A

OIrbMcrRywwCds©

K

1

et

v

0d

Available online at www.sciencedirect.com

Drug and Alcohol Dependence 95S (2008) S60–S73

The impact of two universal randomized first- and second-gradeclassroom interventions on young adult suicide ideation and attempts�

Holly C. Wilcox a,∗, Sheppard G. Kellam b,d, C. Hendricks Brown c, Jeanne M. Poduska b,Nicholas S. Ialongo d, Wei Wang c, James C. Anthony e

a Department of Psychiatry & Behavioral Sciences, Johns Hopkins School of Medicine, 600 North Wolfe Street/CMSC 346,Baltimore, MD 21287, United States

b American Institutes for Research, 921 E. Fort Avenue, Suite 225, Baltimore, MD 21230, United Statesc Department of Epidemiology and Biostatistics, College of Public Health, University of South Florida,

13201 Bruce B Downs Boulvard, Tampa, FL 33612, United Statesd Johns Hopkins University, Bloomberg School of Public Health 624 N. Broadway, 8th Floor, Baltimore, MD 21205, United States

e Department of Epidemiology, College of Human Medicine, Michigan State University B601 West Fee Hall,East Lansing, MI 48824, United States

Received 12 April 2007; received in revised form 8 January 2008; accepted 8 January 2008Available online 7 March 2008

bstract

bjective: This paper reports the impact of two first- and second-grade classroom based universal preventive interventions on the risk of Suicidedeation (SI) and Suicide Attempts (SA) by young adulthood. The Good Behavior Game (GBG) was directed at socializing children for the studentole and reducing aggressive, disruptive behavior. Mastery Learning (ML) was aimed at improving academic achievement. Both were implementedy the teacher.ethods: The design was epidemiologically based, with randomization at the school and classroom levels and balancing of children across

lassrooms. The trial involved a cohort of first-grade children in 19 schools and 41 classrooms with intervention at first and second grades. Aeplication was implemented with the next cohort of first grade children with the same teachers but with little mentoring or monitoring.esults: In the first cohort, there was consistent and robust GBG-associated reduction of risk for suicide ideation by age 19–21 years compared toouths in standard setting (control) classrooms regardless of any type of covariate adjustment. A GBG-associated reduced risk for suicide attemptas found, though in some covariate-adjusted models the effect was not statistically robust. No statistically significant impact on these outcomes

as found for ML. The impact of the GBG on suicide ideation and attempts was greatly reduced in the replication trial involving the second cohort.onclusions: A universal preventive intervention directed at socializing children and classroom behavior management to reduce aggressive,isruptive behavior may delay or prevent onset of suicide ideation and attempts. The GBG must be implemented with precision and continuingupport of teachers.

2008 Elsevier Ireland Ltd. All rights reserved.

idem

cc

eywords: Suicide ideation; Suicide attempts; Adolescence; Developmental ep

. Introduction

Suicide and suicide-related behavior have been studied inpidemiology for more than 150 years. Among the earliest con-ributions were those made by William Farr, the early 19th

� Supplementary data on the second cohort can be accessed with the onlineersion of this paper at http://dx.doi.org/10.1016/j.drugalcdep.2008.01.005.∗ Corresponding author. Tel.: +1 410 502 0629.

E-mail address: hwilcox1@jhmi.edu (H.C. Wilcox).

ito(w(iop

376-8716/$ – see front matter © 2008 Elsevier Ireland Ltd. All rights reserved.oi:10.1016/j.drugalcdep.2008.01.005

iology; Universal prevention programs; Good behavior game

entury father of biostatistics, who sought to test whether sui-ide risk varied with successful adaptation or achievementn relation to education (Farr, 2000). Later in the 19th cen-ury, Emile Durkheim’s sociological theory linked higher ratesf suicide to the absence of shared social values and normsanomie) in the general population and lower rates of suicideith the opposite—greater social integration and shared values

Durkheim, 1897). In the present study we extend that researchnto the domain of experimental and developmental epidemi-logy at the school and classroom levels. The main aim of thisaper is to estimate the impact of two universal classroom-based

ohol D

iAeirsipwto

tpt1cSPfuLL1ae1(cHeca

talyF1viari2ies

cocfivas

eclvStacfitspra

b(ddavadl(ma

cpiomdo1oo1Ssaiaegii(dr(w

H.C. Wilcox et al. / Drug and Alc

nterventions on risk of Suicide Ideation (SI) and/or Suicidettempts (SA), referred to here as suicidality. We report on

vidence from a randomized controlled trial of two preventiventerventions implemented in the first and second grade class-ooms of elementary schools with follow-up assessments ofuicidality when the participants were in young adulthood. Onentervention, the Good Behavior Game (GBG), was directed atroducing shared values, norms, and proper student behaviorithin the first- and second-grade classroom. The second, Mas-

ery Learning (ML), was directed at learning to read, a basic taskf social adaptation in first and second grades.

Suicide ideation is considered by some to be the first stepoward suicide (Gili-Planas et al., 2001); it is one of the strongestredictors of a future suicide attempt in adolescents and is relatedo risk of completed suicide (Brent et al., 1993; Lewinsohn et al.,996; Reinherz et al., 1995). The current public health signifi-ance of suicidal thoughts and behaviors is large in the Unitedtates and globally (Murray and Lopez, 1996; United Statesublic Health Service, 1999). Prevalence estimates derivedrom recent community studies of adolescents indicate thatp to 30% have experienced suicide ideation (Fergusson andynskey, 1995; Ialongo et al., 2002; Juon and Ensminger, 1997;ewinsohn et al., 1996; Reinherz et al., 1995; Velez and Cohen,988). In the United States alone, approximately two milliondolescents attempt suicide each year, resulting in 700,000mergency room visits, and death certifications in 2005 for5–24 year-olds indicated that over 4000 had committed suicideMcIntosh, 2008; Shaffer and Pfeffer, 2001). Prevention of sui-ide has been declared a national priority (United States Publicealth Service, 1999), but there is little definitive evidence on the

fficacy or effectiveness of suicide prevention programs amonghildren and adolescents (Burns and Patton, 2000; Goldsmith etl., 2002; Gould et al., 2003).

Over the past three decades, evidence from developmen-al epidemiological studies has consistently identified specificntecedents measured as early as first grade in the prediction ofater mental and behavioral disorders during the middle schoolears and beyond (Cairns et al., 1989; Ensminger et al., 1983;arrington, 1995; Hawkins et al., 1988, 1992; Kellam et al.,983; Reid, 1993; Reid and Eddy, 1997). Randomized pre-entive intervention trials conducted by our group and othersndicate that school-based universal interventions (i.e., thoseddressing all children, not merely those indicated to be at higherisk) can have beneficial effects on aggressive, disruptive behav-or and achievement (Dolan et al., 1993; Ialongo et al., 1999,001; Kellam et al., 1994a; Reid et al., 1999), off-task behav-or (Brown, 1993b), depressive symptoms in first grade (Kellamt al., 1994b), and delayed onset or reduced risk of tobaccomoking (Kellam and Anthony, 1998).

The work reported in this paper is grounded in lifeourse/social field theory (Kellam et al., 1975), which is builtn the observation that one or more main social fields are criti-ally important in each stage of life. Within each of these social

elds there are defined social task demands to which an indi-idual must respond, a process called social adaptation. Thedequacy of an individual’s response to these demands, termedocial adaptational status (SAS), is rated by natural raters in

asmt

ependence 95S (2008) S60–S73 S61

ach social field, such as parents in the family, teachers in thelassroom, significant peers in the peer group, or, later in theife course, by spouses in the intimate social field or super-isors in the work social field. In keeping with Cicchetti andchneider-Rosen (1984), the theory posits that success in mas-

ering social task demands specific to one stage of developmentnd in one social field will lead to an increase in later suc-esses in the same and other social fields. Life course/socialeld theory also suggests that SAS often is reciprocally related

o psychological well-being (PWB), the internal psychologicaltatus of an individual in regard to constructs such as self-esteem,sychiatric symptoms or disorders, and neurobiological or neu-opsychological conditions. PWB may be an antecedent and/orconsequence of SAS, and vice versa.

The social task demands of the first grade classroom haveeen studied and reported frequently by our research groupe.g., Kellam et al., 1994a,b). In general, these social taskemands consist of socializing appropriately with other chil-ren and the teacher, obeying classroom rules, concentratingnd attending, and learning academic subjects. The two inter-entions used in this study were directed at improving socialdaptation to these demands; therefore, we hypothesized thatecreasing aggressive, disruptive behavior in classrooms wouldead to a reduction in risk of outcomes linked to this antecedentKellam et al., 1994a), and that improving classroom achieve-ent would improve PWB, particularly depressive symptoms

nd disorders in vulnerable children (Kellam et al., 1994b).As the course of development unfolds in relation to suc-

esses or failures at social adaptational tasks in classrooms, ineer groups, and the community, other mediating or moderat-ng issues can stem from as well as further influence the coursef development and the impact of the GBG or ML. In com-unity studies of adolescents, the use of alcohol and illegal

rugs has been identified as an influential factor contributing tobserved increases in the rate of adolescent suicide (Brent et al.,988; Fowler et al., 1986). The prevalence of substance use dis-rders (alcohol and other drugs) in psychological autopsy andther studies has varied between 37% and 66% (Brent et al.,988; Fowler et al., 1986; Runeson, 1989; Shaffer et al., 1988;hafii et al., 1988). Depression, substance abuse, and aggres-ive, disruptive behavior have been found to distinguish suicidettempters from non-attempters in community and clinical stud-es of adolescents and adults (Brent et al., 1993; Garrison etl., 1993; Kessler et al., 1999; Lewinsohn et al., 1996; Petronist al., 1990). Drug and alcohol abuse has been associated withreater frequency of suicide attempts, more serious attemptsn terms of lethality and intent, and increased levels of suicidedeation (Crumley, 1990; Lewinsohn et al., 1996). Garrison et al.1993) found that the relationships between alcohol and illegalrug use and suicidal behaviors were most pronounced with theeported use of the more potentially dangerous or ‘harder’ drugse.g., cocaine) but remained even when the substance of interestas nicotine. The role of early and continuing drug use may be

n important element in the developmental psychopathology ofuicidality and will be considered in this report. Other potentialediators of this association could be constructs often thought

o be associated with early drug and alcohol use, such as aggres-

S ohol D

sCd

1

seaTeegtatccIobaorpb

bwm1idah(

1

ipaBtmuamewMDfd1

1s

lrlwadBDpL11

biSsaSFt11

Mcpddsthda(wiB

SdahirGirm

62 H.C. Wilcox et al. / Drug and Alc

ive, disruptive behavior in later childhood or early adolescence,onduct Disorder symptoms, academic self-competence, self-erogation, parental monitoring, or drug using or deviant peers.

.1. The good behavior game

The GBG was developed by Barrish et al. (1969) and waselected for this trial because of its efficacy in short-term non-xperimental or quasi-experimental designs, and because it wascceptable to the participating school leadership and community.he GBG is a classroom team-based behavior management strat-gy that promotes good behavior by rewarding teams that do notxceed maladaptive behavior standards as set by the teacher. Theoal of the GBG is to create an integrated classroom social sys-em that is supportive of all children being able to learn with littleggressive, disruptive behavior. The methods involve helpingeachers to define unacceptable behaviors clearly and to socializehildren with regulation of teammate’s behavior through a pro-ess of team contingent reinforcement and mutual self-interest.n this trial, children in the GBG classrooms were assigned tone of three heterogeneous teams containing equal numbers ofoys and girls, equal numbers of aggressive, disruptive children,nd equal numbers of shy, socially isolated children. At the startf the game session, the teacher described and posted basic class-oom rules of student behavior. All teams could “win” during aarticular game period, with the criterion for winning the gameeing less than four infractions of acceptable student behavior.

Evidence to date has been supportive of the GBG reducingehavioral problems in the primary and middle school years, asell as reducing risk of tobacco smoking in early adolescence,ainly for boys (Kellam et al., 1994a; Kellam and Anthony,

998). The GBG intervention trial also has shown long-termmpact on drug abuse and dependence and alcohol abuse andependence disorders as well as antisocial personality disordernd regular tobacco smoking, especially among males who hadigher levels of aggressive, disruptive behavior in first gradeKellam et al., 2008).

.2. Mastery learning

ML is a teaching strategy with demonstrated effectiveness inmproving achievement and the underlying theory and researchosit that under appropriate instructional conditions virtuallyll students can learn most of what they are taught (Block andurns, 1976; Bloom, 1976; Dolan, 1986; Guskey, 1997). Prior

o and throughout this trial, all first-grade teachers in the Balti-ore City Public School System were expected to teach reading

sing a ML curriculum. Our ML intervention teachers receivedmore precise, enhanced reading curriculum that they imple-ented in the context of a system-wide ML curriculum (Dolan

t al., 1993; Kellam et al., 1991). This enhanced ML curriculumas used by both first- and second-grade teachers. Short-termL benefits were found in reading achievement (Brown, 1993a;

olan et al., 1993), and children with depressive symptoms in the

all of first grade who gained in achievement showed reducedepressive symptoms by the end of first grade (Kellam et al.,994b).

(aml

ependence 95S (2008) S60–S73

.3. Hypothesized mechanisms for ML and GBG impact onuicidality

The occurrence of suicidality among teenagers has beeninked to the targets of both the ML and GBG interventions. Loweading achievement at school entry and other academic prob-ems have been linked with suicidality, usually via an associationith conduct problems (Beautrais et al., 1996, 1997; Bennett et

l., 2003; Lewinsohn et al., 1994;) or depressive symptoms andisorder (Andrews and Lewinsohn, 1992; Beautrais et al., 1996;rent et al., 1993; Esposito and Clum, 2002; Shaffer et al., 1996).epressive symptoms and disorder are also considered strongredictors of suicidality in youth and young adults (Andrews andewinsohn, 1992; Beautrais et al., 1998; Fergusson and Lynskey,995; Kessler et al., 1999; Petronis et al., 1990; Reinherz et al.,995).

The proximal target of the GBG, aggressive, disruptiveehavior has also been reported as a risk factor for suicidal-ty (Brent et al., 1993) and a number of externalizing behaviors.uch behavior outcomes are developmentally linked to aggres-ive, disruptive behavior and include impulsivity (Stanley etl., 1986), conduct disorder (Andrews and Lewinsohn, 1992;haffer et al., 1996), antisocial behavior (Beautrais et al., 1996;ergusson and Lynskey, 1995; Shafii et al., 1985), juvenile jus-

ice involvement (Beautrais et al., 1997; Fergusson and Lynskey,995) and alcohol and other drug use disorders (Beautrais et al.,996; Brent et al., 1993; Shaffer et al., 1996).

Our previously reported findings support a conclusion that theL intervention affects its intended proximal targets. Specifi-

ally, ML showed a short-term impact on learning and mastery,articularly for those who began school with high levels ofepressive symptoms (Kellam et al., 1994b). By loweringepressive symptoms, ML could hypothetically also lessenuicidality through improved mastery of the student role. Fur-hermore, the ML intervention appeared to have somewhatigher overall achievement results for children who began moreepressed in first grade, and improvement in achievement wasssociated with a lessening of self-reported depressed feelingsKellam et al., 1994b). We therefore hypothesized that MLould reduce suicidality because of the role that mastery plays

n self-esteem, depression, and suicidal thoughts (Battle, 1990;rookover, 1965; Covington, 1989; Holly, 1987).

The GBG impact was also supported by previous findings.tudents in the GBG classrooms showed reduced aggressive,isruptive behavior during elementary and middle school as wells reduced drug and alcohol involvement through young adult-ood (Kellam et al., 2008). Because the GBG intervenes earlyn the child’s life before classroom maladaptive aggressive, dis-uptive behavior becomes intransigent, we hypothesize that theBG should reduce the course of aggressive, disruptive behav-

or and many of its consequences, including suicidality. Severaleasons for this continuity in aggressive, disruptive behavior areade clear in the Patterson–Reid model of conduct disorder

Reid et al., 2002). In the absence of an intervention to reduceggressive, disruptive behavior, aggressive, disruptive childrenay have trouble learning academic subjects, may remain dis-

iked by teachers and their peers, and often limit their friends to

ohol D

odiad2

haubawKWbricytap

2

tooyglwvatbwPmpdadt

akttpGWcosGsstod

tcrcc

srei

a1wiv5so

ai2aip

aams1“wac

r11td1sC(alif

2

wautnopcfi

H.C. Wilcox et al. / Drug and Alc

ther deviant peers (Reid et al., 2002). Over time, such chil-ren are at increased risk of failing in school and engagingn drug use, and impulsive and antisocial behavior (Kellam etl., 1983, 1994a, 1998, 2008). All of these processes and con-itions place youths at higher risk for suicide (Gould et al.,003).

In prior papers we have reported that the GBG interventionad a long-term impact into young adulthood on outcomes suchs drug and alcohol abuse and dependence disorders, tobaccose, antisocial personality disorder and violent and criminalehavior, and service use, particularly among males who weret higher levels of first grade aggressive, disruptive behavior,ith lesser impact among females (Kellam and Anthony, 1998;ellam et al., 2008; Petras et al., 2008; Poduska et al., 2008).e therefore hypothesize that the largest GBG impact will

e seen in highly aggressive, disruptive males. We have alsoeported that females show greater developmental continuityn internalizing symptoms (Kellam et al., 1983). Suicidalityould have internalizing and/or externalizing pathways intooung adulthood; there could also be different developmen-al pathways for females and males. In addition, ML mightlter internalizing while the GBG might impact externalizingathways.

. Methods

The research design was that of an epidemiologically based randomized fieldrial, applied to a defined population of children. The study population consistedf all youths who started first grade in 41 classrooms in 19 elementary schoolsf the Baltimore City Public School System during two successive academicears: 1985–1986 for Cohort 1 first graders and 1986–1987 for Cohort 2 firstraders. In this paper emphasis is placed on the first cohort where the highestevel of mentoring and monitoring for fidelity was conducted. The 19 schoolsere located within five urban areas which varied in socioeconomic status fromery poor to moderate income, and in degree of racial segregation. Within eachrea, three or four schools were sorted into homogenous sets, and selected forhe trial (Kellam et al., 1991). All of the entering first graders in the schoolsecame eligible participants in the study. Of the 1196 pupils in Cohort 1, 66%ere African-American and almost all the remainder were non-Hispanic Whites.arents, and children as they grew older, could refuse to participate in assess-ents at any time. Written consent was obtained from parents for childhood

articipation for those assessments beyond standard school assessment proce-ures. Consent was obtained from each young adult at the time of the youngdult interviews. During the first year of recruitment for each cohort, only 5%eclined to participate or sign consent forms (Kellam et al., 2008); participationo the young adult years is described below.

Within each intervention school, all incoming first grade children weressigned to classrooms so as to balance classrooms within school on gender,indergarten experience, and academic and behavioral performance. Followinghis, classrooms/teachers were assigned at random to intervention condition,hat is, in schools testing ML, they were assigned to either ML or standardrogram (internal ML control). The same procedure was followed within theBG schools, yielding GBG classrooms and internal GBG control classrooms.ithin the control schools, all youth were assigned in a balanced fashion to

lassrooms that served as external controls. In a few instances, the study teambserved imbalance after assignment, so a small number of children were reas-igned immediately to induce balance on entry to first grade. The ML andBG interventions lasted for 2 years. To maintain this design, GBG and ML

tudents moved into second grade in their same classroom units. Similarly,tandard program (control) students were kept together in the control condi-ion through the end of second grade. Transfers from one intervention to thether or from control to either intervention were rare. We have analyzed theata based on original assignment (i.e., in an ‘intent to treat’ design). After

ttd(1

ependence 95S (2008) S60–S73 S63

he 2-year intervention period ended, students were assigned by the school tolassrooms via usual and customary procedures of the school system withoutegard to prior intervention status. The same design was used for the secondohort and teachers delivered the same intervention as they did for the firstohort.

Both ML teachers and GBG teachers each received an initial 40 h of trainingupporting their respective interventions. Standard program (control) teacherseceived a comparable time of in-service training but in areas not affecting thexperimental intervention conditions. Care was taken to provide comparablencentives and reinforcements to all teachers in the five conditions.

Over 15 years after school entry, between 1999 and 2002, the research teamssessed 1918 (83%) in the two young adult follow-up interviews, including54 participants found within incarceration facilities (who were interviewedith additional consent approvals). Four percent were traced but refused to be

nterviewed; another 6% were traced, but for logistical reasons were not inter-iewed (e.g., out of state with no telephone number, military postings overseas);.7% were not found despite extensive tracing efforts. National Death Indexearches through April 2003 found that 41 pupils had died (1.8%), includingne confirmed suicide.

The key response variables in this study were the occurrence of suicidality asssessed by face-to-face interview during young adulthood. Other young adultnformation was also obtained from a separate phone interview (see Kellam et al.,008). The study protocols for the two follow-up assessments were reviewed andpproved by the Institutional Review Boards for protection of human subjectsn research at the Johns Hopkins Bloomberg School of Public Health and thehone interview by the American Institutes for Research as well.

The components of suicidality, suicide ideation and suicide attempt, werenalyzed separately. Age at first suicide ideation and age at first attempt weressessed via standardized questions embedded within the follow-up interview’sulti-item assessment of major depressive disorder based upon an adapted ver-

ion of the NIMH Diagnostic Interview Schedule (DIS) (Robins et al., 1981,994). Suicide ideation was measured in the face-to-face survey by asking,Have you ever felt so low you thought about committing suicide?” Youthsho answered ‘yes’ have been designated as cases of suicide ideation, and were

sked to report the age of first occurrence of ideation. Suicide attempt data wereollected with the question, “Have you ever attempted suicide?”

Other covariates under study were selected on the basis of prior theory andesearch on suicide-related behaviors (e.g., see Brent, 1995; Garrison et al.,991, 1993; Ialongo et al., 2002; Juon and Ensminger, 1997; Lewinsohn et al.,994, 1996). A standardized teacher rating of aggressive, disruptive behavior wasaken in the fall of first grade, which provided an assessment of early aggressive,isruptive behavior or rule-breaking as described elsewhere (e.g., Kellam et al.,975; Werthamer-Larsson et al., 1991). The child’s scaled self-reports of depres-ive symptoms (Child Depression Index-CDI) and anxious symptoms (Revisedhildren’s Manifest Anxiety Scale, RCMAS) in fall of first grade were also used

Ialongo et al., 1994, 2001; Kellam et al., 1994b). Data for each youth’s gender,ge, race/ethnicity, and family financial status (i.e., eligibility for free or reducedunch) were abstracted from the school system’s administrative database. Famil-al suicidality and psychiatric disturbance was assessed retrospectively during aollow-up assessment conducted during early young adulthood via telephone.

.1. Measurement of potential mediators

Early onset alcohol, tobacco, marijuana, and inhalant use (use before age 16)as measured in yearly assessments during 1989–1994 in school via interviewer

dministered self-report methods. The children were first interviewed about theirse of alcohol, tobacco, marijuana, and inhalants in the spring of 1989, whenhey were in third or fourth grades, and thereafter through 1994 in eighth orinth grades. Standardized questions on age at first use were as follows: “Howld were you when you first smoked tobacco?”; “Not counting sips with yourarent’s permission, how old were you the first time you drank beer, wine, wineoolers, liquor or any other drink with alcohol in it?”; “How old were you therst time you smoked marijuana?”; “(Not counting cocaine,) How old were you

he first time you sniffed something to get high or for some other feeling likehat?” Additional information on these standardized assessments of youthfulrug involvement can be obtained from prior publications of this research groupe.g., see Chilcoat et al., 1995; Chilcoat and Anthony, 1996; Kellam and Anthony,998).

S ohol Dependence 95S (2008) S60–S73

samAPPuaSitcwrtihaahmma1oI

ifdvibccC

D((amccba(ey1sr

3

3o

bspga

ptio

nan

des

timat

edri

skof

suic

ide

idea

tion

and

atte

mpt

for

Coh

ort1

Atb

asel

ine

As

youn

gad

ults

Suic

ide

idea

tion

Suic

ide

atte

mpt

Inci

dent

case

sC

umul

ativ

ein

cide

nce

RR

a95

%C

Ip-

Val

ueIn

cide

ntca

ses

Cum

ulat

ive

inci

denc

eR

Ra

95%

CI

p-V

alue

587

(49%

)38

9(4

5%)

580.

151.

030

0.08

1.0

609

(51%

)46

9(5

5%)

650.

140.

9(0

.6–1

.3)

0.66

264

0.14

1.9

(1.2

–3.0

)0.

007*

-His

pani

c)40

5(3

4%)

233

(27%

)52

0.22

2.2

360.

151.

7/e

thni

cgr

oups

791

(66%

)62

5(7

3%)

710.

111.

0(1

.5–3

.3)

<0.

001*

580.

091.

0(1

.1–2

.7)

0.01

4*

lysu

bsid

ized

lunc

h 644

(55%

)43

1(5

0%)

720.

171.

050

0.11

1.0

548

(45%

)42

4(5

0%)

500.

120.

7(0

.5–0

.9)

0.04

1*44

0.10

0.9

(0.6

–1.4

)0.

600

atus

684

(57%

)49

2(5

7%)

820.

171.

061

0.12

1.0

274

(23%

)18

8(2

2%)

240.

130.

7(0

.4–1

.2)

0.21

116

0.09

0.7

(0.4

–1.2

)0.

162

238

(20%

)17

8(2

1%)

170.

100.

5(0

.3–0

.9)

0.02

4*17

0.09

0.7

(0.4

–1.3

)0.

292

skes

timat

esar

eno

tadj

uste

dfo

rco

vari

ates

(gen

der,

age,

and

race

);ad

just

edes

timat

esar

epr

esen

ted

inth

ete

xt.

64 H.C. Wilcox et al. / Drug and Alc

In 1989, when the children were in third or fourth grade, an adapted ver-ion of the Harter Scholastic Competence subscale (Harter, 1985) was used tossess academic self-competence. Self-derogation was measured by a slightlyodified version of the Kaplan–Pokorny Scale (Kaplan and Pokorny, 1969).lso in 1989, overt and covert antisocial behavior scales were adapted from theatterson–Capaldi OSLC measure of Overt Antisocial behavior (Capaldi andatterson, 1989). The Baltimore Conduct Problems and Delinquency Scale wassed in 1991. This self-report measure of delinquent and antisocial behavior isn adaptation of the measure developed by Elliott and Huizinga for the Nationalurvey of Delinquency and Drug Use (Elliott et al., 1985). Drug using peer

nvolvement was assessed in 1989 by an eight item yes/no scale that assessedhe frequency of peers who smoke tobacco, drink alcohol, smoke marijuana, userack, frequency of peers who say it is a bad idea to use drugs, and peers whoould drop people as friends if they started using drugs. Peer involvement in

ule-breaking behavior was assessed in 1989 by a five item scale that assessedhe frequency of peers who cheat on tests, ruin or damage property belong-ng to someone else, steal something worth less than five dollars, threaten toit someone without a reason and suggest that the participant do somethinggainst the law. This scale was a modified version of that described by Capaldind Patterson (1989). Youths were asked in forced choice format to indicateow often their peers have engaged in the various antisocial behaviors. Parentalonitoring was assessed in 1989 (Capaldi and Patterson, 1989). The 10-itemodified scale was originally from the Structured Interview of Parent Man-

gement Skills and Practices—Youth Version (SIPMSP, Capaldi and Patterson,989). Major Depressive Episode was assessed using the Major Depressive Dis-rder section of the CIDI, Version 2.1, which was modeled after the Nationalnstitute of Mental Health-Diagnostic Interview Schedule (DIS).

Survival analysis methods were used to estimate the relative hazard of suicidedeation and attempt for both ML and the GBG, followed by covariate adjustmentor gender, race, baseline aggressive, disruptive behavior, clinical features ofepression and anxiety, parental suicidality and psychiatric disturbance. All sur-ival analyses measured the time to first suicide ideation (or attempt) since entrynto first grade; the two participants who reported age of onset of suicide ideationefore age six were excluded from analyses. For those who did not report sui-idality, the censoring time was age at last assessment. Separate Kaplan–Meierurves for each intervention condition and gender are presented for those inohort 1 along with log-rank test statistics and 95% confidence intervals (CI).

To take into account the fact that time was measured in years, we usediscrete-Time Survival Analyses (DTSA) with covariates, as suggested by Cox

1972), and more recently by Willet and Singer (1993) and Singer and Willet1994). To account for clustering of students within initial elementary schoolsnd within urban areas, we stratified on school. This stratified analysis accom-odates data interdependencies by conditioning on school and thus allows direct

omparison of each of the two active interventions with their respective internalontrols. We have found that school variation was high at baseline across a num-er of measures. Once we condition on school many variables show little vari-tion at the classroom level because children were balanced across classroomsBrown, 1993b; Kellam et al., 2008). Thus, balancing classes within schools is anfficient way to control for school and community characteristics in impact anal-ses (e.g., see Breslow and Day, 1980; Brown and Liao, 1999; Schlesselman,982). This approach is especially useful when suicidality might depend onocially shared determinants that are not easily quantifiable (e.g., widely sharedeligious beliefs and local area norms about suicide-related behavior).

. Results

.1. Characteristics of the population and the epidemiologyf suicidality

Table 1 presents characteristics of the Cohort 1 sample ataseline and follow-up including the cumulative incidence of

uicide ideation and suicide attempt (SA) among young adultarticipants, as well as unadjusted relative risk estimates forender, race, free lunch status, and randomized interventionssignment. The mean age at the face-to-face follow-up inter- Ta

ble

1Sa

mpl

ede

scri

Gen

der

Mal

eFe

mal

e

Rac

e Whi

te(N

onO

ther

raci

al

Free

orpa

rtia

lN

oY

es

Inte

rven

tion

stC

ontr

olM

LG

BG

aR

elat

ive

ri*

p<

0.05

.

H.C. Wilcox et al. / Drug and Alcohol Dependence 95S (2008) S60–S73 S65

e ide

vy

3(wywcsvtcRas

itaAaltSt(tfm

3(cfhrdTt

ia

wAGcsrstgi

3

foaSfdcSac1

patm1nhGGcr

3

Fig. 1. Estimated cumulative probability of reporting suicid

iew was 22 years (range 20–24 years old) for Cohort 1 and 21ears (range 19–23 years old) for Cohort 2.

.1.1. Suicide ideation. In Cohort 1, 123 of 858 participants14%) assessed in young adulthood had experienced SI, whichas twice as common (RR = 2.2) among non-Hispanic Whiteouths as within the other racial/ethnic group, nearly all of whichere African American (p < 0.001). It was also 1.3 times more

ommon among participants who had not qualified for free orubsidized lunch versus those who had (p = 0.041). Those indi-iduals assigned to the GBG intervention were half as likelyo have experienced SI, as compared to those in the controllassrooms (internal GBG, internal ML, and external controls,R = 0.5, p = 0.024). The ideation value for the ML interventionlso was smaller than that for controls, but was not statisticallyignificant (p = 0.211).

In the Cohort 2 sample, 95 of 837 participants (11%) assessedn young adulthood had experienced SI, which was almost threeimes as common (RR = 2.7) among non-Hispanic White youthss within the other racial/ethnic group, nearly all of which werefrican American (p < 0.001). SI was 1.5 times more common

mong participants who had not qualified for free or subsidizedunch versus those who had (p = 0.001). Approximately, 9% ofhose who had received the GBG intervention had experiencedI compared to 12% of those in the control classrooms, but

he relative risk estimate did not reach statistical significanceRR = 0.8, p = 0.408). The ML impact on SI was also not statis-ically significant (RR = 1.1, p = 0.592). Full Cohort 2 analysesor this and other outcomes can be found in the supplementaryaterial on the journal’s website.

.1.2. Suicide attempt. In Cohort 1, 94 of 858 participants11%) assessed in young adulthood had made an SA. Theumulative incidence of SA was nearly twice as high amongemales as males (RR = 1.9, p = 0.007). The rate was also 70%igher among non-Hispanic Whites compared to within the otheracial/ethnic group (p = 0.014). The crude SA cumulative inci-ence estimates did not vary by free or reduced lunch status.hey were slightly lower in both intervention groups compared

o controls (p > 0.15 for both).In the Cohort 2 sample, 61 of 837 participants (7%) assessed

n young adulthood had experienced SA, which was two timess common (RR = 2.0) among non-Hispanic White youths as

DS

ation among females in Cohort 1 for GBG (a) and ML (b).

ithin the other racial/ethnic group, nearly all of which werefrican American (p = 0.010). Those individuals assigned to theBG intervention were not less likely to have experienced SA

ompared to controls, as the relative risk estimate did not reachtatistical significance (RR = 1.0, p = 0.965). The SA relativeisk estimate for the ML intervention was also not statisticallyignificant (RR = 1.1, p = 0.844). In both Cohorts, one-third ofhose who had attempted suicide, ideation was not present sug-esting that the SA was an impulsive act without anticipatorydeation.

.2. Kaplan–Meier life table results

We compared onset of SI by intervention condition amongemales in Cohort 1 (see Fig. 1). There was markedly lowerccurrence of SI in the GBG condition compared to controlsfter age 16. By young adulthood, the cumulative incidence ofI was 19% for females in the control condition versus about 9%or those in the GBG condition. Fig. 1b shows a much smallerifference in SI incidence for ML females versus internal MLontrols. Fig. 2 shows plotted cumulative incidence values forA among females. Here, the patterns were similar to SI withreduction in risk for the GBG but not for ML. About 20% of

ontrols had made at least one SA by young adulthood versus0% for the GBG.

Parallel results for SI and SA among males in Cohort 1 areresented separately in Figs. 3 and 4. Fig. 3a shows that byge 15, the SI incidence for the GBG group decreased relativeo control values. By young adulthood, approximately 24% of

ales who were in control classrooms had experienced SI versus1% of GBG males. Just as for females, the ML impact on SI wasull for males; approximately 15% of boys in ML classroomsad SI versus 13% in the control classrooms. Ten percent ofBG males had made an SA by age 22 versus 18% of internalBG control males. There were no differences between ML and

ontrols for males’ SA risk. For full Cohort 2 analyses pleaseefer to the supplementary material on the journal website.

.3. GBG and ML impact on suicidal ideation

Table 2 presents relative odds of the GBG impact based oniscrete-Time Survival Analysis for SI for Cohort 1. Here, givenI rarity, the relative odds (RO) approximates relative risk (RR).

S66 H.C. Wilcox et al. / Drug and Alcohol Dependence 95S (2008) S60–S73

Fig. 2. Estimated cumulative probability of reporting suicide attempt among females in Cohort 1 for GBG (a) and ML (b).

ide id

FMsi2o4os(m

om

poaidcitwt

Fig. 3. Estimated cumulative probability of reporting suic

our models are presented. Model 1 is unadjusted for covariates;odel 2 adjusts for gender, race, and baseline levels of aggres-

ive, disruptive behavior, depression, and anxiety as measuredn fall of first grade; Model 3 adjusts for covariates in Model

as well as caregiver suicidality and mental illness that werebtained retrospectively at the young adult interviews; Modelis Model 2 plus terms to capture a hypothesized interaction

f design and baseline aggressive, disruptive behavior, depres-ion, and anxiety. Given no gender difference in the GBG impactp = 0.828) for SI and SA, Tables 2 and 3 include data on both

ales and females.The overall summary relative risk estimate for occurrence

f SI indicates an inverse (protective) association with assign-ent to the GBG intervention (RR = 0.4; 95% CI = 0.2, 0.8;

iatf

Fig. 4. Estimated cumulative probability of reporting suicide at

eation among males in Cohort 1 for GBG (a) and ML (b).

= 0.008), as well as a consistent and robust inverse associationf the GBG intervention with SI across the series of covariate-djusted models (RR = 0.4–0.5, see Table 2). The interventionmpact did not vary by baseline mental health indicators; noesign by baseline interaction terms in Model 4 were statisti-ally significant (all p > 0.05). A model was then fit for SI thatncluded covariates with relative risk estimates that were greaterhan 2.0—namely GBG and caregiver suicide ideation, both ofhich showed strong and consistent associations with SI across

he series of models. The best fitting model for SI in Cohort 1

ncluded terms for GBG (RR = 0.5; 95% CI 0.2, 0.9; p = 0.015)nd caregiver SI (RR = 4.3; 95% CI 2.5, 7.2; p < 0.001). In con-rast to findings for the GBG intervention, relative risk estimatesor ML based upon Discrete-Time Survival Analysis for SI for

tempt among males in Cohort 1 for GBG (a) and ML (b).

H.C. Wilcox et al. / Drug and Alcohol Dependence 95S (2008) S60–S73 S67

Table 2Impact of GBG on suicide ideation for males and females in Cohort 1

Predictor or covariate Relative odds estimate (adjusted) 95% CI p-Value

Model 1a

GBG 0.4 0.2–0.8 0.008*

Model 2b

GBG 0.4 0.2–0.8 0.011*

Female 1.0 0.7–1.5 0.828White Non-Hispanic 1.4 0.7–3.1 0.376Aggressive, disruptive behavior 0.9 0.6–1.4 0.624Depressive symptoms 1.2 0.8–2.0 0.385Anxious symptoms 1.2 0.5–2.7 0.647

Model 3c

GBG 0.5 0.2–0.9 0.025*

Female 1.0 0.7–1.50 0.985White Non-Hispanic 1.2 0.5–2.6 0.709Aggressive, disruptive behavior 0.9 0.6–1.5 0.763Depressive symptoms 1.2 0.7–2.0 0.442Anxious symptoms 1.2 0.5–2.6 0.697Caregiver depression 1.2 0.7–2.0 0.494Caregiver ideation 2.5 1.2–5.5 0.018*

Caregiver suicide threat 1.7 0.7–4.1 0.269Caregiver suicide attempt 1.0 0.4–2.7 0.961Caregiver mentally ill 1.2 0.6–2.1 0.624Caregiver mental health treatment 1.2 0.7–2.1 0.573

Model 4d

GBG 0.4 0.2–0.9 0.044*

Female 1.0 0.7–1.5 0.842White Non-Hispanic 1.5 0.7–3.2 0.323Aggressive, disruptive behavior 1.0 0.6–1.8 0.980Depressive symptoms 1.2 0.7–2.1 0.596Anxious symptoms 1.6 0.6–4.1 0.362Aggressive, disruptive behavior × GBG 0.8 0.3–2.9 0.824Depressive symptoms × GBG 1.1 0.2–5.5 0.943Anxious symptoms × GBG 1.4 0.1–14.1 0.776

a Unadjusted.b Adjusted for sex, race, cohort, first grade aggressive, disruptive behavior, and symptoms.c Adjusted for sex, race, cohort, first grade aggressive, disruptive behavior, and symptoms, as well as caregiver variables.

d sym

Cs

f10aaims1a(1fsrt

iafsMoqdpeti

t

d Adjusted for sex, race, cohort, first grade aggressive, disruptive behavior, an* p < 0.05.

ohort 1 were not statistically significant (all p > 0.05, data nothown in table).

Gender-stratified results revealed that the estimates foremales were very similar to the overall estimates. For Cohortfemales, the unadjusted association showed a relative risk of

.40 (95% CI 0.2, 0.9; p = 0.042, data not shown in a table) anddjustment for baseline mental health and demographic char-cteristics did not change the estimate appreciably. The GBGmpact was statistically significant in all models. In the series of

odels, caregiver SI was the only covariate that was statisticallyignificant (RR = 3.4; 95% CI 1.1, 10.7; p = 0.037). In the Cohortmale analyses, the GBG had a marginal but non-robust associ-

tion with SI with or without adjustment for baseline covariatesRR = 0.4, 95% CI 0.2, 1.1, p = 0.067; RR = 0.4, 95% CI 0.2,.1, p = 0.083). The association weakened in models adjusting

or the series of covariates. The effects for males on SI were thuslightly smaller, with larger p-values than those for females. Theelative risk estimate for caregiver SI was even stronger for maleshan females.

Gbtw

ptoms, as well as interactions with GBG.

Relative risk estimates for ML were all non-significant, butn the hypothesized direction (RR < 1.0) for Cohort 1 femalesnd in the opposite direction (RR > 1.0) for Cohort 1 males. Foremales in Cohort 1, one interaction term (depression × ML) wasignificant in Model 4 (RR = 5.7, 95% CI 1.2, 28.6; p = 0.033).

L-assigned females in the top quartile for depression in fallf first grade were found to experience increased risk for subse-uent SI. Of note is the fact that neither baseline aggressive,isruptive behavior nor depressive symptoms by themselvesredicted suicide ideation in males or females. There was novidence of an interaction effect of baseline aggressive, disrup-ive behavior or baseline depressive symptoms with the GBGntervention condition.

Using the same series of models, the association betweenhe GBG and SI was examined in Cohort 2. The impact of the

BG on SI in Cohort 2 was in the same direction as Cohort 1ut not statistically significant compared to internal GBG con-rols (RR = 0.6–0.8, p ≥ 0.2). In the series of adjusted models, SIas four times as common among non-Hispanic White youths

S68 H.C. Wilcox et al. / Drug and Alcohol Dependence 95S (2008) S60–S73

Table 3Impact of GBG on suicide attempt for males and females in Cohort 1

Predictor or covariate Relative odds estimate (adjusted) 95% CI p-Value

Model 1a

GBG 0.5 0.3–0.9 0.041*

Model 2b

GBG 0.5 0.3–1.0 0.065Female 2.2 1.4–3.5 0.001*

White Non-Hispanic 2.0 0.8–5.0 0.142Aggressive, disruptive behavior 1.4 0.9–2.4 0.143Depressive symptoms 1.7 1.0–2.8 0.043*

Anxious symptoms 0.6 0.2–1.6 0.299

Model 3c

GBG 0.6 0.3–1.2 0.130Female 2.2 1.4–3.5 0.001*

White Non-Hispanic 1.8 0.7–4.6 0.243Aggressive, disruptive behavior 1.6 0.9–2.7 0.070Depressive symptoms 1.7 1.0–2.9 0.037*

Anxious symptoms 0.6 0.2–1.7 0.331Caregiver depression 1.4 0.8–2.4 0.289Caregiver ideation 1.7 0.7–4.0 0.250Caregiver suicide threat 0.9 0.3–2.6 0.828Caregiver suicide attempt 1.8 0.6–5.1 0.294Caregiver mentally ill 2.2 1.1–4.2 0.021*

Caregiver mental health treatment 1.3 0.7–2.6 0.404

Model 4d

GBG 0.3 0.1–0.7 0.008*

Female 2.3 1.4–3.6 0.001*

White Non-Hispanic 2.0 0.8–5.1 0.138Aggressive, disruptive behavior 1.1 0.5–2.0 0.884Depressive symptoms 1.5 0.8–2.8 0.199Anxious symptoms 0.4 0.1–1.7 0.204Aggressive, disruptive behavior × GBG 2.2 0.6–8.0 0.220Depressive symptoms × GBG 1.2 0.2–6.2 0.869Anxious symptoms × GBG 6.1 0.5–81.1 0.174

a Unadjusted.b d sym

d symd sym

anSsop

atipewCrGig(m

tr(

3

1et(fsTMt

Adjusted for sex, race, cohort, first grade aggressive, disruptive behavior, anc Adjusted for sex, race, cohort, first grade aggressive, disruptive behavior, and Adjusted for sex, race, cohort, first grade aggressive, disruptive behavior, an* p < 0.05.

s within other racial/ethnic groups (RR = 3.9–4.3, p ≤ 0.006),early all of which were African American. Those with caregiverI were three and a half times more likely to report SI them-elves (RR = 3.5; 95% CI 1.6, 7.4; p = 0.001). The impact of MLn SI was null compared to internal ML controls (RR = 1.2–1.5,≥ 0.2).

Whereas the Cohort 1 female results were consistent withrobust GBG-associated protection against SI, this was not

he case for the Cohort 2 replication. For females, the GBGmpact on SI risk was modest (RR = 0.4–0.8; p ≥ 0.1) as com-ared to female internal GBG controls. White non-Hispanicthnicity and caregiver SI were the only two covariates thatere associated with SI by young adulthood (RR = 4.6; 95%I 1.2, 17.6; p = 0.027; RR = 3.3; 95% CI 1.1, 9.9; p = 0.032,

espectively from the best fitting model). Estimates for theBG impact on SI for Cohort 2 males as compared to male

nternal GBG controls were inverse but weaker, and care-iver SI was the only covariate that was associated with SIRR = 3.6, 95% CI 1.0, 13.1; p = 0.049). For both females andales, results for ML were all non-significant as compared

vamm

ptoms.ptoms, as well as caregiver variables.ptoms, as well as interactions with GBG.

o respective female and male internal ML controls, but theesults were in the opposite direction than the GBG impactRR > 1.0).

.4. GBG and ML impact on suicide attempt

Corresponding estimates for GBG and risk of SA for Cohortare presented in Table 3. The overall summary relative risk

stimate for occurrence of SA is consistent with an inverse (pro-ective) association with assignment to the GBG interventionRR = 0.5, 95% CI 0.3, 0.9; p = 0.041). Subsequently, adjustingor an array of covariates, this association of the GBG retainedtrength, but for some models lost statistical significance (seeable 3, Model 2 GBG RR = 0.5, 95% CI 0.3–1.0, p = 0.065;odel 3 GBG RR = 0.6, 95% CI 0.3, 1.2; p = 0.130). Covariate

erms for baseline by intervention interactions yielded no sign of

ariation in intervention impact as a function of baseline char-cteristics. Female gender, baseline depression, and caregiverental illness retained statistically significance. The best fittingodel for SA in Cohort 1 included terms for GBG (RR = 0.5;

ohol D

9C

wtdgsCdwaStdi(eti

maamccCpw

svScess

3

eawG1uSAPiasTsc

ofarnm

4

ostittoahtwiwTmnacamsy

d2srDwAhbrFp

taesnct

H.C. Wilcox et al. / Drug and Alc

5% CI 0.3, 0.9; p = 0.031) and female gender (RR = 2.0; 95%I 1.3, 3.1; p = 0.002).

We also explored subgroup variation via stratified analyses,ith initial attention to gender differences (data not shown in a

able). There was evidence of a GBG-associated SI and SA riskifferences in females, but the interaction between the GBG andender in the Discrete-Time Survival Analysis models were nottatistically significant by conventional standards (p > 0.05). Allohort 1 female models of GBG impact on SA were in the pre-icted direction. In analyses for Cohort 1 males, all estimatesere in the predicted direction (RR < 1.0) except the model that

djusted for caregiver variables. No ML impact estimates forA were statistically significant although all estimates were in

he hypothesized direction (data not in a table). There was evi-ence that females in the highest aggressive, disruptive quartilen the fall of first grade were at increased risk for suicide attemptRR = 7.7; 95% CI 1.4, 42.5; p = 0.020). There was no othervidence of an interaction effect of baseline aggressive, disrup-ive behavior or baseline depressive symptoms with the GBGntervention condition.

For full Cohort 2 analyses please refer to the supplementaryaterial on the journal website. In Cohort 2 gender stratified

nalyses, neither the GBG nor ML were significantly associ-ted with SA in these models. White non-Hispanic ethnicity forales and caregiver suicide threats for females were the only

ovariates in the series of models that were statistically signifi-ant (RR = 5.7; 95% CI 1.2, 27.0; p = 0.028 and RR = 20.2; 95%I 2.9, 142.5; p = 0.003, respectively). There were estimationroblems in some of the Cohort 2 male analyses, as the dataere too sparse for computation.Analyses were conducted individually for the six GBG

chools to see if the impact of the GBG intervention on SI or SAaried by school in Cohort 1. The impact of the GBG on bothI and SA was particularly strong in one urban school that washaracterized as 99% African American and 93% of studentsligible for free lunch; estimates of the GBG impact in the otherchools were consistently favoring the GBG but of somewhatmaller magnitude.

.5. Examination of mediation

The examination of mediation in our study addresses thextent to which change in one or more mediating variablesccount for observed program effects. The following variablesere tested as possible mediators in the association between theBG and suicide ideation and attempt: alcohol use before age6, tobacco before age 16, marijuana use before age 16, inhalantse before age 16, Conduct Disorder symptoms, Academicelf-Competence, Self-Derogation, Parental Monitoring, Overtggression, Covert Aggression, Drug Using Peers, Devianteers and depressive episode that occurred prior to suicide

deation or attempt. For both continuous and categorical medi-tors, the product of coefficient method was used to test the

ignificance of a mediation effect (MacKinnon et al., 2002).his approach treats mediation as the product of two regres-ion coefficients, the regression of the mediator on treatmentonditions, and the partial regression coefficient of the outcome

a1ii

ependence 95S (2008) S60–S73 S69

n the mediator in Discrete-Time Survival Analyses, adjustedor the treatment conditions. We tested both suicide attemptnd suicide ideation and the analyses were conducted sepa-ately for males and females. According to the method used,one of the variables studied showed evidence of potentialediation.

. Discussion

This paper reports on one of a few but growing set of epidemi-logically based randomized prevention trials to study children’school achievement, behavior, and psychological well-being athe time of entry into first grade and to conduct assessmentsn young adulthood. In this classroom-based randomized trialhe results support the hypothesis that first graders assignedo GBG classrooms experienced subsequent lower incidencef suicidality through childhood, adolescence, and into youngdulthood compared to internal GBG controls. They reportedalf the lifetime rates of ideation and attempts compared toheir matched controls. For ideation, the beneficial GBG effectas consistent regardless of whether baseline covariates were

ncluded. The GBG effect on attempts was less definitive oncee controlled for gender and baseline depressive symptoms.his beneficial impact of the GBG was consistent for Cohort 1ales and females, who show similar relative risks. There was

o consistent finding of positive relations between first gradeggressive, disruptive behavior or depressive symptoms and sui-idality. Caregiver suicide ideation had a strong and consistentssociation with offspring’s suicide ideation, especially amongales, and retrospective reports on caregiver mental illness and

uicide threats were also related to offspring suicide attempts byoung adulthood.

This study is one of the few universal population-based ran-omized trials to evaluate impact on suicidality (Brown et al.,007). Most other trials have been based on selected high-riskamples. Recently, however, there have been trials on a cur-iculum that teaches self-awareness of depression (Aseltine andeMartino, 2004) and on a gatekeeper program to detect thoseho are suicidal (Brown et al., 2006; Wyman et al., 2008).lthough suicidality in young people represents a critical publicealth issue, the relatively low cumulative incidence of suicidalehaviors necessitates relatively large sample sizes or specialandomized designs (Brown et al., 2006; see also Brown andaraone, 2004). This study gains power from a substantial sam-le size and a long follow-up period.

A few study limitations merit attention. The results seen inhis trial can only be directly generalized to the given populationnd time period from which the sample came. This issue of gen-ralizablity can be addressed only through replication in otherchool districts with similar and different social, ethnic, and eco-omic characteristics. A trial of the GBG with a similar cohort ofhildren has been conducted in the Netherlands and the findingsend to provide evidence of effectiveness of the GBG (van Lier et

l., 2005). The beneficial GBG findings for suicidality in Cohortwere not replicated in Cohort 2, although most Cohort 2 find-

ngs were in a consistent direction. The GBG was implementedn the second cohort with less precision because sufficient men-

S ohol D

ttitccc

nbcisbtyrtwiGtwyaFlotea

aohotsoTsmrg

lswatthds

oa

sTasDsisritahddtnomsi

fitsfitaapemrrhTitttwt(swtCtisiaI

70 H.C. Wilcox et al. / Drug and Alc

oring and monitoring procedures were not in place, resulting inwo possible consequences: a diminished impact and/or a shift inmpact (see Kellam et al., 2008). There is also a hypothetical sta-istical explanation for these diminished findings in the secondohort. Classroom variability and heterogeneity across the threeontrol conditions were markedly higher in the second cohortompared to the first; hence, statistical power was reduced.

Another limitation involves measurement. Suicide ideationecessarily requires some degree of self-report assessmentecause many people with suicidal ideation do not come to clini-al attention and this aspect of psychological well-being requiresnterpersonal communication if it is to be known by others. Theelf-report character of the study data and the constrained num-er of survey items on suicidality are both limitations, as washe reliance upon retrospective assessment of age of onset duringoung adulthood. Also, since the caregiver data were collectedetrospectively from the youths, it is possible that the statis-ical model for estimating the GBG impact was mis-specifiedhen terms for caregiver mental illness and suicidality were

ncluded. These terms might well attenuate the impact of theBG through biased underreporting or partial mediation since

hese events could occur after intervention. Relationships foundith caregiver reports might be explained by the fact that suicidalouth would be more likely to know about their parents’ psychi-tric histories than youth who did not show suicidal behavior.inally, the minimal relationship that we found between base-

ine aggressive, disruptive behavior and depressive symptomsn the one hand and suicidality on the other, in contrast tohe known relation between these factors, could be partiallyxplained by the long time interval from elementary school todulthood.

Our analytical method in this multi-level study relied onn analysis that stratified by school, which has the advantagef holding fixed contextual characteristics related to neighbor-ood and school that were not possible to include explicitly inur analyses. While classroom characteristics were not formallyaken into account in this analysis, our conditioning on schoolhould address this factor since within schools there was onlyne control classroom and at most two intervention classrooms.hus, estimates from this analysis, comparing suicidality withinchools should be very similar to estimates from random effectsodels that formally take classroom variation into account. Such

andom effects models also appropriately take into account theroup-based randomization in this study.

It was difficult to examine some of the assumptions under-ying our analysis. We addressed these interdependencies via aurvival analysis approach conditioning on the schools withinhich classrooms were nested. However, multilevel models

lso may prove to be useful to quantify intervention effects inhe presence of interdependencies (e.g., the alternating logis-ic regressions). In addition, we examined some mediationalypotheses, but this trial had limited capacity to carry out moreetailed mediational models across the life course to examine

pecific mechanisms leading to reduced suicidality.

The GBG has been found to have short- and long-term impactn externalizing behaviors among males in Cohort 1 and tomore limited extent in Cohort 2 especially among aggres-

tt

v

ependence 95S (2008) S60–S73

ive, disruptive first grade males (Kellam et al., 1994a, 2008).he long-term impact of the GBG on alcohol and drug abusend dependence disorders (Kellam et al., 2008), mental healthervice use (Poduska et al., 2008), and Antisocial Personalityisorder in young adulthood (Petras et al., 2008) was con-

istently strong particularly among higher risk males, whereasmpact on anxiety and depressive disorders has been found non-ignificant in intent-to-treat analyses (Kellam et al., 2008). Thiseport on suicidality showed a stronger and more consistent GBGmpact in Cohort 1, similar to the findings in our other studies inhis issue; although this suicidality study showed similar impactmong both females and males. The only other outcome thateld for both genders was alcohol abuse and dependency disor-ers, which for females was not related to their early aggressive,isruptive behavior. The reasons for this difference in impact ofhe GBG may be related to alcohol use disorders and suicidalityot being as anti-social as the drug and ASPD/violence impactsf the GBG also reported in this issue. Clearly, the develop-ental epidemiology of these issues and gender differences in

uch studies are a vital next stage in understanding impact andncreasing it for the GBG and other interventions.

The results are generally consistent with life course/socialeld theory regarding the importance of early mastery of social

ask demands in the classroom in promoting later successfulocial adaptation not only in school but also in other main socialelds. This work resonates with that of Durkheim (1897) in

hat when teachers cannot manage their classrooms there is anbsence of clear norms (anomie). The GBG intervention wasimed at setting the social contextual norms, with a goal ofupils’ mastery of the student role. The GBG consists of: (1)stablishing and posting classroom rules, teaching children theeaning of the rules and providing examples of breaking the

ules; (2) the teacher assigning students to teams within the class-oom setting consisting of equal numbers of girls and boys andeterogeneous with regard to aggressive, disruptive behavior.he teacher thereby establishes both classroom norms of behav-

or and group membership of each child; these are reinforced byhe children in each team influencing each other to follow theeacher established rules. We hypothesize that this deliberateeacher led classroom structure around norms and membershipas how the GBG might have influenced both the behavior and

he psychological well-being of the children. van Lier et al.2005) found that, in classrooms with the GBG where the grouptructure was established by the teachers, children who startedith more aggressive, disruptive behavior not only improved, but

hey associated with children later on who were better behaving.hildren are strongly affected by their social contexts, and the

eacher’s ability to improve the social adaptational process mayndeed promote mastery in children, thereby leading to improvedense of self and self-esteem, and thereby raising confidencen one’s competence to deal with future demands in schoolnd beyond (Harris et al., 1990; Kellam et al., 1994a, 1998).n addition, the GBG reduced aggressive, disruptive and off-

ask behavior and presumably therefore allowed teachers moreime/opportunity to teach (Brown, 1993b).

The Good Behavior Game and Mastery Learning are “uni-ersal interventions” received by all children in all classrooms

ohol D

aietfarpfi4retmitp

nTaTcititisfi

C

i

A

tdPArtIdftpyaw

baD

dtf

lsfid

wcMpiBiCchilnStLfi

dt

d

m

A

i

R

A

A

B

B

B

B

H.C. Wilcox et al. / Drug and Alc

nd are not focused upon youths at higher risk. As such, thenterventions are generally economical and do not require muchxtra teacher or student burden; they can be implemented byeachers during the regular school day. The absence of findingsor ML suggests that the GBG effect was specific to the GBG,nd cannot be attributed to added attention in the GBG class-ooms. However, the GBG intervention must be carried out withrecision. The results were more modest in a second cohort ofrst graders with the same teachers, but without the support of0 h of mentoring and monitoring that these same teachers hadeceived during the prior year with the first cohort. Based onxtensive analyses of alternative explanations, we hypothesizehat the absence of retraining and monitoring of program imple-

entation during the second cohort resulted in more modestmpact. In addition, the prevalence of suicidality was lower inhe second cohort as compared to the first cohort, reflecting, inart, that Cohort 1 was an average of 1 year older.

The results point toward the need for training programs forew teachers and in-service training for more experienced ones.eachers are not as often trained in classroom behavior man-gement as strongly as these results suggest they should be.he results also underline the vital importance of the first gradelassroom as a stage setting the course for children’s life coursen school and beyond (Kellam et al., 2008). The implications ofhese findings are that the first grade classroom can have a criticalmpact on the developmental course of psychopathology. Whilehis seems an apparent concept, the strength and clarity of thempact of the precisely directed early GBG intervention empha-izes the validity of early interventions in this critical socialeld.

onflict of interest

All other authors declare that they have no conflicts ofnterest.

cknowledgements

Since 1984, the partnership between our research team andhe Baltimore City Public School System (BCPSS) has con-ucted the three generations of randomized field trials. Aliceinderhughes, Superintendent of BCPSS, Dr. Leonard Wheeler,rea Superintendent, principals, and teachers played essential

oles as partners. We would like to acknowledge the contribu-ions of Dr. Jaylan Turkkan who led the Good Behavior Gamentervention, and Dr. Lawrence Dolan and Dr. Carla Ford whoirected the Mastery Learning Intervention. We are gratefulor the Prevention Science Methodology Group who offeredhoughtful feedback on the analytic strategy employed in thisaper. We would like to extend our deepest gratitude to theouths and families who participated in this project. We wouldlso like to thank Amelia Mackenzie for her important editorialork.

Role of Funding Source: Funding for this study was provided

y NIMH Grants R01 MH42968, P50 MH38725, R01 MH40859nd T32 MH018834, and NIDA grants R01 DA09897 and R01A004392; the NIMH and NIDA had no further role in study

B

ependence 95S (2008) S60–S73 S71

esign; in the collection, analysis and interpretation of data; inhe writing of the report; or in the decision to submit the paperor publication.

Contributors: Dr. Wilcox led data analyses, managed theiterature searches, participated in project management andupervision of the NIDA young adult assessment, wrote therst draft and incorporated coauthor feedback on all subsequentrafts.

Dr. Kellam was P.I. of the main grants that supported thisork throughout the period of the trial and the young adult out-

omes, with the important exceptions of the Prevention Scienceethodology grant led by Dr. Brown and many of the grants sup-

orting studies and measures of drug use from early childhoodnto early adulthood and suicidality led by Dr. Anthony. Bothrown and Anthony collaborated with Kellam in the initial form-

ng of the designs and periodic follow-up. Kellam led the Lifeourse/Social Field and Developmental Epidemiological con-eptual frame that underlies the work, and led the trial reportedere including the measures and choice of the intervention andts lead staff persons and the precision of its implementation. Heed the community and institutional base building and its mainte-ance including the core partnership with Baltimore City Publicchool System. He collaborated in defining the research ques-

ion reported in this paper and the analytic strategies employed.astly, he participated in the writing of the manuscript and itsnal assessment prior to submission.

Drs. Brown, Anthony, and Wang provided supervision onata analyses and interpretation. Brown and Wang also under-ook mediation analysis.

Drs. Poduska and Ialongo participated in manuscript writing,ata design and collection, and project supervision.

All authors contributed to and have approved the finalanuscript.

ppendix A. Supplementary data

Supplementary data associated with this article can be found,n the online version, at doi:10.1016/j.drugalcdep.2008.01.005.

eferences

ndrews, J.A., Lewinsohn, P.M., 1992. Suicidal attempts among older adoles-cents: prevalence and co-occurrence with psychiatric disorders. J. Am. Acad.Child Adolesc. Psychiatry 31, 655–662.

seltine, R.H., DeMartino, R., 2004. An outcome evaluation of the SOS suicideprevention program. Am. J. Publ. Health 94, 446–451.

arrish, H., Saunders, M., Wolf, M., 1969. Good Behavior Game: effects ofindividual contingencies for group consequences on disruptive behavior ina classroom. J. Appl. Behav. Anal. 2, 119–124.

attle, J., 1990. Self-Esteem: The New Revolution. James Battle & Assoc.,Alberta, Canada.

eautrais, A.L., Joyce, P.R., Mulder, R.T., 1996. Risk factors for serious sui-cide attempts among youths aged 13 through 24 years. J. Am. Acad. ChildAdolesc. Psychiatry 35, 1174–1182.

eautrais, A.L., Joyce, P.R., Mulder, R.T., 1997. Precipitating factors and life

events in serious suicide attempts among youths aged 13 through 24 years.J. Am. Acad. Child Adolesc. Psychiatry 36, 1543–1551.

eautrais, A.L., Joyce, P.R., Mulder, R.T., 1998. Psychiatric illness in a NewZealand sample of young people making serious suicide attempts. N. Z.Med. J. 111, 44–48.

S ohol D

B

B

B

B

B

B

B

B

B

B

B

B

B

B

B

C

C

C

C

C

C

C

C

DD

D

E

E

E

F

F

F

F

G

G

G

G

G

G

H

H

H

H

H

I

I

I

I

72 H.C. Wilcox et al. / Drug and Alc

ennett, K.J., Brown, K.S., Boyle, M., Racine, Y., Offord, D., 2003. Does lowreading achievement at school entry cause conduct problems? Soc. Sci. Med.56, 2443–2448.

lock, J., Burns, R., 1976. Mastery learning. In: Shulman, L. (Ed.), Review ofResearch in Education, vol. 4. F.E. Peacock, Itasca, IL, pp. 3–49.

loom, B.S., 1976. Human Characteristics and School Learning. McGraw-Hill,New York.

rent, D.A., 1995. Risk factors for adolescent suicide and suicidal behavior:mental and substance abuse disorders, family environmental factors, andlife stress. Suicide Life Threat. Behav. 25, 52–63.

rent, D.A., Johnson, B., Bartle, S., Bridge, J., Rather, C., Matta, J., Connolly, J.,Constantine, D., 1993. Personality disorder, tendency to impulsive violence,and suicidal behavior in adolescents. J. Am. Acad. Child Adolesc. Psychiatry32, 69–75.

rent, D.A., Perper, J.A., Kolko, D.J., Zelenak, J.P., 1988. The psychologicalautopsy: methodological considerations for the study of adolescent suicide.J. Am. Acad. Child Adolesc. Psychiatry 27, 362–366.

reslow, N.E., Day, N.E., 1980. Statistical methods in cancer research, vol. 1.The analysis of case–control studies. IARC Sci. Publ. 32, 5–338.

rookover, W.B., 1965. Self-Concept of Ability and School Achievement. Officeof Research and Public Information. Michigan State University, East Lans-ing, MI.

rown, C.H., 1993a. Analyzing preventive trials with generalized additive mod-els. Am. J. Community Psychol. 21, 635–664.

rown, C.H., 1993b. Statistical methods for preventive trials in mental health.Stat. Med. 12, 289–300.

rown, C.H., Faraone, S.V., 2004. Prevention of schizophrenia and psychoticbehavior: definitions and methodologic issues. In: Stone, W.S., Faraone,S.V., Tsuang, M.T. (Eds.), Early Clinical Intervention and Prevention inSchizophrenia. Humana Press, Totowa, NJ, pp. 255–284.

rown, C.H., Liao, J., 1999. Principles for designing randomized preventivetrials in mental health: an emerging developmental epidemiology paradigm.Am. J. Community Psychol. 27, 673–710.

rown, C.H., Wyman, P.A., Brinales, J.M., Gibbons, R.D., 2007. The role ofrandomized trials in testing interventions for the prevention of youth suicide.Int. Rev. Psychiatry 19, 617–631.

rown, C.H., Wyman, P.A., Guo, J., Pena, J., 2006. Dynamic wait-listed designsfor randomized trials: new designs for prevention of youth suicide. Clin.Trials 3, 259–271.

urns, J.M., Patton, G.C., 2000. Preventive interventions for youth suicide: arisk factor-based approach. Aust. N. Z. J. Psychiatry 34, 388–407.

airns, R.B., Cairns, B.D., Neckerman, H.J., 1989. Early school dropout: con-figurations and determinants. Child Dev. 60, 1437–1452.

apaldi, D.M., Patterson, G.R., 1989. Psychometric Properties of FourteenLatent Constructs from the Oregon Youth Study. Springer-Verlag, New York.

hilcoat, H.D., Anthony, J.C., 1996. Impact of parent monitoring on initiationof drug use through late childhood. J. Am. Acad. Child Adolesc. Psychiatry35, 91–100.

hilcoat, H.D., Dishion, T.J., Anthony, J.C., 1995. Parent monitoring and theincidence of drug sampling in urban elementary school children. Am. J.Epidemiol. 141, 25–31.

icchetti, D., Schneider-Rosen, K., 1984. Toward a transactional model of child-hood depression. In: Cicchetti, D., Schneider-Rosen, K. (Eds.), ChildhoodDepression: A Developmental Perspective. Jossey-Bass, San Francisco, pp.5–28.

ovington, M., 1989. Self-esteem and failure in school. In: Mecca, A., Smelser,N.J., Vasconcellos, J. (Eds.), The Social Importance of Self-Esteem. UCPress, Berkeley, CA.

ox, D.R., 1972. Regression models and life-tables. J. R. Stat. Soc. B 34,187–220.

rumley, F.E., 1990. Substance abuse and adolescent suicidal behavior. JAMA263, 3051–3056.

olan, L., 1986. Mastery learning as a preventive strategy. Outcomes 5, 20–27.

olan, L.J., Kellam, S.G., Brown, C.H., Werthamer-Larsson, L., Rebok, G.W.,

Mayer, L.S., Laudolff, J., Turkkan, J., Ford, C., Wheeler, L., 1993. The short-term impact of two classroom-based preventive interventions on aggressiveand shy behaviors and poor achievement. J. Appl. Dev. Psychol. 14,317–345.

J

ependence 95S (2008) S60–S73

urkheim, E., 1897. In: Spaulding, J.A., Simpson, G. (Eds.), Suicide: A Studyof Sociology. (Translated, 1951). Free Press, Glencoe, IL.

lliott, D.S., Huizinga, D., Ageton, S.S., 1985. Explaining Delinquency andDrug Use. Sage Publications, Beverly Hills, CA.

nsminger, M.E., Kellam, S.G., Rubin, B.R., 1983. School and family origins ofdelinquency: comparisons by sex. In: Van Dusen, K.T., Mednick, S.A. (Eds.),Prospective Studies of Crime and Delinquency. Kluwer-Nijhoff Publishing,Boston, pp. 73–97.

sposito, C., Clum, G.A., 2002. Psychiatric symptoms and their relation tosuicidal ideation in a high risk adolescent community sample. J. Am. Acad.Child Adolesc. Psychiatry 41, 44–51.

arr, W., 2000. Vital statistics: memorial volume of selections from the reportsand writings. 1885. Bull. World Health Org. 78, 88–95.

arrington, D.P., 1995. The development of offending and antisocial behav-ior from childhood: key findings from the Cambridge Study in DelinquentDevelopment. J. Child Psychol. Psychaitry 360, 929–964.

ergusson, D.M., Lynskey, M.T., 1995. Childhood circumstances, adolescentadjustment, and suicide attempts in a New Zealand birth cohort. J. Am.Acad. Child Adolesc. Psychiatry 34, 612–622.

owler, R.C., Rich, C.L., Young, D., 1986. San Diego Suicide Study II: substanceabuse in young cases. Arch. Gen. Psychiatry 43, 962–965.

arrison, C.Z., Addy, C.L., Jackson, K.L., McKeown, R.E., Waller, J.L., 1991.The CES-D as a screen for depression and other psychiatric disorders inadolescents. J. Am. Acad. Child Adolesc. Psychiatry 30, 636–641.

arrison, C.Z., McKeown, R.E., Valois, R.F., Vincent, M.L., 1993. Aggression,substance use, and suicidal behaviors in high school students. Am. J. PublicHealth 83, 179–184.

ili-Planas, M., Roca-Bennasar, M., Ferrer-Perez, V., Bernardo-Arroyo, M.,2001. Suicidal ideation, psychiatric disorder, and medical illness in a com-munity epidemiological study. Suicide Life Threat. Behav. 31, 207–213.

oldsmith, S.K., Pellmar, T.C., Kleinman, A.M., Bunney, W.E., 2002. ReducingSuicide: A National Imperative. National Academies Press, Washington, DC,pp. 1–516.

ould, M.S., Greenberg, T., Velting, D.M., Shaffer, D., 2003. Youth suicide riskand preventive interventions: a review of the past 10 years. J. Am. Acad.Child Adolesc. Psychiatry 42, 386–405.

uskey, T.R., 1997. Implementing Mastery Learning, second ed. WadsworthPublishing Co., Belmont, CA.

arris, T.O., Brown, G.W., Bifulco, A.T., 1990. Depression and situational help-lessness/mastery in a sample selected to study childhood parental loss. J.Affect. Disord. 20, 27–41.

arter, S., 1985. Manual for the Self-Perception Profile for Children. Universityof Denver, Denver, CO.

awkins, J.D., Catalano, R.F., Miller, J.Y., 1992. Risk and protective factorsfor alcohol and other drug problems in adolescence and early adulthood:implications for substance abuse prevention. Psychol. Bull. 112, 64–105.

awkins, J.D., Doueck, H.J., Lishner, D.M., 1988. Changing teaching practicesin mainstream classrooms to improve bonding and behavior of low achievers.Am. Educ. Res. J. 25, 31–50.

olly, W., 1987. Self-Esteem: Does It Contribute to Students’ Academic Suc-cess? Oregon School Study Council. University of Oregon, Eugene, OR.

alongo, N.S., Edelsohn, G., Kellam, S.G., 2001. A further look at the prognosticpower of young children’s reports of depressed mood and feelings. ChildDev. 72, 736–747.

alongo, N., Edelsohn, G., Werthamer-Larsson, L., Crockett, L., Kellam, S.,1994. The significance of self-reported anxious symptoms in first-gradechildren. J. Abnorm. Child Psychol. 22, 441–455.

alongo, N., McCreary, B.K., Pearson, J.L., Koenig, A.L., Wagner, B.M.,Schmidt, N.B., Poduska, J., Kellam, S.G., 2002. Suicidal behavior amongurban. African American young adults. Suicide Life Threat. Behav. 32,256–271.

alongo, N.S., Werthamer, L., Kellam, S.G., Brown, C.H., Wang, S., Lin, Y.,1999. Proximal impact of two first-grade preventive interventions on the

early risk behaviors for later substance abuse, depression, and antisocialbehavior. Am. J. Community Psychol. 27, 599–641.

uon, H.S., Ensminger, M.E., 1997. Childhood, adolescent, and young adultpredictors of suicidal behaviors: a prospective study of African Americans.J. Child Psychol. Psychiatry 38, 553–563.

ohol D

K

K

K

K

K

K

K

K

K

K

L

L

M

M

M

P

P

P

R

R

R

R

R

R

R

R

S

S

S

S

S

S

S

S

U

v

V

W

W

H.C. Wilcox et al. / Drug and Alc

aplan, H.B., Pokorny, A.D., 1969. Self-derogation and psychosocial adjust-ment. J. Nerv. Ment. Dis. 149, 421–434.

ellam, S.G., Anthony, J.C., 1998. Targeting early antecedents to preventtobacco smoking: findings from an epidemiologically based randomizedfield trial. Am. J. Public Health 88, 1490–1495.

ellam, S.G., Branch, J.D., Agrawal, K.C., Ensminger, M.E., 1975. MentalHealth and Going to School: The Woodlawn Program of Assessment, EarlyIntervention, and Evaluation. University of Chicago Press, Chicago.

ellam, S.G., Brown, C.H., Poduska, J., Ialongo, N., Wang, W., Toyinbo, P.,Petras, H., Ford, C., Windham, A., Wilcox, H.C., 2008. Effects of a uni-versal classroom behavior management program in first and second gradeson young adult behavioral, psychiatric, and social outcomes. Drug AlcoholDepend. 95, S5–S28.

ellam, S.G., Brown, C.H., Rubin, B.R., Ensminger, M.E., 1983. Paths leadingto teenage psychiatric symptoms and substance abuse: developmental epi-demiological studies in Woodlawn. In: Guze, S.B., Earls, F.J., Barrett, J.E.(Eds.), Childhood Psychopathology and Development. Raven Press, NewYork, pp. 17–51.

ellam, S.G., Ling, X., Merisca, R., Brown, C.H., Ialongo, N., 1998. The effectof the level of aggression in the first grade classroom on the course andmalleability of aggressive behavior into middle school. Dev. Psychopathol.10, 165–185 [See also Kellam, S.G., Ling, X., Merisca, R., Brown, C.H.,Ialongo, N., 2000. The effect of the level of aggression in the first gradeclassroom on the course and malleability of aggressive behavior into mid-dle school: results of a developmental epidemiology-based prevention trial:Erratum. Dev. Psychopathol. 12, 107].

ellam, S.G., Rebok, G.W., Ialongo, N., Mayer, L.S., 1994a. The course andmalleability of aggressive behavior from early first grade into middle school:results of a developmental epidemiologically-based preventive trial. J. ChildPsychol. Psychiatry 35, 259–282.

ellam, S.G., Rebok, G.W., Mayer, L.S., Ialongo, N., Kalodner, C.R., 1994b.Depressive symptoms over first grade and their response to a developmentalepidemiologically based preventive trial aimed at improving achievement.Dev. Psychopathol. 6, 463–481.

ellam, S.G., Werthamer-Larsson, L., Dolan, L.J., Brown, C.H., Mayer, L.S.,Rebok, G.W., Anthony, J.C., Laudolff, J., Edelsohn, G., Wheeler, L., 1991.Developmental epidemiologically-based preventive trials: baseline model-ing of early target behaviors and depressive symptoms. Am. J. CommunityPsychol. 19, 563–584.

essler, R.C., Borges, G., Walters, E.E., 1999. Prevalence of and risk factors forlifetime suicide attempts in the National Comorbidity Survey. Arch. Gen.Psychiatry 56, 617–626.

ewinsohn, P.M., Rohde, P., Seeley, J.R., 1994. Psychosocial risk factors forfuture adolescent suicide attempts. J. Consult. Clin. Psychol. 62, 297–305.

ewinsohn, P.M., Rohde, P., Seeley, J.R., 1996. Adolescent suicidal ideation andattempts: prevalence, risk factors, and clinical implication. Clin. Psychol.Sci. Prac. 3, 25–46.

acKinnon, D.P., Lockwood, C.M., Hoffman, J.M., West, S.G., Sheets, V., 2002.A comparison of methods to test mediation and other intervening variableeffects. Psychol. Methods 7, 83–104.

cIntosh, J.L., 2008. U.S.A. Suicide: 2003 Official Final Data. AmericanAssociation of Suicidology. http://www.suicidology.org/associations/1045/files/2003data.pdf.

urray, C.J., Lopez, A.D. (Eds.), 1996. The Global Burden of Disease: A Com-prehensive Assessment of Mortality and Disability from Diseases, Injuries,and Risk Factors in 1990 and Projected to 2020. Harvard School of PublicHealth. Harvard University Press, Cambridge, MA.

etras, H., Kellam, S.G., Brown, C.H., Muthen, B., Ialongo, N.S., Poduska,J.M., 2008. Developmental epidemiological courses leading to antisocialpersonality disorder and violence and criminal behavior: effects by youngadulthood of a universal preventive intervention in first-and second-gradeclassrooms. Drug Alcohol Depend 95, S45–S59.

etronis, K.R., Samuels, J.F., Moscicki, E.K., Anthony, J.C., 1990. An epi-

demiologic investigation of potential risk factors for suicide attempts. Soc.Psychiatry Psychiatr. Epidemiol. 25, 193–199.

oduska, J., Kellam, S., Wang, W., Brown, C.H., Ialongo, N., Toyinbo,P., 2008. Impact of the Good Behavior Game, a universal classroom-based behavior intervention, on young adult service use for problems

W

ependence 95S (2008) S60–S73 S73

with emotions, behavior, or drugs or alcohol. Drug Alcohol Depend 95,S29–S44.

eid, J.B., 1993. Prevention of conduct disorder before and after schoolentry: relating interventions to development findings. Dev. Psychopathol.5, 243–262.

eid, J.B., Eddy, J.M., 1997. The prevention of antisocial behavior: some con-siderations in the search for effective interventions. In: Stoff, D.M., Breiling,J., Maser, J.D. (Eds.), Handbook of Antisocial Behavior. John Wiley & Sons,New York, pp. 343–356.

eid, J.B., Eddy, J.M., Fetrow, R.A., Stoolmiller, M., 1999. Description andimmediate impacts of a preventive intervention for conduct problems. Am.J. Community Psychol. 27, 483–517.

eid, J.B., Patterson, G.R., Snyder, J.J. (Eds.), 2002. Antisocial Behavior inChildren and Adolescents: A Developmental Analysis and the Oregon Modelfor Intervention. American Psychological Association, Washington, DC.

einherz, H.Z., Giaconia, R.M., Silverman, A.B., Friedman, A., Pakiz, B., Frost,A.K., Cohen, E., 1995. Early psychosocial risks for adolescent suicidalideation and attempts. J. Am. Acad. Child Adolesc. Psychiatry 34, 599–611.

obins, L.N., Helzer, J.E., Croughan, J., Ratcliff, K.S., 1981. National Instituteon Mental Health Diagnostic Interview Schedule: its history, characteristics,and validity. Arch. Gen. Psychiatry 38, 381–389.

obins, L.N., Helzer, J.E., Croughan, J., Ratcliff, K.S., 1994. National Instituteon Mental Health Diagnostic Interview Schedule (DIS). In: Mezzich, J.E.,Jorge, M.R., Salloum, I.M. (Eds.), Psychiatric Epidemiology: Assessment,Concepts and Method. The Johns Hopkins University Press, Baltimore, MD.

uneson, B., 1989. Mental disorder in youth suicide. DSM-III-R Axes I and II.Acta. Psychiatr. Scand. 79, 490–497.

chlesselman, J.J., 1982. Case–Control Studies: Design, Conduct, Analysis.Oxford University Press, New York.

haffer, D., Garland, A., Gould, M., Fisher, P., Trautman, P., 1988. Preventingteenage suicide: a critical review. J. Am. Acad. Child Adolesc. Psychiatry27, 675–687.

haffer, D., Gould, M.S., Fisher, P., Trautman, P., Moreau, D., Kleinman, M.,Flory, M., 1996. Psychiatric diagnosis in child and adolescent suicide. Arch.Gen. Psychiatry 53, 339–348.

haffer, D., Pfeffer, C., 2001. Practice parameters for the assessment and treat-ment of children and adolescents with suicidal behavior. J. Am. Acad. ChildAdolesc. Psychiatry 40, 24S–51S.

hafii, M., Carrigan, S., Whittinghill, J.R., Derrick, A., 1985. Psychologicalautopsy of completed suicide in children and adolescents. Am. J. Psychiatry142, 1061–1064.

hafii, M., Steltz-Lenarsky, J., Derrick, A.M., Beckner, C., Whittinghill, J.R.,1988. Comorbidity of mental disorders in the post-mortem diagnosis of com-pleted suicide in children and adolescents. J. Affect. Disord. 15, 227–233.

inger, J.D., Willet, J.B., 1994. Designing and analyzing studies on onset, cessa-tion, and relapse: using survival analysis in drug abuse prevention research.NIDA Res. Monogr. 142, 196–263.

tanley, M., Stanley, B., Traskman-Bendz, L., Mann, J.J., Meyendorff, E., 1986.Neurochemical findings in suicide completers and suicide attempters. Sui-cide Life Threat. Behav. 16, 286–300.

nited States Public Health Service, 1999. The Surgeon General’s Call to Actionto Prevent Suicide. United States Public Health Service, Washington, DC.

an Lier, P.A.C., Vuijk, P., Crijnen, A.A.M., 2005. Understanding mechanismsof change in the development of antisocial behavior: the impact of a universalintervention. J. Abnorm. Child Psychol. 33, 521–535.

elez, C.N., Cohen, P., 1988. Suicidal behavior and ideation in a communitysample of children: maternal and youth reports. J. Am. Acad. Child Adolesc.Psychiatry 27, 349–356.

erthamer-Larsson, L., Kellam, S.G., Wheeler, L., 1991. Effect of first gradeclassroom environment on child shy behavior, aggressive behavior, and con-centration problems. Am. J. Community Psychol. 19, 585–602.

illet, J.B., Singer, J.D., 1993. Investigating onset, cessation, relapse, and recov-ery: why you should, and how you can, use discrete-time survival analysis

to examine event occurrence. J. Consult. Clin. Psychol. 61, 952–965.

yman, P.A., Brown, C.H., Inman, J., Cross, W., Schmeelk-Cone, K., Guo, J.,Pena, J.B., 2008. Randomized trial of a gatekeeper program for suicide pre-vention: 1-year impact on secondary school staff. J. Consult. Clin. Psychol.76 (1), 104–115.