Figure 1. Past-year prevalence of major mood disorders (major depressive disorder [MDD], mania) and anxiety disorders (posttraumatic stress disorder [PTSD], generalized anxiety disorder [GAD], and panic disorder) by sex.
Figure 2. Past-year prevalence of substance use disorders (alcohol disorder [Alc], drug disorder) and any disruptive behavior disorder (DBD) by sex and race/ethnicity.
Teplin LA, Welty LJ, Abram KM, Washburn JJ, Dulcan MK. Prevalence and persistence of psychiatric disorders in youth after detention: a prospective longitudinal study. Arch Gen Psychiatry. 2012. doi:10.1001/archgenpsychiatry.2011.2062.
eTable 1. Sample characteristics at baseline, time 1, and time 2
eTable 2. Prevalence of disorder at baseline, time 1, and time 2, by race/ethnicity in males
eTable 3. Prevalence of disorder at baseline, time 1, and time 2, by race/ethnicity in females
eTable 4. Prevalence of substance use disorders among participants living in the community at time 2
eTable 5. Persistence of disorders from baseline to time 1 and from baseline to time 2, by race/ethnicity in males
eTable 6. Persistence of disorders from baseline to time 1 and from baseline to time 2, by race/ethnicity in females
eMethods. Expanded notes on study methods eReferences.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Teplin LA, Welty LJ, Abram KM, Dulcan MK, Washburn JJ. Prevalence and Persistence of Psychiatric Disorders in Youth After Detention: A Prospective Longitudinal Study. Arch Gen Psychiatry. 2012;69(10):1031–1043. doi:10.1001/archgenpsychiatry.2011.2062
Context Psychiatric disorders are prevalent among incarcerated juveniles. Most juveniles eventually return to their communities, where they become the responsibility of the community mental health system. However, no large-scale study has examined psychiatric disorders after youth leave detention.
Objective To examine changes in the prevalence and persistence of psychiatric disorders during the 5 years after detention, focusing on sex and racial/ethnic differences.
Design Prospective longitudinal study with up to 5 interviews (1829 youth: 1172 males and 657 females). To ensure representation of key demographic subgroups, the randomly selected sample was stratified by sex, race/ethnicity (African American, non-Hispanic white, and Hispanic), age, and legal status (juvenile or adult court).
Setting The Northwestern Juvenile Project, sampling youth from the Cook County Juvenile Temporary Detention Center, Chicago, Illinois.
Participants Detained youth, aged 10 to 18 years at baseline interview.
Main Outcome Measures At baseline, the Diagnostic Interview Schedule for Children Version 2.3. At follow-up interviews, the Diagnostic Interview Schedule for Children Version IV (Child and Young Adult versions) and the Diagnostic Interview Schedule Version IV (substance use disorders and antisocial personality disorder).
Results Five years after baseline, more than 45% of males and nearly 30% of females had 1 or more psychiatric disorders with associated impairment. More than 50% of males and more than 40% of females had 1 or more psychiatric disorders without impairment. Substance use disorders were the most common; males, however, had higher rates over time (5 years after baseline, adjusted odds ratio [AOR], 2.61; 95% CI, 1.96-3.47). Non-Hispanic whites and Hispanics also had higher rates of substance use disorders vs African Americans (AOR, 1.96; 95% CI, 1.54-2.49 and AOR, 1.59; 95% CI, 1.24-2.03). Females had higher rates of major depression over time (AOR, 1.59; 95% CI, 1.22-2.08).
Conclusions Although prevalence rates of most psychiatric disorders declined as youth aged, a substantial proportion of delinquent youth continue to have disorders. There are notable sex and racial/ethnic differences in the prevalence and persistence of psychiatric disorders in this population.
Psychiatric disorders are prevalent among incarcerated juveniles.1-14 A 2008 literature review14 concluded that psychiatric disorders are substantially more common in adolescents in detention than among adolescents in the general population. The Northwestern Juvenile Project15 found that at intake to detention, even after excluding conduct disorder, more than 60% of juvenile detainees met diagnostic criteria with diagnosis-specific impairment for 1 or more psychiatric disorders. Among youth incarcerated for 9 months, Karnik et al16 found even higher rates: approximately 90% of detainees had a psychiatric disorder other than conduct disorder or oppositional defiant disorder. Using the lower rate,15 more than 55 000 of the 92 854 youth held in US correctional facilities each day17 have 1 or more psychiatric disorders.
Many psychiatric disorders are likely to persist as these juveniles become young adults. Risk factors for psychiatric disorders are common among delinquent youth: maltreatment,18-20 dysfunctional families,19,21 family substance abuse,18 and brain injury.22 Because delinquent youth have few protective factors to offset these risks, many are vulnerable to continued psychiatric morbidity as they age.18
Despite their importance, few longitudinal studies have examined the prevalence and persistence of psychiatric disorders after youth leave detention. Instead, studies23-26 of delinquent youth have focused on the association between psychiatric disorders and criminal recidivism, antisocial behavior, or social functioning. We could find only one longitudinal study of the persistence and prevalence of psychiatric disorders in detained youth. Harrington et al27 found that 2 years after detention, many mental health problems persisted or had worsened. Their sample, however, excluded females, was 80% white, and was too small (N = 97) to permit detailed analyses. Moreover, the study was conducted in the United Kingdom, limiting generalizability to juvenile detainees in the United States.
The related literature—longitudinal studies of “high-risk” youth—also provides little information. Youth with histories of detention have been included in studies of high-risk youth: the homeless,28,29 youth living in impoverished or high-crime neighborhoods,30-34 and the offspring of parents who used substances or had psychiatric disorders.29,35-46 Yet, none of these studies distinguished between youth with and without histories of detention (summary tables available on request).
Delinquent youth are also underrepresented in general population longitudinal studies, such as household-based surveys (eg, Great Smoky Mountains Study of Youth,47,48 National Comorbidity Surveys,49-51 and National Epidemiologic Survey on Alcohol and Related Conditions52) and studies drawn from school-based samples (eg, Oregon Adolescent Depression Project53,54 and Pittsburgh Youth Study55) and general pediatric clinics.56,57 Household surveys typically exclude persons in correctional institutions.47,50,58-61 School-based samples omit youth who are truant or miss school because they are detained.53-55,62-65 Samples drawn from general pediatric clinics omit youth who do not receive medical care.56,57 Even if sampled initially, youth may be lost to follow-up when they are incarcerated because they cannot be found and because studying prisoners requires special procedures and approvals from the Secretary of Health and Human Services.66
In sum, to our knowledge, no large-scale longitudinal study has examined the prevalence and persistence of psychiatric disorders after youth leave detention. This omission is critical. Juvenile detainees have a median length of stay of only 2 weeks.67 Thus, juvenile detainees become a community public health problem when they are released and may continue to burden society as they age. Epidemiologic studies are the first step to improve prevention and treatment in correctional facilities and in the community.68 Data are also needed to address health disparities, a priority of Healthy People 202069 and the Institute of Medicine.70 Of the approximately one-half million incarcerated youth and young adults (aged 24 years or younger), nearly two-thirds are African American or Hispanic,17,71 compared with one-third in the general population.72
This study presents data from the Northwestern Juvenile Project, the first comprehensive longitudinal study of psychiatric disorders in youth after they leave detention. The sample is large (N = 1829), racially/ethnically diverse, and includes males and females. We examined changes in the prevalence and persistence of disorders during the first 5 years after detention, focusing on sex and racial/ethnic differences.
We recruited a stratified random sample of 1829 youth at intake to the Cook County Juvenile Temporary Detention Center (CCJTDC) in Chicago, Illinois, between November 20, 1995, and June 14, 1998. The CCJTDC is used for pretrial detention and for offenders sentenced for fewer than 30 days. To ensure adequate representation of key subgroups, we stratified our sample by sex, race/ethnicity (African American, non-Hispanic white, Hispanic, or other), age (10-13 years or 14-18 years), and legal status (processed in juvenile or adult court). Final sampling fractions for strata ranged from 0.11 to 0.69.
All detainees who were awaiting the adjudication or disposition of their case were eligible to participate in the study. Among these, 2275 detainees were randomly selected; 4.2% (34 youth and 62 parents or guardians) refused to participate. There were no significant differences in refusal rates by sex, race/ethnicity, or age. Twenty-seven youth left the detention center before an interview could be scheduled; 312 left CCJTDC while we attempted to locate their caretakers for consent. Eleven others were excluded from the sample because they were unable to complete the interview. The final sample size was 1829: 1172 males and 657 females; 1005 African Americans, 296 non-Hispanic whites, 524 Hispanics, and 4 other race/ethnicity; and age range, 10 to 18 years (mean, 14.9 years; median, 15 years) (eTable 1). Face-to-face structured interviews were conducted at the detention center in a private area, most within 2 days of intake.
Our design included (1) follow-up interviews at 3 and 4½ years after baseline for the entire sample and (2) 2 additional interviews at 3½ and 4 years after baseline for a random subsample of 997 individuals (600 males and 397 females). For each follow-up, we interviewed participants regardless of where they were living: in the community (approximately two-thirds of interviews), at correctional facilities (nearly 30% of interviews), or by telephone if they lived more than 2 hours away (<5% of interviews).
Participants were paid $25 for the 2- to 3-hour baseline interview and $50 for each of the 3- to 4-hour follow-up interviews. Most interviewers had graduate degrees in psychology or an associated field and had experience interviewing at-risk youth; one-third were fluent in Spanish. All interviewers were trained for at least 1 month. For each wave, consistency across interviewers was established and maintained via scripted mock interviews after training and midwave. Additional information on our methods is in the eMethods and is also published elsewhere.15,73,74
For all interviews, participants signed either an assent form (if they were aged <18 years) or a consent form (if they were aged ≥18 years). The Northwestern University Institutional Review Board and the Centers for Disease Control and Prevention Institutional Review Board approved all study procedures and waived parental consent for persons younger than 18 years, consistent with federal regulations regarding research with minimal risk.75 We nevertheless attempted to contact parents of minors to obtain their consent and to provide them with information on the study and used an independent participant advocate to represent the minors' interests75 (eMethods).
We used the Diagnostic Interview Schedule for Children Version 2.3 (DISC-2.3),76,77 the most recent English and Spanish versions then available. This version of the DISC, based on the DSM-III-R, assesses the presence of disorders in the past 6 months. The DISC-2.3 is highly structured, contains detailed symptom probes, has acceptable reliability and validity, and requires relatively brief training.76,78-81 Because the DISC-2.3 did not include posttraumatic stress disorder (PTSD), we used the module from the DISC-IV when it became available, 13 months after the study began.73 Additional detail on baseline diagnostic decisions can be found elsewhere.15,73,74
We administered the DISC-IV (Child and Young Adult versions), based on the DSM-IV, to assess for schizophrenia, mood disorders, anxiety disorders, attention-deficit/hyperactivity disorder, and disruptive behavior disorders in the past year.82,83 Impairment was defined as moderate impairment in at least one area of functioning (criterion A).84 To facilitate comparison with studies that do not impose impairment criteria, we present prevalence rates of disorders with and without the impairment criterion; our analyses of changes in disorders over time use disorders with impairment.
To assess substance use disorders and antisocial personality disorder at follow-up, we administered the Diagnostic Interview Schedule Version IV (DIS-IV).85,86 We used the DIS-IV to assess substance use disorders because the DISC-IV is not sufficiently detailed for our population. Antisocial personality disorder was assessed for participants 18 years or older (who were no longer eligible for diagnoses of childhood disruptive behavior disorders). Disorders are assessed for the year before the interview. Consistent with the National Comorbidity Survey Replication (NCS-R),50 participants who met the criteria for substance use disorder or antisocial personality disorder with “partial recovery” were scored as having the disorder. We did not implement DSM exclusionary criteria.
Our diagnostic measures changed over time for 3 reasons: (1) the release of the DISC-IV (based on DSM-IV criteria) midstudy, (2) some participants turned 18 years old and were therefore ineligible for childhood disruptive behavior disorders, and (3) our need to use a more comprehensive measure of substance use disorder (DIS-IV) for the follow-up interviews.
To check that changes in prevalence rates over time were not the result of changes in measurement, we conducted sensitivity analyses. We created a set of adjusted scoring algorithms to maximize comparability among the DISC-2.3, DISC-IV, and DIS-IV criteria while minimizing alterations (Author's Appendix [http://hdl.handle.net/2166/repo/teplin-0001]). All analyses were conducted twice, with and without these adjusted criteria. Because there were no substantive differences in findings and to enable comparisons with other studies, we present the results using the original, unadjusted diagnoses.
Because some participants were interviewed more often than others, we used a subset of interviews to summarize prevalence rates at 3 time points: baseline (time 0), time 1, and time 2.
Time 1 is the first follow-up interview but excludes interviews that occurred more than 18 months after the interview due date. Using a narrower window would restrict the generalizability of our findings because, in this high-risk and highly mobile population, participants can be difficult to track. The median time between baseline and time 1 was 3.0 years (mean [SD], 3.2 [0.3] years; range, 2.7-4.5 years). For simplicity, we refer to the time 1 interview as occurring approximately 3 years after baseline. As reported in eTable 1, which summarizes sample demographics and retention rates, 90.7% of the participants had a time 1 interview.
For each participant, time 2 consists of the earliest follow-up interview that occurred approximately 4.5 years after baseline. As with time 1, we excluded interviews that occurred more than 18 months after this due date. The median time between baseline and the time 2 interview was 4.7 years (mean [SD], 4.9 [0.4] years; range, 4.3-6.0 years). To ensure that prevalence rates reflect temporally distinct cross-sections of the sample, we required at least 16 months between the time 1 and time 2 interviews. We subsequently refer to the time 2 interviews as occurring approximately 5 years after baseline. As reported in eTable 1, 85.3% of participants had a time 2 interview.
To generate prevalence rates that reflect CCJTDC's population, each participant was assigned a sampling weight augmented with a nonresponse adjustment to account for missing data.87 Taylor series linearization was used to estimate standard errors.88,89
We used all available interviews, with an average of 3.9 interviews per person (range, 1-5 interviews per person); 88.7% of the participants completed all scheduled interviews. Using generalized estimating equations (GEEs),90 we fit marginal models examining (1) differences in the prevalence of disorders by sex and race/ethnicity over time and (2) changes in disorders as the youth aged. Disorder was modeled as binomial with a logit link function. We used a robust sandwich estimator with an unstructured correlation matrix; in the few instances in which models failed to converge, we specified an exchangeable correlation structure. All statistically significant (P < .05) odds ratios contrasting sex and racial/ethnic differences are noted and were determined using GEEs. Unless otherwise noted, these odds ratios contrast sex and race/ethnicity over time. Only 2 sets of odds ratios—those examining persistence of disorders and those examining prevalence of disorders among participants living in the community—contrast sex and race/ethnicity at single time points (time 1 or time 2). These were estimated using logistic regression.
All GEE models included covariates for sex, race/ethnicity (African American, Hispanic, or non-Hispanic white), and aging (time since baseline). We also included age at baseline (10-18 years) and legal status at detention (processed in juvenile or adult court) because they were stratification characteristics. The 4 participants who identified themselves as other race/ethnicity at baseline were excluded. We estimated models with quadratic terms for aging when the linear term was statistically significant. When main effects were significant, we estimated models with the corresponding interaction terms. Only interaction terms or quadratic aging terms that reached statistical significance were included in the final models. For models with significant interactions between sex and aging, we report model-based odds ratios for sex differences at 3 and 5 years after baseline. The GEE models were estimated with sampling weights to account for study design. We examined the sensitivity of our models to attrition by using weights that combined the probability of being sampled with the probability of dropping out.91 Results (available on request) were substantially similar.
The GEE analyses were conducted on disorders (with the impairment criterion) except for (1) schizophrenia, because there were too few cases; (2) hypomania, because participants with this disorder were significantly more likely to be missing follow-up data than were those without the disorder (33.3% vs 11.0%, P = .01); and (3) diagnoses assessed only for juveniles (attention-deficit/hyperactivity disorder, conduct disorder, and oppositional defiant disorder) or adults (antisocial personality disorder), because nearly all participants became 18 years old or older during the follow-up period. Instead, we estimated changes over time in any disruptive behavior disorder, defined as conduct disorder or oppositional defiant disorder (for participants aged <18 years) or antisocial personality disorder (for participants aged ≥18 years).
Because detention may alter access to substances, all models describing substance use disorders included linear and quadratic terms for time incarcerated prior to the interviews. When substance use disorder was measured at a follow-up interview, we used number of days incarcerated in the past year because disorders were assessed in the past year. When substance use disorder was measured at the baseline interview (and therefore assessed in the past 6 months), we used time incarcerated in the past 90 days, the best available estimate. All analyses were conducted using commercial software (Stata 11; StataCorp) with its survey routines.92
Table 1 and Table 2 report prevalence rates of disorders—computed with and without the impairment criterion—at baseline, time 1, and time 2 for males and females. At time 2, more than 45% of the males and nearly 30% of the females had any disorder with impairment. More than 50% of the males and more than 40% of the females had 1 or more psychiatric disorders without impairment. Even excluding disruptive behavior disorders, 36.9% of males and 25.3% of females had any disorder.
Prevalence rates of disorder by race/ethnicity for males and females are given in eTable 2 and eTable 3. Among males, 44.3% of African Americans, 49.8% of Hispanics, and 63.9% of non-Hispanic whites had any disorder at time 2. More than one-quarter of African American females and more than one-third of Hispanic and non-Hispanic white females had any disorder.
Other than mania, prevalence rates decreased as the participants aged. For every additional year, there was an approximately 10% decrease in the odds of any mood disorder (adjusted odds ratio [AOR], 0.91; 95% CI, 0.85-0.97) and its subcategories: any major mood disorder (AOR, 0.89; 95% CI, 0.82-0.96) and major depression (AOR, 0.91; 95% CI, 0.84-0.98). The prevalence of dysthymia also decreased with age (AOR, 0.57; 95% CI, 0.46-0.70).
Over time, females had higher rates of any mood disorder (AOR, 1.33; 95% CI, 1.05-1.68) and its subcategories: any major mood disorder (AOR, 1.54; 95% CI, 1.19-2.00) and major depression (AOR, 1.59; 95% CI, 1.22-2.08). Figure 1 illustrates prevalence rates of major mood disorders over time by sex.
The only significant racial/ethnic difference was for mania, which was more prevalent among minorities over time (African American vs non-Hispanic white: AOR, 6.92; 95% CI, 1.78-26.88; Hispanic vs non-Hispanic white: AOR, 8.01; 95% CI, 2.05-31.38).
The prevalence of panic disorder increased slightly overall (AOR, 1.30 per year; 95% CI, 1.03-1.65). Figure 1 shows changes in prevalence rates over time by sex.
Females had higher rates of any anxiety disorder (AOR, 1.42; 95% CI, 1.06-1.91). Although rates of PTSD appeared to decrease, especially for females, this trend was not statistically significant (AOR, 0.93; 95% CI, 0.81-1.05).
Compared with non-Hispanic whites, Hispanics had more than twice the odds of any anxiety disorder (AOR, 2.18; 95% CI, 1.37-3.49) and its subcategory PTSD (AOR, 2.82; 95% CI, 1.66-4.80). Compared with African Americans, Hispanics had 3.82 times the odds of panic disorder (95% CI, 1.54-9.47) and 1.54 times the odds of PTSD (95% CI, 1.01-2.34). In addition, African Americans were more likely than non-Hispanic whites to have PTSD (AOR, 1.84; 95% CI, 1.10-3.07), although non-Hispanic whites were more likely than African Americans to have panic disorder (AOR, 3.56; 95% CI, 1.28-9.94).
The prevalence of any disruptive behavior disorder decreased over time, but the rate of decrease depended on sex. Figure 2 illustrates these sex differences over time.
Males and females did not have significantly different rates of any disruptive behavior disorder at baseline, but prevalence decreased faster among females than among males. Three years after baseline, males had 1.82 times the odds of any disruptive behavior disorder (95% CI, 1.46-2.26) compared with females; 5 years after baseline, males had 2.95 times the odds (95% CI, 2.16-4.02).
As illustrated in Figure 2, non-Hispanic whites had the highest rates of any disruptive behavior disorder over time (non-Hispanic white vs African American: AOR, 2.34; 95% CI, 1.80-3.04; non-Hispanic white vs Hispanic: AOR, 1.56; 95% CI, 1.16-2.11), followed by Hispanics (Hispanic vs African American: AOR, 1.50; 95% CI, 1.10-2.04).
Substance use disorders were the most prevalent disorders in our sample. The prevalence generally decreased over time, but the rate of decrease depended on sex. Figure 2 illustrates sex and racial/ethnic differences over time.
At baseline, compared with females, males had about one-third greater odds of any substance use disorder (AOR, 1.34; 95% CI, 1.05-1.71) and its subcategory, drug use disorder (AOR, 1.37; 95% CI, 1.07-1.75); rates of alcohol use disorder were not significantly different. By the follow-up interviews, however, the disparities between males and females increased substantially because prevalence rates decreased faster for females than for males. Three years after baseline, compared with females, males had approximately twice the odds of any substance use disorder (AOR, 2.00; 95% CI, 1.64-2.43) and its subcategories, drug use disorder (AOR, 2.01; 95% CI, 1.62-2.49) and alcohol use disorder (AOR, 1.97; 95% CI, 1.53-2.53). Five years after baseline, males had more than 2.5 times the odds of these disorders compared with females (any substance use disorder: AOR, 2.61; 95% CI, 1.96-3.47; drug use disorder: AOR, 2.60; 95% CI, 1.88-3.58; and alcohol use disorder: AOR, 2.87; 95% CI, 2.00-4.13). Although the prevalence rates of most disorders decreased for males and females (Figure 2), 3 years after baseline, rates of alcohol disorder were no longer decreasing among males (AOR, 1.03 per year; 95% CI, 0.96-1.11).
Even after adjusting for time spent in correctional facilities, substance use disorders were more common among non-Hispanic whites and Hispanics than among African Americans (Figure 2). Compared with African Americans, non-Hispanic whites had nearly twice the odds of any substance use disorder and its subcategory, drug use disorder (AOR, 1.96; 95% CI, 1.54-2.49 and AOR, 1.91; 95% CI, 1.50-2.43, respectively), and 1.57 times the odds of alcohol disorder (95% CI, 1.15-2.13). Hispanics had approximately 1.5 times the odds of any substance use disorder and its subcategory, drug use disorder, compared with African Americans (AOR, 1.59; 95% CI, 1.24-2.03 and AOR, 1.46; 95% CI, 1.12-1.92, respectively).
We also analyzed substance use disorders without impairment criteria, an approach used by recent studies of the general population.50,58 Results were substantially similar (available on request).
Because substance use is restricted in jails and prisons, we examined rates of substance use disorders only among participants who had lived in the community the entire year before time 2 (345 males and 479 females). These prevalence rates, and the demographic differences, were substantially similar to those in the entire sample (eTable 4).
Table 3 reports the persistence of disorders: among participants who had the disorder at baseline, the proportion who still had the disorder at time 1 or time 2. For most disorders, rates of persistence were higher at time 1 than at time 2.
Regardless of sex, approximately 1 in 5 participants had a mood disorder that persisted to time 2. Substance use disorders were among the most persistent disorders for both males and females but were significantly more likely to persist among males. Any disruptive behavior disorder was also among the most persistent disorders in males and at time 2 was significantly more likely to persist in males than in females.
Among males, there were no significant racial/ethnic differences in the persistence of disorders (eTable 5). Among females, there were several significant differences (eTable 6). At time 1, any substance use disorder and its subcategory, alcohol use disorder, had approximately 3 times the odds of persisting among non-Hispanic whites and Hispanics than among African Americans. At time 2, drug use disorders also had greater odds of persisting among non-Hispanic whites than among African Americans.
Although prevalence rates of most psychiatric disorders declined over time, a substantial proportion of delinquent youth continued to have disorders as they aged. For some youth, detention may coincide with a period of crisis that subsequently abates. Many youth, however, continue to struggle: 5 years after detention, when participants were ages 14 to 24 years, nearly half of males and nearly 30% of females had 1 or more psychiatric disorders with associated impairment.
Substance use and disruptive behavior disorders continued to be the most common disorders. For many delinquent youth—especially males—externalizing disorders were not limited to adolescence. Five years after baseline, males had 2 to 3 times the odds of having substance use and disruptive behavior disorders compared with females. Furthermore, the disparity between males and females increased over time. Males were also more likely than females to have persistent substance use disorders and disruptive behavior disorder.
The observed sex differences in externalizing disorders are consistent with those in the general population, where males are up to 10 times more likely than females to persist in antisocial behavior from childhood to adulthood.93 Why might males fare worse than females? First, delinquent males are less likely to receive services.94 Second, delinquent males may have fewer opportunities to assume age-appropriate social roles (ie, jobs, postsecondary schooling)—turning points that might reduce problem behaviors.95 Finally, precocious transition to adult social roles, such as parenthood, may be associated with worse outcomes for males than for females.96-98
As in the general population,99 females had higher rates of internalizing disorders than males. The persistence of mood disorders, about 20%, was similar in males and females.
Rates of substance use disorders and disruptive behavior disorders were lower in African Americans than non-Hispanic whites. These findings may reflect underlying racial/ethnic disparities in the legal system71,100-102 and the different pathways by which non-Hispanic whites and racial/ethnic minorities enter the system. We found racial/ethnic differences in substance use disorders even after taking into account that African Americans spend more time in correctional facilities, where access to alcohol and drugs is restricted.100,101 Our findings add to the growing debate about how the “war on drugs” has affected the disproportionate incarceration of African Americans. Many investigators suggest that disproportionate confinement of racial/ethnic minorities for drug offenses is the result, in part, of disparate enforcement of drug laws in African American communities rather than higher rates of use or dealing.103-107 Our findings appear consistent with this view.
It is difficult to compare specific prevalence rates in our sample with those in the general population because differences in instrumentation and sample demographics limit meaningful comparisons. The National Comorbidity Survey Replication (NCS-R) provides the most comparable data to our time 2 interview. Although the NCS-R used different, and often less stringent, impairment criteria and did not assess the same disorders (eg, antisocial personality disorder), it provides DSM-IV diagnoses for a similarly aged sample (18-24 years).108,109 The most marked discrepancies between our findings and those of NCS-R were for drug use disorders, regardless of sex and race/ethnicity. For example, we found that approximately 20% of males had a drug use disorder compared with approximately 7% in the NCS-R; nearly 14% of Hispanic females and nearly 25% of Hispanic males had a drug use disorder compared with less than 5% of Hispanics in the NCS-R.99
Changes in prevalence over time mirror those in the general population for most disorders. As summarized in the recent literature review by Costello et al,110 many disorders in the general population decrease from adolescence to young adulthood, with the exception of panic disorders and substance use disorders, which increase48,111-116; findings on depression have been equivocal.48,111-116 Again, our sample is most notably different from the general population in substance use disorders: our rates decreased over time. Perhaps substance abuse peaks earlier in delinquent youth, coinciding with the general course of delinquent behavior.117,118 In contrast, general population youth may experience events that increase the likelihood of substance abuse as they age,119,120 such as living in college dormitories, freedom from social controls, and delays in assuming adult responsibilities such as parenting—events less likely to be experienced by delinquent youth.121
In terms of persistence, the most recent comparable investigation48 that was conducted in the United States using a sample of similar age- and DSM -based criteria (albeit different measures) found lower rates of persistence of depression and disruptive behavior disorders than in our sample. (Persistence of substance use disorders is not comparable because the definitions differed; Copeland et al48 used more liberal impairment criteria and included nicotine.)
Our data are subject to the limitations of self-report. Moreover, it was not feasible to study more than one jurisdiction; because the prevalence of psychiatric disorders may vary across jurisdictions,14,122,123 generalizability may be limited. We cannot know whether psychiatric disorders increase the likelihood of arrest and detention or vice versa. Findings might have been marginally different had we been able to use identical measures and time frames at the baseline and follow-up interviews. Rates would likely have been higher had caretakers been available to provide independent reports at baseline.15 At the follow-up interviews, many participants were not eligible to have caretakers interviewed because the participants were older than 17 years or did not live with a caretaker. Although retention rates were high, participants who missed interviews might be more likely to have had disorders than those who were found. Our findings do not take into account mental health services that might have been provided. Despite these limitations, our findings have implications for future research and mental health policy.
Retain incarcerated persons in longitudinal studies of psychiatric disorders. Most large-scale longitudinal studies of the general population do not retain persons who become incarcerated by follow-up (eg, the National Epidemiologic Survey on Alcohol and Related Conditions [B. Grant, PhD, PhD, email communication, August 13, 2010]) or reinterview too few to analyze (the Epidemiologic Catchment Area Study [W. Eaton, PhD, oral communication, August 11, 2010]). Thus, these samples are biased: they systematically exclude persons who, as our study suggests, are likely to have psychiatric disorders and poor outcomes. Excluding incarcerated persons biases prevalence rates, especially for African American males. At any given time, nearly 1 in 9 African American males aged 25 to 34 years are incarcerated.71 To address health disparities, we must include correctional populations, which have increased from approximately 2 million in 2000 to nearly 2.4 million in 2008.71,124
Add variables on incarceration history to general population studies. Although many studies examine the prevalence of psychiatric disorders in incarcerated populations,13-16,74,122,125-127 few focus on the effect of incarceration on psychiatric disorders. We suggest that general population epidemiologic surveys add the following variables: number of incarcerations, age at time of incarceration, length of incarcerations, and experiences in “community corrections” (parole, probation, and community supervision). This strategy would generate necessary information on how disproportionate confinement of racial/ethnic minorities affects health disparities in psychiatric disorders and related outcomes.
Include females in longitudinal studies of delinquents. The sex differences observed in our study underscore that findings for males may not generalize to females. Yet, most longitudinal studies of delinquents exclude females or sample too few to analyze sex differences (summary tables available on request). Future studies must include females and collect data on pregnancy, childbirth, and child-rearing. We will then have the requisite empirical foundation to improve sex-specific mental health services, especially needed now that females constitute an increasing proportion (now 30%128) of juvenile arrests.
Examine variables that affect trajectories of disorder in high-risk youth. Few studies of high-risk youth examine trajectories of disorder; fewer still examine how potentially modifiable risk and protective factors predict trajectories of disorder. Future studies should investigate how social, cognitive, and biological factors interact to affect trajectories. For example, advances in neuroscience research provide unique opportunities to investigate how developmental differences in emotion regulation interact with “turning points” to alter trajectories.129-131
Focus on delinquent males. In recent years, innovative programs funded by the Office of Juvenile Justice and Delinquency Prevention132—such as Girl Talk,133 and Girl Scouts in Detention Centers—addressed the needs of delinquent females.134-136 The mental health system must now improve services for males, who constitute 70% of juvenile arrests and 85% of youth in correctional facilities.17,128 Our findings demonstrate that interventions for substance use and disruptive behavior disorders are especially needed. Comprehensive interventions, such as functional family therapy,137 multidimensional treatment foster care,138 and multisystemic therapy,139 can be effective. Continued development and dissemination of these programs can further reduce illegal behaviors and provide cost-effective alternatives to incarceration.140
Assess and treat substance use disorders in correctional facilities and after release. Irrespective of sex or race/ethnicity, alcohol and drug use disorders were among the most common and persistent disorders. However, the need for services far exceeds their availability. Approximately one-half of youth in juvenile correctional facilities141,142 and approximately three-quarters of youth in adult jails and prisons who need substance abuse treatment do not receive it.141 Incarcerated adults fare much worse: summarizing Department of Justice statistics,143,144 a recent study published in JAMA145 concluded that 80% to 85% of adult prisoners who needed treatment for drug abuse did not receive it. After release back to the community, services may be difficult to obtain. The Substance Abuse and Mental Health Services Administration reports, for example, that fewer than 10% of juveniles146 and adults147 with an “alcohol use problem” received specialty services in the past year.
Despite the potential of health care reform, the law may not improve mental health services for persons like our participants, who may frequently cycle through correctional facilities. Health care reform has no benefit to prisoners because federal law prohibits Medicaid payments for individuals—juvenile or adult—who are inmates of public institutions.148 Incarceration disrupts community treatment and Medicaid benefits.149
Moreover, the Patient Protection and Affordable Care Act (PPACA) has limitations. First, although the Supreme Court upheld most provisions of the PPACA,150 the law does not improve the deteriorating public health infrastructure,151,152 where many impoverished persons would receive treatment153,154 after being released from corrections. Second, although the PPACA expands Medicaid benefits to persons with incomes up to 133% of the federal poverty level, the Supreme Court ruled that states could opt out.155 The effect of the Supreme Court's ruling is not yet known. Many persons like our participants are poor enough to qualify for Medicaid even without the expansion. Moreover, states have strong financial incentives to expand Medicaid.156,157 Thus, as of this writing, it is unclear how many additional people would actually receive health care coverage. Since the Supreme Court's decision, the Congressional Budget Office158 projected that 30 million “nonelderly residents”—such as undocumented immigrants and persons with low incomes who fail to enroll in Medicaid152,153—will remain uninsured in 2016. Therefore, we must improve services and access to care in corrections and in the community, where virtually all detained juveniles and incarcerated adults will eventually return.
Correspondence: Linda A. Teplin, PhD, Health Disparities and Public Policy, Department of Psychiatry and Behavioral Sciences, Northwestern University Feinberg School of Medicine, 710 N Lake Shore Dr, Ste 900, Chicago, IL 60611 (firstname.lastname@example.org).
Submitted for Publication: April 4, 2011; final revision received November 18, 2011; accepted November 23, 2011.
Author Contributions: Dr Teplin had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All other authors had full access to all study data.
Financial Disclosure: Dr Dulcan receives book royalties from American Psychiatric Publishing, Inc. She sat on the Attention-Deficit/Hyperactivity Disorder and Reader Disorder and Strattera advisory boards of Eli Lilly and Company, most recently in 2009. She is a consultant for Care Management Technologies, Inc.
Funding/Support: This work was supported by grants R01DA019380, R01DA022953, and R01DA028763 from the National Institute on Drug Abuse; grants R01MH54197 and R01MH59463 from the National Institute of Mental Health (Division of Services and Intervention Research and Center for Mental Health Research on AIDS); and grants 1999-JE-FX-1001, 2005-JL-FX-0288, and 2008-JF-FX-0068 from the Office of Juvenile Justice and Delinquency Prevention. Major funding was also provided by the National Institute on Alcohol Abuse and Alcoholism; the National Institutes of Health (NIH) Office of Behavioral and Social Sciences Research, Substance Abuse and Mental Health Services Administration (Center for Mental Health Services, Center for Substance Abuse Prevention, and Center for Substance Abuse Treatment); the NIH Center on Minority Health and Health Disparities; the Centers for Disease Control and Prevention (National Center for Injury Prevention and Control and National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention); the NIH Office of Research on Women's Health; the NIH Office of Rare Diseases, Department of Labor, Department of Housing and Urban Development; the William T. Grant Foundation; and the Robert Wood Johnson Foundation. Additional funds were provided by the John D. and Catherine T. MacArthur Foundation, The Open Society Institute, and The Chicago Community Trust.
Additional Contributions: Celia Fisher, PhD, provided invaluable advice on the project. Elena de la Garza Bassett, MS; Lynda Carey, MA; Jennifer Hershfield, MA; Kathryn Jacoby, BA; and Nichole Olson, MS, provided research assistance and editorial assistance. We thank our participants for their time and willingness to participate, as well as the Cook County Juvenile Temporary Detention Center, Cook County Department of Corrections, and Illinois Department of Corrections for their cooperation. We also thank Mark Heyrman, JD, and the anonymous reviewers, who provided insightful and incisive comments on earlier drafts.
Create a personal account or sign in to: