behavior;
  • sleep;
  • suicide

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Aims

This study clarifies the associations among sleep duration, non-restorative sleep, suicidal ideation and suicide attempts in a representative sample of Korean adolescents.

Methods

Analyses are based on data from the 2007 Korean Youth Risk Behavior Web-based Survey. The survey used a cross-sectional, national and representative sample consisting of 78 843 students (grades 7–12) who were selected using a stratified, clustered, multistage sampling method. Logistic regression analysis was conducted to test the association between sleep and suicide variables while controlling for demographic characteristics and other potential risk factors of suicide.

Results

Fewer than 4 h of sleep and a lack of feeling refreshed after sleeping increased the likelihood of suicidal ideation but not of suicide attempts.

Conclusions

Non-restorative sleep as well as short sleep duration are significantly associated with suicidal ideation in adolescents. This finding highlights the need to assess for both non-restorative sleep and short sleep duration when screening suicide risk in adolescents. Future research should examine the moderating or mediating effects of individual and environmental characteristics on the association between sleep and actual suicide attempt.

SUICIDE RATES MARKEDLY increase in adolescence compared with childhood. Moreover, suicide is the second most common cause of adolescent death in the world.[1, 2] In 2007, 4.6 out of every 100 000 Korean adolescents aged 10–19 years committed suicide, making suicide the second highest cause of adolescent death in Korea that year.[3] Because suicide attempts may have fatal consequences, acknowledging the risk factors of attempted suicide and preventing suicide are important tasks for clinicians and public health practitioners. Various studies have investigated the risk factors of adolescent suicide. Numerous studies have identified the association between psychiatric disorders and suicidal risk.[4-9] Psychiatric disorders such as depression, anxiety and substance abuse are among the most significant risk factors for suicide, and sleep difficulty is the most common symptom that patients with these psychiatric disorders experience.[10-14]

Epidemiology studies on adolescents suggest that there is an association between sleep problems and suicide. In a study including 1600 adolescents aged 13–16 years, those with suicidal ideation tended to sleep <7 h per night.[15] That study also found that 31% of the adolescents who easily felt fatigued experienced suicidal ideation, whereas only 8% of those who did not feel fatigued reported suicidal ideation. Another study surveyed 763 high school students with regard to sleep variables, including difficulty falling or staying asleep, need for more sleep, early awakening and chronic sleeping pill intake. Researchers found that 40.8% of adolescents had sleep problems, and poor sleepers had more suicidal ideations and suicide attempts than good sleepers (38% vs 15% and 9% vs 1%, respectively).[16] In a school-based survey of 5423 adolescents, suicidal ideation was related to insomnia (odds ratio [OR] = 3.4) and hypersomnia (OR = 2.8).[17] That study also found a significant relationship between suicidal ideation and insomnia (OR = 1.5), even after adjusting for mood disturbance.

Additional studies, however, are necessary to verify the association between suicide and sleep in adolescents. Although quantitative aspects of sleep and suicide have been studied previously, few studies have examined the qualitative aspects of sleep. Non-restorative sleep (NRS) is related to qualitative aspects of sleep. NRS is defined as a subjective feeling of lack of refreshment upon awakening.[18] Because short sleep duration does not necessarily mean insufficient sleep, there is a need to enhance the understanding of the relationship between sleep and suicide by investigating the level of refreshment after sleeping. NRS is frequently observed in depressive disorders, bipolar disorder and anxiety disorders,[19, 20] each of which are associated with a high risk of suicide. This study investigated the associations among sleep characteristics (i.e. NRS and sleep duration), suicidal ideation and suicide attempts in a representative sample of Korean adolescents.

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Data and sample

Data were drawn from the 2007 Korean Youth Risk Behavior Web-based Survey (KYRBWS) conducted by the Korean Centers for Disease Control and Prevention (KCDC). The KYRBWS is a cross-sectional survey of Korean middle and high school students (grades 7–12) designed to provide annual data on adolescent physical and mental health.[21] A sample was selected using a stratified multistage cluster method according to geographic area, sex and school type to represent all middle and high school students in Korea. A total of 78 843 students across 800 schools participated in the study. The survey procedures were designed to protect student privacy by allowing anonymous and voluntary participation. After the survey had been fully explained and all participants had provided written informed consent (both directly and from their parents or legal guardians), participants completed an online, self-report questionnaire in a school computer room. A teacher who was trained to assist students participating in the survey provided the students with an introduction regarding the correct procedure for completing the questionnaire. The final data were obtained from 74 698 respondents and the overall response rate was 94.8%. The survey was conducted in September 2007. The KCDC provided ethics approval.

Measures

Suicidal ideation and suicide attempts

The survey asked two questions regarding suicidal ideation and attempts: ‘Have you ever seriously thought about committing suicide in the past 12 months?’ and ‘Have you attempted suicide in the past 12 months?’ Only respondents who answered ‘yes’ to the former question were required to answer the latter question.

Sleep variables

With regard to sleep duration, respondents answered the question, ‘How many hours per day did you sleep on weeknights over the last week?’ Respondents chose one of six answers categorized by hour (e.g. <4, 4–<5, 5–<6, 6–<7, 7–<8 and ≥8). With regard to the level of refreshment experienced after sleeping, respondents answered the question, ‘Did you feel refreshed after sleeping over the last week?’ Answer choices included ‘never’, ‘rarely’, ‘sometimes’ and ‘always.’

Control variables

The control variables included in the model were gender, grade, family living structure, economic status, academic performance, cigarette smoking, alcohol drinking and subjective feeling of depression. The family living structure variable had two choices: living with at least one parent or living without parents. Economic status and academic performance each had three levels: good, fair and poor. Respondents checked ‘yes’ for cigarette smoking and alcohol drinking if they had done either over the past month and ‘no’ if they had not. The following question assessed subjective feelings of depression: ‘Did you experience depression or frustration for 2 weeks over the last year?’

Statistical analysis

The descriptive statistics for suicide, sleep and the control variables are presented as frequencies, weighted percentages and 95% confidence intervals (CI). Percentages are weighted as part of the complex survey design so that they are representative of Korean students in grades 7–12. Unadjusted bivariate logistic regression analysis was conducted to determine the association between sleep and suicide variables. Next, we conducted a multiple logistic regression analysis to control for the confounding effects of potential risk factors, such as gender, grade, family living structure, economic status, academic performance, cigarette smoking, alcohol drinking and subjective feelings of depression. Because sleep duration and refreshment of sleep are closely related to each other, we treated sleep duration and refreshment of sleep as a confounding factor for each other. The results are presented as unadjusted and adjusted OR with 95%CI. The significance level was set at P < 0.05. Missing data were excluded from the analyses. All analyses were conducted on weighted data using the SURVEYFREQ and SURVEYLOGISTIC procedures in SAS 9.0 (SAS Institute, Cary, NC, USA) to accommodate the complex survey design and to represent Korean students in grades 7−12.

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

Prevalence

The frequencies, weighted percentages and 95%CI for all variables included in the full analytic models are listed in Table 1. Of the 74 698 respondents included in the analyses, 23.7% (n = 17 783) experienced suicidal ideation at least once in the past year. Among students who answered ‘yes’ to the suicidal ideation item, 24.6% (n = 4460) actually attempted suicide at least once over the past year. A total of 2969 respondents (4.2%) reported that they slept for 4 h at most and 12.3% (n = 9153) of the respondents did not feel refreshed at all after sleeping.

Table 1. Factors of suicidal ideation and suicide attempt
 Subjects (n = 74 698)
nWeighted %a (95%CI)
  1. a

    Percentages are weighted as part of the complex survey design to be representative of students in grades 7–12 in Korea.

  2. CI, confidence interval.

Suicide variables  
Suicidal ideation17 78323.7 (23.1–24.2)
Suicide attempts/ideation4 460/17 78324.6 (23.7–25.4)
Sleep variables  
Sleep duration (h/night)  
<42 9694.2 (4.0–4.4)
4 – <59 69313.2 (12.7–13.7)
5 – <617 59624.0 (23.5–24.5)
6 – <722 64330.3 (29.8–30.8)
7 – <815 18919.8 (19.3–20.4)
≥86 6088.4 (8.1–8.8)
Restorative sleep  
Never9 15312.3 (12.0–12.7)
Rarely20 87228.1 (27.6–28.6)
Sometimes25 32033.7 (33.2–34.2)
Always19 35325.9 (25.3–26.4)
Control variables  
Boys39 46653.1 (50.1–56.1)
Grade  
7th13 03517.4 (16.8–18.0)
8th12 84917.6 (16.9–18.2)
9th12 93617.9 (17.2–18.5)
10th12 80917.1 (16.4–17.8)
11th11 59615.1 (14.5–15.7)
12th11 47314.9 (14.3–15.6)
Living without parents1 9542.3 (2.1–2.4)
Economic status  
Lower18 89623.8 (23.2–24.5)
Middle35 52347.4 (46.9–47.9)
Upper20 27928.8 (28.0–29.6)
Academic performance  
Lower27 40136.6 (36.1–37.1)
Middle19 89626.4 (26.0–26.8)
Upper27 40137.0 (36.5–37.5)
Cigarette smoking10 03813.3 (12.7–14.0)
Alcohol drinking21 07827.8 (27.1–28.5)
Subjective feeling of depression30 95141.3 (40.7–41.8)

Suicidal ideation

Bivariate logistic regression analysis showed that a short sleep duration and a lack of refreshment after sleeping were associated with a higher rate of suicidal ideation. In particular, suicidal ideation had a strong association with receiving <4 h sleep (OR = 2.01, 95%CI: 1.78–2.26) and ‘never’ feeling refreshed after sleeping (OR = 3.19, 95%CI: 2.97–3.42). According to multiple logistic regression analysis, suicidal ideation was significantly related to receiving <4 h of sleep (OR = 1.43, 95%CI: 1.28–1.61) and ‘never’ feeling refreshed after sleeping (OR = 1.91, 95%CI:1.78–2.04; Table 2).

Table 2. Association of suicidal ideation with sleep duration and restorative sleep
 Suicidal ideation
Unadjusted OR (95%CI)PAdjusted ORa (95%CI)P
  1. a

    Adjusted for gender, grade, living structure, economic status, academic performance, cigarette smoking, alcohol drinking and subjective feeling of depression. Additionally, variables of sleep duration and restorative sleep were included in multivariate analysis as confounding factors.

  2. CI, confidence interval; OR, odds ratio.

Sleep duration (h/night)    
<42.01 (1.78–2.26)<0.0011.43 (1.28–1.61)<0.001
4 – <51.39 (1.26–1.53)<0.0011.12 (1.02–1.22)0.061
5 – <61.14 (1.05–1.24)0.0021.03 (0.95–1.12)0.488
6 – <70.99 (0.92–1.07)0.8510.97 (0.90–1.05)0.463
7 – <80.90 (0.82–0.98)0.0120.94 (0.87–1.02)0.128
Restorative sleep    
Never3.19 (2.97–3.42)<0.0011.91 (1.78–2.04)<0.001
Rarely1.81 (1.71–1.93)<0.0011.36 (1.28–1.44)<0.001
Sometimes1.38 (1.30–1.46)<0.0011.22 (1.15–1.29)<0.001

Suicide attempts

According to the bivariate analysis, suicide attempts were significantly related to ‘never’ feeling refreshed after sleeping (OR = 1.42, 95%CI: 1.24–1.63). Multiple logistic regression analysis, however, showed that receiving <4 h of sleep and not feeling refreshed after sleeping did not significantly increase the likelihood of suicide attempts (Table 3).

Table 3. Association of suicide attempt with sleep duration and restorative sleep
 Suicide attempt
Unadjusted OR (95%CI)PAdjusted ORa (95%CI)P
  1. a

    Adjusted for gender, grade, living structure, economic status, academic performance, cigarette smoking, alcohol drinking and subjective feeling of depression. Additionally, variables of sleep duration and restorative sleep were included in multivariate analysis as confounding factors.

  2. CI, confidence interval; OR, odds ratio.

Sleep duration (h/night)    
<41.08 (0.88–1.33)0.4801.19 (0.98–1.44)0.079
4 – <50.71 (0.60–0.85)<0.0010.91 (0.77–1.06)0.218
5 – <60.65 (0.56–0.76)<0.0010.81 (0.70–0.93)0.003
6 – <70.69 (0.59–0.80)<0.0010.80 (0.70–0.92)0.001
7 – <80.66 (0.56–0.77)<0.0010.74 (0.64–0.86)<0.001
Restorative sleep    
Never1.42 (1.24–1.63)<0.0011.13 (0.99–1.28)0.140
Rarely1.03 (0.91–1.17)0.6230.93 (0.81–1.04)0.229
Sometimes1.02 (0.90–1.16)0.7450.99 (0.89–1.11)0.957

With regard to control variables, being male, living without parents, low economic status, low academic performance, cigarette smoking, alcohol drinking and subjective feelings of depression were significantly associated with suicidal ideation and suicide attempts.

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

The primary study results are as follows. First, the less that adolescents sleep, the more they tend to think about suicide. In particular, <4 h of sleep per night increased the likelihood of suicidal ideation but did not increase the likelihood of suicide attempts. Second, the lack of refreshment after sleeping was related to suicidal ideation, but was not significantly related to suicide attempts.

The present study found that only sleep duration of ≤4 h was related to suicidal ideation after adjusting for control variables; this period of sleep is relatively short compared with previous epidemiological findings regarding adolescents. This difference can at least be partially attributed to the differences in average sleeping time per nation. On average, Korean adolescents sleep 6.3–6.5 h,[22] which is lower than that for US (7.4 h)[23] and Chinese (7.6 h)[24] adolescents. A study conducted on Chinese adolescents reported that a sleep duration of ≤8 h was related to increases in suicidal ideation.[25] Eight hours of sleep is more than the average Korean adolescent receives. Short sleep duration is known to lead to increased suicidal ideation, but objectively defining ‘short’ is difficult. Thus, future studies should compare an individual's sleep duration relative to their age and the average sleep duration in their region.

The present results differ from those of a previous study which reported a dose-dependent relationship between sleep duration and suicidal ideation.[26] In the present bivariate analysis, sleep duration of 6 h, 5 h and ≤4 h significantly increased the OR of suicidal ideation in a dose-dependent manner. After controlling other confounding factors, however, only extremely short sleep duration (e.g. ≤4 h) remained in the multivariate model. Although we cannot fully explain this, we suggest two possible reasons for this dose-independent association. First, confounding factors in the analytic model have a stronger effect on suicidal ideation than sleep duration. Thus the effects of confounding factors dilute the effects of sleep duration on suicidal ideation. Second, it is possible that individual psychopathology influences sleep time estimation. Depressive individuals with insomnia, who are at high risk for suicidal ideation, have a tendency to underestimate their total sleep duration.[27, 28] This would have led the present group at high risk for suicidal ideation to report a sleep duration that was shorter than their actual sleep duration. Thus underestimation of sleep duration may have contributed to the result that extremely short sleep duration was the only significant factor left in multiple logistic analysis.

The present study indicates that NRS is associated with suicidal ideation even after adjusting for other suicide risk factors. Sleep helps individuals recover from the physical and mental exhaustion experienced through daily work. NRS describes a state in which the restorative function has deteriorated.[18, 29] Because every individual requires a different amount of sleep, short sleep duration does not necessarily mean insufficient sleep. Accordingly, when assessing suicide risk, both sleep duration and NRS must be assessed. Polysomnography is the most accurate method for estimating the quality of sleep, but NRS assessment is a simple and useful method to determine the quality of sleep in a large-scale epidemiologic study. Nevertheless, there is a lack of consensus regarding the definition of NRS; thus, the conceptualization and operationalization of NRS differs across studies.[18] Criteria should be established so that NRS can be assessed more objectively in future studies.

In the present study, sleep duration and NRS did not increase the likelihood of suicide attempts after adjusting control variables. Bernert et al. conducted a study of 176 adult outpatients to investigate the correlation between sleep disturbance and suicide.[30] They found that insomnia and nightmares were related to suicides and depression but, after controlling for depression and age, only nightmares were related to suicide and the relationship between insomnia and suicide was no longer significant. In addition, Liu reported that there was no significant correlation between sleep duration and suicide attempts after adjusting for depression in adolescents.[25] The present findings are consistent with these studies, but sleep disturbances increase suicidal ideation, which is a strong predictor of suicide attempt.

Sleep disturbances may lead to suicide via three pathways.[31] First, sleep disturbance causes impairment in judgment, poor impulse control, increased irritability and decreased tolerance of frustration; any of these factors may increase suicide risk. Second, sleep disturbance acts as a stressor that causes psychopathologies such as depression or bipolar disorder, which may increase suicide risk. Third, sleep disturbance may lead to increased suicide risk by interacting with other present suicide risk factors. The present results do not conclusively show that insufficient sleep leads to an increased risk of suicide attempts. Vulnerability and resilience may moderate or mediate the association between sleep and suicide as well as between sleep and other suicide risk factors. Future research should examine the moderating or mediating effects of individual and environmental characteristics on the association between sleep and suicide.

Several limitations should be considered when interpreting the present results. First, this study is cross-sectional; a longitudinal study is needed to establish a causal role between sleep and suicide. Second, this study may be skewed by recall bias because it is based on survey respondents' self-reports. Objective assessments using polysomnography or actigraphy would increase the accuracy of sleep data; nevertheless, self-report questionnaires appear to be the most effective approach for collecting large-scale epidemiologic data. Third, the questions concerning sleep duration and NRS were confined to ‘the past week’, whereas questions regarding suicidal ideation and attempts were confined to ‘the past year’. Thus, ascertaining the chronological order between sleep disturbance and suicidal ideation is difficult. Fourth, this study did not control for the presence or absence of the psychiatric disorders associated with a high risk of suicide, such as depression, bipolar disorder, or anxiety disorders. This study investigated only subjective feelings of depression, which were highly correlated with suicidal ideation and attempts. Because of the large-scale nature of this survey, individual evaluation of the psychopathology of all respondents would have been difficult.

Despite these limitations, this study had several strengths. First, this survey was conducted with 74 698 adolescents and the systematic sampling method used reflected the characteristics of the Korean adolescent population. Both large sample size and the representative sampling method improve the precision of study results. Second, this study enhances the current understanding of the association between sleep and suicide by assessing NRS. When clinically assessing suicide risk, the level of refreshment after sleeping should be evaluated in addition to the number of sleep hours. Third, this study accounts for differences in average sleep duration by nation; this difference should be considered in future studies.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References

The authors would like to thank Mrs Eun Hee Yoo, clinical psychologist, for her valuable help with the manuscript. This study was supported by research funds from Dong-A University. None of the authors have a potential conflict of interest.

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. References