Suicide is the 10th leading cause of death in the US, and one of the three leading causes of death for those aged 15 to 35 years.
The trajectory to suicide is complex and multidimensional with many identified risk factors, including mental disorders, poor physical health, previous suicidal behavior, family history of suicide, family violence, and trauma/stressful life events.
There is abundant evidence on the relationship between mental disorders and suicidal behavior.3,4 Hawton and van Heeringen noted that the presence of major depressive disorder (MDD), eating disorders, adjustment disorder, anxiety disorders, and many of the personality disorders all contribute and predispose to an increased risk of suicidal behavior.3
Exposures to stressful life events (SLEs) are highly associated with the early onset of suicidal behaviors among youth and are often accompanied with comorbid mental disorders.5 Past research has also demonstrated that mental disorders mediate the effect between SLEs and suicidality. Some studies have demonstrated that SLEs commonly occur in the year preceding the onset of mental disorders, including but not limited to: substance-use disorders, depression, bipolar, and posttraumatic stress disorder (PTSD).6,7 These disorders in turn are linked to suicide attempts.
Beautrais and colleagues stated that life events and mental disorders both independently and together contribute to one’s susceptibility of suicidal behaviors.8 Whether there exists an independent relationship between SLEs and suicide attempts in the absence of mental disorders is not well understood and remains an area requiring further study.
Several limitations in the existing literature contribute to the absence of a clear understanding on the relationship between suicide attempts and SLEs. Past research has examined the relationship between recent SLEs and suicidal behavior using restricted samples, such as youth or suicidal patients seeking treatment in the emergency department (ED),2,9 limiting its general application to the larger population.
Research consistently demonstrates a positive dose-response relationship between SLEs and broad measures of suicidal behavior, but has not examined suicide attempts specifically.10 No previous study has examined a comprehensive range of SLEs; the focus has been only on a particular type of SLE, such as PTSD-related trauma16 or financial stress.11,12
Additionally, most studies have only controlled for the effect of some, but not all, Axis I and Axis II mental disorders. Finally, the majority of studies have focused only on lifetime exposures to, rather than recent measures of, SLEs, while it has been demonstrated that past-year SLEs, as opposed to those from childhood and lifetime, are better predictors of suicidal behavior.13 These studies have shown that suicide attempts among those who have experienced SLEs in the past year were approximately five times higher than those who reported no such experience.5
National Sample Data
To address the shortcomings in the existing literature, the primary goal of the current study is to explore whether past-year exposure to SLEs is independently related to past-year suicide attempts in a nationally representative sample, the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) Wave 2. There are several advantages to using the NESARC data. First, a large, national representative dataset that surveyed the noninstitutionalized adult population in the US, NESARC14 overcomes the selection bias and limited size of previous study samples.
Second, it provides a comprehensive assessment of DSM-IV Axis I and II mental disorders. Lastly, the survey assessed a variety of SLEs and their occurrences in the past-year. Based on the results of previous studies, we hypothesized that a relationship exists between SLEs and suicide attempts, independent of the effect of mental disorders.10,12
In NESARC Wave 1 (2001–2002), 43,093 individuals were assessed. In Wave 2 (2004–2005), 34,653 individuals of the original Wave 1 sample were reassessed (ages ≥20 years).
Analyses conducted in the current study were based on the Wave 2 sample, which had a response rate of 81%. The overall response rate for both Wave 1 and 2 was 70.9%. All respondents in Wave 2 were asked about SLEs and history of suicide attempts.13–15 Statistical weights were employed to ensure sample representativeness of the data to the US population. For details about the NESARC, please refer to the original reports.14,16
Stressful Life Events
Thirty-four SLEs were assessed in the NESARC study. Respondents were asked whether they had experienced any of these in the past year. Upon initial frequency analysis, nine of the SLEs were eliminated because no individuals were exposed to those events in the past-year. Thus only 26 items were included in further analysis. Table 1 (see page 103) presents all 25 SLEs examined in the study. To facilitate comparison, and maintain an adequate sample size, these events were grouped into seven categories based on previous studies:17 assaultive violence; financial stress; shocking or traumatic event; relationship, friendship, or interpersonal stressors or changes; property loss/damage; legal problems; and learned trauma.
Table 1. Past-Year Prevalence of Stressful Life Events
Past-Year Suicide Attempt
Lifetime suicide attempt was assessed by the survey question, “In your entire life, did you ever attempt suicide?” The occurrence of a past-year suicide attempt was not directly assessed by NESARC; however, the survey did include the age of the most recent suicide attempt. To determine whether the attempt occurred in the past year, we calculated the difference between the age of the respondent and the age of the most recent attempt. A difference of zero or one was categorized as a past-year suicide attempt and was included in subsequent analyses.
Eight sociodemographic variables were included in the analysis: income; race/ethnicity; education; marital status; age; sex; region; and urban status (urbanicity). Income ($0–19,999; $20,000–34,999; $35,000–59,999; $60,000+) and region (Northeast, Midwest, South, and West) were both divided into four categories each. Ethnicity was composed of five categories, including, “white,” “black,” “American Indian or Alaskan,” “Asian or Hawaiian,” and “Hispanic.” Education was trichotomously divided into “less than high-school graduate,” “high school-only graduate,” and “some college or higher.”
Participants were categorized into one of the six marital status categories: married; living with someone as if married; widowed; divorced; separated; and never married. Age was entered into the analyses as a continuous variable. Sex was dichotomously categorized as “male” versus “female.”
DSM-IV18 Axis I and Axis II mental disorders were assessed with the Alcohol Use Disorder and Associated Disabilities Interview Schedule IV (AU-DADIS-IV).19 It has been demonstrated that the AUDADIS-IV is a reliable and valid measure of mental disorders.13,14
The current study examined past-year Axis I mental disorder diagnoses, including mood disorders (major depressive disorder, dysthymic disorder, bipolar disorder, and hypomania); anxiety disorders (panic disorder without agoraphobia, panic disorder with agoraphobia, agoraphobia without history of panic disorder, social phobia, specific phobia, generalized anxiety disorder, and posttraumatic stress disorder); and substance-use disorders (alcohol abuse and dependence, nicotine dependence, drug abuse and dependence).
An “any Axis I disorder” aggregate variable was used as a covariate in the analyses. Likewise, an “any Axis II disorder” aggregate variable was created and included individuals who met criteria for any lifetime personality disorder (paranoid, schizoid, schizotypal, antisocial, narcissistic, histrionic, borderline, obsessive-compulsive, avoidant, or dependent).
We reasoned that lifetime measures of personality disorders were acceptable to use in place of past-year measures because of the likelihood that these disorders are often long-standing, although we understand that some researchers might argue otherwise.
Analyses of Data
Due to missing data, 1,375 people were excluded from all analyses; leaving a sample of 33,278. Appropriate statistical weights provided by the NESARC were applied and Taylor Series Linearization was performed as a variance estimation technique to account for the complex design of the survey. All analyses were conducted using SUDAAN version 10.0.20–22 Cross-tabulations were conducted to determine frequency rates for each individual SLE and the seven types of SLEs.
Logistic regression analyses were conducted to determine the relationship between suicide attempts and both the individual SLEs and SLE categories, after adjusting odds ratios (AOR) for the effects of sociodemographic variables (AOR1) and mental disorders (AOR2).
AORs were also calculated in a multivariable model that contained all seven SLE categories, thus simultaneously adjusting for the effects of other SLEs. Population-attributable fractions (PAFs) of past-year SLEs on past-year suicide attempts were calculated for the SLE categories significantly associated with suicide attempts in the multivariable model. PAFs were calculated following the equation proposed by comparable population health and biostatistics research:4,23
Where P is the proportion of individuals in the entire population who were exposed to the specific category of SLE, AOR is the calculated adjusted odds of suicide attempts within the given SLE category.
Accordingly, the PAF represents the percentage of suicide attempts in the population in the past year that may be attributed to the occurrence of the SLE. Although the original formula involves calculating PAF using relative risk (RR) instead of odds ratios, but research has indicated that odds ratios are close approximations of RR, which can be substituted into this formula.23
To assess the presence of a dose-response relationship between SLEs and suicide attempt, a count variable was created by summing the total number of SLEs that each individual was exposed to in the past year. The reference group was those who experienced no SLEs in the past year. This dose-response relationship was analyzed while adjusting for sociodemographic variables and mental disorders.
Table 1 illustrates that any shocking or traumatic events, and any relationship, friendship, or interpersonal stressors or changes were the most prevalent SLE categories. In addition, death of a family member or close friend, and changes in work or job responsibilities were the most prevalent individual SLEs.
Table 2 shows cross-tabulations and the first two series of logistic regressions of the 25 individual and seven types of past-year SLEs with past-year suicide attempts.
Table 2. Association Between Exposure to Stressful Life Events and Suicide Attempt
After adjusting for sociodemographic factors, all individual SLEs examined were significantly associated with having made a past-year suicide attempt with the exception of natural disasters. After further adjustment for mental disorders, many SLEs were no longer significantly associated with past-year suicide attempts. The AORs ranged from 1.67 [95% CI, 1.05–2.64 (moving or having someone new live with him or her)] to 4.66 [95% CI, 1.48–14.6 (being mugged or threatened with a weapon)].
The results of the multivariable model are presented in Table 3. Only two SLE categories remained significantly associated with past-year suicide attempts after accounting for the impact of each of the other SLE categories: any assaultive violence (AOR=3.02; 95% CI, 1.34–6.83) and any financial stress (AOR=2.38; 95% CI, 1.29–4.39).
Table 3. Multivariable Model to Examine the Unique Associations Between Types of Stressful Life Events and Suicide Attempts
Finally, a dose-response with regards to SLE on presence of a suicide attempt was examined through two means: first as a continuous variable and secondly as a categorical variable, both using “No SLE” as the reference group.
Analyses revealed evidence for a dose-response relationship both as a continuous variable (AOR=1.01; 95% CI, 1.00–1.02) and grouped by categories (one SLE [AOR=2.13; 95% CI, 0.87–5.22]; two SLEs [AOR=2.64; 95% CI, 1.02–6.83]; and three or more SLEs [AOR=7.08; 95% CI, 3.05–16.42]).
To our knowledge, this is the first study examining a broad number of past-year SLEs and their relationship with past-year suicide attempts in a nationally representative sample. Several important findings emerged.
First, consistent with our hypothesis, the relationship between SLEs and suicide attempts persisted even after adjusting for mental disorders and sociodemographic variables. This finding demonstrates that SLEs are associated with suicide attempts regardless of an individual’s current demographic and mental health status.
Second, certain specific types of SLEs are more strongly associated with suicide attempts than others. Financial stress and assaultive violence, in particular, were strongly associated with suicide attempts.
Our findings revealed that financial stress has an especially strong association with suicide attempts with a high PAF value of approximately 21%. This finding bolsters results from previous findings that recent negative financial change is correlated with suicidal behavior.12
Using the same dataset, Bolton and Robinson examined the PAFs for mental disorders on suicide attempts. In comparison to the PAF values for DSM-IV mental disorders, financial stress demonstrated a higher PAF value on suicide attempts than almost all mental disorders with the exception of MDD (PAF=26.6). Assaultive violence also had a higher PAF value than most anxiety disorders, psychotic disorders, and personality disorders. This finding highlights SLEs as important areas of assessment in suicide attempt risk.
Third, this study also provided evidence of a dose-response relationship between SLEs and suicide attempts. The experience of multiple SLEs in the past year is more strongly associated with attempting suicide than if exposed to no SLE. This effect is especially prominent when there is an exposure to three or more SLEs.
This is the first study to examine the dose-response relationship between SLEs and suicide attempt in the past-year time frame.24 Taken together, these results suggest that while the type of SLE is important, the number of SLEs experienced in the last year is also a correlating factor in suicide attempt risk.
Our results should be interpreted carefully while considering several limitations, such as that although all SLEs and suicide attempts occurred in the past year, the order of onset of these events could not be determined. Thus, we cannot firmly conclude whether past-year exposure of SLEs triggered the suicide attempt.
Another limitation is that it is possible an unmeasured common third variable, such as subsyndromal personality characteristics that do not meet diagnostic criteria may have been at the root of the SLEs and suicide attempts.25
Although there is some evidence indicating that cross-sectional data can be used as a valid evaluation for PAFs,26 we still need to bear in mind that PAF calculations are based on an assumption of causality, which cannot be addressed with this study design. Additionally, SLEs were based on self-reports, which are susceptible to recall biases.27 However, this method has been validated in research conducted on proxy-based data that has shown a significant overlap between self-reports and informant–reports, especially regarding recent life events.29
Lastly, the focus of the study was among those who made at least one suicide attempt, but we should not completely neglect those who responded “no” to suicide attempts.
Even though an attempt was not made, we cannot conclude that these individuals did not experience distress as a result of these SLEs. Subsequently, the calculations for past-year measures were conducted to the best of our abilities with the given data; we acknowledge possible complications with the time of interview and birthdates.
Despite the above limitations, this study provides important evidence that exposure to SLEs in the past year may be associated with suicide attempts even after accounting for mental disorders. In particular, past-year assaultive violence and financial stress were correlated with suicide attempts, even when accounting for presence of other SLEs. Family members and health care providers should be aware of these findings to ensure that recent occurrences of these kinds of SLEs prompt appropriate assessment and intervention for suicidal behaviors.
- World Health Organization. Suicide Prevention (SUPRE). 2010. World Health Organization. Available at: www.who.int/mental_health/prevention/suicide/suicideprevent/en. Accessed Feb. 9, 2012.
- Moscicki EK. Epidemiology of completed and attempted suicide: toward a framework for prevention. Clin Neurosci Res. 2001;1:310–323. doi:10.1016/S1566-2772(01)00032-9 [CrossRef]
- Hawton K, van Heeringen K. Suicide. Lancet. 2009;373(9672):1372–1381. doi:10.1016/S0140-6736(09)60372-X [CrossRef]
- Bolton JM, Robinson J. Population-attributable fractions of Axis I and Axis II mental disorders for suicide attempts: findings from a representative sample of the adult, noninstitutionalized US population. Am J Public Health. 2010;100(12):2473–2480. doi:10.2105/AJPH.2010.192252 [CrossRef]
- Fergusson DM, Woodward LJ, Horwood LJ. Risk factors and life processes associated with the onset of suicidal behaviour during adolescence and early adulthood. Psychol Med. 2000;30(1):23–39. doi:10.1017/S003329179900135X [CrossRef]
- Horesh N, Apter A, Zalsman G. Timing, quantity and quality of stressful life events in childhood and preceding the first episode of bipolar disorder. J Affect Disord. 2011;134(1–3):434–437. doi:10.1016/j.jad.2011.05.034 [CrossRef]
- Taylor R, Page A, Morrell S, Harrison J, Carter G. Mental health and socio-economic variations in Australian suicide. Soc Sci Med. 2005;61(7):1551–1559. doi:10.1016/j.socscimed.2005.02.009 [CrossRef]
- Beautrais AL. Suicide and serious suicide attempts in youth: a multiple-group comparison study. Am J Psychiatry. 2003;160(6):1093–1099. doi:10.1176/appi.ajp.160.6.1093 [CrossRef]
- Kumar CT, Mohan R, Ranjith G, Chandrasekaran R. Characteristics of high intent suicide attempters admitted to a general hospital. J Affect Disord. 2006;91(1):77–81. doi:10.1016/j.jad.2005.12.028 [CrossRef]
- Stein DJ, Chiu WT, Hwang I, et al. Cross-national analysis of the associations between traumatic events and suicidal behavior: findings from the WHO World Mental Health Surveys. PLoS One. 2010;5(5):e10574. doi:10.1371/journal.pone.0010574 [CrossRef]
- Sareen J, Afifi TO, McMillan KA, Asmundson GJ. Relationship between household income and mental disorders: findings from a population-based longitudinal study. Arch Gen Psychiatry. 2011;68(4):419–427. doi:10.1001/archgenpsychiatry.2011.15 [CrossRef]
- Belik SL, Cox BJ, Stein MB, Asmundson GJ, Sareen J. Traumatic events and suicidal behavior: results from a national mental health survey. J Nerv Ment Dis. 2007;195(4):342–349. doi:10.1097/01.nmd.0b013e318060a869 [CrossRef]
- Rich CL, Warstadt GM, Nemiroff RA, Fowler RC, Young D. Suicide, stressors, and the life cycle. Am J Psychiatry. 1991;148(4):524–527.
- Grant BF, Harford TC, Muthén BO, Yi HY, Hasin DS, Stinson FS. DSM-IV alcohol dependence and abuse: further evidence of validity in the general population. Drug Alcohol Depend. 2007;86(2–3):154–166. doi:10.1016/j.drugalcdep.2006.05.019 [CrossRef]
- Dawson DA, Saha TD, Grant BF. A multidimensional assessment of the validity and utility of alcohol use disorder severity as determined by item response theory models. Drug Alcohol Depend. 2010;107(1):31–38. doi:10.1016/j.drugalcdep.2009.08.019 [CrossRef]
- Grant BF, Chou SP, Goldstein RB, et al. Prevalence, correlates, disability, and co-morbidity of DSM-IV borderline personality disorder: results from the Wave 2 National Epidemiologic Survey on Alcohol and Related Conditions. J Clin Psychiatry. 2008;69(4):533–545. doi:10.4088/JCP.v69n0404 [CrossRef]
- Fetzner MG, McMillan KA, Sareen J, Asmundson GJ. What is the association between traumatic life events and alcohol abuse/dependence in people with and without PTSD? Findings from a nationally representative sample. Depress Anxiety. 2011;28(8):632–638. doi:10.1002/da.20852 [CrossRef]
- Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994.
- Ruan WJ, Goldstein RB, Chou SP, et al. The alcohol use disorder and associated disabilities interview schedule-IV (AUDADIS-IV): reliability of new psychiatric diagnostic modules and risk factors in a general population sample. Drug Alcohol Depend. 2008;92(1–3):27–36. doi:10.1016/j.drugalcdep.2007.06.001 [CrossRef]
- Software for the Statistical Analysis of Correlated Data (SUDAAN) [computer program]. Version 8.0. Research Triangle Park, NC: Research Triangle Institute; 2000.
- Shah BV, Barnswell BG, Bieler GS. SU-DAAN User’s Manual: Software for Analysis of Correlated Data. Research Triangle Park, NC: Research Triangle Institute; 1995.
- Rockhill B, Newman B, Weinberg C. Use and misuse of population attributable fractions. Am J Public Health. 1998;88(1):15–19. doi:10.2105/AJPH.88.1.15 [CrossRef]
- Young TK. Population Health: Concepts and Methods. 2nd ed. New York, NY: Oxford University Press; 2005.
- Belik SL, Stein MB, Asmundson GJ, Sareen J. Relation between traumatic events and suicide attempts in Canadian military personnel. Can J Psychiatry. 2009;54(2):93–104.
- Baumeister RF. Suicide as escape from self. Psychol Rev. 1990;97(1):90–113. doi:10.1037/0033-295X.97.1.90 [CrossRef]
- Eide GE, Heuch I. Attributable fractions: fundamental concepts and their visualization. Stat Methods Med Res. 2001;10(3):159–193. doi:10.1191/096228001680195148 [CrossRef]
- Ebner-Priemer UW, Kuo J, Welch SS, et al. A valence-dependent group-specific recall bias of retrospective self-reports: a study of borderline personality disorder in everyday life. J Nerv Ment Dis. 2006;194(10):774–779. doi:10.1097/01.nmd.0000239900.46595.72 [CrossRef]
- Dejong TM, Overholser JC. Assessment of depression and suicidal actions: agreement between suicide attempters and informant reports. Suicide Life Threat Behav. 2009;39(1):38–46. doi:10.1521/suli.2009.39.1.38 [CrossRef]