Posttraumatic Stress Disorder and Trauma in Youth in Juvenile Detention

Objective  To determine prevalence estimates of exposure to trauma and 12-month rates of posttraumatic stress disorder (PTSD) among juvenile detainees by demographic subgroups (sex, race/ethnicity, and age).

Design  Epidemiologic study of juvenile detainees. Master's level clinical research interviewers administered the PTSD module of the Diagnostic Interview Schedule for Children, version IV (DISC-IV), to randomly selected detainees.

Setting  A large, temporary detention center for juveniles in Cook County, Illinois (which includes Chicago and surrounding suburbs).

Participants  Randomly selected, stratified sample of 898 African American, non-Hispanic white, and Hispanic youth (532 males, 366 females, aged 10-18 years) arrested and newly detained.

Main Outcome Measures  Diagnostic Interview Schedule for Children, version IV.

Results  Most participants (92.5%) had experienced 1 or more traumas (mean, 14.6 incidents; median, 6 incidents). Significantly more males (93.2%) than females (84.0%) reported at least 1 traumatic experience; 11.2% of the sample met criteria for PTSD in the past year. More than half of the participants with PTSD reported witnessing violence as the precipitating trauma.

Conclusion  Trauma and PTSD seem to be more prevalent among juvenile detainees than in community samples. We recommend directions for research and discuss implications for mental health policy.

Each year, approximately 2.4 million youth are arrested, accountingfor 17% of all arrests.¹ On a typical day,approximately 109 000 youth are detained.² Thenumber of youth in the juvenile justice system with psychiatric disordersis a major public health problem. Two thirds of males and three quarters offemales in juvenile detention have 1 or more psychiatric disorders.^3,4

The related literature suggests that posttraumatic stress disorder (PTSD)is more common in youth in the juvenile justice system than in community samples.^4-15 Lifetimediagnoses of PTSD in community samples of youth range from 6.3%¹⁶ to7.8%¹⁰; current diagnoses are 3.5%.¹⁷ Prevalence of PTSD among youth in the juvenile justicesystem varies considerably, depending on the type of sample, the measure used,and the time frame assessed (within the past year, within the past month,or at the time of the interview).^{4,13-15,18-20} Forexample, rates among males are 2.3% among American Indian detainees (pastyear)¹⁹; 4.8% among youth in secure placement(past month)⁴; 24.2% among felons in securecustody (at the time of the interview)¹⁴; and32.3% among incarcerated youth (at the time of the interview).¹³ Farfewer data are available on females in the juvenile justice system. Ducloset al¹⁹ found 0 cases of PTSD within the pastyear among 64 female American Indian juvenile detainees. In contrast, Cauffmanet al¹⁵ found 47 (48.9%) of 96 incarceratedfemales met criteria for PTSD in the past 3 months.

Unfortunately, most of these studies are too small,^13-15,19 toounique,¹⁹ or lack sufficiently standardizeddiagnostic assessments¹⁴ to generate reliableestimates. To date, no large-scale study has examined the prevalence of traumaand PTSD across demographic subgroups that make up increasing proportionsof the juvenile justice population, that is, African Americans, Hispanics,females, and younger children. This omission is critical. Posttraumatic stressdisorder is associated with severe functional impairment¹⁶ andother psychiatric problems.^8,10,21 Leftuntreated, PTSD may become chronic,^8,10,22,23 withenormous personal and societal costs.²⁴

In this article we present the prevalence of PTSD and trauma among juveniledetainees. This study has 2 advantages: (1) a stratified, random sample, largeenough to compare sex, racial/ethnic, and age groups; and (2) a standardizedmeasure of PTSD, the Diagnostic Interview Schedule for Children, version IV(DISC-IV).

Participants were part of the Northwestern Juvenile Project, a longitudinalstudy of 1829 youth (aged 10-18 years) arrested and detained between November20, 1995, and June 14, 1998, at the Cook County (Illinois) Juvenile TemporaryDetention Center (CCJTDC) in Chicago. The random sample was stratified bysex, race/ethnicity (African American, non-Hispanic white, Hispanic), age(10-13 years or ≥14 years), and legal status (processed as a juvenile oras an adult) to obtain enough participants to examine key subgroups (eg, females,Hispanics, and younger children).

The CCJTDC received approximately 8500 admissions each year during thetime data were collected (John Howard Association, unpublished data, 1992).The CCJTDC is used solely for pretrial detention and for offenders sentencedfor fewer than 30 days. All detainees younger than 17 years are held at theCCJTDC, including youth processed as adults (automatic transfers to adultcourt). Youth up to the age of 21 years may be detained in the CCJTDC if theyare being prosecuted for an arrest that occurred when they were younger than17 years.

Like juvenile detainees nationwide,²⁵ approximately90% of the CCJTDC detainees are male; most are racial/ethnic minorities. Thepopulation of the CCJTDC is 77.9% African American, 5.6% non-Hispanic white,16.0% Hispanic, and 0.5% other racial/ethnic groups. The age and offense distributionsof the CCJTDC detainees are also similar to detained juveniles nationwide.²⁵

We chose the detention center in Cook County (which includes Chicagoand surrounding suburbs) for 3 reasons. First, nationwide, most juvenile detaineeslive in and are detained in urban areas.²⁶ Second,Cook County is ethnically diverse and has the third largest Hispanic populationin the United States.²⁷ Studying Hispanicsis important because they are the largest minority group in the United States²⁸ and they are overrepresented in the justice systems.²⁵ Third, the detention center's size (daily censusof approximately 650 youth and intake of 20 youth per day) ensured that enoughparticipants would be available.

No single site can represent the entire country because jurisdictionsmay have different options for diversion.^29,30 Nevertheless,Illinois' criteria for detaining juveniles are similar to other states'.²⁹ All states allow pretrial detention if the juvenileneeds protection, is likely to flee, or is considered a danger to the community.^29,30

Detainees were eligible to be sampled, regardless of their psychiatricmorbidity, state of drug or alcohol intoxication, or fitness to stand trial.Within each stratum, we used a random numbers table to select names from theCCJTDC's intake log. Throughout the study, we tracked how many participantswere still needed to fill each cell. Project staff sampled the rarest categoriesfirst. The final sampling fractions ranged from 0.018 to 0.689. (Additionalinformation on the sample is available from us.)

Studying detained youth requires special procedures because they areminors, because they are detained, and because many do not have a parent orguardian who can provide appropriate consent.³¹ Projectstaff approached participants on their units, explained the project, and assuredthem that anything they told us (except acute suicidal or homicidal risk)would be confidential. Participants signed either an assent form (if theywere <18 years) or a consent form (if they were ≥18 years). Federalregulations allow parental consent to be waived if the research involves minimalrisk (45 CFR §46.116[c], 45 CFR §46.116[d], and 45 CFR §46.408[c]).^31,32 The Northwestern University InstitutionalReview Board, the Centers for Disease Control and Prevention InstitutionalReview Board, and the US Office of Protection from Research Risks waived parentalconsent. However, as ethicists recommend, we nevertheless tried to contactparents to provide them an opportunity to decline participation and to offerthem additional information (45 CFR §46.116[d][4]).^33,34 Despiterepeated attempts to contact the parent or guardian, for 43.8% of the participants,none could be found. In lieu of parental consent, youth assent was overseenby an independent participant advocate representing the interests of the participants.Federal regulations allow for a participant advocate if parental consent isnot feasible (45 CFR §46.116[d]).³³

We began collecting data on PTSD 13 months after the larger study began.Of the 1148 names selected, 34 detainees (3.0%) refused to participate inthe study. There were no significant differences in refusal rates by sex,race/ethnicity, or age. Two youth were released before finishing the interview;189 youth left the CCJTDC while we were locating their caretakers for consentor before we could schedule an interview; 25 youth were released after consentwas obtained but before the interview commenced. The final sample size was898 and comprised 532 males (59.2%) and 366 females (40.8%); 490 African Americans(54.6%), 154 non-Hispanic whites (17.1%), 252 Hispanic (28.1%), and 2 others(0.2%). Participants ranged in age from 10 to 18 years; the mean was 14.8years; the median was 15 years.

Participants were interviewed in a private area, almost always within2 days of intake. Most interviews lasted 2 to 3 hours, depending on how manysymptoms were reported. We used both male and female interviewers. Femaleparticipants were always interviewed by female interviewers. Interviewerswere trained for at least 1 month; most had a master's degree in psychologyor an associated field and experience interviewing high-risk youth. One thirdof our interviewers were fluent in Spanish. We maintained consistency throughoutthe study by monitoring scripted interviews with mock participants. Additionalinformation on our methods is published elsewhere.^3,35,36

We used the DISC-IV, based on DSM-IV criteria,to assess PTSD. (Other disorders, presented elsewhere,^3,35 wereassessed using the DISC, version 2.3.) Our data are based on the youth's self-reportbecause it was not feasible to interview caretakers. Like other measures ofPTSD in children,³⁷ there are still insufficientdata on the DISC-IV's reliability and validity, in part, because the PTSDdiagnosis is relatively new.³⁷ Studies documentingthe module's reliability and validity are in progress (P. Fisher, PhD, e-mail,July 11, 2003). Despite the lack of psychometric data on the PTSD module ofthe DISC-IV, we chose it for several reasons. The DISC is the most widelyused diagnostic instrument for child and adolescent research.³⁸ Itis especially useful for large-scale epidemiologic studies because it is relativelybrief; it can be administered by nonclinicians; it is designed to assess youthwho have and have not been traumatized; and it generates DSM-IV disorders using computerized scoring.

The PTSD module assesses whether youth have ever experienced any ofthe 8 traumatic experiences listed in Table1. Participants then identify the event that was "the most difficultfor you in your entire life." The DISC-IV assesses PTSD diagnosis within thepast year for this "worst" trauma.

Because we stratified our sample by sex, race/ethnicity, age, and legalstatus, we weighted all prevalence estimates to reflect the population ofthe CCJTDC. All reported standard errors and inferential tests have been correctedfor design characteristics with Taylor series linearization^40,41 usingthe survey estimation procedures of Stata Statistical Software:Release 8.0.³⁹ Tests of prevalence uselogit models, and tests of means of counts use Poisson regression models.³⁹ We used 2-tailed tests throughout. To reduce theprobability of type I errors, we used layered Bonferroni corrections.⁴² Our level of significance for each group of testswas P<.05.

Table 1 shows that 92.5%of the sample had experienced at least 1 trauma; 84.0% had experienced morethan 1 trauma (not shown); the mean number of traumatic incidents equals 14.6; the median equals 6 incidents (not shown). Significantly more males (93.2%)than females (84.0%) reported a traumatic experience. There were no significantdifferences in overall prevalence of trauma across race/ethnicity for malesand females. Among both male and female detainees, significantly more youth14 years or older (94.2% males and 86.5% females) reported trauma than youthaged 10 to 13 years (82.4% males [F_1,523 = 7.20, P = .008] and 59.1% females [F_1,363 = 14.56, P<.001]); analyses of age are available from us.

Table 1 also shows that,among both males and females, the 3 most frequently reported traumas werehaving "seen or heard someone get hurt very badly or be killed" (reportedby 74.9% of males and 63.5% of females), having been "threatened with a weapon"(reported by 59.3% of males and 47.3% of females), and being in a situationwhere "you thought you or someone close to you was going to be hurt very badlyor die" (reported by 53.5% of males and 49.1% of females). Significantly moremales than females reported having "been in a bad accident." On the otherhand, significantly more females than males reported being "forced to do somethingsexual that you did not want to do." Among males, non-Hispanic whites weremore likely to have "been attacked physically or beaten badly" than were AfricanAmericans. Among females, Hispanics were more likely to have been attackedphysically or beaten badly than were African Americans.

Table 2 reports PTSD diagnosesin the past year for the entire sample (11.2%). There were no significantdifferences in PTSD diagnosis by sex or across race/ethnicity for males andfemales.

We examined precipitating traumas for persons diagnosed as having PTSD.Among male participants "having seen or heard someone get hurt very badlyor be killed" was the most frequent precipitating trauma for PTSD, significantlyhigher among males (58.9%) than females (23.5%) (F_1,98 = 6.46, P = .01). Among female participants, thinking "you or someoneclose to you was going to be hurt very badly or die" was the most frequentprecipitating trauma, significantly higher among females (27.8%) than males(9.5%) (F_1,98 = 6.31, P = .01). (Theseanalyses are available from us.) Other precipitating traumas were too rareto analyze further.

We also examined the age at which the participants had experienced theirworst, precipitating trauma. Most participants (88.7%) reported that theirworst traumas occurred within 2 years prior to the interview. However, beingforced to do something sexual—when that was identified as the worsttrauma—occurred 5 years before the interview for most participants.(These analyses are available from us.)

Exposure to trauma is a fact of life for delinquent youth. More than90% of our sample experienced at least 1 traumatic event; more than half (56.8%)were exposed 6 or more times. These findings are comparable to reports fromsmaller studies of youth in correction facilities.^7,13,15,18

It is difficult to compare our findings to community studies becausepublished findings vary, depending on the sample (eg, urban, suburban, minority)and which traumas were assessed. Yet, our overall prevalence of trauma issubstantially higher than most studies of youth and young adults (ages 15-24years), especially for severe and violent trauma.^{5,6,16,17,43-45}

Witnessing violence, the most common trauma, is far more common in oursample (63.5% of the females and 74.9% of the males) than in most communitystudies of youth and young adults (4.9%-40.1%)^{5,6,10,16,43,46} Ourfindings are most comparable to studies of urban teenagers.^44,45,47-49 Livingwith widespread or chronic community violence in the inner city has been comparedwith living in a war zone.⁴⁷

More than 1 (11.2%) of 10 detainees had PTSD during the year prior tothe interview. These estimates are lower than those reported by Burton etal¹⁴ (24%, current disorder), Cauffman et al¹⁵ (48.9% of females, past 3 months), and Steiner etal¹³ (31.7% of males, current disorder), perhapsbecause our instruments and methods are different. Burton et al¹⁴ useda symptom checklist administered to small groups. Cauffman et al¹⁵ andSteiner et al¹³ used the PTSD module of the Revised Psychiatric Diagnostic Interview.⁵⁰ The Revised Psychiatric Diagnostic Interview assesses symptomsof PTSD independent of a particular trauma. In contrast, the DISC, like mostinstruments, assesses PTSD based on the participant's perceived worst trauma.

The prevalence of PTSD in our sample was higher than reported by Garlandet al²⁰ (3.1%, past year) and Wasserman etal⁴ (4.8%, males only, past month), who usedthe DISC, and Duclos et al¹⁹ (1.3%, past year),who used a modified version of the Composite International Diagnostic Interview(CIDI); the CIDI is similar in structure to the DISC. Our findings and thoseof prior studies may differ because of the point at which the sample was drawn.We sampled youth right after they were detained, and before their adjudicationhearings. Garland et al²⁰ and Wasserman etal⁴ sampled convicted juveniles in secure placement.The findings of Duclos et al¹⁹ may be differentfrom ours because their sample was composed only of American Indian detainees.

The prevalence of PTSD in our sample (during the 12 months prior tothe interview) exceeds lifetime estimates of PTSD reported in community samplesof youth and young adults (3.5%-9.2%).^5,6,10,16 Morethan half of our participants with PTSD had reported witnessing violence asthe precipitant. Our findings might reflect that our participants, like mostjuvenile detainees nationwide, live in urban areas that have high rates ofviolence.^51,52 Alternatively,our findings are consistent with research linking traumatic victimizationin childhood and subsequent psychosocial problems, such as delinquency, perpetrationof violence, and drug use.^{13,16,17,47,48,53-58}

Why were rates of PTSD not higher, given the extent of exposure to traumain our sample? There are several possibilities. First, the types and patternsof traumas reported by youth in community samples and our sample differ. Wedo not know the conditional risk of specific traumas in our sample becausethe DISC-IV, like most instruments, assesses PTSD only for the worst trauma.⁶ Second, traumatic experiences can precipitate otherconditions besides PTSD, for example, disruptive behavior disorders, otherinternalizing disorders, some personality disorders, and physical illnesses.³⁷ These disorders are common in our sample.^3,35 Third, prevalence of PTSD may havebeen higher had we asked about a wider range of traumas, used more than 1screen question to ask about sexual abuse or other types of intimate violence,or conducted interviews using techniques that allow for anonymity (eg, theVoice DISC).^4,59,60

Most demographic differences corroborated prior investigations of communitysamples.^{6,10,16,46,61} Althoughmale detainees were significantly more likely than female detainees to haveexperienced trauma, female detainees were as likely to have PTSD as were maledetainees. In community samples, females are twice as likely as males to developPTSD following exposure to trauma.⁶

Like prior studies in the community, we found few racial/ethnic differencesin rates of trauma or PTSD.^5,10 Thosefew differences pertained to the type of trauma reported most frequently.In our sample, African American males were more likely to have witnessed violencethan were non-Hispanic whites, consistent with the high levels of violenceexposure among inner-city, minority youth.⁶² Non-Hispanicwhite males were more likely to have experienced actual and threatened violencethan other males. Among females, Hispanics were most likely to have experiencedviolent victimization.

Our findings are drawn from 1 site and may pertain only to youth inurban detention centers with similar demographic composition. Moreover, ourfindings are based on a sample of pretrial detainees and may not be generalizableto adjudicated juveniles serving sentences.

Because it was not feasible to interview caretakers, our data are subjectto the reliability and validity of the youths' self-report. However, youthand their caretakers are comparable reporters of youths' anxiety disorders.⁶³ Recall of traumas may be affected by arrest and detention;yet, recall of events by youth may be less subject to the distortions of timethan recall by adults.⁵ Moreover, the DISC—likemost measures—probes for PTSD for the single-worst trauma; hence, weare unable to estimate the age of onset of PTSD or the vulnerability to PTSDby type of trauma.⁶ Despite these limitations,our study has implications for research on PTSD and for mental health policy.

We suggest 3 directions for future research.

1. Studies of vulnerability to PTSDin high-risk youth. Although more than 90% of our sample were exposedto 1 or more traumas, only 11.2% of the sample met criteria for PTSD in thepast year. We need to determine the relative risk of PTSD for types of trauma(eg, witnessing murder, being shot, witnessing ongoing domestic violence,sudden loss of a loved one) among youth who are frequently exposed to traumaand violence, such as our participants. Such studies could document factorsthat increase resilience to PTSD among high-risk youth and guide preventionstrategies.^64,65

2. Studies of chronic communityviolence and its relationship to PTSD. Community violence is epidemicin inner cities.⁶² Research suggests that chronicexposure to violence may have more deleterious effects on children than acuteviolence.⁶² We must study the effects of chroniccommunity violence on high-risk youth as they become adults.¹⁷ Longitudinalstudies should examine the role that witnessing violence plays in perpetuatingthe cycle of violence.

3. Definition of trauma and diagnosisof PTSD. There is a scarcity of research on the validity and reliabilityof diagnostic measures of PTSD, in part because the diagnosis is relativelynew.³⁷ Moreover, the definitions of traumain DSM-IV are somewhat ambiguous⁶⁶;hence, there is little consistency among diagnostic instruments that measuretraumas. For example, most measures assess violent victimization (DISC-IV,CIDI 2.1)⁶⁷; others also assess trauma fromperpetration of violence (R. C. Kessler, PhD, National Comorbidity Study—Replication,unpublished data, 2001-2002). Some measures assess sexual victimization byany perpetrator (DISC-IV, CIDI 2.1); others specifically ask about victimizationby family members (DIS-IV,⁶⁸ National ComorbidityStudy—Replication). These differences reduce the validity and reliabilityof diagnoses. We need a consensually understood and empirically validatedframework to define and measure traumatic events.^24,69-71

The mental health system must

1. Improve services for victimsof trauma. Exposure to trauma is a serious public health problem amonghigh-risk youth. Yet, services are insufficient.⁴³ Timelyinterventions may avert subsequent and often chronic social problems commonamong traumatized youth.^8,16,62 Tothe extent that PTSD is correlated with subsequent violent perpetration, effectivetreatment is also a matter of public safety.^15,72,73

2. Improve the detection of PTSD. The Surgeon General's report on children's mental health suggeststhat emergency medical providers must address the mental health needs of youthwho have experienced trauma.⁷⁴ Posttraumaticstress disorder is frequently overlooked even in the best psychiatric settings.^75,76 Because PTSD frequently co-occurswith other psychiatric disorders,^10,16 itcan be difficult to detect without systematic screening.

3. Avoid retraumatizing youth. Theconditions of confinement often exacerbate symptoms of mental disorder, includingPTSD.⁷⁷ Juvenile justice providers must alsoreduce the likelihood that youth will be retraumatized during routine processing.Symptoms of PTSD may be exacerbated by such common practices as handcuffsand searches.^78,79 In detentioncenters, psychiatric crises are often handled by isolating and restrainingsymptomatic detainees. These practices can trigger or escalate symptoms ofPTSD (eg, severe anxiety, aggression, and numbing of emotions).^78,79 Psychiatristscan help to develop strategies to manage emergencies more humanely—and,ultimately, more cost-effectively.

Our nation's delinquent children are among the most traumatized. Wemust balance the resources used to punish with resources needed to heal thetraumas endured by vulnerable youth.

Corresponding author and reprints: Linda A. Teplin, PhD, Psycho-LegalStudies Program, Department of Psychiatry and Behavioral Sciences, FeinbergSchool of Medicine, Northwestern University, 710 N Lake Shore Dr, Suite 900,Chicago, IL 60611 (e-mail: psycho-legal@northwestern.edu).

Submitted for publication May 9, 2003; final revision received November4, 2003; accepted November 18, 2003.

This study was supported by grants R01MH54197 and R01MH59463 from theNational Institute of Mental Health, Bethesda, Md; and grant 1999-JE-FX-1001from the Office of Juvenile Justice and Delinquency Prevention. Major fundingwas also provided by the National Institute on Drug Abuse, Bethesda; the Centerfor Mental Health Services, Rockville, Md; the Centers for Disease Controland Prevention (CDC) National Center for HIV, STD, and TB Prevention, Atlanta,Ga; CDC National Center on Injury Prevention and Control, Atlanta; the NationalInstitute on Alcohol Abuse and Alcoholism, Bethesda; the Center for SubstanceAbuse Prevention, Rockville; the Center for Substance Abuse Treatment, Rockville;the National Institutes of Health's (NIH) Center on Minority Health and HealthDisparities, Bethesda; the NIH Office of Research on Women's Health, Bethesda;the NIH Office of Rare Diseases, Bethesda; the William T. Grant Foundation,New York, NY; and the Robert Wood Johnson Foundation, Princeton, NJ. Additionalfunds were provided by The John D. and Catherine T. MacArthur Foundation,Chicago, Ill; the Open Society Institute, New York, NY; and the Chicago CommunityTrust, Chicago, Ill. We thank all our agencies for their collaborative spiritand steadfast support.

Many more people than the authors contributed to this project. Ann Hohmann,PhD, and Kimberly Hoagwood, PhD, provided technical support in the design;Heather Ringeisen, PhD, provided helpful advice. Grayson Norquist, MD, andDelores Parron, PhD, provided steadfast support throughout. Celia Fisher,PhD, guided our human subjects procedures. We thank Gail Wasserman, PhD, andthe reviewers of the ARCHIVES for their insightful comments on earlier versionsof the manuscript, and Prudence Fisher, PhD, for her helpful advice on theDISC-IV PTSD module. We thank all project staff, especially Amy M. Lansing,PhD, for supervising the data collection, and Laura Coats, our expert editorand research assistant. We greatly appreciate the cooperation of everyoneworking in the Cook County justice systems, especially David H. Lux, our projectliaison. Without the county's cooperation, this study would not have beenpossible. Finally, we thank our participants for their time and willingnessto participate.