This survey was a cooperative effort by the Georgia Department of Human Resources (DHR) and the Georgia Department of Children and Youth Services (DCYS). For DHR it is part of a comprehensive statewide assessment of substance abuse treatment need supported by CSAT. For DCYS it is part of the integrated effort of the Georgia Policy Design Academy on the Mental Health Needs of Youth in the Juvenile Justice System to determine the need for mental health services among juvenile offenders.

This survey was designed for three purposes:

• to estimate the prevalence of substance use and other DSM-III-R psychiatric disorders among juvenile offenders entering Regional Youth Detention Centers (RYDCs).
• to evaluate the efficacy of the SCL-90-R questionnaire for screening new admissions for psychiatric problems.
• to examine the association between psychiatric disorders and a history of child abuse in this population.

Study Sample: The study is based on a random sample of 693 of the 8490 youth admitted to the 20 Regional Youth Detention Centers (RYDCs) operated by the Georgia Department of Children and Youth Services between March 13,1995 and September 30, 1995, stratified by RYDC and sex. Because of recidivism and transfers, 11,632 admissions occurred among these 8490 offenders during the study period. The sample does not fully represent all juvenile offenders in Georgia because it was not possible to sample from the Fulton County RYDC, which is operated by the county, not DCYS. The Fulton county RYDC has most of the city of Atlanta as a cachement area and received 20.3 percent of statewide admissions to RYDCs in fiscal year 1994. The interviewed sample was 69.0% male, 38.8% white, 58.0% African American, and 3.2% youth of other ethnicity. The mean age was 15.4 (s.d. = 1.3).

During the study period, 795 subjects were approached. Of these, interviews were not completed on 82 (10.3%). The major reason for non-completion was that the youth was released or called to court during the interview. A small proportion of incomplete interviews were due to factors such as the youth becoming highly oppositional or emotional, or discovery that the subject was not eligible (e.g. had misrepresented age to police to avoid placement in adult jail). Of the 713 completed interviews, 20 (2.8%) were excluded. The reasons for exclusion were: missing release dates, which resulted in a missing sampling weight (9), corrupted interview files (2), interview of an ineligible subject (3), and subject interviewed on multiple occasions (6). The overall completion rate with respect to eligible subjects approached was 88.2%.

Sampling Methods: The objective was for each RYDC to be represented in proportion to its share of statewide admissions and for females to comprise 33% of the sample rather than the 23% of admissions. All youth present in an RYDC who had been admitted to that RYDC within 72 hours of the start of the interviewing day were eligible for participation. Because the interview was treated as part of intake evaluation and served as a basis for referral for clinical services, interviews were initiated with all sampled youth except those who were released between the time they were selected for participation and were called to speak to the interviewer. There were no other exclusions.

Interviewers traveled to facilities cyclically throughout the data collection period. Each day, when an interviewer arrived at a facility, two lists (sampling frames) were constructed. Each sex specific list contained all juveniles of that sex who had been admitted within the past 72 hours. Then a simple random sample was taken from each gender-specific list, with a higher sampling fraction for females than for males to achieve the desired oversampling of females. The sampling fractions were determined by the practical constraint of how many interviews could be conducted in one day. The juveniles selected for the sample were interviewed in the detention facility on the same day that they were selected for the sample.

Interview Instruments: The interview consisted of three interview instruments. In order of presentation in the interview, they were:

• the Childhood Trauma Questionnaire¹ (CTQ), which asks about history of child abuse.
• the SCL-90-R ⁵ which is a 90 item checklist of psychiatric symptoms which is widely used for screening in adults.
• the Diagnostic Interview Schedule for Children Version 2.3 ^6.12.15-17 (DISC) which is a widely used structured interview for diagnosing DSM-III-R psychiatric disorders in epidemiological studies of children and adolescents. Conservative criteria for diagnosis were used. For each disorder, diagnosis required all symptom and duration criteria plus a report of impairment of home, school or social functioning which did not consider present incarceration. Arrest and incarceration probably contributed little to symptom and diagnostic prevalence because the sample consisted of new admissions and the minimum required duration for the presence of any symptom is 14 days.

All components of the interview were integrated into a Computer-Assisted Interviewing program and required approximately 2-1/2 hours to complete.

Interviewer Training and Supervision: Emory University is one of two certified DISC training centers in the country. Interviewers hired at the start of the study attended a three-day training in the DISC interview and CAI software. A fourth day of training was included for administering the CTQ, SCL-90 and for administrative reporting activities.

Training of interviewers also included strategies for converting highly oppositional and frivolous subjects. In cases where, in the judgment of the interviewer, valid interview results could not be obtained due to antagonism, opposition, frivolity/falsification and/or language disability, the reasons were noted in the clinical report and in an interview termination report. When scheduling permitted, at least one attempt to convert oppositional subjects deemed capable of completing the interview was made.

After initial training, each interviewer went through a two-week period of extended training during which he or she conducted supervised interviews at an RYDC in their home area. This extended training also served as a pilot study of interviewing and scheduling activities in the RYDC environment.

Replacement interviewers hired subsequently each received two-days of individual training following the DISC training manual and conducted a minimum of five directly supervised interviews prior to beginning independent interviewing.

Supervision of interviewers was performed from Emory by two experienced research coordinators. The coordinators visited Centers with ongoing interviews on a rotating basis. During visits they monitored interviews, collect audio tapes and interview data and discussed questions and/or problems with the interviewer. The coordinators spoke with each interviewer either in person or by telephone at least twice per week of active interviewing.

Statistical Methods: Sampling weights were computed based on RYDC, gender, number of admissions during the study period and duration of stay (poststratification to adjust for potential undersampling of youth who were released in under two days). All major analyses were computed using SUDAAN ²⁰, which is designed for the analysis of unequal probability samples.

The prevalence of juvenile offenders with at least one psychiatric disorder was 61 ± 4% (95% confidence interval). 30 ± 4% of the youth had at least one anxiety disorder, 19 ± 3% had a mood disorder, 35 ± 4% had a disruptive behavior disorder and 30 ± 5% had a substance use disorder. 44% of youth admitted to the RYDCs had two or more DISC diagnoses. The prevalence of disorders was much higher than in general population studies, with especially high rates of anxiety disorders, disruptive behavior and substance use disorders. Females had higher prevalence of anxiety and mood disorders than males, but substance use disorders were more prevalent among males than females. Males and females had similar prevalence for the presence of any disorder. African American subjects had significantly lower prevalence than whites for all disorders.

The use of alcohol and other drugs, especially marijuana (Table 1) is extremely high for any population, but especially so for one averaging 15.4 years old or 9^th to 10^th grade. It is noteworthy that self-reports of use of drugs other than alcohol and marijuana were almost exclusively limited to white youth and that African-American youth also report significantly less use of alcohol and marijuana. Part of this difference may be due to reporting bias (i.e. African Americans underreport and/or whites overreport). Based on interviewer notes and comments, we believe that at least part of the ethnic difference is real. The juvenile SANTA study, which includes a urine drug screen, will provide confirmation of these results.

The prevalence of DSM-III-R substance use diagnoses (Table 2) indicates a very high level of problems associated with alcohol and drug use in this population. Compared to the results from the NIMH Methods for the Epidemiology of Child and Adolescent Mental Disorders (MECA) study¹⁸ (Figure 1), the rate of substance use disorders was nearly 8 times greater among juvenile offenders than among children and adolescents in the general population. The difference between offenders and the general population was greatest for substance use disorders. However, the DISC using only the child as an informant, as in this study and our reanalysis of the MECA data, greatly underestimates the prevalence of disruptive behavior disorders, such as Attention-Deficit Disorder and Conduct Disorder.

Juvenile offenders with substance use disorders exhibited an extremely high prevalence of comorbid psychiatric disorders (Table 3). Although a high prevalence of comorbid disruptive behavior disorders was expected because alcohol and drug use can be a symptom of conduct disorder, the levels of comorbidity for anxiety and mood disorders are also extremely high. Given this level of comorbidity, providing juvenile offenders with treatment for substance abuse problems without considering their nearly universal comorbidity may be ineffective.

Finally, this survey indicated that contact of the diagnosed population with services for substance use problems is very limited. Based on the question "Have you ever talked to a professional like a psychiatrist, psychologist, social worker or counselor about (your substance use problems)?", only 16 ± 6% responded positively: 19 ± 10% and 24 ± 16% for white males and females, respectively, and 11 ± 9% for African American males. None of the 14 African American females with substance use diagnoses reported talking to a professional about the problem.

Although SCL-90-R scores were highly correlated with the presence of a psychiatric disorder, associations with substance use disorders were comparatively weak. Furthermore, the results of sensitivity analyses indicate that the SCL-90-R possesses inadequate screening characteristics (sensitivity and specificity) to serve as a useful screening instrument for psychiatric problems in this population.

The Childhood Trauma Questionnaire was significantly correlated with 4 of the 5 diagnostic categories: all except substance use disorders. It is surprising that CTQ reports of child abuse are not associated with substance use diagnoses, since there are very consistent reports of strong associations between child abuse and substance use disorders in the literature^{2-4 7,14,17}. It is possible that informing subjects that we were required by law to report child abuse substantially affected the validity of responses to the CTQ, resulting in underreporting of history and weakening or entirely masking an underlying association of child abuse with substance use diagnoses.

This study was made possible through funding from CSAT contract number 270-94-0010, State Demand and Needs Assessment Studies: Alcohol and Other Drugs and through direct and in-kind contributions from the Georgia Department of Human Resources, the Georgia Department of Children and Youth Services, the Georgia Department of Corrections and the Georgia Department of Education. The contributions from State agencies were made possible by and coordinated through the Georgia Policy Design Academy on the Mental Health Needs of Youth in the Juvenile Justice System.

The interviewers were Pamela Bell, Karen Clark, Gina Coleman, Wes Evans, Eddye Garrett, Susan Haley, John Johnson, Reginald Martin, Ernest Pierce, LaTricia Rumph and Charlie Wells. The field supervisors were Alicia Edwards, Bill Ilott and Lisa Medellin-Young.

Finally, we owe a debt of deepest gratitude to the staff and directors of the RYDCs whose support, cooperation and hospitality were outstanding.