Abstract
The rise in the U.S. prison population over the past 40 years has heightened scrutiny of the incarceration of children and adolescents. Correlates of later reincarceration in this group, especially correlates relating to psychiatric and substance use disorders, are understudied in the U.S. population. We aimed to establish the prevalence and correlates of the reincarceration as adults of people incarcerated before age 18. Data were derived from clinical interviews and from validated diagnostic and psychometric instruments. They were obtained as part of a cross-sectional representative survey of the civilian U.S. population, the National Epidemiological Survey on Alcohol and Related Conditions (NESARC-III). We identified 1,543 adults (4.3% of the NESARC sample) who had been incarcerated before they were 18. Of these, 55.9 percent had subsequently been incarcerated as adults. In addition to variables that have been repeatedly identified in criminological research (less education, past antisocial behavior, and parental imprisonment), substance use disorder, bipolar disorder, and longer childhood incarceration were independently associated with incarceration as an adult. The possibility that psychiatric treatment could reduce reincarceration in this group warrants longitudinal and experimental research.
Press coverage of the incarceration of children and adolescents1 comes in the context of long-standing concerns over the personal and societal cost of the threefold increase in U.S. incarceration rates since the 1980s.2 Although most of this increase took place prior to 2010, the further detention of many of these children and adolescents when they become adults raises questions concerning the effectiveness of deterrence and the unwanted effects of incarceration. From a psychiatric perspective, it raises the question also of the extent to which mental ill-health contributes to the risk of subsequent incarceration.
There are few data describing the psychiatric risk factors that are associated with the reincarceration of children as adults. Research on criminal recidivism suggests that parental criminality,3 a genetic predisposition to antisocial behavior, neuropsychological deficits, and adverse family and neighborhood environments in childhood are likely to be important.4 Those adverse environments include low socioeconomic status, parental conflict, harsh discipline, low levels of parental support, and high neighborhood levels of crime.5
The roles of cause and effect can be difficult to distinguish in statistical associations6,7 but there are reasons to suspect that psychiatric factors are important precursors of reincarceration also. Children and adolescents in detention represent a vulnerable group with high rates of trauma, sexual abuse, suicidal ideation, and substance use.8 Lifetime incarceration rates for people with any DSM-5 mental disorder are nearly five times higher than those of people without such a condition,9 even controlling for substance use disorder.10 People who offend frequently, and hence could be expected to be incarcerated more, have higher rates of psychological problems.11,12
The role of depression, although seemingly limited, appears to be complicated. One study that showed no difference, in terms of mental health, between high-rate adolescent offenders who recidivated as adults and those who did not also showed higher levels of “depression/anxiety” to be associated with lower rates of violent recidivism.13 The relationship between mental disorder and offending risk can vary with the temporal trajectory of that offending. Reising and colleagues found mental health problems to be associated with offending that persists into adulthood, but not with offending that ceases in adolescence.14
Researchers continue to point to the benefits of identifying groups of offenders at high risk of reoffending in order to focus services on them.15 In this study we included as variables both known social risk factors for reincarceration and factors related to mental health, such as reliably ascertained diagnosis and a history of mental health treatment. The sample consisted of participants in a large cross-sectional representative survey of the U.S. population who described being incarcerated before the age of 18. We identified those who described subsequently being reincarcerated as adults. We describe the rate and correlates of reincarceration both overall and separately for subjects with extended (1 week or more) and brief (less than 1 week) childhood incarceration.
Methods
Data Source and Study Sample
The data were collected in the third iteration of the National Epidemiological Survey on Alcohol and Related Conditions (NESARC-III), a cross-sectional representative survey of the civilian U.S. population sponsored by the National Institute on Alcohol Abuse and Alcoholism (NIAAA).16 The NESARC-III conducted in-person interviews with U.S. adults, including the residents of group and rest homes, between April 2012 and June 2013. Ethnic minorities were oversampled to ensure adequate numbers for statistical analysis. Individuals who were institutionalized at the time of the study (in nursing homes, prisons, hospitals, or shelters) were excluded, as were active-duty military personnel.
The NESARC-III sample size was 36,309, and data were available on childhood incarceration for 36,293 subjects. Consent procedures were approved by the National Institutes of Health. All data were deidentified prior to their use in the present study. Details of the NESARC-III methodology have been published previously.10,17
Measures
The NESARC-III interview questions are publicly available (Ref. 18, Sections 1-18). The sample used here was generated using the response to the NESARC-III interview item, “Before you were 18, were you ever in jail, prison, or a juvenile detention center?” The dependent variable, incarceration as an adult, was generated using the response to the NESARC-III interview item, “Since you were 18, were you ever in jail, prison, or a correctional facility?”
The independent sociodemographic variables comprised age, sex, self-defined ethnicity, and marital status. Variables relating to a subject’s behavioral background comprised dichotomous questions concerning a parental history of alcohol or drug abuse, imprisonment, psychiatric hospitalization, suicide attempts, and completed suicide. The questions were “Before you were 18 years old, was a parent or other adult living in your home a problem drinker or alcoholic?”; “Before you were 18 years old, did a parent or other adult living in your home have some similar problems with drugs?”; “Before you were 18 years old, did a parent or other adult living in your home go to jail or prison?”; “Before you were 18 years old, was a parent or other adult living in your home treated or hospitalized for a mental illness?”; “Before you were 18 years old, did a parent or other adult living in your home attempt suicide?”; and “Before you were 18 years old, did a parent or other adult living in your home actually commit suicide?”
A history of childhood neglect or abuse was rated by adding the responses on separate five-point scales (endpoints “never” and “very often”) for each of the following questions, prefaced with “Before you were 18 years old”: “How often were you made to do chores that were too difficult or dangerous for someone your age?”; “How often were you left alone or unsupervised when you were too young to be alone, that is, before you were 10 years old?”; “How often did you go without things you needed like clothes, shoes or school supplies because a parent or other adult living in your home spent the money on themselves?”; “How often did a parent or other adult living in your home make you go hungry or not prepare regular meals?”; “How often did a parent or other adult living in your home ignore or fail to get you medical treatment when you were sick or hurt?”; “How often did a parent or other adult living in your home swear at you, insult you or say hurtful things?”; “How often did a parent or other adult living in your home threaten to hit you or throw something at you, but didn’t do it?”; “How often did a parent or other adult living in your home act in any other way that made you afraid that you would be physically hurt or injured?”; and “How often did a parent or other adult living in your home push, grab, shove, slap or hit you?”
The same five-point scale was used to rate sexual abuse, for which the items were each prefaced with “Before you were 18 years old”: “How often did an adult or other person touch or fondle you in a sexual way when you didn’t want them to or when you were too young to know what was happening?”; “How often did an adult or other person have you touch their body in a sexual way when you didn’t want to or were too young to know what was happening?”; “How often did an adult or other person attempt to have sexual intercourse with you when you didn’t want them to or you were too young to know what was happening?”; and “How often did an adult or other person actually have sexual intercourse with you when you didn’t want them to or you were too young to know what was happening?”
“Supportive family” was rated using the summed positive responses to five question that concerned a subject’s life before age 18 and that included “[I] felt I was part of a close‐knit family” and “My family was a source of strength and support.” The subject interviews also provided data on educational attainment (less than high school diploma, high school diploma or “GED,” college attendance but did not complete, completed college), military service and combat exposure, a history of homelessness (lifetime and before the age of 15), and the length of any incarceration in jail, prison, or juvenile detention center before the age of 18 (using the NESARC item “About how long altogether were you in jail or a juvenile detention center before you were 18?”). To create the variable used here, we dichotomized pre-18 incarceration at the median (one week).
Mental health diagnoses for mood disorders, anxiety disorders, posttraumatic stress disorder (PTSD), eating disorders, substance use disorders, and personality disorders were generated using the Alcohol Use Disorder and Associated Disabilities Interview Schedule (AUDADIS)-5,19 a structured interview that generates DSM-5 categories. Test-retest reliability for the diagnostic categories generated by the AUDADIS-5 ranges from fair to excellent and is similar to that for the DSM categories generated by other structured interviews.19 We used DSM-5 categories covering a subject’s lifetime.
In this analysis, as has been done elsewhere,20 we further used the response on two additional NESARC-III items: “In the last 12 months, did a doctor or other health professional tell you that you had schizophrenia or a psychotic illness or episode?” and “Did this happen before 12 months ago?” to generate a further variable, lifetime “schizophrenia/psychosis.” Subjects were asked whether they had ever attended treatment for mental health symptoms or substance abuse. They were asked whether they had sought help from members of clergy for drug-related problems; used self-help groups, such as Narcotics Anonymous, for drug treatment; or made a suicide attempt.
We used two variables to assess current supports and stressors. “Religion importance” was rated (from 1-4) from the question “How important is religion to you?” Higher scores corresponded to greater importance. Social support was assessed using the Interpersonal Support and Evaluation List-12 (ISEL-12), a 12-item instrument with a potential range of 12-48.21 Higher scores indicated greater social support. We assessed subjects’ functioning at the time of interview using current income and employment. Employment was evaluated using the response to a question addressing a variety of nonmutually exclusive work experiences in the last 12 months (work full time, disabled or unemployed, retired, employed part time, and unemployment).
Procedure and Statistical Analyses
In NESARC subjects who described having been incarcerated before the age of 18 (n = 1,543; henceforth, “the sample”), we examined the rates and bivariate correlations of being reincarcerated after the age of 18 and, using NESARC data weighted to adjust for nonresponse, generated effect sizes (risk ratios and Cohen’s d) for reincarceration. We focused on effect sizes because significance testing (generating P values) is less informative with large sample sizes where small, unimportant effects can be statistically significant.22
We entered all variables with substantial effect sizes on bivariate analysis (criteria: risk ratio > 1.5 or <.67; Cohen’s d > .2 or <-.2)23 along with length of childhood incarceration (less than one week, the median duration of incarceration, versus longer periods) into multivariate analyses to identify factors independently associated with adult reincarceration. We thus studied the role of length of childhood incarceration in three ways: bivariate analysis, multivariate analysis, and by examining the rates and correlates of reincarceration in brief and longer term childhood incarceration subgroups separately. All statistical tests were performed using the statistical software SAS version 9.4.
Results
Description of Sample
People incarcerated before age 18 (n = 1,543) comprised 4.3 percent of NESARC subjects (N = 36,293). Extrapolated to the 2020 U.S. census (population 18 or over = 258,343,28124), they thus represent over 10 million adults living in the United States.
The mean age at interview was 42.6 years; 72.9 percent were male, 64.2 percent were white, and 16.1 percent black; 48.7 percent were married; 27.8 percent had not graduated from high school and 33.6 percent had graduated from high school, but not attended college. For 19.2 percent, one or both parents had problems with drug use, and for 27.3 percent, one or both parents had spent time in prison. The sample had experienced more child neglect (16.9 versus 12.1) and child sexual abuse (4.98 versus 4.41) than nonincarcerated NESARC subjects.
Not including personality disorder or substance use, 48.6 percent met criteria for any lifetime psychiatric disorder and 64.3 percent met criteria for any lifetime substance use disorder diagnosis. A total of 16.4 percent had made a suicide attempt.
Analysis of Entire Sample
Of the 1,543 subjects, 863 (55.9%) were subsequently incarcerated as adults (see Table 1, columns A and D). On bivariate analysis, subsequent incarceration as an adult was substantially associated both with social factors from subjects’ background (parental drug use and imprisonment, not completing college, not having been exposed to military combat, childhood and lifetime homelessness, and not being widowed or retired) and with both aggregated and specific variables related to their mental health (schizophrenia or psychosis, bipolar disorder, eating disorder, panic disorder, schizotypal and antisocial personality disorder, any drug use diagnosis and five specific drug use disorders (marijuana, opioid, cocaine, sedative, and stimulant)). Finally, on bivariate analysis, reincarceration was also substantially associated with reports of attending substance use treatment as well as having contacts with clergy and self-help groups to address one’s problems (see Table 1, columns A, D, and the first of the columns labeled effect size). Shorter term childhood incarceration was associated with lower risk of reincarceration with a risk ratio of .71 (42.3% versus 57.6%).
Correlates of Subsequent Incarceration (Bivariate; N = 1,543)
On multivariate analysis, reincarceration in the entire sample of 1,543 was independently associated with variables from the subjects’ background (a parental history of imprisonment, having been incarcerated for a week or more in childhood, and not having graduated from college) as well as with variables relating to psychopathology (antisocial personality disorder, bipolar disorder, and substance use disorder as evidenced by more than one substance use diagnosis or by reporting having attended substance use treatment; see Table 2).
Correlates of Incarceration in Adulthood (Multivariate)
Analysis of Subgroups
Of the subjects who had been incarcerated as children for less than a week, 48.2 percent (365 of 757) went on to be incarcerated as adults. In contrast, 63.4 percent (498) of the 786 subjects who had been incarcerated as children for a week or more went on to be incarcerated as adults (Table 1, columns B, C, E, and F). On bivariate analysis, risk factors for later incarceration among those who had been incarcerated for less than a week included serious mental illness (schizophrenia or psychosis and bipolar disorder; see Table 1 columns B and E and the second effect size column), whereas risk factors for later incarceration among those who had been incarcerated as children for longer than a week included failing to complete college and lower income (see Table 1, columns C and F and the third effect size column). For both subgroups, however, multivariate analysis pointed to the over-riding importance of substance use as a substantial correlate of future incarceration (see Table 2).
Discussion
Overview
In a large and representative sample of the U.S. population, 1,543 adults (4.3%) reported having been incarcerated before they were 18. Of these, 55.9 percent had subsequently been reincarcerated as adults. The risk of adult reincarceration increased with time spent incarcerated as a child (48.2% for less than one week and 63.4% for incarceration that lasted longer than that). In addition to variables that have been repeatedly identified in criminological research (lack of education, past antisocial behavior, and parental imprisonment), substance use disorder, bipolar disorder, and longer childhood incarceration were independently associated with incarceration as an adult. Restricting the analysis to subjects whose childhood incarceration had lasted for more than one week did not identify additional independent correlates of incarceration in adulthood.
The population prevalence of adolescence-limited and life course persistent antisocial behavior in the United States has been estimated at 11.6 percent and 7.4 percent, respectively, for males and 11.4 percent and 6.9 percent, respectively, for females.25 The 680 who were incarcerated as children, but not as adults, and the 863 who were incarcerated both as children and adults represent, respectively, 1.9 percent and 2.4 percent of NESARC subjects. That these proportions are so much smaller than those reported for antisocial behavior overall likely reflects the fact that most antisocial behavior does not result in incarceration. As suggested in the Methods section, even compared with other young people who engage in antisocial behavior without being incarcerated, subjects in this study sample represent a high-risk group.
The correlates of adult incarceration that we report here differ in some respects from those reported elsewhere. Although males and members of black and minority ethnic populations are over-represented in incarcerated populations,26 sociodemographic factors were not associated with reincarceration in this sample. This may be because the sample excluded those currently incarcerated, who remain disproportionately male and black. Although some aspects of a disadvantaged upbringing, namely parental drug abuse and parental imprisonment, were associated with subsequent incarceration (and parental imprisonment independently so on multivariate analysis), others, including child neglect or abuse, child sexual abuse, and a subject’s perception of the lack of a supportive family in childhood, were not. It is unlikely that seeking treatment for substance use, including from clergy and self-help groups, is itself criminogenic; this association is likely explained by seeking treatment acting as an indicator of substance use as well as other problems that would lead a respondent to seek treatment.
Future Research and Policy Implications
Previous research has contrasted persistent antisocial behavior and reincarceration with desistance, a decline in antisocial behavior that has been consistently observed to be a feature of aging and maturation, particularly in males.27 Although cross-sectional data such as these do not allow causal inferences, these results are consistent with different mental health diagnoses having different effects on desistance.
Schizophrenia or psychosis and bipolar disorder, the only diagnoses studied here that focus on psychotic symptoms, were associated with reduced desistence and with the largest effect sizes on bivariate analysis. In the case of bipolar disorder, the association with reduced desistance was also present on multivariate analysis. We are not aware of this association being demonstrated previously. It may be that impaired psychosocial function, a known correlate of psychosis in this sample,20 disrupts family and social networks that would otherwise protect against legally significant outcomes. If that is the case, effective treatment of psychotic symptoms, in addition to improving health, might have prevented the criminal behavior that led to reincarceration. The possibility warrants investigation in studies with longitudinal designs.
Treatment of substance use disorder has already been shown to promote desistance.28,29 Future research should focus on the best ways of making effective substance use treatment available to this high-risk group, perhaps most effectively in the community prior to childhood incarceration.30 The point of release is already known to be a time of vulnerability, particularly with regard to opiate overdose,31 and these results suggest that coordinating community substance use services when people reenter the community from prison may also have the longer term benefit of preventing further incarceration.32
Limitations
Several aspects of the Methods warrant caution regarding these results. First, this is a cross-sectional view of individuals participating in processes, such as reincarceration and desistance, that are by their natures longitudinal. Although most cases of serious mental illness and substance use disorder are present by age 18,33 any causal inferences should await the confirmation of these findings in studies with longitudinal designs. Second, the NESARC-III did not include people who were incarcerated at the time the subject interviews were conducted or those who were living in nursing homes, hospitals, and shelters. The exclusion of those in jails and prisons will have lowered the rates of reincarceration reported here and may have modified the observed correlates of that reincarceration. The exclusion of those in other residential settings may have lowered the prevalences that we found for medical illness, substance abuse, and other forms of mental disorder. Future research seeking to represent the U.S. population should seek to include people in all of these settings.
Third, although the interview items used here have been tested and found reliable, they are based on self-report and did not include collateral data. Some items might have been worded differently had they been written with the present study in mind. The NESARC inquiry regarding adult incarceration, for instance, “Since you were 18, were you ever in jail, prison, or a correctional facility?” does not preclude the possibility that the subject’s incarceration had commenced prior to their becoming 18 and, therefore, represented a continuous period of detention, not reincarceration. We would note, however, that no additional findings emerged when we restricted the analysis to subjects detained in childhood for less than a week, a restriction which will have excluded most of these “childhood into adulthood” cases.
Fourth, beyond looking separately at subjects who experienced brief and longer term childhood incarceration, we have not sought to repeat the analyses for subgroups, for instance, men and women. Fifth, although the NESARC database suited our focus on mental health variables, it was not designed to cover the full range of factors known to affect the risk of incarceration and reincarceration. For instance, the database does not include data on neuropsychological deficits or a subject’s neighborhood environment in childhood.29 Also, although we were able to study both child sexual abuse and to include a measure of neglect that addressed physical abuse, detained youth and incarcerated adults have high rates of many kinds of trauma. These variables and others that have been linked to incarceration, such as attention deficit hyperactivity disorder (ADHD), should be included in future studies that might usefully also generate separate variables for neglect and physical abuse.
Conclusion
This is the first study of which we are aware using population survey methodology to describe the mental health correlates of reincarceration as adults of people who were incarcerated as children. The findings include a previously unrecognized statistical association between psychosis and subsequent incarceration as an adult. This association warrants testing in studies with longitudinal designs.
Footnotes
Disclosures of financial or other potential conflicts of interest: None.
- © American Academy of Psychiatry and the Law
