Effectiveness of a school-based group psychotherapy program for war-exposed adolescents: a randomized controlled trial

Copyright @ 2008 American Academy of Child and Adolescent Psychiatry. Unauthorized reproduction of this article is prohibited.

Effectiveness of a School-Based Group PsychotherapyProgram for War-Exposed Adolescents: A Randomized

Controlled TrialCHRISTOPHER M. LAYNE, PH.D., WILLIAM R. SALTZMAN, PH.D.,

LANDON POPPLETON, B.A., GARY M. BURLINGAME, PH.D., ALMA PASALIC, M.S.,ELVIRA DURAKOVIC, PH.D., MIRJANA MUSIC, M.A., NIHADA CAMPARA, M.D.,

NERMIN ÐAPO, PH.D., BERINA ARSLANAGIC, M.D., ALAN M. STEINBERG, PH.D., AND

ROBERT S. PYNOOS, M.D., M.P.H.

ABSTRACT

Objective: To evaluate the comparative effectiveness of a classroom-based psychoeducation and skills intervention (tier

1) and a school-based trauma- and grief-focused group treatment (tier 2) of a three-tiered mental health program for

adolescents exposed to severe war-related trauma, traumatic bereavement, and postwar adversity.Method: A total of 127

war-exposed and predominantly ethnic Muslim secondary school students attending 10 schools in central Bosnia who

reported severe symptoms of posttraumatic stress disorder (PTSD), depression, or maladaptive grief and significant

impairment in school or relationships were randomly assigned to one of two experimental conditions. These included either

an active-treatment comparison condition (tier 1), consisting of a classroom-based psychoeducation and skills intervention

alone (n = 61, 66% girls, mean age 16.0 years, SD 1.13) or a treatment condition composed of both the classroom-based

intervention and a 17-session manual-based group therapy intervention (tier 2), trauma and grief component therapy for

adolescents (n = 66, 63% girls, mean age 15.9 years, SD 1.11). Both interventions were implemented throughout the

school year. Distressed students who were excluded from the study due to acute risk for harm (n = 9) were referred for

community-based mental health services (tier 3). Results: Program effectiveness was measured via reductions in

symptoms of PTSD, depression, and maladaptive grief assessed at pretreatment, posttreatment, and 4-month follow-up.

Analysis of mean-level treatment effects showed significant pre- to posttreatment and posttreatment to 4-month follow-up

reductions in PTSD and depression symptoms in both the treatment and comparison conditions. Significant pre- to

posttreatment reductions in maladaptive grief reactions were found only in the treatment condition. Analyzed at the

individual case level, the percentages of students in the treatment condition who reported significant (p < .05) pre- to

posttreatment reductions in PTSD symptoms (58% at posttreatment, 81% at 4-month follow-up) compare favorably to

those reported in controlled treatment efficacy trials, whereas the percentages who reported significant reductions in

depression symptoms (23% at posttreatment, 61% at follow-up) are comparable to, or higher than, those found in

community treatment settings. Lower but substantial percentages of significant symptom reduction were found for PTSD

Accepted April 9, 2008.Drs. Layne, Saltzman, Steinberg, and Pynoos are with the UCLA/Duke

National Center for Child Traumatic Stress; Dr. Burlingame and Mr. Poppletonare with Brigham Young University; Drs. Durakovic and Ðapo are with theUniversity of Sarajevo; Ms.Muxic is with Student Policlinic; Dr. Campara is withTravnik Hospital; Ms. Paxalic is with Sarajevo University Medical Center; andDr. Arslanagic is with UNICEF Bosnia and Herzegovina.

Financial support was provided byUNICEF Bosnia andHerzegovina, the BrighamYoung University Family Studies Center, the David M. Kennedy Center forInternational Studies, the Bing Fund, and Tony Bennett. The authors thank Drs.Michael Lambert and Joseph Olsen for statistical consultation, John-Paul Legerski and

Benjamin Carter for help with the literature review, and Preston Finley for manuscriptpreparation. The authors are also indebted to Rune Sturland, M.S., formerly ofUNICEF Bosnia and Herzegovina, for his pioneering work and support.

Article Plus (online-only) materials for this article appear on the Journal`sWebsite: www.jaacap.com.

This article is the subject of an editorial by Dr. Judith A. Cohen in this issue.Correspondence to Dr. Christopher M. Layne, UCLA National Center for

Child Traumatic Stress, 11150West Olympic Boulevard, Suite 650, Los Angeles,CA 90064; e-mail: [email protected].

0890-8567/08/4709-1048�2008 by the American Academy of Child andAdolescent Psychiatry.

DOI: 10.1097/CHI.0b013e31817eecae

1048 WWW.JAACAP.COM J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 47:9, SEPTEMBER 2008


(33% at posttreatment, 48% at follow-up) and depression symptoms (13% at posttreatment; 47% at follow-up) in students

in the comparison condition. The odds of significant symptom reduction were higher for PTSD symptoms at both

posttreatment and 4-month follow-up and for maladaptive grief at posttreatment (no follow-up was conducted on

maladaptive grief). Rates of significantly worsened cases were generally rare in both the treatment and comparison

conditions. Conclusions: A three-tiered, integrative mental health program composed of schoolwide dissemination of

psychoeducation and coping skills (tier 1), specialized trauma- and grief-focused intervention for severely traumatized and

traumatically bereaved youths (tier 2), and referral of youths at acute risk for community-based mental health services (tier

3) constitutes an effective and efficient method for promoting adolescent recovery in postwar settings. J. Am. Acad. Child

Adolesc. Psychiatry, 2008;47(9):1048Y1062. Key Words: posttraumatic stress disorder, depression, grief, school-based

intervention. Clinical trial registration informationVRandomized Controlled Trial of the Effectiveness of Group Treatment

with War-Exposed Bosnian Adolescents. URL: http://www.clinicaltrials.gov. Unique identifier: NCT00480480.

Children and adolescents have been directly exposed toviolent conflict at increasing rates.1 The United NationsChildren`s Fund (UNICEF) reports that civilian warcasualties have increased from 5% to 90% during the past20 years, with children accounting for at least half ofthose seriously injured.2 Commonly reported reactions inwar-exposed youths include symptoms of posttraumaticstress disorder (PTSD), anxiety, and depression, griefreactions, somatic complaints, social withdrawal, rest-lessness, externalizing behavior, separation anxiety, age-inappropriate dependence on caregivers, and behavioral,academic, and interpersonal difficulties.3Y9

The 1991Y1995 civil war in the former Yugoslaviaexemplifies the brutal nature and ravaging effects ofmodern armed conflict. In Bosnia the war claimedapproximately 100,000 lives, with civilians comprising alarge proportion of war-related casualties.9 Protractedsieges of cities, ethnic cleansing, shelling of denselypopulated areas, prison camps, detention centers, andgenocidal campaigns resulted in multiple traumaexposures, traumatic losses, and severe hardship on amassive scale, including the forced displacement ofnearly 1.5 million people.3,10 Surveys of children andadolescents across the region revealed high rates ofviolent deaths of parents or family members andassociated distress reactions including PTSD, depres-sion, and grief.9Y12

Beginning in 1992, UNICEF Bosnia and Herzego-vina (B&H) engaged members of the UCLA TraumaPsychiatry Program to provide consultation and trainingto Bosnian government representatives and educators.In 1996, shortly after the formal cessation of hostilities,the UCLA team conducted a needs assessment thatreviewed psychosocial services provided to Bosnianchildren and adolescents during and after the war.13 The

report concluded that many youths were experiencingpersisting severe symptoms of PTSD, traumatic grief,and functional impairment and were at risk forbecoming a Blost generation^ due to major disruptionsin their developmental trajectories.Team members then partnered with UNICEF B&H,

government officers from the Federation of B&H andthe Republika Srpska, and local universities and schoolsystems, with the mission of implementing school-basedmental health services for war-exposed adolescents. Theprimary aims of these school-based services includedreducing mental distress and promoting adaptivefunctioning in a culturally sensitive and developmentallyappropriate manner, building local professional capacityand program sustainability, and creating safeguards toreduce the risk for iatrogenic outcomes.This randomized controlled treatment outcome study

investigated the comparative effectiveness of the firsttwo tiers of a three-tiered school- and community-basedmental health program as implemented with Bosniansecondary students exposed to severe trauma, traumaticbereavement, and adversity. The two tiers under studyconsisted of a classroom-based psychoeducation andskills intervention (tier 1) and a classroom-basedintervention paired with a manual-based 17-sessiontrauma- and grief-focused group treatment (tier 2)consisting of trauma and grief component therapy foradolescents (TGCT).14 The content of the tier 1classroom-based intervention, consisting of psychoedu-cation, training in skills for managing trauma and lossreminders, and other coping skills, was taken fromselected modules of TGCT.TGCT is an assessment-driven, manual-based psy-

chotherapy protocol specifically designed for adolescentswhose histories of exposure to trauma, traumatic loss, and

TRAUMA AND GRIEF COMPONENT THERAPY

WWW.JAACAP.COM 1049J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 47:9, SEPTEMBER 2008


severe adversity place them at high risk for severe persistingdistress, functional impairment, and developmental dis-ruption. TCGT is informed by a developmental modelof child traumatic stress15 and is based on a wellness-oriented public health framework.16 TCGT is a flexibleintervention, specific components of which are pre-scribed as indicated by assessment results. Treatmentcomponents are drawn from a broad evidence base ofempirically based therapeutic models and approaches17Y20

and consist of adolescent-appropriate therapeutic strate-gies designed to promote specific adolescent develop-mental tasks that are adversely influenced by exposure totrauma, traumatic loss, and severe adversity.The primary aims of TCGT are to reduce distress and

dysfunction, enhance current positive adaptation, andpromote healthy developmental progression and pre-paration for the roles and responsibilities of adulthoodand mature citizenship.14,21,22 Special features of theintervention include the following:

• Providing psychoeducation concerning trauma and lossreminders and the traumatic stress and grief reactionsthat they evoke (including common adolescent con-cerns about going crazy, being different, or beingirreparably damaged)

• Enhancing coping with intense negative emotions toaddress adolescent social withdrawal, preoccupationwith desires for revenge, and temptations to reducedistress through substance use or other risky behaviors

• As appropriate, conducting in-depth trauma processing,giving special attention to existential dilemmas (i.e.,painful quandaries during traumatic events in whichone is forced to choose between self-protection andassisting others in distress)22 and intervention thoughts(i.e., distressing ongoing preoccupations about how thetraumatic injury or loss could have been prevented,protected against, or repaired, including those linked toissues of human accountability and accompanyingintense negative emotions and cognitions, includingguilt, shame, bewilderment, rage, and helplessness)

• As appropriate, grief processing designed to helptraumatic grief reactions to recede while addressingthe impact of the loss on adolescent identity, life plansand purpose, and future outlook

• Building social support skills to bridge social estrange-ment and facilitate coping with adversities

• Enhancing problem-solving skills, including thoseneeded to address lost developmental opportunities,

age-appropriate developmental tasks (e.g., forming age-appropriate friendships and romantic attachments,preparing for a professional career and for family life,developing personal values and ethics, developingadaptive schemas of the self, others, and the world),and coping with changes in family responsibilities

• Engaging adolescent capacities for insight to under-stand the links between present behavior and traumaand loss reminders and to mentally discriminatebetween reminder-laden situations and past traumaticcircumstances

• Reappraising traumatic expectations (i.e., trauma-induced pessimistic alterations in basic belief systemsand core values that are dysfunctional given one`spresent life context)22 that interfere with life plans andactivities that promote a healthy transition toadulthood

Versions of TCGT have been implemented andevaluated with adolescents in various field settings,including postearthquake Armenia (using a 5-yearfollow-up),23 among U.S. students exposed to commu-nity violence,24 in postwar Bosnia,25 and in New YorkCity following the September 11th terrorist attacks,20

showing evidence of significant symptom improvementin all of the studies. However, before this study, norandomized controlled evaluation of a manual-basedversion of TCGT had yet been conducted.Between 1997 and 2001, a total of 33 secondary

schools throughout B&H implemented TCGT throughtheir participation in the UNICEF School-Based Psy-chosocial Program for War-Exposed Adolescents. Theeffectiveness of TCGT with war-exposed Bosnianadolescents was first evaluated in the Republika Srpskaduring the 2000Y2001 school year using an open-trialdesign.25 Analyses revealed significant pre- to posttreat-ment reductions in PTSD, depression symptoms, andgrief reactions. Reductions in distress symptoms werepositively correlated with classroom rule compliance andschool interest and inversely correlated with anxiety andsocial withdrawal. These findings pointed to the potentialbenefit of TCGT in a postwar setting and to the need formore rigorous evaluation.

STUDY OBJECTIVES

The purpose of this study was to evaluate, using a ran-domized controlled design, the comparative effectiveness

LAYNE ET AL.



of two tiers of a three-tiered school- and community-based intervention program for adolescents exposed tosevere trauma, traumatic bereavement, and adversity.The program was implemented by local school andcommunity professionals with war-exposed adolescentsattending 10 secondary schools located in postwarcentral Bosnia during the 2000Y2001 school year.Effectiveness was evaluated with respect to pre- toposttreatment reductions in PTSD symptoms, depres-sion symptoms, and maladaptive grief reactions and themaintenance or continuation of treatment gains at 4-month follow-up. The first level of intervention, tier 1,consisted of participation in a classroom-based psychoe-ducation and skills intervention implemented by thetrained group counselors and served as an active-treatment comparison condition. The second level ofintervention, tier 2, consisted of participation in group-based TCGT in addition to the tier 1 classroom-basedintervention and served as the treatment condition.Given that tier 2 intervention consisted of moreintensive psychotherapeutic services than tier 1, hypoth-esis 1 was that group � time interactions would befound for all four treatment outcome measures, suchthat the treatment condition would show significantlygreater pre- to posttreatment reductions compared tothe comparison condition. Specifically, we used randomassignment to conditions in an attempt to producenonsignificant between-group differences on all mea-sures at pretreatment, but hypothesized that thetreatment condition would have significantly lowermean distress scores on all measures at posttreatment.Hypothesis 2 was that, where appropriate to calculate,significant main effects for time would be found forboth the treatment and comparison conditions on all ofthe outcome measures, indicating that participation inboth conditions would be associated with pre- toposttreatment reductions on the study measures.Hypothesis 3 was that participation in the treatmentcondition would be associated with an incrementalbenefit over the comparison condition on all of theoutcome measures, as manifested by a larger odds ratio(OR) in the treatment condition for the proportion ofcases reporting significant pre- to posttreatment reduc-tions in distress (measured by the Reliable ChangeIndex, described below). Hypothesis 4 was that anytherapeutic gains observed within each of the twogroups at posttreatment would be maintained or furtherimprove at 4-month follow-up.

METHOD

Participants

Participants were adolescent students attending 10 secondaryschools located throughout Central Bosnia. The schools comprisedthe total set of schools implementing the UNICEF program in theFederation of B&H during that school year. The treatmentcondition included 66 students, approximately 63% girls and34% boys (sex not reported for two), ranging in age from 13 to 18years (X

–= 15.9, SD 1.11) and from the first to the third year of high

school. The comparison condition included 61 students, approxi-mately 66% girls and 34% boys, ranging in age from 14 to 19 years(X–= 16.0, SD 1.13) and from the first to the third year of school.

Because this region is predominantly populated by Bosnian-speakingethnic Muslims, it is likely that the study participants were almostexclusively ethnic Muslims.

Study Procedures

In fall 2000, in accordance with the implementation proto-col,26,27 16 trained school counselors (composed of psychologistsand pedagogues; the duties of pedagogues most closely resemblethose of guidance counselors in U.S. high schools) administered aself-report risk screening survey to selected classrooms in theirschools known to have high proportions of severely affected war-exposed students. Approximately 100 students per school completedthe survey. Figure 1 presents a CONSORT flowchart of studentprogress throughout the study. Based on their responses, approxi-mately 20 students per school were identified based on threeinclusion criteria: significant trauma exposure before, during, and/orafter the war (e.g., serious physical injury, life threat, witnessingdeath or serious injury, traumatic bereavement); significant currentdistress, especially severe persisting symptoms of PTSD, depression,or traumatic grief; and significant functional impairment, includingfamily or peer relationships and school performance. With respect toexclusion criteria, students who did not meet the three inclusioncriteria or who did but showed signs of psychosis, represented animminent threat to themselves or others, were unable to attendgroup meetings, were judged not to be appropriate for group-basedintervention due to highly disruptive behavioral or substance abuseproblems, or reluctance to participate in a group setting wereexcluded from participation in the study and, as indicated, providedwith appropriate referrals, with high-risk cases being referred to tier 3community-based traditional psychiatric treatment23 (Fig. 1).The school counselors then invited the students at potential high

risk to participate in a semistructured individual screening interviewin which the counselor reviewed the student`s screening surveyresponses and verbally assessed the students` present relationships,school performance, and interest in group participation. In cases inwhich a history of bereavement was reported, the interviewingcounselor administered the UCLA Grief Inventory (describedbelow) to assess grief reactions in greater depth. Students meetingthe inclusion criteria were then randomly assigned to either thetreatment or the comparison condition at their school by theparticipating school counselors (who randomly drew names ofprogram-eligible students out of a box), with between 6 and 10students per group. Neither the counselors nor participating studentswere blinded to students` experimental condition.The counselors then conducted pregroup individual interviews with

(only) the students assigned to the treatment condition. Interview goalsincluded building rapport, enhancing realistic positive expectations,




assessing coping resources, and increasing students` insight into ways inwhich past traumatic experiences and losses continued to influence theirlives.23 Following the pregroup interviews, the counselors implementedthe group treatment at their schools. Between 17 and 20 weekly groupsessions were held at each school throughout the school year (fromNovember to May), with intermittent breaks for school holidays andexamination weeks. Each group session lasted 60 to 90 minutes.Approval for the study and subsequent use of the data set, including

procedures used to obtain informed consent from caregivers andinformed assent from students, was given locally by an ad hocinstitutional review board made up of representatives from UNICEFB&H and its local partners, and subsequently by the institutionalreview board of Brigham Young University. Data were collectedat pretreatment (October 2000) and at posttreatment (June 2001).In addition, all of the schools were invited to voluntarily participatein a 4-month follow-up questionnaire consisting of only the PTSDand depression measures (October 2001). Given UNICEF budgetconstraints and shortages in local resources (e.g., some counselorsretired during the summer), 5 of the 10 schools volunteered toparticipate in the 4-month follow-up (Fig. 1).

Tier 2 Treatment Protocol: TCGT

TCGT is based on an evidence-informed developmentalpsychopathology model of childhood traumatic stress15 and consistsof a pregroup interview and four treatment modules. A detailed

description of TCGT components is provided in Figure 2, availableonline through the Article Plus feature on the Journal`s Web site atwww.jaacap.com. Module I components include psychoeducationconcerning common distress reactions and the trauma and lossreminders that may evoke them, emotional and behavioral regulationskills, social support skills, and group cohesion-building exercises.Module II components focus on processing youths` traumaticexperiences, including distressing cognitions and mental preoccupa-tions. Module III components focus on promoting beneficialgrieving and include psychoeducation, processing grief reactions,retrieving or constructing a nontraumatic image of the deceased,and reminiscing. Components of module IV include promotingdevelopmental progression, reappraising traumatic expectations,building problem-solving skills, strengthening social connections,reducing risky behavior, addressing lost developmental opportunitiesand current developmental tasks, engaging in constructive socialaction, and planning for the future. Modules are designed to be usedflexibly based on youths` difficulties and strengths. A group-basedmodality was chosen based on evidence from the group psychother-apy outcome literature that group-based treatments for a broad rangeof patient populations are generally equivalent in treatmenteffectiveness, and more cost-effective compared to individuallybased interventions,28 evidence that group processes (e.g., cohesion,member-to-member interaction, group peer-referenced social com-parisons) provide agents of therapeutic change that are not availablein individual treatment modalities,28 and the capacity of skillfully led

Fig. 1 Participant progress (CONSORT flowchart) for trauma and grief component therapy for Adolescents (TGCT).

LAYNE ET AL.



groups to harness the increased receptiveness to peer influencescharacteristic of adolescence to enhance therapeutic benefit.29

Because all of the students in each school had been exposed tomajor trauma and many had been traumatically bereaved, all fourTCGT modules were implemented in each school. In module III(which involves grief processing), nonbereaved students were invitedto process other types of losses, including involuntary separationsfrom family members or close friends, highly valued possessions,their former way of life, and lost developmental opportunities.

Training, Supervision, and Monitoring Fidelity

Training in administering the screening protocol (including the riskscreening survey and the screening interview) and in conductingTCGTwas provided in four 2-day training seminars led by the authors(C.L., W.S., G.B.) in collaboration with local mental healthprofessionals (E.D., M.M., N.C.) who served as supervisors to theschool counselors and as expert cultural consultants. This trainingincluded a 2-day seminar devoted to implementing TCGT in atherapeutic small group setting.30Group supervisionmeetings were ledregionally by the local supervisor every 2 to 4 weeks, as needed,throughout program implementation. Meetings were based on currentstandards of clinical supervision31 and used a variety of methods,including didactic presentations and school-specific consultation bysupervisors, role playing, group discussions, and interschool counselorteam sharing to consolidate the content of the training seminars, refinetherapeutic skills, monitor and promote protocol adherence, andfacilitate collegial networking and support. Counselors prepared forsupervision meetings by creating detailed notes of their groups`beneficial (e.g., group cohesion) and potentially detrimental (e.g.,marginalization of members) processes and challenges to protocolimplementation. These notes were then reviewed in detail at themeeting, followed by consultation on adhering to the protocol andtraining in specific skills with which to implement it.

Tier 1 Intervention: Classroom SkillsYBased

Psychoeducation and Skills

Because they attended the same schools, students in both thetreatment and comparison conditions received a tier 1 classroom-based psychoeducation and skills intervention, which was imple-mented throughout the school year. This intervention consisted ofthe broad dissemination of information contained in modules I andIV of the treatment manual (including psychoeducation regardingcommon distress reactions, coping skills to manage trauma and lossreminders, relaxation training, skills to self-regulate emotions andbehavior, social support skills, and problem-solving skills; see Figure2 online through the Article Plus feature on the Journal`s Web siteat www.jaacap.com). Materials were presented in various forumsthroughout the school, including lectures in psychology andpedagogy classes the counselors taught at their schools, presentationsin head teacher (homeroom) classes, and informal and formal sharingof knowledge and skills by group members with their classmates.30

Instruments

Posttraumatic stress symptoms were measured by the Posttrau-matic Stress Disorder Reaction Index (RI),32 a 17-item self-reportscale of symptom frequency during the previous month. Itemscorrespond to DSM-IV PTSD criteria and are rated on a 5-pointLikert-type scale ranging from never (0) to almost always (4). Thetotal-scale score has shown good internal consistency (" = .87),

criterion-referenced validity in reference to measures of depression,anxiety, somatic complaints, traumatic grief, and existential grief(0.30Y0.70), and test-retest reliability (0.75) among Bosnianadolescents from data collected in 2002 and 2003.17

Depression symptoms were measured using the 18-item self-report Depression Self-Rating Scale (DSRS).33 The original 3-pointscale (designed for children) was modified to a 5-point frequencyscale ranging from never (0) to almost always (4) to increase thescale`s sensitivity to clinical change in adolescent populations. Thisadapted version has shown good internal consistency (" = .85);criterion-referenced validity in reference to measures of posttrau-matic stress, anxiety, somatic complaints, traumatic grief, andexistential grief (0.37 to 0.62); and acceptable 2-week test-retestreliability (0.64) in Bosnian adolescents.17

Maladaptive grief reactions were measured using the self-reportUCLA Grief Inventory.14 Frequency ratings are made on a 5-pointscale ranging from never (0) to almost always (4). Two subscales wereused: Traumatic Grief (in which distress reactions to the circum-stances of the death interfere with adaptive grief processes [sixitems]), and Existential Grief (characterized by the loss of perceivedpurpose and meaning to one`s life following bereavement [sixitems]). These subscales have shown good internal consistency(" = .74 and .89), 2-week test-retest reliability (0.73 and 0.84),and criterion-referenced validity in reference to measures of post-traumatic stress, depression, anxiety, and somatic complaints(0.23Y0.58) in Bosnian adolescent samples.17

Statistical Analyses

Three statistical procedures were used to compare the relativeeffectiveness of the treatment and comparison conditions, two ofwhich were mean based and the third of which was individual casebased. The first procedure was multivariate analysis of variance(MANOVA) performed on the condition means, with time(pretreatment versus posttreatment, posttreatment versus 4-monthfollow-up) as a repeated measure and group (treatment versuscomparison) as a between-subjects fixed factor. The second, used asan index of treatment effect sizes, involved calculating conditionmean difference scores and associated significance tests andconfidence intervals. The third procedure involved the use of arecent advancement in case-based psychotherapy outcome research,the Reliable Change Index (RCI).34,35 The RCI is based on the SE ofthe difference between two test scores and denotes whetherdifferences in test scores (with chance of error typically calculatedat p < .05) reflect statistically reliable (i.e., significant) change insteadof random fluctuation. RCI values consist of a difference score (e.g.,pretreatment minus posttreatment, pretreatment minus follow-up)divided by the SE of the difference set at p < .05 and can be used toclassify study participants according to treatment response on a givenoutcome variable. Those whose difference scores are positive andexceed the SE are Breliably improved^ cases, those whose differencescores are negative and exceed the SE are Breliably deteriorated^cases, and participants whose difference scores do not exceed the SEare Btreatment nonresponders.^ An RCI score was calculated foreach participating student on each measured outcome variable.

RESULTS

Exposure to Trauma and Severe Adversity

Approximately 73% of the students participatingreported experiencing direct life threat arising from close




proximity to exploding shells or rifle fire, 36% reportedwitnessing during the war violent death or seriousinjury, 12% reported witnessing torture, and 46%reported the serious injury of a person to whom theywere close. Furthermore, 55% reported abandoningtheir homes, and 54% reported having to changeschools because of the war. Moreover, 14% reported theviolent death during the war of a nuclear familymember, and 73% reported the violent death of atleast one person to whom they were close. Approxi-mately 12% reported receiving confirmation that adisappeared person to whom they were close was dead,24% reported that a primary wage earner wasunemployed, 20% reported lacking sufficient moneyfor basic necessities, 39% reported having heavy familyresponsibilities, and 20% reported intense interpersonalconflict in their homes.

Pre- to Posttreatment Changes in PTSD

and Depression Symptoms

Separate MANOVAs were conducted on measures ofposttraumatic stress and depression symptoms (mea-sured by the RI and DSRS, respectively). Consistentwith hypothesis 1, a significant group � time interac-tion was found (F1,125 = 6.77, p = .01; Table 1).However, subsequent analyses did not clarify the sourceof the interaction. Post hoc analyses with Bonferroni

correction revealed that the mean posttraumatic stressscores of the treatment and comparison conditionconditions did not significantly differ either at pretreat-ment (p 9 .01) or at posttreatment (p 9 .01).Contrasting within experimental conditions, the meanposttraumatic stress scores of both the treatmentcondition and the comparison condition decreasedsignificantly between pretreatment and posttreatment(both p values <.01). Furthermore, a test of the meandifference scores reached significance in both thetreatment and comparison conditions, indicating thatthe average posttraumatic stress symptom scoredecreased significantly between pretreatment and post-treatment in both groups. Consistent with hypothesis 2,a significant main effect was found for time (F1,125 =54.40, p < .05), indicating that the marginal mean(i.e., mean posttraumatic stress score averaged acrossboth groups) decreased significantly between pretreat-ment (X

–= 33.99, SD 13.13) and posttreatment (X

–=

25.60, SD 13.05).Consistent with hypothesis 1, a significant group �

time interaction was found for DSRS scores (F1,123 =6.16, p < .05; Table 1). Standard post hoc analyses withBonferroni correction did not clarify the source of theinteraction in that the means of the treatment andcomparison conditions did not significantly differ eitherat pretreatment (p 9 .01) or posttreatment (p 9 .01).

TABLE 1Mean, Pre-/Posttreatment Difference Scores, and Confidence Intervals of PTSD and Depression Scores

Group

Posttraumatic Stressa Depressionb

PretreatmentMean (SD)

PosttreatmentMean (SD)

MeanDifferenceScorec (SD)

95% CI forDifference Score Pretreatment

Mean (SD)PosttreatmentMean (SD)

MeanDifferenceScore (SD)

95% CI forDifferenceScore

Lower Upper Lower Upper

Treatment 36.37(14.27),n = 66

24.52(13.61),n = 66

j11.85***(13.96),n = 66

j15.28 j8.42 32.61(11.39),n = 65

29.93(6.51),n = 65

j2.69*(10.64),n = 65

j5.33 j0.06

Comparison 33.02(10.27),n = 61

27.35(12.31),n = 61

j5.67**(12.71),n = 61

j8.93 j2.42 28.61(9.86),n = 60

30.52(5.74),n = 60

1.91(10.05),n = 60

j0.68 4.51

Note: All missing data managed by decision rule: If =80% of responses to scale items are present, calculate mean item score and multiply bytotal number of items; otherwise exclude case from all analyses on that variable. Analyses were then performed using students with complete dataon the variables under study. CI = confidence interval.

aUCLA Reaction Index-Revised.bDepression Self-Rating Scale.cDifference score = [(posttreatment score) - (pretreatment score)].*p < .05; **p < .01; ***p < .001.

LAYNE ET AL.



However, consistent with hypothesis 1, a test of themean difference scores reached significance only in thetreatment condition, indicating that the average depres-sion symptom score decreased significantly betweenpretreatment and posttreatment only in the treatmentcondition. No test of main effect for time was conducteddue to the interaction.

Pre- to Posttreatment Changes in Grief Reactions

Restricting the analysis to students who reported atleast one bereavement and who completed the GriefIndex, separate MANOVAs were then conducted on thetwo grief subscales. A significant group � time in-teraction was found for traumatic grief (F1,51 = 19.69,p < .01; Table 2). Post hoc analyses with Bonferronicorrection revealed that despite randomization, themean of the treatment condition was significantly higherthan the mean of the comparison condition at pre-treatment (p < .01), whereas the means of the treatmentcondition and comparison condition did not signifi-cantly differ at posttreatment (p 9 .01). Contrastingwithin experimental conditions, the mean of thetreatment condition was significantly higher at pretreat-ment than at posttreatment (p < .01), whereas the meanof the comparison condition did not significantly differbetween pretreatment and posttreatment (p 9 .01).Consistent with this finding, a test of the meandifference scores reached significance only in thetreatment condition (p < .001), indicating that theaverage traumatic grief score decreased significantly

between pre- and posttreatment only in the treatmentcondition. No test of main effect for time was conducteddue to the interaction.A significant group � time interaction was found for

existential grief (F1,53 = 7.96, p < .01; Table 2). Post hocanalyses with Bonferroni correction revealed that despiterandomization, the mean of the treatment condition wassignificantly higher than that of the comparisoncondition at pretreatment (p < .01), whereas themeans of the treatment and comparison conditionsdid not significantly differ at posttreatment (p 9 .01).Contrasting within conditions, the mean of thetreatment condition was significantly higher at pretreat-ment than at posttreatment (p < .01), whereas the meanof the comparison condition did not significantly differbetween pre- and posttreatment (p 9 .01). Consistentwith this finding, a test of the mean difference scoresreached significance only in the treatment condition (p <.001), indicating that the average existential grief scoredecreased significantly between pre- and posttreatmentonly in the treatment condition. No test of main effectfor time was conducted due to the interaction.

Pre- to Posttreatment RCI Values and Odds Ratios

Table 3 provides RCI values for reliably improvedand reliably deteriorated cases within each condition.Reliable improvement in the treatment conditionranged from 58% of cases for PTSD symptoms to23% of cases for depression symptoms. Reliableimprovement in the comparison condition ranged

TABLE 2Mean, Pre-/Posttreatment Difference Scores, and Confidence Intervals of Grief Scores

Group

Traumatic Griefa Existential Griefb


PosttreatmentMean (SD)

MeanDifferenceScorec (SD)

95% CI forDifferenceScore Pretreatment

Mean (SD)PosttreatmentMean (SD)

MeanDifferenceScore(SD)

95% CI forDifferenceScore


Treatment 13.65(5.37),n = 40

8.76(4.71),n = 40

j4.89***(4.87),n = 40

j6.45 j3.33 13.94(5.51),n = 40

10.22(6.06),n = 40

j3.72***(4.93),n = 40

j5.29 j2.14

Comparison 9.58(4.23),n = 24

10.90(4.86),n = 24

1.32(4.25),n = 24

j0.48 3.11 10.92(4.60),n = 26

11.45(4.63),n = 26

0.52(4.57),n = 26

j1.32 2.37

Note: CI = confidence interval.aUCLA Grief Index, Traumatic Grief subscale.bUCLA Grief Index, Existential Grief subscale.cDifference score = [(posttreatment score) - (pretreatment score)].*p < .05; **p < .01; ***p < .001.




from 33% of cases for PTSD symptoms to 8% of casesfor traumatic and existential grief. Reliable deteriorationwas generally rare in the treatment condition, rangingfrom 9% of cases in depression scores to 0% in the griefsubscales; deterioration occurred more frequently in thecomparison condition, ranging from 17% of cases indepression scores to 8% of cases in traumatic grief.Table 4 presents ORs and associated 95% confidence

intervals, which were computed to compare relativerates of reliable improvement between the treatment

and comparison conditions. ORs were calculated bytaking the ratio of the odds of improvement in thetreatment condition to the odds of improvement in thecomparison condition. An OR of 1.0 indicates an equallikelihood of reliable change in both the treatment andcomparison conditions, an OR 9 1.0 indicates greaterlikelihood of reliable change in the treatment condition,and an OR < 1.0 indicates lower likelihood of reliablechange in the treatment condition. Consistent withhypothesis 3, the odds of reliable pre- to posttreatment

TABLE 3Reliable Change Indices (RCIs) (p < .05)a for Pre-to Posttreatment

Outcome Variable

Significantly Improved Cases Significantly Deteriorated Cases

Treatment Condition Comparison Condition Treatment Condition Comparison Condition

nNo. Sig.RCIs

%RCIs n

No. Sig.RCIs

%RCIs n

No. Sig.RCIs

%RCIs n

No. Sig.RCIs

%RCIs

Posttraumatic stressb 66 38 58 61 20 33 66 3 5 61 6 10Depressionc 65 15 23 60 8 13 65 6 9 60 10 17Traumatic griefd 40 16 40 24 0 0 40 0 0 24 2 8Existential griefe 40 13 33 26 2 8 40 0 0 26 4 15

aFollowing common practice, RCI values were calculated using Cronbach ! and pooled pretreatment SDs for that variable.bUCLA Reaction Index-Revised.cDepression Self-Rating Scale.dUCLA Grief Index, Traumatic Grief subscale.eUCLA Grief Index, Existential Grief subscale.

TABLE 4Odds Ratios for Reliable Improvement and Reliable Deterioration in the Treatment Condition Relative to the Comparison Condition

Outcome Measure

Odds Ratio for Reliable Improvementa Odds Ratio for Reliable Deteriorationb

Pre-to Posttreatment 95% CI Pre-to Posttreatment 95% CI

Posttraumatic stressc 1.76 1.16Y2.66 0.46 0.12Y1.77Depressiond 1.73 0.79Y3.79 0.55 0.21Y1.43Traumatic griefe f f g g

Existential griefh 4.23 1.04Y17.21i g g

Pre-to 4-Mo Follow-up 95% CI Pre-to 4-Mo Follow-up 95% CIPosttraumatic stressc 1.67 1.11Y2.51 g g

Depressiond 1.31 0.82Y2.08 g g

Note: CI = confidence interval.aOdds ratio calculated as [(% reliable improvers in treatment group)/(% reliable improvers in comparison group)].bOdds ratio calculated as [(% reliable deteriorators in treatment group)/(% reliable deteriorators in comparison group)].cUCLA Reaction Index-Revised total scale score.dDepression Self-Rating Scale.eUCLA Grief Index, Traumatic Grief subscale.fOdds ratio not calculated because comparison group had no reliably improved cases.gOdds ratio not calculated because treatment group had no reliably deteriorated cases.hUCLA Grief Index, Existential Grief subscale.iBroad width of confidence interval is likely attributable to small sample sizes in both groups.

LAYNE ET AL.



improvement were significantly greater (p < .05) in thetreatment condition for all of the outcome variables withthe exception of depression, which was in the expecteddirection (OR 1.73).

4-Month Follow-up Analyses

Given the limited scope of the 4-month follow-up,preliminary analyses were conducted to investigatepotential nonrandom attrition at the follow-up amongstudents who had completed the initial pre- toposttreatment evaluation (Fig. 1). A group (treatmentversus comparison) by completer status (follow-upcompleter versus follow-up noncompleter) MANOVAwas computed in which pre- and posttreatment RIscores served as dependent variables. A significant Wilks4 was found for completer status (F2,121 = 11.03, p <.05). Post hoc analyses revealed that the meanpretreatment RI score of follow-up completers (X

–=

39.61, SD 10.63) was significantly higher than that ofnoncompleters (X

–= 30.03, SD 12.68; p < .05). No

group � completer status interaction was found for theRI (p 9.05). A similar MANOVA was computed inwhich pre- and posttreatment DSRS scores served asdependent variables. A significant Wilks 4 was found forcompleter status (F2,120 = 6.45; p < .05). Post hocanalyses revealed that the mean DSRS score of follow-upcompleters was significantly higher than that ofnoncompleters at both pretreatment (X

–= 33.90, SD

10.79 and X–= 27.43, SD 9.93, respectively) and

posttreatment (X–= 31.42, SD 6.48 and X

–= 28.98, SD

5.53, respectively; p < .05). No group � completerstatus interaction was found for the DSRS (p 9 .05).Subsequent analyses were restricted to students who hadcompleted the pre- to posttreatment evaluation and the4-month follow-up.

Maintenance of Treatment Gains: Posttreatment to 4-Month

Follow-up PTSD and Depression Symptom Scores

A MANOVA was first computed on RI scores withtime (posttreatment versus 4-month follow-up) as arepeated measure and group (treatment versus compari-son) as a between-subjects fixed factor. The test forgroup x time interaction did not reach significance (p 9.05), nor did the test of main effect for group (p 9 .05).Consistent with hypothesis 4, a significant main effectfor time was found (F1,61 = 4.82; p < .05), indicatingthat the marginal mean (average posttraumatic stresssymptom score pooled across groups) decreased to

a modest but significant degree between posttreatment(X–= 25.60, SD 13.03) and 4-month follow-up (X

–=

24.12, SD 11.57). A similar MANOVA conducted onDSRS scores found a nonsignificant group � timeinteraction (p 9 .05); the test of a main effect for groupwas also nonsignificant (p 9 .05). Consistent with hy-pothesis 4, a significant main effect for time was found(F1,62 = 75.40; p < .01), indicating that the marginalmean (average depression symptom score pooledacross groups) decreased to a significant and substantialdegree between posttreatment (X

–= 28.66, SD 5.84)

and 4-month follow-up (X–= 20.61, SD 11.98).

Total Pretreatment to 4-Month Follow-up Improvement:

Mean Difference Scores, RCI Values, and ORs

To investigate potential longer term impacts ofprogram participation, analyses were conducted examin-ing pretreatment to 4-month follow-up treatment out-comes (Table 5). Consistent with hypothesis 4, a test ofthe mean difference scores on the RI reached significancein both the treatment and comparison conditions (both pvalues <.001), indicating that in each group the averageposttraumatic stress symptom score decreased signifi-cantly between pretreatment and 4-month follow-up.Also consistent with hypothesis 4, a test of the meandifference scores on the DSRS reached significance inboth the treatment condition (p < .001) and compari-son condition (p < .01), indicating that in each groupthe average depression symptom score decreasedsignificantly between pretreatment and follow-up.Individual case-based analysis also provided evidence in

support of hypotheses 3 and 4. Table 6 presents RCIvalues for reliably improved and reliably deteriorated casesas calculated using pretreatment to 4-month follow-upscores. Consistent with hypothesis 4, percentages ofreliable improvers ranged from 61% of cases fordepression symptoms to 81% of cases for PTSDsymptoms in the treatment condition and from 47% ofcases for depression symptoms to 48% for PTSDsymptoms in the comparison condition. No cases ofreliable deterioration were found in the treatmentcondition, whereas 7% were found in the comparisoncondition for both the posttraumatic stress and depressionsymptom measures. The lower part of Table 4 presentsOR values for reliable change between pretreatment and4-month follow up. Consistent with hypothesis 3, theORof reliable improvement was significantly greater (p < .05)in favor of the treatment condition for posttraumatic




stress symptoms; however, the OR of reliable improve-ment in depression symptoms did not reach significancebut was in the expected direction (OR 1.31). OR valuesfor reliable deterioration were also in the expecteddirection, but neither could be calculated because thetreatment group had no reliably deteriorated cases or didnot reach significance (p 9 .05).

DISCUSSION

This study compared the effectiveness of the first twotiers of a three-tiered school- and community-basedmental health intervention for war-exposed adolescents,as delivered on a large scale in a war-ravaged nation bylocally trained and supervised school counselors.5 Thesuccessful implementation of both the tier 1 comparison

condition (composed of classroom-based psychoeduca-tion and skills) and the tier 2 treatment condition(composed of TGCT plus the tier 1 classroom-basedintervention) levels of intervention in a postwarenvironment marked by widespread unemployment,political instability, inadequate physical facilities, andsupply shortages support the feasibility of implementingmultitiered interventions in resource-poor regions.Underscoring program sustainability, the Federal Min-istry incorporated implementing the program into thecounselors` job requirements and contracted localmental health professionals to provide supervision.The resulting professional network interlinked partici-pating schools with local mental health clinics and thelocal university and allowed the program to runautonomously throughout the school year.

TABLE 6Reliable Change Indices (RCIs) for Pretreatment to 4-Month Follow-up

Outcome Variable

Significantly Improved Cases Significantly Deteriorated Cases

Treatment Condition Comparison Condition Treatment Condition Comparison Condition

nNo. Sig.RCIs

%RCIs n

No. Sig.RCIs

%RCIs n

No. Sig.RCIs

%RCIs n

No. Sig.RCIs

%RCIs

Posttraumatic stressa 36 29 81 29 14 48 36 0 0 29 2 7Depressionb 36 22 61 30 14 47 36 0 0 30 2 7

aUCLA Reaction Index-Revised total scale score.bDepression Self-Rating Scale.

TABLE 5Means, Pretreatment to 4-Month Follow-up Treatment Difference Scores, and Confidence Intervals of Posttraumatic Stress and

Depression Scores

Group

Posttraumatic Stressa Depressionb


Follow-upTreatmentMean(SD)

MeanDifferenceScore (SD)c

95% CI forDifference Score


Follow-upTreatmentMean(SD)

MeanDifferenceScore (SD)c

95% CI forDifference Score


Treatment 42.09(10.56),n = 36

23.54(10.64),n = 36

j18.54***(9.65),n = 36

j21.81 j15.28 37.49(10.28),n = 36

22.14(11.63),n = 36

j15.35***(11.28),n = 36

j19.17 j11.53

Comparison 36.35(9.50),n = 29

24.67(9.65),n = 29

j11.68***(14.12),n = 29

j17.05 j6.31 29.71(9.60),n = 30

21.33(13.89),n = 30

j8.38**(12.32),n = 30

j12.98 j3.78

Note: CI = confidence interval.aUCLA Reaction Index-Revised, total scale score.bDepression Self-Rating scale.cPretreatment to 4-month follow-up difference score.*p < .05 **p < .01 ***p < .001.

LAYNE ET AL.



Mean-based analyses (MANOVA andmean differencescores) of the effectiveness of both tiers of interventionproduced generally favorable results, with significant pre-to posttreatment reductions in PTSD and depressionsymptoms in both the treatment and comparison con-ditions and significant reductions in maladaptive griefreactions in the treatment condition. Moreover, signifi-cant symptom improvement between posttreatment and4-month follow-up was observed in both treatment andcomparison conditions on measures of PTSD anddepression symptoms. Given the generally slow rate ofpostwar recovery in the region during this period, it isunlikely that the reductions in self-reported symptomsobserved in both conditions are attributable to signif-icant changes in social, economic, or political factors.

Individual case-based analyses using the RCI revealedmore differentiated, although still generally favorable,effects between the two experimental conditions. Theseanalyses showed rates of significant improvement inPTSD symptoms in the treatment condition thatcompare favorably to general improvement rates inrandomized controlled trials of psychotherapies withadults as conducted under carefully controlled experi-mental settings (reliable change rates of which varybetween 57.6% and 67.2%).36 Findings that the meanPTSD symptom score decreased from the thresholdsevere range to the mild range on the UCLA RI32 in thetreatment condition sheds light on the clinical sig-nificance of these symptom reductions. Althoughsmaller, the rates of significant improvement for thegrief and depression subscales in the treatment condi-tion compare favorably to rates of general improvementproduced in community-based treatment settings withadults (reliable change rates of which range from 29.1%to 44.4%).37 These results underscore the effectivenessof group treatment for severely traumatized andtraumatically bereaved adolescents that addresses post-traumatic stress, depression, and grief reactions,22Y28

and contribute to the literature on trauma-focusedgroup-based treatments for youths.38 The results alsosuggest that TCGT is safe to implement with thepersonnel, training, supervision, and referral system usedand provides significant benefit with low iatrogenic risk.

Underscoring the potential effectiveness of tier 1(classroom-based) interventions, RCI analyses revealedcomparatively lower but substantial rates of reduction inPTSD symptoms (33% at posttreatment, 48% atfollow-up) and depression symptoms (13% at posttreat-

ment, 47% at follow-up) in the comparison condition.These findings are generally comparable to ratesobtained in community-based treatment settings withadults.37 In addition, results of a qualitative programevaluation conducted at posttreatment supported thefeasibility, impact, and sustainability of mounting a tier1 intervention in a postwar setting.25 The qualitativeevaluation found evidence of high program penetrationin participating schools. Counselors regularly usedpsychoeducational and coping skills components ofTGCT and disseminated them in a variety of forums,including classroom presentations, posters, staff meet-ings, and parent/teacher conferences. Student groupmembers also reported sharing program skills with theirclassmates, friends, and families. Notably, the quanti-tative program evaluation protocol required at least fourcounselor contacts with students in the comparisoncondition to monitor their status and potential need forspecialized services. Given evidence of substantialimprovement in the comparison condition, these resultssuggest that school-wide classroom-based programs thatprovide psychoeducational and skills-related informa-tion paired with supportive contacts with at-riskstudents may be adequate to substantially reduce distressin many war-exposed youths with modest risk foriatrogenic outcomes. Concordant evidence is furnishedby a program that provided these core recovery skills(without specialized trauma or grief processing) tomoderately distressed youths exposed to the September11th terrorist attacks.20

Taken together, the favorable results obtained in boththe treatment and comparison conditions suggest that amultitiered mental health intervention may be aneffective and efficient method for delivering beneficial,low-risk services to war-exposed youths living inresource-poor regions. These tiers are specificallycomposed of tier 1, the broad dissemination ofpsychoeducational and skills-based components drawnfrom TGCT modules I and IV; tier 2, specializedmental health interventions (consisting of TGCTmodules I, II, and IV as needed for severely traumatizedyouths with severe PTSD symptoms, paired withmodule III for traumatically bereaved youths experienc-ing maladaptive grief reactions; see Figure 2 onlinethrough the Article Plus feature on the Journal`s Website at www.jaacap.com); and tier 3, referral of youthsat acute risk for community-based mental healthservices.26




Moreover, these results point to the feasibility andutility of instituting multiple safeguards to reduceiatrogenic risks. Safeguards included using both a riskscreening survey and screening interview to identifyappropriate cases,24,27 the creation of a professionalsupport network that facilitated referrals of cases at acuterisk to community-based agencies, training in grouptherapy skills,28 the use of group co-leader teams whereresources permitted, and a pregroup interview thathelped incoming members to select traumatic experi-ences appropriate for group work.22 Given thesesafeguards and the relatively low percentages of reliabledeteriorators in both tiers of intervention, conservativeprocedures may be appropriate in selecting adolescentsmost likely to benefit from tier 2 interventions.Specifically, risk identification and triage may give thehighest priority to traumatically bereaved youths whoare experiencing concurrent severe levels of PTSD andmaladaptive grief reactions.39 Given their incrementalbenefit compared to tier 1 interventions, tier 2interventions may, as resources allow, be provided toyouths experiencing intense distress (particularly PTSDand depression symptoms) and marked functionalimpairment without concurrent maladaptive grief.This evaluation of a school-based program designed

to assist youths in greatest need, in a naturalistic setting,using service providers who form part of youths` naturalecologies, is consistent with the tenets of Beffectiveness^research.40Y42 The primary aims of effectiveness researchinclude evaluating outcomes under real-world practiceconditions, with real-world clients to enhance thegeneralizability of results to actual clinical practice.Few intervention studies with war-traumatized youthsare reported in the literature; those published havegenerally been conducted with refugee or immigrantpopulations outside their countries of origin.5

The present study had several limitations. First, thetier 2 intervention also included exposure to the tier 1(classroom-based) arm of the intervention. The expo-sure of group members to TGCT treatment compo-nents in both group and classroom settings makes itchallenging to contrast the two active treatmentconditions and their respective effects. More informa-tion relating to how specific components of TGCT weredisseminated in tier 1 activities would have been helpfulin determining the specific components and depth ofcoverage that may be necessary and sufficient to producetherapeutic benefit. Second, the study design focused

primarily on one of the three aims of TGCT: thereduction of distress. Thus, particularly given previousfindings that symptom reductions are associated withimproved school-related behavior and attitudes,25 theinclusion of measures of functioning (e.g., schoolperformance), as well as markers of developmentalprogression (e.g., graduating to the next grade level),would have captured a broader range of potentialoutcomes of TGCT. Third, the significantly higherpretreatment scores on the grief variables observed in thetreatment condition, despite randomization, in combi-nation with the relatively small number of students inboth groups who completed the grief measures, under-scores the need for conservative interpretation ofanalyses involving the grief variables and for studyreplication. Fourth, the pre-/posttreatment design didnot clarify the sequencing of changes in variables overtime, limiting the ability to observe the course oftreatment response and the identification of mediatorsof improvement. Fifth, the follow-up assessment waslimited in scope (retaining roughly half of the studentswho completed posttreatment measures), breadth(measuring only PTSD and depression symptoms),and duration (4 months). These observations point tothe need for conservative interpretation of the follow-upanalyses. (Nevertheless, analyses suggested that attritionrates were not systematically related to experimentalcondition, but rather that more distressed students had ahigher likelihood of completing the follow-up.) More-over, the significant improvements in PTSD anddepression symptoms observed in both conditions at4-month follow-up suggest that ongoing improvementmay persist well past termination and underscore theneed for longer term follow-up.Future studies would benefit from the inclusion of

measures from multiple informants, as well as measuresthat tap functioning in a broader range of domains (e.g.,risky behavior, peer and family relationships, academicperformance, school behavior).43 Including measures ofdistress, functioning, and developmental progressionwill not only better capture the three primary aims ofTGCT but it is also consistent with recent calls in thechild and adolescent psychotherapy literature to evaluateboth the statistical and clinical significance of treatmentoutcomes; that is, whether treatment leads to mean-ingful improvements in patient`s functioning andquality of life.35Y37,44 In addition, future studies mayinclude follow-ups that make use of intent-to-treat

LAYNE ET AL.



analysis (including subjects who either drop out or areexcluded due to high risk and referred to tier 3interventions) and that span time frames of sufficientduration to capture developmental impact. Accordingly,longer follow-up periods (see, for example, the 5-yearfollow-up study of Goenjian et al.21) containing multipledata collection intervals are needed to better understand thecausal pathways leading to specific treatment outcomes,44

improved functioning, and amelioration of disturbances indevelopmental trajectories.16 Studies are also needed thatseek to identify predictors45 and moderators of treatmentresponse (e.g., factors linked to reliably improved versusreliably deteriorated versus treatment nonresponder out-comes)44 and mediators and other mechanisms oftherapeutic change35 and that evaluate the cost-effective-ness of specific tiers of intervention at the individual caselevel.46 Last, future treatment outcome studies mayprofitably incorporate recent advances in the family-basedtreatment of traumatized youths because this modalityshows promise for increased generalization and mainte-nance of improvement over time.47

The term lost generationwas originally used in referenceto young adults whose lives weremarred in enduring waysby World War I,48 a war precipitated in Sarajevo nearly acentury ago. We hope that field-tested, culturallysensitive, developmentally appropriate interventions willreduce the risk for losing present and future generations.Adolescence is a critical period of preparation for the rolesof adulthood that is susceptible to significant develop-mental disruption in the face of trauma, loss, and severeadversity. Protecting and promoting adolescent develop-ment and the transition to young adulthood is of vitalimportance to societies striving to recover from majorupheavals, for it is through upcoming generations who aresufficiently recovered and prepared to engage in produc-tive careers, stable marriages, successful child-rearing,prosocial activities, and other roles and responsibilities ofmature citizenship that healthy societies are best rebuiltand sustained.

Disclosure: The authors report no conflicts of interest.

REFERENCES

1. Wexler ID, Branski D, Kerem E. War and children. JAMA.2006;296:579Y581.

2. Bellamy C. The State of the World`s Children 2005. New York: UnitedNations Children`s Fund; 2004.

3. Allwood MA, Bell-Dolan D, Husain SA. Children`s trauma and

adjustment reactions to violent and nonviolent war experiences. J AmAcad Child Adolesc Psychiatry. 2002;41:450Y457.

4. Barath A. Psychological status of Sarajevo children after the war:1999Y2000 survey. Croat Med J. 2002;43:213Y220.

5. Ehntholt KA, Yule W. Practitioner review: assessment and treatment ofrefugee children and adolescents who have experienced war-relatedtrauma. J Child Psychol Psychiatry. 2006;47:1197Y1210.

6. Fazel M, Wheeler J, Danesh J. Prevalence of serious mental disorder in7000 refugees resettled in western countries: a systematic review. Lancet.2005;365:1309Y1314.

7. Macksoud M, Aber J. The war experience and psychosocial developmentof children in Lebanon. Child Dev. 1996;67:70Y88.

8. Nader K, Pynoos R, Fairbanks L, Al-Ajeel M, Al-Asfour A. A preliminarystudy of PTSD and grief among the children of Kuwait following theGulf crisis. Br J Clin Psychol. 1993;32:407Y416.

9. Shaw JA. Children exposed to war/terrorism. Clin Child Fam Psychol.2003;6:237Y246.

10. Smith P, Perrin S, Yule W, Hacam B, Stuvland R. War exposure amongchildren from Bosnia-Hercegovina: psychological adjustment in acommunity sample. J Trauma Stress. 2002;15:147Y156.

11. Goldstein RD, Wamler NS, Wise PH. War experiences and distresssymptoms of Bosnian children. Pediatrics. 1997;100:873Y878.

12. Husain SA, Nair J, Holcomb W, Reid JC, Vargas V, Nair SS. Stressreactions of children and adolescents in war and siege conditions. Am JPsychiatry. 1998;155:1718Y1719.

13. Saltzman WR, Pynoos RS, Layne CM. Survey of Mental Health Programsfor Youth in Post-War Bosnia-Herzegovina. Sarajevo, Bosnia: UNICEFBosnia and Herzegovina; 1996.

14. Layne CM, Saltzman WR, Steinberg AM, Pynoos RS. Trauma and GriefComponent Therapy for Adolescents: Group Treatment Manual. Sarajevo,Bosnia: UNICEF Bosnia and Herzegovina; 2000.

15. Pynoos RS, Steinberg AM, Wraith R. A developmental model ofchildhood traumatic stress. In: Cicchetti D, Cohen D, eds. Manual ofDevelopmental Psychopathology. Vol 2, Risk, Disorder, and Adaptation.New York: John Wiley; 1995:72Y95.

16. Layne CM, Beck CJ, Rimmasch H, Southwick JS, Moreno MA,Hobfoll SE. Promoting Bresilient^ posttraumatic adjustment in child-hood and beyond: Bunpacking^ life events, adjustment trajectories,resources, and interventions. In: Brom D, Pat-Horenczyk R, Ford J, eds.Treating Traumatized Children: Risk, Resilience, and Recovery. New York:Routledge; 2009, pp. 13Y47.

17. Layne CM, Warren JS, Hilton S, et al. Measuring adolescents` supportnetworks within war-related contexts: development of the multi-sectorsocial support inventory. In: Barber BK, ed. Adolescents and Violence.New York: Oxford University Press; 2007, pp. 145Y176.

18. Hobfoll SE, Watson P, Bell CC, et al. Five essential elements ofimmediate and mid-term mass trauma intervention: empirical evidence.Psychiatry. 2007;70:283Y315.

19. Saxe GN, MacDonald HZ, Ellis BH. Psychosocial approaches forchildren with PTSD. In: Friedman MJ, Keane TM, Resick PA, eds.Handbook of PTSD: Science and Practice. New York: Guilford Press;2007:359Y375.

20. CATS Consortium. Implementing CBT for traumatized children andadolescents after September 11th: lessons learned from the Child andAdolescent Trauma Treatments and Services (CATS) Project. J ClinChild Adolescent Psychol. 2007;36:581Y592.

21. Goenjian AK, Walling D, Steinberg AM, et al. A prospective study ofposttraumatic stress and depressive reactions among treated anduntreated adolescents 5 years after a catastrophic disaster. Am J Psychiatry.2005;162:2302Y2308.

22. Layne CM, Pynoos RS, Cardenas J. Wounded adolescence: school-basedgroup psychotherapy for adolescents who sustained or witnessed violentinjury. In: Shafii M, Shafii S, eds. School Violence: Assessment, Manage-ment and Prevention. Washington, DC: American Psychiatric Publishing;2001:163Y186.

23. Saltzman WR, Layne CM, Steinberg AM, Pynoos RS. Trauma/grief-focused group psychotherapy with adolescents. In: Schein LA, Spitz HI,Burlingame GM, Muskin PR , eds. Group Approaches for the PsychologicalEffects of Terrorist Disasters. New York: Haworth; 2006:669Y730.




24. Saltzman WR, Pynoos RS, Layne CM, Steinberg A, Aisenberg E.Trauma/grief-focused intervention for adolescents exposed to commu-nity violence: results of a school-based screening and group treatmentprotocol. Group Dyn Theor Res. 2001;5:291Y303.

25. Layne CM, Pynoos RS, Saltzman WR, et al. Trauma/grief-focused grouppsychotherapy: school-based post-war intervention with traumatizedBosnian adolescents. Group Dyn Theor Res. 2001;5:277Y290.

26. Saltzman WR, Layne CM, Steinberg AM, Arslanagic B, Pynoos RS.Developing a culturally-ecologically sound intervention program foryouth exposed to war and terrorism. Child Adolesc Psychiatr Clin N Am.2003;12:319Y342.

27. Saltzman WR, Layne CM, Steinberg AM, Pynoos RS. School-basedtrauma- and grief-focused intervention for adolescents exposed tocommunity violence. Prev Res. 2003;10:8Y11.

28. Davies DR, Burlingame GM, Layne CM. Integrating small group processprinciples into trauma-focused group psychotherapy: what should agroup trauma therapist know? In: Schein LA, Spitz HI, Burlingame GM,Muskin PR, eds. Group Approaches for the Psychological Effects of TerroristDisasters. New York: Haworth; 2006:385Y424.

29. Akos P, Hamm JV, Mack SG, Dunaway M. Utilizing the developmentalinfluence of peers in middle school groups. J Spec Group Work. 2007;32:51Y60.

30. Cox J, Davies DR, Burlingame GM, Campbell JE, Layne CM,Katzenbach RJ. Effectiveness of a trauma/grief-focused group interven-tion: a qualitative study with war-exposed Bosnian adolescents. Int JGroup Psychother. 2007;53:319Y345.

31. Lansen J, Haans T. Clinical supervision for trauma therapists. In:Wilson JP, Drozdek B, eds. Broken Spirits: The Treatment of TraumatizedAsylum Seekers, Refugees, War and Torture Victims. New York: Brunner-Routledge; 2004:317Y353.

32. Steinberg AM, Brymer M, Decker K, Pynoos RS. The UCLA PTSDreaction index. Curr Psychiatr Rep. 2004;6:96Y100.

33. Birelson P. The validity of depressive disorder in childhood anddevelopment of a self-rating scale: a research project. J Child PsycholPsychiatry. 1981;22:73Y88.

34. Tingey RC, Lambert MJ, Burlingame GM, Hansen NB. Assessingclinical significance: proposed extensions to method. Psychother Res.1996;6:109Y123.

35. Lunnen C, Ogles BM. A multi-perspective, multi-variable evaluation ofreliable change. J Consult Clin Psych. 1998;66:400Y410.

36. Hansen NB, Lambert MJ, Forman EM. The psychotherapy dose-response effect and its implications for treatment delivery services. ClinPsychol Sci Pract. 2002;9:329Y343.

37. Lutz W, Lambert MJ, Harmon SC, Tschitsaz A, Schurch E, Stulz N. Theprobability of treatment success, failure and durationVwhat can belearned from empirical data to support decision making in clinicalpractice? Clin Psychol Psychotherapy. 2006;13:223Y232.

38. Stein BD, Jaycox LH, Kataoka SH, et al. A mental health intervention forschoolchildren exposed to violence. JAMA. 2003;290:603Y611.

39. Pfeffer CR, Altemus M, Heo M, Jiang H. Salivary cortisol andpsychopathology in children bereaved by the September 11, 2001 terrorattacks. Biol Psychiatry. 2007;61:957Y965.

40. Glasgow RE, Lichtenstein E, Marcus AC. Why don`t we see moretranslation of health promotion research to practice? Rethinking theefficacy-to-effectiveness transition. Am J Public Health. 2003;93:1261Y1267.

41. Weisz JR, Weiss B, Donenberg GR. The lab versus the clinic: effects ofchild and adolescent psychotherapy. Am Psychol. 1992;47:1578Y1585.

42. Wells KB. Treatment research at the crossroads: the scientific interfaceof clinical trials and effectiveness research. Am J Psychiatry. 1999;156:5Y10.

43. Weisz JR, Kazdin AE. Concluding thoughts: present and future ofevidence-based psychotherapies for children and adolescents. In: KazdinAE, Weisz JR, eds. Evidence-Based Psychotherapies for Children andAdolescents. New York: Guilford Press; 2003:439Y451.

44. Kazdin AE. The meanings and measurement of clinical significance.J Consult Clin Psychol. 1999;3:332Y339.

45. Burlingame G, Strauss B, Joyce A, MacKenzie K, Ogrodniczuk J, TaylorS. CORE Battery: A Revision and Update. New York: American GroupPsychotherapy Association; 2006.

46. Hoch JS, Smith MW. A guide to economic evaluation: methods for cost-effectiveness analysis of person-level data. J Traum Stress. 2006;19:787Y797.

47. Saltzman WR, Babayan T, Lester P, Beardslee W, Pynoos RS. Familytreatment for child traumatic stress: a review and report on recentinnovations. In: Brom D, Pat-Horenczyk R, Ford J, eds. TreatingTraumatized Children: Risk, Resilience, and Recovery. New York:Routledge; In press.

48. Hemingway E. The Sun Also Rises. New York: Charles Scribner`s Sons;1926.

LAYNE ET AL.


Effectiveness of a school-based group psychotherapy program for war-exposed adolescents: a randomized controlled trial

Documents