Suicide Attempt Characteristics, Diagnoses, and Future Attempts: Comparing Multiple Attempters to Single Attempters and Ideators

Suicide Attempt Characteristics, Diagnoses, and Future Attempts: Comparing Multiple Attempters to Single Attempters and Ideators REGINA MIRANDA, PH.D., MICHELLE SCOTT, PH.D., ROGER HICKS, M.B.A., HOLLY C. WILCOX, PH.D., JIMMIE LOU HARRIS MUNFAKH, B.A., AND DAVID SHAFFER, M.D.

ABSTRACT Objective: To compare psychiatric diagnoses and future suicide attempt outcomes of multiple attempters (MAs), single attempters (SAs), and ideators. Method: Two hundred twenty-eight teens who reported recent ideation or a lifetime suicide attempt in a screening of 1,729 high school students completed the Adolescent Suicide Interview, which provided information on attempt number and characteristics and mood, anxiety, and substance use disorder modules of the Diagnostic Interview Schedule for Children; 191 were reinterviewed 4 to 6 years later to ascertain interval attempts and psychiatric disorder. Between screening and follow-up, 33 (17%) teens made an attempt, 12 of whom were previously classified as lifetime MAs (more than one attempt) and six as SAs. Results: MAs more often met criteria for any one of the DSM diagnoses assessed at baseline (mood, anxiety, or substance use disorder; 71%), compared with SAs (39%) and ideators (41%), and at follow-up (mood, anxiety, substance use, or disruptive behavior disorder; 69%) compared with SAs (36%) (p < .05). As reported at baseline, MAs (versus SAs) more often wished to die during their attempt (53% versus 23%), less often planned their attempt for intervention (44% versus 76%), and more often regretted recovery (26% versus 7%; p < .05). Baseline MAs had significantly higher odds of making a later attempt compared to ideators (odds ratio 4.0, 95% confidence interval 1.5Y10.2) and SAs (odds ratio 4.6, 95% confidence interval 1.0Y20.2). No participants committed suicide during follow-up. SAs who made another attempt (versus those who did not) more often met criteria for a baseline anxiety disorder and more often wished to die during their baseline attempt. Conclusions: MAs more strongly predict later suicidality and diagnosis than SAs and ideation. Forms that assess past suicide attempts should routinely inquire about frequency of attempts. The similarity between the present findings and those of clinical samples suggests that screening may yield a representative sample of suicide attempters and ideators. J. Am. Acad. Child Adolesc. Psychiatry, 2008; 47(1):32Y40. Key Words: nonreferred suicide attempters, multiple attempters, suicidal intent, longitudinal.

A history of a suicide attempt has been identified as one of the best predictors of a future attempt among clinical and community samples of teenagers and

among adult clinical samples for completed suicide.1Y5 Furthermore, there is evidence that having made more than one suicide attempt increases the prediction of

Accepted August 13, 2007. Dr. Miranda is with the Department of Psychology, Hunter College, City University of New York; Drs. Miranda, Scott, and Shaffer, and Mr. Hicks and Ms. Munfakh are with the Division of Child and Adolescent Psychiatry, Columbia University, College of Physicians and Surgeons; Dr. Wilcox is with the Johns Hopkins University School of Medicine. Statistical analyses were conducted by Regina Miranda, Ph.D., with consultation from J. Blake Turner, Ph.D. Portions of this article were presented at the 53rd Annual Meeting of the American Academy of Child and Adolescent Psychiatry, San Diego, October 24Y29, 2006. This work was funded by grant R49/CCR 202598 from the Centers for Disease Control and Prevention, NIMH grants P30 MH 43878 and

ST32MH-16434, and grants from the American Mental Health Foundation and the Carmel Hill Fund at Columbia University. The authors thank Madelyn Gould, Matthew Nock, Mark Olfson, J. Blake Turner, and Eva De Jaegere for comments on previous versions of this manuscript. Correspondence to Dr. Regina Miranda, Division of Child and Adolescent Psychiatry, Columbia University, College of Physicians and Surgeons, 1051 Riverside Drive, Unit 78, New York, NY 10032; e-mail: mirandar@ childpsych.columbia.edu. 0890-8567/07/4701-0032Ó2007 by the American Academy of Child and Adolescent Psychiatry. DOI: 10.1097/chi.0b013e31815a56cb

32

WWW.JAACAP.COM

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 47:1, JANUARY 2008

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

MULTIPLE ADOLESCENT SUICIDE ATTEMPTERS

risk for further attempts.6,7 A cross-national European sample of 1,264 clinically referred teens found that adolescent repeat attempters had more than three times higher odds of making another suicide attempt within 1 year compared with first-ever attempters.7 In a follow-up study of 180 adolescents ages 12 to 18, Goldston and colleagues6 found that a history of more than one previous suicide attempt predicted increased risk for a subsequent attempt compared to history of only one previous attempt or no attempt 5 years after hospitalization. Because these were clinical studies and thus susceptible to referral bias, it would be of interest to determine whether the same findings held for nonreferred samples. The authors are not aware of any similar investigations with nonreferred samples. Several possible mechanisms have been proposed to explain the increased risk conferred by multiple attempts. One possibility is that multiple attempts reflect persistent risk factors (e.g., a chronic or recurring psychiatric disorder or adverse psychosocial conditions). Cross-sectional studies of adolescents have demonstrated a relationship between multiple attempts and diagnosis. Esposito et al.8 studied 74 single attempters (SAs) and 47 multiple attempters (MAs; ages 12Y18) seen in an emergency department after a suicide attempt and found higher rates of mood disorder diagnosis among MAs. Rosenberg and colleagues,9 analyzing data drawn from the Youth Risk Behavior Survey, found that high school students (ages 13Y18) who endorsed multiple suicide attempts in the previous 12 months had more than seven times higher odds of concurrently reporting depressed mood, more than two times higher odds of using alcohol or marijuana, and more than three times higher odds of using other illegal drugs compared with SAs. D_Eramo and colleagues10 found that among consecutive admissions to an adolescent psychiatric inpatient unit, MAs had higher rates of psychiatric comorbidity compared to SAs, ideators, and nonsuicidal teens between the ages of 12 and 17. Joiner11 suggests that multiple suicide attempts increase the risk for subsequent attempts because practice allows MAs to acquire the ability to engage in more serious suicidal behavior. Longitudinal evidence of increases in proficiency or changes in methods over time among MAs is not available, although a study of low-income African American (adult) female

J. AM . ACAD. CHILD ADOLESC. PSYCH IATRY, 47:1, JANUARY 2008

attempters examined in an emergency department found that MAs showed higher degrees of planning, lethality of intent, and feelings of regret about recovering from their attempt.12 Confirming and understanding the association could be of public health and clinical value because it may point to clinical features of an attempt that are indicative of greater future risk. The present report examined the following questions: whether MAs would be distinguished from SAs and ideators by higher rates of psychiatric diagnoses, both at baseline and follow-up; whether MAs would be distinguished from SAs in the characteristics of their most recent suicide attempt, assessed at baseline; and whether a history of multiple suicide attempts, compared to a history of only one suicide attempt and ideation, would predict a future suicide attempt, above and beyond demographic variables and a psychiatric diagnosis. These questions were examined in a nonreferred sample of adolescent MAs, SAs, and suicide ideators identified in a screening and followed up 4 to 6 years later as part of a larger study designed to assess the predictive significance of the characteristics of an attempt and of ideation. Analysis of this nonreferred sample also offered the opportunity to determine whether characteristics of suicidal teens identified in clinical samples are also present in a screening sample, which would provide evidence that a sample identified by screening is more generally representative of suicide attempters and ideators. METHOD Participants and Procedure Baseline Assessment. Participants were 228 teenagers, ages 12 to 18 (mean 15.5, SD 1.3), who endorsed a lifetime suicide attempt or recent suicidal ideation during a two-stage screening procedure.13 The majority of teens (63%) were female, and ethnic breakdowns were as follows: 51% white, 15% African American, 20% Hispanic, 9% Asian, and 5% other ethnicities. Teens were recruited from a high school screening (N = 1,729) between 1991 and 1994, from a population of 2,583 students from seven high schools in the New York metropolitan area. Teens who endorsed a lifetime attempt or recent suicidal ideation (past 3 months) were recruited to be interviewed further about their suicide attempt history.13 Forms were sent home to parents describing the study and giving them an opportunity to participate. Research staff obtained written assent from teens whose parents had not denied consent and who were interested in taking part in the study. No action was taken after the screen (e.g., referral for treatment) without active

WWW.JAACAP.COM

33

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

MIRANDA ET AL.

parental consent. Individuals were positive for suicidality if they indicated that they had ever made a suicide attempt, thought about suicide, or both, on the Columbia Suicide Screen (CSS; N = 298; 42 attempt only, 190 ideation only, and 66 both), which was administered to the entire sample of students,13 or if during a later interview they verbally reported suicidal thoughts or past attempts to the study interviewer or to a clinician. A total of 228 teenagers completed the Adolescent Suicide Interview (ASI) to assess whether they had made a suicide attempt. Of the 298 teens who endorsed suicidal ideation or an attempt on the CSS, those who completed the ASI were representative in terms of age, but more ASI completers (versus noncompleters) were male (35% versus 18%) and fewer were female (65% versus 82%) than would be expected by chance, 2 = 9.10; p < .01, and more were Hispanic (20% versus 11%) than expected, 2 = 3.92; p < .05. Two hundred twenty-four teens completed the Diagnostic Interview Schedule for Children (DISC) to assess for a selected psychiatric diagnosis,14 and 80 teens who endorsed a suicide attempt on the ASI were interviewed about characteristics of their most recent attempt. Follow-up Assessment. Four to 6 years later, 191 (84%; 67 attempters, 124 ideators) of the 228 teenagers were reinterviewed as part of a broader follow-up study (Shaffer and colleagues, in preparation) of 641 (of the original 1,729) teenagers who completed both the CSS and DISC at baseline (552 of 641 teens provided follow-up data: 13 were not located, 65 refused the interview, 4 were unavailable, 4 were deceased for reasons other than suicide, and 3 provided incomplete interviews). Individuals were contacted via telephone, letters, telegrams, or home visits and were readministered an adapted version of the ASI by telephone (or in person, when reached by a home visit) to assess whether they had made a suicide attempt since the original interview. Consent was obtained by telephone from adult participants using a telephone consent script, and both parental consent and adolescent assent were obtained by telephone from participants younger than age 18. Interviews were conducted by lay interviewers under the supervision of a psychiatrist. The followup sample did not differ significantly from the initial sample in sex, ethnicity, DISC diagnosis, and screen endorsement of ideation or an attempt. The average age of participants at follow-up was 20.8 years (SD 1.4). It should be noted that previous research suggests no iatrogenic risk for asking teenagers questions about suicidal ideation or behavior during a high school screening.15 The materials and consent procedures used in this study were approved by the institutional review boards of the New York State Psychiatric Institute, the New York State Board of Education, and the Archdiocese of New York. Measures CSS. Two questions from the CSS assessed for history of suicidality: lifetime suicide attempts (BHave you ever tried to kill yourself?[) and suicidal ideation in the past 3 months (BDuring the past 3 months, have you thought about killing yourself?[).13 These items were scored using a binary scale (i.e., yes versus no) and were embedded within a larger 32-question health survey. DISC. The DISC, version 2.3,14 is an instrument administered by lay interviewers to establish probable psychiatric diagnoses, consistent with DSM-III-R criteria in children and adolescents ages 6 to 17 years, including whether symptoms cause clinically significant impairment.14 Modules reflecting symptoms of unipolar

34

WWW.JAACAP.COM

depression, anxiety, and substance use, assessed for the previous 6 months, were administered at baseline. Diagnoses at follow-up were assessed using the DISC-IV, a computer-based adaptation of the DISC 2.3, administered by lay interviewers.16 All of the modules reflecting DSM-IV diagnostic criteria (including mood, anxiety, substance use, and disruptive behavior disorders), assessed for the previous year, were administered. ASI. The ASI is a semistructured interview designed by Shaffer and colleagues to obtain information about the total number of suicide attempts that a teenager has made and the circumstances surrounding the adolescent_s most recent attempt. It consists of a standard glossary of terms and rating scales with descriptive anchor points. The probes used to obtain information and the order in which questions are posed are left to the interviewer. Respondents were asked an initial probe question (e.g., BI want to ask you about the last time you tried to kill yourself. Can you tell me what you actually did?[), and, when they had completed that account, they were asked additional questions to complete the rating scale attached to 36 questions covering characteristics of suicide attempts, including method, warnings provided before the attempt, length of planning, preparatory behavior, isolation, timing of the attempt, wish to die, and feelings after the attempt. Suicide attempt at follow-up, assessed blind to baseline attempt history, was determined using an adapted version of the ASI that began by inquiring about lifetime suicide attempt history, participants_ most recent suicide attempt, and whether the most recent attempt occurred after the screening. Participants who indicated that their most recent suicide attempt had occurred after the screening were considered to have made an attempt at followup. Agreement in endorsement of a lifetime suicide attempt between the ASI and the CSS was high (. = .77; p < .01). Agreement in endorsement of a lifetime attempt on the ASI between baseline and follow-up was adequate (. = .56; p < .01). Data Analysis Respondents were classified as being SAs versus MAs based on their baseline response to a probe on the ASI inquiring about total number of suicide attempts (i.e., BHow many times, in your whole life, have you tried to kill yourself?[). A respondent_s classification was not changed if the number of attempts reported as having occurred (before the follow-up period) varied between information provided at baseline and follow-up (this was the case for one participant). Individuals who did not endorse an attempt on the ASI were classified as ideators. Differences among MAs, SAs, and ideators in rates of diagnosis and between MAs and SAs in baseline suicide attempt characteristics were examined via 2 statistics ( p values determined using the Fisher exact test when expected cell sizes were <5 in 2  2 analyses). Logistic regression analyses were conducted to predict diagnosis at baseline and follow-up, adjusting for age (measured continuously), sex (with males as the reference group), and ethnicity (with white participants as the reference group), and to determine whether a history of multiple attempts would statistically predict characteristics of adolescents_ most recent suicide attempt at baseline, adjusting for demographic variables. Finally, two logistic regression analyses were used to test the predictions that MAs, but not SAs, would have significantly higher odds of making a subsequent suicide attempt, compared to ideators, and that MAs would have significantly higher odds of making a subsequent attempt, compared to SAs. Both regressions adjusted for demographic variables and the presence of a baseline DISC diagnosis. All of the statistical tests were two-tailed.

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 47:1, JANUARY 2008

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

MULTIPLE ADOLESCENT SUICIDE ATTEMPTERS

were no statistically significant differences between SAs and MAs in rates of a mood or substance use disorder. There were no baseline differences in diagnosis between SAs and ideators ( p >.10).

RESULTS Demographic Differences by Attempt Status

Most teens who completed the ASI at baseline indicated that they had not made a suicide attempt in the past (i.e., ideation only; n = 148; 65%); however, 20% (n = 45) indicated they had made one attempt (i.e., a single attempt), and 15% (n = 35) indicated they had made two or more attempts (i.e., multiple attempts). The average number of attempts endorsed by MAs at baseline was 2.83 (SD 1.07; range 2Y5). There was no significant omnibus sex difference in attempt status (Table 1). However, compared to ideators, a higher proportion of teens who reported any past attempt (single or multiple) were female (2 = 5.04; p < .05). A lower proportion of SAs were white and a higher proportion was Hispanic versus other ethnicities. Finally, SAs were on average older than either ideators or MAs.

Diagnostic Characteristics at Follow-up

At follow-up, MAs were significantly more likely than SAs (but not more likely than ideators) to meet criteria for a disruptive behavior disorder (2 = 6.60; adjusted p < .05) and for any type of DISC diagnosis (i.e., mood, anxiety, substance use, or disruptive behavior disorder; 2 = 6.94, p < .01). They were marginally more likely than SAs to meet criteria for a substance use disorder (2 = 3.81; p = .05), but not after adjusting for demographic characteristics and not when compared to ideators. MAs were not more likely to meet criteria specifically for a mood, anxiety, or substance use disorder compared to ideators and SAs (Table 2). There were no statistically significant differences in diagnoses at follow-up between SAs and ideators.

Diagnostic Characteristics at Baseline

MAs compared to ideators had higher rates of any DISC diagnosis (2 = 10.22, p < .01), whether mood (2 = 7.71, p < .01), anxiety (2 = 8.48, p < .01), substance use (2 = 5.32, p < .05), or more than one of these three types of diagnoses (2 = 10.70, p < .01; Table 2). MAs had significantly higher rates of an anxiety disorder (2 = 6.09, p < .05) and of any DISC diagnosis (2 = 8.42, p < .01) than SAs. However, there

Suicide Attempt Characteristics Reported at Baseline

Teens reported making their attempts by ingesting a substance (56%; n = 43), using a cutting instrument (29%; n = 23), or some other method (18%; n = 14). Females were more likely to attempt by ingestion (64%) compared with males (33%; 2 = 5.93; p < .05), whereas males were more likely than females to use a

TABLE 1 Demographic Characteristics at Baseline of 228 ASI Completers by Attempt Status Ideation Onlya (n = 148) Single (n = 45) Multiple (n = 35) Age, Mean 15.4 (SD 1.3) Sex Male, n = 85 Female, n = 143 Ethnicity White, n = 116 Black, n = 34 Hispanic, n = 46 Asian, n = 21 Other, n = 11

Age, Mean 16.0 (SD 1.4)

Age, Mean 15.3 (SD 1.3)

F = 4.81

p < .01



pb

2

N

None, %

N

Single, %

N

Multiple, %

63 85

43 57

11 34

24 76

11 24

31 69

5.45

.07

80 22 23 14 9

54 15 16 10 6

15 8 17 4 1

33 18 38 9 2

21 4 6 3 1

60 11 17 9 3

7.31 .63 10.83 .03 1.47

.03 .73 <.01 .98 .48

Note: ASI = Adolescent Suicide Interview. a No attempts. b p Value estimated for a 2 test with 2 df. Ethnicity 2 analyses for each ethnic group use; all other groups as the reference category.

J. AM . ACAD. CHILD ADOLESC. PSYCH IATRY, 47:1, JANUARY 2008

WWW.JAACAP.COM

35

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

MIRANDA ET AL.

TABLE 2 DISC Diagnostic Characteristics of Single and Multiple Attempters

Baseline diagnosis Any mood (n = 217) Any anxiety (n = 191) Any substance (n = 217) Any dx (n = 224) >1 dx type (n = 224) Follow-up diagnosis Any mood (n = 184) Any anxiety (n = 188) Any substance (n = 186) Any disruptive (n = 177) Any dx (n = 186) >1 dx type (n = 186)

Ideation (n = 145),a No. (%)

Single (n = 44),a No. (%)

Multiple (n = 35), No. (%)

31 (22) 40 (33) 15 (11) 59 (41) 25 (17) Ideation (n = 123)a No. (%)

12 (29) 12 (32) 4 (10) 17 (39) 10 (23) Single (n = 36) No. (%)

15 (46) 19 (61) 9 (26) 25 (71) 15 (43) Multiple (n = 29)a No. (%)

12 (10) 22 (18) 44 (36) 8 (7) 62 (51) 19 (16)

1 (3) 7 (19) 9 (25) 0 (0) 13 (36) 4 (11)

4 (14) 7 (24) 14 (48) 5 (18) 20 (69) 7 (24)

Multiple vs. Single Attempter 

2

p

OR

95% CI

2.29 6.09 3.57 8.42 3.65

.13 .01 .06 <.01 .06

2.7 4.7 4.9 6.1 2.9

0.9Y8.3 1.4Y15.7 1.0Y23.4 2.0Y18.8 0.9Y8.7

2.90 .21 3.81 6.60 6.94 1.94

.16b .65 .05 .02b <.01 .20b

5.0 1.4 2.5 NA 3.4 3.1

0.4Y58.6 0.4Y4.9 0.7Y8.4 NA 1.1Y10.7 0.7Y14.4

Note: Baseline diagnoses: mood = major depressive disorder or dysthymic disorder; Anxiety = panic disorder, agoraphobia, social phobia, generalized anxiety disorder, or overanxious disorder; substance use = alcohol abuse/dependence, marijuana abuse/dependence, or other substance use/dependence. Diagnoses assessed at follow-up: mood = major depressive disorder, or dysthymic disorder, or mania; anxiety = panic disorder, social phobia, generalized anxiety disorder, obsessive-compulsive disorder, or posttraumatic stress disorder; substance use = alcohol abuse/dependence, nicotine dependence, marijuana abuse/dependence, or other substance use/dependence; disruptive = attention-deficit/ hyperactivity disorder, oppositional defiant disorder, or conduct disorder. Odds ratios are adjusted for age, sex, and ethnicity. DISC = Diagnostic Interview Schedule for Children; OR = odds ratio; CI = confidence interval; dx = diagnosis. a Three of 148 ideators and one of 45 single attempters did not complete the DISC at baseline. One ideator and two MAs did not provide DISC data at follow-up. b p Value determined using the Fisher exact test due to expected cell values <5.

method other than ingestion or cutting (36%) compared with females (10%; 2 = 7.48; adjusted p < .05). Most attempters spent less than 1 hour planning their attempt (70%), did not engage in any preparatory behavior (e.g., writing a suicide note) before their attempt (92%), and gave no warnings before the attempt (71%). Males were more likely than females to spend Q1 hour planning their attempt (50% and 22%, respectively [2 = 5.72; p < .05]), and girls were more likely than boys to time their attempts for intervention (70% and 40%, respectively [2 = 5.46; p < .05]). There were no other sex differences in attempt characteristics examined. MAs more often reported wishing to die versus not wishing to die compared with SAs (2 = 6.98; p < .01), and they were more often uncertain or sorry to have recovered from their most recent attempt compared to SAs (26% and 7%, respectively [2 = 5.20; p < .05]; Table 3). SAs more often reported planning their attempt so that intervention was either possible or likely (76%) compared with MAs (44%;

36

WWW.JAACAP.COM

2 = 7.72; p < .01). There were no significant differences between groups on method of attempt, previous warnings or threats, time spent planning, and preparatory behaviors. After adjusting for age, sex, and ethnicity, MAs had more than three times higher odds of endorsing a wish to die (or being uncertain) versus not wishing to die at the time of the attempt (odds ratio [OR] 3.4, 95% confidence interval [CI] 1.1Y10.5; p < .05) and more than four times higher odds of reporting that they were sorry or uncertain about having recovered (OR 4.4, 95% CI 1.0Y20.2; p = .05). They also had significantly lower odds of timing their most recent attempt so that intervention was possible or likely than SAs (OR 0.3, 95% CI 0.1Y0.9; p < .05). The width of the CIs in these analyses can likely be attributed to imprecision of measurement due to sample size. Suicide Attempts at Follow-up

A total of 33 (17%) of 191 participants in the followup sample (31 MAs, 36 SAs, and 124 ideators, as

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 47:1, JANUARY 2008

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

MULTIPLE ADOLESCENT SUICIDE ATTEMPTERS

TABLE 3 Suicide Attempt Characteristics as Reported at Baseline Single (n = 45), Multiple (n = 35), No. (%) No. (%) Method (includes 3 participants with multiple methods) Ingested substance (n = 77) Cutting instrument (n = 80) Other method (n = 80) Warning/threats (n = 80) None Veiled threat(s) Specific threat(s) Planning (n = 76) <1 hour 1Y24 hours >24 hours Preparatory behavior (n = 79) Isolation: alone (n = 78) Timing so that intervention possible/likely (n = 73) Wish to die at time of attempt (n = 78) Did not want to die Uncertain/did not care whether lived/died Wanted to die Regretted recovery or uncertain (n = 78)

23 (55) 11 (24) 7 (16)

20 (57) 12 (34) 7 (20)

33 (73) 6 (13) 6 (13)

24 (69) 4 (11) 7 (20)

32 (78) 5 (12) 4 (10) 2 (5) 11 (25) 31 (76)

21 (60) 8 (23) 6 (17) 4 (11) 11 (32) 14 (44)

18 (41) 16 (36) 10 (23) 3 (7)

7 (21) 9 (27) 18 (53) 9 (26)

2

p

.04 .93 .27 .66

.83 .34 .60 .72

2.92

.23

1.32 .51 7.72 7.93

.40a .47 <.01 .02

6.98b 5.20

<.01 .03a

Note: The number of participants who responded to each Adolescent Suicide Interview question is indicated in parentheses in each row after the item label. a p Value determined using the Fisher exact test due to expected cell values <5. b Chi square is for 2  2 comparison of Bwanted to die[ vs. Bdid not want to die.[

classified at baseline) endorsed having made a suicide attempt after the initial suicide interview. Of these, 12 had previously been classified as MAs, 6 had been SAs, and 15 were ideators. Thus, 17% of SAs became MAs during follow-up, whereas 39% of MAs made a subsequent attempt (2 = 4.12; p < .05). The majority of attempters reported ingestion of a substance (n = 22; 67%). The remainder endorsed the use of a cutting instrument (n = 3; 9%); a handgun (n = 3; 9%); hanging, choking, or asphyxiation (n = 3; 9%); or another method (n = 2; 6%). MAs had significantly higher odds of endorsing a subsequent attempt at follow-up compared to ideators (OR 4.0, 95% CI 1.5Y10.2; p < .01) and compared to SAs (OR 4.6, 95% CI 1.0Y20.2; p < .05), after adjusting for demographic variables and the presence of any baseline DISC diagnosis (Table 4). However, SA status (compared to ideation) did not significantly predict a future attempt. Being Hispanic versus white significantly increased the odds of a subsequent attempt (OR 3.0, 95% CI 1.0Y8.6; p < .05).

J. AM . ACAD. CHILD ADOLESC. PSYCH IATRY, 47:1, JANUARY 2008

SAs Who Became MAs

Differences between baseline SAs who became MAs at follow-up (i.e., new MAs; n = 6) and baseline SAs who did not make a subsequent attempt (n = 30) were examined in both diagnostic and baseline suicide attempt characteristics. Compared to SAs who made no further attempts, new MAs were significantly more likely to meet criteria for an anxiety diagnosis at baseline (80% versus 28%; 2 = 4.85; adjusted p < .05), but not at follow-up. In addition, at the time of their baseline attempt, they were distinguished by more often endorsing either wishing to die or being uncertain about their wish to die (100% versus 48%; 2 = 5.43; adjusted p < .05). New MAs were also compared to MAs who made another attempt at follow up (i.e., multiple reattempters; n = 12) on diagnostic characteristics (as reported at baseline and follow-up) and attempt characteristics (as reported at baseline). Multiple reattempters showed a nonsignificant trend to more often provide warnings/ threats before their attempt (42%) than did new MAs

WWW.JAACAP.COM

37

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

MIRANDA ET AL.

TABLE 4 Logistic Regressions Predicting Attempt at Follow-up From Baseline Attempt Status, Adjusting for Demographic Variables and Baseline DISC Diagnosis Attempt at Attempt at Follow-up, Follow-up, Full Sample Attempters (N = 190) Only (N = 67) Predictors

OR

95% CI

Age Female Ethnicity Black Hispanic Asian or other DISC diagnosis Attempt statusa Single attempt Multiple attempts

0.9 0.9

0.7Y1.3 Age 0.4Y2.4 Female Ethnicity 0.4Y5.2 Black 1.0Y8.6 Hispanic 0.7Y7.6 Asian or other

1.5 3.0* 2.3

Predictors

OR 95% CI 1.1 0.7Y1.7 2.7 0.4Y16.4 3.0 0.5Y18.2 3.4 0.7Y17.7 0.6 0.1Y7.6

2.3

0.9Y5.4 DISC 3.0 0.7Y12.4 diagnosis Attempt statusb 1.3 0.4Y3.9 4.0** 1.5Y10.2 Multiple 4.6* 1.0Y20.2 attempts

Note: Age was entered as a continuous variable (and thus the odds ratio represents the change in the estimated odds of an attempt for every unit increase in age). Ethnicity was entered as a categorical variable, with whites as a reference group. Diagnostic Interview Schedule for Children (DISC) diagnosis is dichotomous, with a value of 1 indicating the presence of any diagnosis and a value of 0 indicating no diagnosis. a Reference group = ideators. b Reference group = single attempters. * p = .05. * *p < .01.

(0%; 2 = 3.46; adjusted p = .11) and to less likely time their attempt for intervention (33% versus 80%; 2 = 3.09; adjusted p = .13). In addition, multiple reattempters more often met criteria for a DISC diagnosis in the previous year at follow-up (75%) compared to new MAs (17%) (2 = 5.51; p < .05). New MAs showed no significant diagnostic differences at follow-up compared to ideators who became SAs (i.e., new SAs). However, there was a trend for them to more often meet criteria for an anxiety disorder at baseline (80%) than new SAs (25%; 2 = 4.41; adjusted p = .10). DISCUSSION

We describe a long-term follow-up study of adolescents who identified themselves as having made a suicide attempt and/or ideating about suicide during a

38

WWW.JAACAP.COM

high school screening. Findings from this nonreferred sample confirm previous results from clinical samples that multiple suicide attempters are more likely to make subsequent attempts compared to SAs and to teens without a history of an attempt,6,7 suggesting that samples identified by screening may be representative of suicide attempters and ideators more generally. In addition, MAs presented with a different diagnostic profile than SAs and ideators and were distinguished from SAs in the characteristics of their most recent suicide attempt, as reported at baseline; they more often wished to die, their attempts were timed so that intervention was less likely, and they more often regretted recovery. Single attempts with a wish to die at the time of the attempt, along with an anxiety disorder at baseline, distinguished SAs who became MAs from those who did not make another attempt. These findings suggest that psychiatric history (specifically, an anxiety disorder), along with a definite or uncertain wish to die at the time of an attempt, confers risk for future attempts among individuals with a history of any attempt. It is possible that MAs are at higher risk for a future attempt due to some underlying condition (e.g., psychiatric diagnosis), whereas SAs or ideators may be dealing with acute stressors that resolve more readily,17 as suggested by the present findings of no differences between SAs and ideators in psychiatric diagnoses and in risk for a future attempt. Rudd,18 drawing on the work of Beck,19 has proposed a fluid vulnerability theory suggesting that some individuals, particularly MAs, remain at heightened risk for suicidal episodes before and after acute exacerbations or stresses. This theory proposes that MAs have a higher, chronic baseline risk, resulting from developmental or historical factors, that manifests itself across various domains. For instance, dispositional traits, such as aggression and impulsivity, have distinguished MAs from SAs in previous research.20,21 Additional factors, such as skills deficits (e.g., problem-solving), may confer risk for both psychiatric diagnosis and suicide attempts independently.22,23 Assessment of suicide risk should take into account not only acute risk factors (e.g., current suicidal ideation) but also assess for chronic risk.19 Finally, Joiner11 suggests that individuals may acquire the ability to engage in suicide attempts through a history of engaging in self-harm behavior or through conditions that have previously habituated them to the experience

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 47:1, JANUARY 2008

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

MULTIPLE ADOLESCENT SUICIDE ATTEMPTERS

of self-harm. Such individuals are at highest risk for future attempts when their desire for suicide is high. Thus, treatments with MAs should include interventions that decrease the wish to die.11 The present findings that MAs predict risk for future attempts above and beyond the presence of a diagnosis lend support to the idea of higher risk conferred by a history of multiple attempts. Furthermore, wish to die during an attempt emerged as a characteristic that distinguished teens at heightened risk for a repeat attempt, thus highlighting it as an area of potential clinical focus. It is always possible that the present data underestimate the number of actual MAs in the sample because information about number of past attempts was obtained via face-to-face interview rather than selfreport, and there was no third-party verification of attempts. Underestimation of number of suicide attempts would result in misclassification of single attempts and would work against the present findings, possibly attenuating the differences between groups. In addition, the present sample does not include teens who were not present during the screening, which may have resulted in further underestimation of past suicide attempts among truant adolescents with an elevated risk for attempts. However, such teens may already be overrepresented in clinical samples of suicide attempters, whereas the present study included teens who would not have otherwise come to the attention of a clinician (M. Scott, personal communication, 2007). Second, the sample may overestimate number of male and Hispanic suicide attempters, given the overrepresentation of male and Hispanic teens who completed the ASI, relative to the sample of ideators/ attempters identified by the CSS. Third, only unipolar mood, anxiety, and substance use disorders were included in the baseline diagnostic assessment. Inclusion of a broader range of diagnoses may have allowed for differentiation between ideators, SAs, and MAs on externalizing disorders at baseline and would have enabled these to be ruled out as alternative explanations for these findings. However, assessment of these three diagnostic categories at baseline is appropriate, given that they have previously been found to confer risk for either suicide attempts or completed suicide.24Y26 In addition, measurement of some aforementioned risk factors (e.g., impulsivity, problem-solving deficits), may assist in explaining the findings of this study.

J. AM . ACAD. CHILD ADOLESC. PSYCH IATRY, 47:1, JANUARY 2008

Finally, the present analyses would benefit from increased statistical power. None of the present comparisons of suicide attempt characteristics would survive Bonferroni corrections, which were not made in the interest of avoiding type II error. A history of multiple attempts provides useful information about psychopathology and suicide attempt characteristics, including degree of intent and feelings after the attempt. From a practical point of view, a history of multiple attempts may thus allow clinicians to discover, at the time of examination, teens who are most in need of further evaluation and intervention. The fact that a history of multiple attempts predicts subsequent attempts and is associated with a higher prevalence of psychiatric diagnosis may be useful for interventions aimed at preventing future attempts among adolescents. Assessment and intervention with ideators and first-time attempters may focus on resolving acute stressors, whereas treatments for MAs may focus on monitoring both acute and chronic risk, including factors that affect suicide intent. Future research should address whether SAs and MAs experience different types of stressors, along with whether experiencing a particular stressor, even without a diagnosis, confers risk for suicide attempts. At minimum, these findings strongly suggest that initial inquiries regarding suicide history should include questions about multiple attempts, as identifying whether an individual has a history of more than one suicide attempt may improve the ability to predict whether the individual will make a future suicide attempt. Furthermore, this study provides some validation for screening as a method of identifying suicide attempters who demonstrate similar characteristics to teens identified in clinical studies. Disclosure: The authors report no conflicts of interest.

REFERENCES 1. Pfeffer CR, Klerman GL, Hurt SW, et al. Suicidal children grow up: rates and psychosocial risk factors for suicide attempts during follow-up. J Am Acad Child Adolesc Psychiatry. 1993;32:106Y113. 2. Lewinsohn PM, Rohde P, Seeley JR. Psychosocial risk factors for future adolescent suicide attempts. J Consult Clin Psychol. 1994;62:297Y305. 3. Wichstro¨m L. Predictors of adolescent suicide attempts: a nationally representative longitudinal study of Norwegian adolescents. J Am Acad Child Adolesc Psychiatry. 2000;39:603Y610. 4. Brown GK, Beck AT, Steer RA, et al. Risk factors for suicide in psychiatric outpatients: a 20-year prospective study. J Consult Clin Psychol. 2000;68:371Y377.

WWW.JAACAP.COM

39

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

MIRANDA ET AL. 5. Harris EC, Barraclough B. Suicide as an outcome for mental disorders: a meta-analysis. Br J Psychiatry. 1997;170:205Y228. 6. Goldston DB, Daniel SS, Reboussin DM, Reboussin BA, Frazier PH, Kelley AE. Suicide attempts among formerly hospitalized adolescents: a prospective naturalistic study of risk during the first 5 years after discharge. J Am Acad Child Adolesc Psychiatry. 1999;38:660Y671. 7. Hulte´n A, Jiang G-X, Wasserman D, et al. Repetition of attempted suicide among teenagers in Europe: frequency, timing, and risk factors. Eur Child Adolesc Psychiatry. 2001;10:161Y169. 8. Esposito C, Spirito A, Boergers J, et al. Affective, behavioral, and cognitive functioning in adolescents with multiple suicide attempts. Suicide Life Threat Behav. 2003;33:389Y399. 9. Rosenberg HJ, Jankowski MK, Sengupta A, et al. Single and multiple suicide attempts and associated health risk factors in New Hampshire adolescents. Suicide Life Threat Behav. 2005;35:547Y557. 10. D_Eramo KS, Prinstein MJ, Freeman J, et al. Psychiatric diagnoses and comorbidity in relation to suicidal behavior among psychiatrically hospitalized adolescents. Child Psychiatry Hum Dev. 2004;35:21Y35. 11. Joiner TE. Why People Die by Suicide. Cambridge, MA: Harvard University Press; 2005;46Y93, 203Y222. 12. Kaslow NJ, Jacobs CH, Young SL. Suicidal behavior among lowincome African American women: a comparison of first-time and repeat suicide attempters. J Black Psychol. 2006;32:349Y365. 13. Shaffer D, Scott M, Wilcox H, et al. The Columbia Suicide Screen: validity and reliability of a screen for youth suicide and depression. J Am Acad Child Adolesc Psychiatry. 2004;43:71Y79. 14. Shaffer D, Fisher P, Dulcan MK, et al. The NIMH Diagnostic Interview Schedule for Children, version 2.3 (DISC-2.3): description, acceptability, prevalence rates, and performance in the MECA Study (Methods for the Epidemiology of Child and Adolescent Mental Disorders Study). J Am Acad Child Adolesc Psychiatry. 1996;35:865Y877. 15. Gould MS, Marrocco FA, Kleinman M, et al. Evaluating iatrogenic risk of youth suicide screening programs: a randomized controlled trial. JAMA. 2005;293:1635Y1643. 16. Shaffer D, Fisher P, Lucas CP, et al. NIMH Diagnostic Interview for

40

WWW.JAACAP.COM

17.

18.

19. 20. 21. 22. 23. 24. 25. 26.

Children, version IV (NIMH DISC-IV). Description, differences from previous versions, and reliability of some common diagnoses. J Am Acad Child Adolesc Psychiatry. 2000;39:28Y38. Goldston DB, Daniel SS, Reboussin DM, Reboussin BA, Frazier PH, Kelley AE. Psychiatric diagnoses of previous suicide attempters, firsttime attempters, and repeat attempters on an adolescent inpatient psychiatry unit. J Am Acad Child Adolesc Psychiatry. 1998;37:924Y932. Rudd MD. Fluid vulnerability theory: a cognitive approach to understanding the process of acute and chronic suicide risk. In: Cognition and Suicide: Theory, Research, and Therapy, Ellis TE, ed. Washington, DC: American Psychological Association; 2006;355Y367. Beck AT. Beyond belief: a theory of modes, personality, and psychopathology. In: Frontiers of Cognitive Therapy, Salkovskis PM, ed. New York: Guilford; 1996;1Y25. Gispert M, Davis MS, Marsh L, et al. Predictive factors in repeated suicide attempts by adolescents. Hosp Community Psychiatry. 1987;38: 390Y393. Stein D, Apter D, Ratzoni G, et al. Association between multiple suicide attempts and negative affects in adolescents. J Am Acad Child Adolesc Psychiatry. 1998;37:488Y494. Jeglic EL, Sharp IR, Chapman JE, et al. History of family suicide behaviors and negative problem solving in multiple suicide attempts. Arch Suicide Res. 2005;9:135Y146. McAuliffe C, Corcoran P, Keeley HS, et al. Problem-solving ability and repetition of deliberate self-harm: a multicentre study. Psychol Med. 2006;36:45Y55. Foley DL, Goldston DB, Costello EJ, et al. Proximal psychiatric risk factors for suicidality in youth: the Great Smoky Mountains Study. Arch Gen Psychiatry. 2006;63:1017Y1024. Brent DA, Baugher M, Bridge J, et al. Age and sex-related risk factors for adolescent suicide. J Am Acad Child Adolesc Psychiatry. 1999;38: 1497Y1505. Shaffer D, Gould M, Fisher P, et al. Psychiatric diagnosis in child and adolescent suicide. J Am Acad Child Adolesc Psychiatry. 1996;53: 339Y348.

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 47:1, JANUARY 2008

Copyright @ 2007 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.