Insomnia Symptoms, Nightmares, and SuicideRisk: Duration of Sleep Disturbance Matters

MICHAEL R. NADORFF, PHD, SARRA NAZEM, MS, AND AMY FISKE, PHD, CBSM

Duration of insomnia symptoms or nightmares was investigated to see ifit was related to suicide risk independent of current insomnia symptoms, night-mares, anxiety symptoms, depressive symptoms, and posttraumatic symptoms.The cross-sectional study involved analyses of survey responses from under-graduate students who endorsed either insomnia symptoms (n = 660) or night-mares (n = 312). Both insomnia symptom and nightmare duration weresignificantly associated with suicide risk independent of current insomnia symp-toms or nightmares, respectively. Relations were also significant after control-ling for anxiety symptoms, depressive symptoms, and posttraumatic symptoms.Results suggest that duration of sleep disturbance is relevant when assessing sui-cide risk.

Suicide is a large and growing problem inthe United States. In 2009, suicide was thetenth leading cause of death in the country,accounting for 36,909 deaths. Although sui-cide rates had remained fairly steady formany years, 2009 was the fourth straightyear in which suicide rates increased, withthe suicide rate for that year, 12.0/100,000,being the highest suicide rate in a decade(McIntosh, 2012). Thus, the quest to deter-mine what factors are associated with sui-cidal behaviors is imperative.

Prior studies have shown that insom-nia symptoms and nightmares are related to

suicidal ideation, attempts, and suicides(Barbe et al., 2005; Hall, Platt, & Hall, 1999;McGirr et al., 2007; Nadorff, Nazem, &Fiske, 2011; Sj€ostr€om, Hetta, & Wærn,2009; Tanskanen et al., 2001). For example,symptoms of insomnia were found to be sig-nificantly more common in a cross-sectionalexamination of depressed youth (ages 7–17)who reported suicidal ideation than indepressed youth who did not report suicidalideation (Barbe et al., 2005). Insomnia symp-toms were also commonly reported by thosewho have made medically serious suicideattempts (Hall et al., 1999). Furthermore,two prospective studies and one retrospectivestudy found that insomnia symptoms wereassociated with death by suicide (Fawcettet al., 1990; Fujino, Mizoue, Tokui, & Yo-shimura, 2005; McGirr et al., 2007).

Evidence suggests that the relationbetween insomnia symptoms and suicidalbehaviors may be explained by depressivesymptoms. In a cross-sectional survey of583 college students, we found that the rela-tion between insomnia symptoms and sui-cidal ideation was fully mediated bydepressive symptoms (Nadorff et al., 2011).

MICHAEL R. NADORFF, Department ofPsychology, Mississippi State University, Missis-sippi State, MS, USA; SARRA NAZEM and AMY

FISKE, Department of Psychology, West VirginiaUniversity, Morgantown, WV, USA.

This research was supported in part by agrant from the National Institute of MentalHealth, grant number R15 MH80399 to AmyFiske.

Address correspondence to MichaelNadorff, Department of Psychology, MississippiState University, P.O. Box 6161, MississippiState, MS 39762; E-mail: mnadorff@psychology.msstate.edu

Suicide and Life-Threatening Behavior 1© 2012 The American Association of SuicidologyDOI: 10.1111/sltb.12003

Nonetheless, research suggests that insomniasymptoms often precede depressive symp-toms (Perlis et al., 2006), suggesting thatinsomnia symptoms may be a critical factorin the development of depression and sui-cidal ideation.

Similar findings have emerged in stud-ies that investigated the relation betweennightmares and suicidal behaviors. Night-mares have been shown to be associated withsuicidal ideation (e.g., Cukrowicz et al.,2006), suicide attempts (Sj€ostr€om, Wærn, &Hetta, 2007), and death by suicide (Tanska-nen et al., 2001). For example, in a studyexamining the relation between nightmaresand death by suicide, when compared withindividuals without nightmares, those whoreported experiencing occasional nightmareswere at 57% greater risk to die by suicide,and those who reported experiencing fre-quent nightmares were at 107% greater riskof suicide (Tanskanen et al., 2001). How-ever, unlike insomnia symptoms, nightmaresremained a significant predictor of suicidalbehavior independent of psychopathology.In a cross-sectional sample of college stu-dents, nightmares were associated with sui-cidal ideation after controlling fordepressive symptoms, anxiety symptoms,and posttraumatic symptoms (Nadorffet al., 2011). Additionally, nightmares wererelated to high suicidality scores in individu-als who had made a medically serious suicideattempt (Sj€ostr€om et al., 2007); individualswho reported having nightmares were at 4.8times greater risk of re-attempting suicidethan those who did not report experienc-ing nightmares, even after controlling forAxis I disorders (Sj€ostr€om et al., 2009).

What remains unclear from thisresearch, however, is why insomnia symptomsand nightmares are associated with suicidalbehaviors. One aspect of insomnia symp-toms and nightmares that may be relevantis the duration of symptoms. A qualitativestudy by Kyle, Espie, and Morgan (2010)found that individuals who experienced per-sistent insomnia experienced a cumulativenegative effect on work and social activities.In other words, the longer an individual

had insomnia, the more severe the negativeconsequences. Participants also reportedfeeling isolated, as well as lacking life aspi-rations.

The Interpersonal-Psychological The-ory of Suicide (Joiner, 2005) postulates thatfeelings of isolation are associated with desirefor suicide, which may help explain the rela-tion between insomnia symptoms and sui-cide. Additionally, insomnia has a negativeconsequence on one’s work and social activi-ties, which may lead the individual to believethey are a burden (Roth & Ancoli-Israel,1999). These hypotheses are supported byBryan’s 2011 study, which found insomniaseverity was positively associated withthwarted belongingness (r = .40) as well asperceived burdensomeness (r = .29), sug-gesting that insomnia symptom severity maybe predictive of these two critical suicide ide-ation risk factors.

A similar relation may exist with regardto nightmares and suicidal behaviors suchthat longer duration of nightmares may beassociated with more hopelessness and isola-tion, as the nightmare sufferer may feelunable to escape the distress and impairmentcaused by the nightmares, which could leadto feelings of hopelessness. To our knowl-edge, however, duration of insomnia symp-toms or nightmares has not been examined inrelation to suicidal outcomes.

PRESENT STUDY

The purpose of the present study wasto further characterize the associationbetween insomnia symptoms, nightmares,and suicide risk using a cross-sectionaldesign. We aimed to extend the findings ofour previous work (Nadorff et al., 2011) ina new sample by examining duration as oneaspect of insomnia symptoms and night-mares that may play a role in the relationbetween insomnia symptoms, nightmares,and suicide risk. Our first hypothesis wasthat insomnia symptom duration would beassociated with suicide risk independentof current level of insomnia symptoms.

2 INSOMNIA SYMPTOMS, NIGHTMARES, AND SUICIDE RISK

Similarly, our second hypothesis predictedthat nightmare duration would be associ-ated with suicide risk independent of cur-rent level of nightmare symptoms. Lastly,our third hypothesis predicted that night-mare duration and insomnia symptom dura-tion would be significantly related tosuicide risk after controlling for currentdepressive symptoms, anxiety symptoms,and posttraumatic symptoms, in addition tocurrent insomnia symptoms and night-mares.

METHOD

Participants

Participants were drawn from a surveyof 673 students from a large, public univer-sity in the mid-Atlantic United States. Theparticipants for the current analyses includedtwo overlapping subsets: those who endorsedany insomnia symptoms (n = 660); and thosewho endorsed any nightmares (n = 312).The sample of individuals who indorsedinsomnia symptoms was 72% female (1 miss-ing sex), with an age range of 18 to 36 years(Mean age = 20.0 years, SD = 2.4 years). Inkeeping with the characteristics of the region,most participants (92.1%) were Caucasian.The sample of individuals who endorsednightmares was 79% female, with an agerange of 18 to 36 years (Mean age = 20.0years, SD = 2.3 years). A large majority ofthe participants (92.9%) were Caucasian. Par-ticipants were recruited using the SONA sys-tem, which is an online survey manager, andwere given course credit for participating inthe study.

Measures

The Disturbing Dreams and NightmareSeverity Index. The Disturbing Dreamsand Nightmare Severity Index (DDNSI;Krakow et al., 2002), a revised version ofthe Nightmare Frequency Questionnaire(Krakow et al., 2000), was used to measure

nightmare severity and frequency for thelast year. It measures the number of nightswith nightmares per week (0–7 nights) andnumber of total nightmares per week (up to14 nightmares). The DDNSI also measuresthe severity and intensity of the nightmareson a Likert-type scale, ranging from noproblem (0) to extremely severe problem (6), aswell as how often nightmares result inawakenings, ranging from never/rarely (0) toalways (4). A score greater than 10 mayindicate the presence of a nightmare disor-der (Krakow et al., 2002). In the overallsample, the mean was 4.14 (SD = 5.84) withacceptable reliability (a = .90).

The Insomnia Severity Index. TheInsomnia Severity Index (ISI; Bastien, Valli-eres, & Morin, 2001) is a 7-item, self-reportscale that assesses the individual’s insomniasymptoms for the past 2 weeks. Each itemis scored on a 0–4 scale with total scoresranging from 0 to 28. Scores of 0–14 areconsidered to reflect no insomnia or sub-clinical insomnia, 15–21 are considered anindicator of moderate insomnia, and 22–28are considered to indicate severe insomnia.The ISI has been shown to have adequatetest–retest reliability over 3 months andconcurrent validity with sleep diaries andpolysomnography (Bastien et al., 2001;Savard, Savard, Simard, & Ivers, 2005). TheISI has also been used to determine thepresence or absence of clinically significantinsomnia symptoms using a cutoff of 15(Bernert, Turvey, Conwell, & Joiner, 2007;Tang, Wright, & Salkovskis, 2007). In theoverall sample, the mean was 8.04 (SD = 4.65)with acceptable reliability (a = .84).

Insomnia and Nightmare Duration. Insomnia symptom duration was assessed byasking participants: “If you have an insom-nia problem, how long have you had it for(please specify months and years)?” Thesedata were converted to months, resulting ina total number of months the personreported having insomnia. The data werechecked to ensure that there were no out ofrange responses. Looking at those who hadat least a score of 1 on the ISI, insomniasymptom duration ranged from 0 to

NADORFF ET AL. 3

252 months, with the mean being5.24 months (SD = 21.86). Similarly, night-mare duration was assessed using one of thequestions from the DDNSI that does notcontribute to the total score for the mea-sure. Participants were asked: “Please esti-mate the number of months or years youhave had disturbing dreams and/or night-mares. How many months? How manyyears?” (Krakow et al., 2002). As with theinsomnia symptom duration, the data wereconverted to months to provide the numberof months the participant reported havinghad nightmares. The data were checked toensure that no responses were out of range.For those who had at least a score of 1 onthe DDNSI, nightmare duration rangedfrom 0 to 360 months, with the mean being73.48 months (SD = 79.17).

The Center of Epidemiological StudiesDepression Scale. The Center of Epidemio-logical Studies Depression Scale (CES-D;Radloff, 1977) is a 20-item, self-report mea-sure of depressive symptoms. The CES-D isscored on a 4-point scale (0–3) with scoresranging from 0 to 60. A score of 16 is com-monly used as the cutoff indicating thatdepressive symptoms are clinically significant(Radloff, 1977). The CES-D has acceptableinternal consistency for both the general(a = .85) and clinical (a = .90) populations(Radloff, 1977). It has been demonstrated tobe a valid screening measure for detectingdepressive symptoms (Weissman, Sholoms-kas, Pottenger, Prusoff, & Locke, 1977). Inthe overall sample, the mean was 15.47(SD = 9.97) with acceptable reliability(a = .91). Although this mean is high, theproportion of participants over cutoff is con-sistent with findings from an earlier studyconducted at the same university (Nadorffet al., 2011), as well as studies conducted atother universities (e.g., Garlow et al., 2008).To minimize multicollinearity, the item “Mysleep was restless” (Radloff, 1977) wasexcluded from our regression analyses.

The PTSD Checklist-Civilian Ver-sion. The PTSD Checklist-Civilian Version(PCL-CL; Weathers, Litz, Herman, Huska,& Keane, 1993) is a measure of posttraumatic

stress disorder (PTSD) symptom severity. Itconsists of 17 questions which ask partici-pants to rate how much each symptom both-ers them on a scale ranging from 1 (not at all)to 5 (extremely), with total scores rangingfrom 17 to 85. There is no single, agreed-oncutoff for the PCL, but 44 and 50 have eachbeen used as cutoff scores in prior research(Ruggiero, Del Ben, Scotti, & Rabalais,2003). The PCL has high internal consistency(a = .94) and has demonstrated convergentvalidity with respect to a clinician-adminis-tered PTSD scale (Peterlin, Tietjen, Meng,Lidicker, & Bigal, 2008). In the overall sam-ple, the mean was 32.66 (SD = 11.87) withacceptable reliability (a = .92). To minimizemulticollinearity, the items “Repeated, dis-turbing dreams of a stressful experience” and“Trouble falling or staying asleep” (Weatherset al., 1993) were excluded from our regres-sion analyses.

The Zung Self-Rating Anxiety Scale. TheZung Self-Rating Anxiety Scale (SAS; Zung,1971) is a 20-item measure that assesses anxi-ety symptoms over the last several days. Eachitem is rated on a 4-point Likert-type scale.Response options range from 1 (None or a lit-tle of the time) to 4 (Most of the time). Thetotal score ranges from 20 to 80 with scoresof 50 or higher indicating clinically signifi-cant anxiety symptoms (Zung, 1986). TheSAS has been shown to discriminate betweena normal adult sample and those with anxietydisorders (Zung, 1986). In the overall sam-ple, the mean was 34.70 (SD = 8.59) withacceptable reliability (a = .87). To minimizemulticollinearity, the items “I fall asleep eas-ily and get a good night’s rest” and “I havenightmares” (Zung, 1971) were excludedfrom our regression analyses.

The Suicidal Behaviors Questionnaire-Revised. The Suicidal Behaviors Ques-tionnaire-Revised (SBQ-R; Osman et al.,2001), revised from the SBQ (Linehan,1981), is a 4-item, self-report measuredesigned to assess levels of suicidal risk.The SBQ’s four items are summed to createa score ranging from 3 to 18. The first iteminquires about past suicidal thoughts andsuicide attempts, the second and third items

4 INSOMNIA SYMPTOMS, NIGHTMARES, AND SUICIDE RISK

assess past suicidal ideation and threats, andthe fourth item asks about future suicidalbehavior. It has acceptable internal consis-tency with an alpha of 0.88 in a clinicalsample and 0.87 in a nonclinical sample(Osman et al., 2001). A cutoff score of 7 forthe general population or 8 for psychiatricinpatients may be used to determine clini-cally significant levels of suicide risk (Os-man et al., 2001). In the overall sample, themean was 4.60 (SD = 2.64) with acceptablereliability (a = .85).

Procedure

Data were collected using the SONAonline survey system. Participants loggedonto the SONA system and selected thestudy. Participants were asked to read acover sheet informing them of the purposeof the project and their rights as a partici-pant. At the conclusion of the survey,participants were shown a referral sheet withcontact information for local mental healthservices should they desire counseling fol-lowing the study. Participants were awardedcourse credit for participating in the study.

Analyses

Because data were collected via anonline survey, special steps were taken to

check data quality prior to conducting anal-yses. Specifically, participants who took10 minutes or less to complete the survey(duration M = 28.7, SD = 14.4) wereremoved from the data set. Of the original721 participants in the data set, nine wereremoved for this reason. An additional 39records were removed from the dataset dueto potentially biased response set, inconsis-tent responding, or missing data for thedependent variable. As a result, data from673 individuals comprised the overall studysample, from which participants were drawnfor the current study. Means, standard devi-ations, and proportions of the sample aboveclinical cutoff on all study measures for boththe insomnia and nightmare samples can befound in Tables 1 and 2, and the correla-tions among study variables are found inTable 3.

The study measures were also exam-ined for normality of their distributions. TheSBQ deviated from normality (skew = 2.0,kurtosis = 3.8). A log transformation wasemployed to restore normality. Followingthe transformation, the skew (1.29) andkurtosis (0.58) were reduced.

For the insomnia symptom and night-mare duration variables, insomnia symptomduration was only included for participantswith at least a score of 1 on the ISI, andnightmare duration was only included for

TABLE 1

Descriptive Statistics and the Proportion of the Sample over the Clinical Cutoff for Those Reportingat Least One Insomnia Symptoma

Measure n Mean SD Clinical cutoff % Over clinical cutoff

Insomnia symptoms 660 8.17 4.58 15 9.9%Insomnia symptom durationb 654 5.16 21.85 N/A N/ANightmares 654 4.14 5.81 11 15.1%Nightmare durationc 283 73.38 79.17 N/A N/ADepressive symptoms 636 15.53 9.97 16 44.3%PTSD symptoms 648 32.72 11.81 50 11.6%Anxiety symptoms 649 35.76 8.57 50 7.7%Suicide risk 652 4.63 2.65 7 16.6%

aStatistics are before items involving sleep were removed from the measures of depressive symp-toms, PTSD symptoms, and anxiety symptoms.

bNumber of months. Excludes individuals reporting no insomnia symptoms.cNumber of months. Excludes individuals reporting no nightmares.

NADORFF ET AL. 5

participants with at least a score of 1 on theDDNSI. This resulted in different samplesizes being used for analyses utilizinginsomnia or nightmare duration. Descrip-tive statistics and correlations of study mea-sures for both samples can be found inTables 1 through 3.

The high correlations among depres-sive symptoms, anxiety symptoms, andposttraumatic symptoms raised questionsregarding potential multicollinearity. Tohelp reduce multicollinearity, the sleep

items were removed from the covariate vari-ables (i.e., the CESD, SAS, and PCL) tominimize their shared variance with ourindependent variables. Following removal,to test whether multicollinearity was anissue in this sample, the variance inflationfactor (VIF) and tolerance were calculated.The VIF statistics for all variables wereunder 10 (range = 1.07–3.14), and tolerancewas above 0.1 (range = .32–.98), indicatingthat multicollinearity was not a significantissue for this sample.

TABLE 2

Descriptive Statistics and the Proportion of the Sample over the Clinical Cutoff for IndividualsReporting Nightmaresa

Measure n Mean SD Clinical cutoff % Over clinical cutoff

Insomnia symptoms 310 9.45 4.64 15 12.6%Insomnia symptom durationb 302 7.12 24.80 N/A N/ANightmares 312 8.85 5.60 11 32.05%Nightmare durationc 288 73.93 79.52 N/A N/ADepressive symptoms 302 17.99 10.10 16 55.6%PTSD symptoms 306 37.16 12.23 50 18.6%Anxiety symptoms 306 38.94 8.75 50 12.8%Suicide risk 307 5.14 3.01 7 22.2%

aStatistics are before items involving sleep were removed from the measures of depressive symp-toms, PTSD symptoms, and anxiety symptoms.

bNumber of months. Excludes individuals reporting no insomnia symptoms.cNumber of months. Excludes individuals reporting no nightmares.

TABLE 3

Correlations of Study Variables

Variable 1 2 3 4 5 6 7 8

1. Insomnia symptoms – .22** .36** .00 .52** .50** .61** .26**2. Insomnia duration .24** – .13** .07 .14** .10* .12** .23**3. Nightmare symptoms .32** .14* – .03 .33** .44** .46** .25**4. Nightmare duration �.00 .07 .05 – .08 .08 .01 .16**5. Depressive symptoms .52** .17** .37** .09 – .74** .78** .47**6. PTSD symptoms .44** .12* .42** .08 .72** – .77** .43**7. Anxiety symptoms .62** .13* .41** .01 .79** .75** – .35**8. Suicide risk .23** .31** .21** .17** .50** .42** .34** –

Note. Correlations for the insomnia symptom sample are above the diagonal and correlationsfor the nightmare sample are below the diagonal. Sample sizes ranged from 665 to 291, with the miss-ing data being primarily due to reporting nightmare duration for those individuals without nightmaresas missing. Correlations are after overlapping items were removed from the control variables. PTSD;posttraumatic stress disorder.

*p < .05, **p < .01.

6 INSOMNIA SYMPTOMS, NIGHTMARES, AND SUICIDE RISK

An ANOVA was utilized to test foreffects of sex and a regression analysis forage. Neither sex, F (1,662) = .74, p = .39,nor age, b = .002, p = .96, was significantlyrelated to suicide risk. Interaction effectswere calculated between age, sex, and all ofthe independent variables, with none reach-ing significance. Thus, sex and age werenot included as covariates in the analyses.All analyses were conducted using SAS v.9.2 statistical software (SAS Institute Inc.,Cary, NC, USA).

RESULTS

The first hypothesis, which predictedthat insomnia symptom duration would besignificantly associated with suiciderisk independent of level of current insom-nia symptoms, was tested using a linearregression. The overall regression wassignificant, N = 646, F(2,643) = 33.31, p <.01, R2 = .09, and both the level of insom-nia symptoms, b = .23, p < .01, and insom-nia symptom duration, b = .16, p < .01,were significantly associated with suiciderisk. Thus, the first hypothesis was sup-ported.

The second hypothesis, which pre-dicted that nightmare duration would besignificantly associated with suicide riskindependent of level of current nightmares,was tested using a linear regression. Theoverall regression was significant, N = 283,F(2,280) = 9.36, p < .01, R2 = .06, and bothcurrent level of nightmares, b = .18,p < .01, and nightmare duration, b = .16,p < .01, were significantly associated withsuicide risk. Thus, the second hypothesiswas also supported.

To test the third hypothesis, whichpredicted that insomnia symptoms andnightmare duration would be significantlyrelated to suicide risk after controlling fordepressive symptoms, anxiety symptoms, andposttraumatic symptoms, current insomniaand current nightmares, a multiple linearregression was utilized in a subset of partic-

ipants who endorsed both insomnia symp-toms and nightmares (see Table 4).Depressive symptoms, anxiety symptoms,posttraumatic symptoms, insomnia, andnightmares were added in the first step of theregression, whereas insomnia symptom dura-tion and nightmare duration were addedin the second step. The overall regressionwas significant in step one, N = 288,F(5,282) = 21.85, p < .01, R2 = .28. Depres-sive symptoms and posttraumatic symptomswere positively associated with suicide risk,and symptoms of anxiety were negativelyassociated with suicide risk. Frequency andseverity of insomnia symptoms and night-mares were not significantly related to sui-cide risk. In step two, the results were thesame except both insomnia symptom dura-tion and nightmare duration were signifi-cantly positively associated with suicide risk,and anxiety symptoms were no longerrelated to suicide risk. The R2 changebetween the first and second step was signifi-cant, R2 change = .050, F(1,258) = 19.00,p < .01. Therefore, the third hypothesis wassupported, as insomnia symptoms andnightmare duration were both associated

TABLE 4

Insomnia and Nightmare Duration PredictingSuicide Risk, Controlling for Other Symptomsof Psychopathology Including Insomnia andNightmare Symptoms*

Predictors R2 b t p

Step 1 .28 <.01Insomnia symptoms .00 .07 .95Nightmare symptoms .01 .24 .81Depressive symptoms .50 5.86 <.01Anxiety symptoms �.18 �2.03 .04PTSD symptoms .20 2.52 .01Step 2 .33Insomnia symptomduration

.23 4.26 <.01

Nightmare duration .13 2.41 .02

Significant results shown in boldface type.*Statistics are after items involving sleep

were removed from the measures of depressivesymptoms, PTSD symptoms, and anxietysymptoms.

NADORFF ET AL. 7

with suicide risk after controlling for cur-rent depressive symptoms, anxiety symp-toms, posttraumatic symptoms, insomniasymptoms, and nightmares.

DISCUSSION

The current study is significant as, tothe best of our knowledge, it is the firststudy examining duration of insomniasymptoms or nightmares in relation to sui-cide risk. The findings demonstrate that theduration of both insomnia symptoms andnightmares are associated with suicide riskindependent of each other, as well as cur-rent symptoms of psychopathology, and arethus important factors to consider in rela-tion to suicide risk.

Our findings are consistent with theresearch examining the relation betweenpersistent insomnia, nightmares, and sui-cidal behavior. Sj€ostr€om et al. (2009) fol-lowed suicide attempters for 2 years todetermine whether sleep problems increasedthe risk of future attempts. They foundstronger associations between persistentinsomnia and nightmare difficulties (definedas having insomnia or nightmares both atbaseline and 2 month follow-up followingthe attempt) and suicide attempts whencompared with the relations between base-line levels of insomnia symptoms and night-mares in relation to suicide attempts. Thus,the findings of the our study build on thefindings of Sj€ostr€om et al., suggesting thatsymptom duration is important in relationto suicide risk.

The finding that insomnia symptomduration is associated with suicide risk aftercontrolling for depressive symptoms is par-ticularly interesting, as the relation betweeninsomnia symptoms and suicide risk hasbeen found to be explained by depressivesymptoms in previous research (Cukrowiczet al., 2006; Nadorff et al., 2011). Thesefindings suggest that there may be differentmechanisms underlying the relationsbetween suicidality and insomnia symptomduration versus current insomnia symptoms.

For instance, long standing symptoms ofinsomnia may be more likely to reflect asleep disorder than to reflect symptoms ofdepression, anxiety, or PTSD, which mayexplain why these variables did not explainthe relation between insomnia symptomduration and suicide risk.

The results of the current study areconsistent with the previous literature dem-onstrating that nightmares are associatedwith suicide risk independent of symptomsof psychopathology (Cukrowicz et al., 2006;Nadorff et al., 2011). However, the currentstudy adds to the literature by demonstratingthat the duration of nightmare problems mayalso be important in explaining suicide risk.This finding has implications for clinicalpractice, as clinicians have in the past beeninstructed not to use nightmare treatments,such as Imagery Rehearsal Therapy, withsuicidal individuals due to concern that thenightmares may get worse before they getbetter (Talbot, 2009). However, the currentstudy suggests that not treating the night-mares may also lead to increased suicide riskas duration was associated with higher scoreson the SBQ-R. Therefore, research investi-gating strategies to treat nightmares in indi-viduals with suicide risk may be warranted,so long as proper precautions are taken toprotect the participants (e.g., establishing asafety plan, inpatient treatment, or frequentoutpatient visits).

Limitations

There are several limitations of ourstudy. First, the current study utilizes a col-lege population, which limits the study’sexternal validity. Nonetheless, it is animportant population to study, as suicide isthe third leading cause of death for college-age individuals (Centers for Disease Con-trol & Prevention, 2012). Second, as theSBQ-R includes questions about lifetimesuicidal ideation and behavior, it may beless sensitive than other measures to recentchanges in suicide risk. However, the SBQ-R is a commonly used, validated measure of

8 INSOMNIA SYMPTOMS, NIGHTMARES, AND SUICIDE RISK

suicide risk (Osman et al., 2001), and wassensitive enough to detect an effect in thecurrent study. It may also be argued thatthe association between insomnia symptomduration, nightmare duration, and suiciderisk may be inflated because all three arelifetime measures. However, the fact thatcurrent insomnia symptoms and nightmaresare more strongly correlated with the SBQ-R than either insomnia symptom durationor nightmare duration suggests that the factthat they are lifetime measures does notinflate their relation to suicide risk. Third,the use of a cross-sectional design is a limi-tation of this study; however, to the best ofour knowledge, the present study is the firstto link sleep difficulty duration to suiciderisk, which will hopefully lay the ground-work for prospective studies. The use of aone-item retrospective measure of insomniaand nightmare duration is a limitation as itdoes not provide a way for the authors toensure the veracity of the duration reports.However, there are no validated measuresof insomnia and nightmare duration in theliterature. Further, although the use of aretrospective measure is a limitation, itallowed us to lay the groundwork for futureexploration by establishing that there is arelation between duration of insomniasymptoms, nightmares, and suicide risk.

Overall, the findings from this studyare an important first step, as the relationsbetween insomnia symptom duration, night-mare duration, and suicide risk have not beenexamined previously. Thus, the currentresults provide an avenue for future work.

Future Directions

Based on the results of this study, sev-eral areas of future work are suggested. First,prospective research is needed to furtherexamine the relation between insomnia andnightmare duration and suicide risk.Researchers may want to examine whethernightmare or insomnia symptom durationare associated with greater hopelessness orlower feelings of belongingness, as thesevariables may mediate the relation between

insomnia symptom duration, nightmareduration, and suicide risk. Second, futurestudies should examine the relation betweeninsomnia symptom duration, nightmareduration, and suicide risk in different agegroups, especially those at higher risk, todetermine whether these findings generalizebeyond a college-age sample. Additionally,future research utilizing diagnoses instead ofsymptoms of disorders would be valuable tohelp determine whether insomnia symptomsand nightmares due to a psychiatric disorderdiffer from those due to the presence of asleep disorder in relation to suicidal behav-ior. Finally, prospective or retrospectiveresearch examining insomnia symptom andnightmare duration in suicide decedents mayincrease the external validity of the currentstudy beyond suicide risk.

Our finding that insomnia symptomsand nightmare duration were significantlyassociated with suicide risk independent ofdepressive symptoms, anxiety symptoms,and posttraumatic symptoms raise the ques-tion of whether treating sleep disorders maybe important for reducing suicide risk inconjunction with other treatments. Researchhas found that adding cognitive behavioraltherapy for insomnia to antidepressanttreatment leads to higher rates of remissionfrom depression (Manber et al., 2008). Sim-ilarly, sleep treatments, such as cognitivebehavioral therapy for insomnia and imag-ery rehearsal therapy for nightmares, couldbe combined with other treatmentsdesigned to reduce suicide risk in individu-als with both suicide risk and sleep difficul-ties. Based on our findings, researchinvestigating the efficacy of adding sleeptreatment to standard treatment for suiciderisk would be very interesting.

REFERENCES

BARBE, R. P., WILLIAMSON, D. E., BRIDGE,J. A., BIRMAHER, B., DAHL, R. E., AXELSON, D.A., et al. (2005). Clinical differences between sui-cidal and nonsuicidal depressed children andadolescents. The Journal of Clinical Psychiatry, 66,492–498.

NADORFF ET AL. 9

BASTIEN, C. H., VALLIERES, A., & MORIN,C. M. (2001). Validation of the Insomnia Sever-ity Index as an outcome measure for insomniaresearch. Sleep Medicine, 2, 297–307.

BERNERT, R., TURVEY, C., CONWELL, Y.,& JOINER, T. (2007). Sleep disturbance as aunique risk factor for completed suicide. Sleep,30, A344.

BRYAN, C. J. (2011). The clinical utility ofa brief measure of perceived burdensomenessand thwarted belongingness for the detection ofsuicidal military personnel. Journal of ClinicalPsychology, 67, 981–992.

Centers for Disease Control and Preven-tion. (2012). Web-based Injury Statistics Query andReporting System. Retrieved February 10, 2012, fromhttp://www.cdc.gov/injury/wisqars/index.html.

CUKROWICZ, K. C., OTAMENDI, A., PINTO,J. V., BERNERT, R. A., KRAKOW, B., & JOINER, T.E. JR. (2006). The impact of insomnia and sleepdisturbances on depression and suicidality.Dreaming, 16, 1–10.

FAWCETT, J., SCHEFTNER, W. A., FOGG,L., CLARK, D. C., YOUNG, M. A., HEDEKER, D.,et al. (1990). Time-related predictors of suicidein major affective disorder. The American Journalof Psychiatry, 147, 1189–1194.

FUJINO, Y., MIZOUE, T., TOKUI, N., &YOSHIMURA, T. (2005). Prospective cohort studyof stress, life satisfaction, self-rated health,insomnia, and suicide death in Japan. Suicide andLife-Threatening Behavior, 35, 227–237.

GARLOW, S. J., ROSENBERG, J., MOORE, J.D., HAAS, A. P., KOESTNER, B., HENDIN, H.,et al. (2008). Depression, desperation, and sui-cidal ideation in college students: Results fromthe American Foundation for Suicide PreventionCollege Screening Project at Emory University.Depression and Anxiety, 25, 482–488.

HALL, R. C., PLATT, D. E., & HALL, R.C. (1999). Suicide risk assessment: A review ofrisk factors for suicide in 100 patients who madesevere suicide attempts: Evaluation of suicide riskin a time of managed care. Psychosomatics, 40,18–27.

JOINER, T. (2005). Why people die by suicide.Cambridge: Harvard University Press.

KRAKOW, B., HOLLIFIELD, M., SCHRADER,R., KOSS, M., TANDBERG, D., LAURIELLO, J., et al.(2000). A controlled study of imagery rehearsalfor chronic nightmares in sexual assault survivorswith PTSD: A preliminary report. Journal ofTraumatic Stress, 13, 589–609.

KRAKOW, B., MELENDREZ, D. C., JOHNSTON,L. G., CLARK, J. O., SANTANA, E. M., WARNER,T. D., et al. (2002). Sleep dynamic therapy forCerro Grande Fire evacuees with posttraumaticstress symptoms: A preliminary report. The Journalof Clinical Psychiatry, 63, 673–684.

KYLE, S. D., ESPIE, C. A., & MORGAN, K.(2010). “… Not just a minor thing, it is some-thing major, which stops you from functioningdaily”: Quality of life and daytime functioning ininsomnia. Behavioral Sleep Medicine, 8, 123–140.

LINEHAN, M. M. (1981). The SuicidalBehaviors Questionnaire (SBQ). Unpublishedmanuscript, University of Washington.

MANBER, R., EDINGER, J. D., GRESS, J. L.,SAN PEDRO-SALCEDO, M. G., KUO, T. F., & KA-

LISTA, T. (2008). Cognitive behavioral therapyfor insomnia enhances depression outcome inpatients with comorbid major depressive disorderand insomnia. Sleep, 31, 489–495.

MCGIRR, A., RENAUD, J., SEGUIN, M.,ALDA, M., BENKELFAT, C., LESAGE, A., et al.(2007). An examination of DSM-IV depressivesymptoms and risk for suicide completion inmajor depressive disorder: A psychologicalautopsy study. Journal of Affective Disorders, 97,203–209.

MCINTOSH, J. L. (2012). U.S.A. suicide2009: Official final data. Retrieved February 14,2012, from http://www.suicidology.org.

NADORFF, M. R., NAZEM, S., & FISKE, A.(2011). Insomnia symptoms, nightmares, and sui-cidal ideation in a college student sample. Sleep,34, 93–98.

OSMAN, A., BAGGE, C. L., GUTIERREZ, P.M., KONICK, L. C., KOPPER, B. A., & BARRIOS, F.X. (2001). The Suicidal Behaviors Questionnaire–Revised (SBQ-R): Validation with clinical andnonclinical samples. Assessment, 8, 443–454.

PERLIS, M. L., SMITH, L. J., LYNESS, J. M.,MATTESON, S. R., PIGEON, W. R., JUNGQUIST, C.R., et al. (2006). Insomnia as a risk factor foronset of depression in the elderly. BehavioralSleep Medicine, 4, 104–113.

PETERLIN, B. L., TIETJEN, G., MENG, S.,LIDICKER, J., & BIGAL, M. (2008). Post-traumaticstress disorder in episodic and chronic migraine.Headache, 48, 517–522.

RADLOFF, L. S. (1977). The CES-D Scale:A self-report depression scale for research in thegeneral population. Applied Psychological Measure-ment, 1, 385–401.

ROTH, T., & ANCOLI-ISRAEL, S. (1999).Daytime consequences and correlates of insomniain the United States: Results of the 1991 NationalSleep Foundation Survey II. Sleep, 22 (Suppl 2),S354–S358.

RUGGIERO, K. J., DEL BEN, K., SCOTTI, J.R., & RABALAIS, A. E. (2003). Psychometric prop-erties of the PTSD Checklist-Civilian Version.Journal of Traumatic Stress, 16, 495–502.

SAVARD, M., SAVARD, J., SIMARD, S., &IVERS, H. (2005). Empirical validation of theInsomnia Severity Index in cancer patients. Psy-cho-Oncology, 14, 429–441.

10 INSOMNIA SYMPTOMS, NIGHTMARES, AND SUICIDE RISK

SJ€oSTR€oM, N., HETTA, J., & WÆRN, M.(2009). Persistent nightmares are associated withrepeat suicide attempt: A prospective study. Psy-chiatry Research, 170, 208–211.

SJ€oSTR€oM, N., WÆRN, M., & HETTA, J.(2007). Nightmares and sleep disturbances inrelation to suicidality in suicide attempters. Sleep,30, 91–95.

TALBOT, M. (2009, November 16). Night-mare scenario, can we learn to rewrite our baddreams. New Yorker, 43–51.

TANG, N. K. Y., WRIGHT, K. J., &SALKOVSKIS, P. M. (2007). Prevalence andcorrelates of clinical insomnia co-occurring withchronic back pain. Journal of Sleep Research, 16,85–95.

TANSKANEN, A., TUOMILEHTO, J., VIINA-

MAKI, H., VARTIAINEN, E., LEHTONEN, J., & PUSKA,P. (2001). Nightmares as predictors of suicide.Sleep, 24, 844–847.

WEATHERS, F. W., LITZ, B. T., HERMAN,D. S., HUSKA, J. A., & KEANE, T. M. (1993,

October). The PTSD checklist (PCL); reliability,validity, and diagnostic utility. Paper presented atthe International Society for Traumatic StressStudies, San Antonio, TX.

WEISSMAN, M. M., SHOLOMSKAS, D.,POTTENGER, M., PRUSOFF, B. A., & LOCKE, B.Z. (1977). Assessing depressive symptoms infive psychiatric populations: A validationstudy. American Journal of Epidemiology, 106,203–214.

ZUNG, W. W. (1971). A rating instrumentfor anxiety disorders. Psychosomatics, 12, 371–379.

ZUNG, W. W. (1986). Prevalence of clini-cally significant anxiety in a family practice set-ting. The American Journal of Psychiatry, 143,1471–1472.

Manuscript Received: March 3, 2012Revision Accepted: August 31, 2012

NADORFF ET AL. 11