Our analysis of nationwide inpatient data from the US hospitals is a first attempt to understand the demographic predictors for conduct disorder-related hospitalization in adolescents and association with spectrum of SUD. Our study findings of male preponderance (64.4%) in the conduct disorder inpatients corroborate the findings reported by McCabe KM et al. [27] who found that boys are more likely to have childhood-onset conduct disorder than girls because they had a more severe risk profile. Two previous studies have also concluded that boys are two to three times more likely to be diagnosed with conduct disorder than girls [28,29]. In our study, we found a higher proportion of conduct disorder in adolescents aged 12–15 years (62.6%) which correlates with the study by Patel et al. [28] which found that 55% of the children admitted to hospital for conduct disorder were adolescents (12 to 18 years).

Regardless of the higher prevalence of conduct disorder in Whites, we found that African Americans and Native American/Asian adolescents have higher odds for hospitalization and an inpatient study conducted by Patel et al. [28] that studied the demographic predictors for conduct disorder-related hospitalization also showed similar findings that African Americans were two times more likely to be admitted for conduct disorder than other race/ethnicities. Few other studies have shown that African American children are more likely to be diagnosed with conduct disorder with a lifetime prevalence of 8.1% [30,31]. Self-reported and observed aggressive and externalizing behaviors are more prevalent among Black youths [32,33], which suggests that CD diagnoses are likely to be higher as well [34,35]. Two possible reasons have been suggested as explanations for this apparent disproportional involvement of Black youth in aggressive activities (other than bias in reporting of criminal involvement). One is the fact that Blacks are more likely than Whites to live in lower SES environments [36,37], which are characterized by greater exposure to aggressive models [38], and higher rates of CD [39,40]. A second reason is a fact that Black youths are exposed to an additional source of stress that seldom affects White youths, and that is racial discrimination [41]. This stress has been linked with both internalizing and externalizing behaviors in adolescents [42]. More generally, stress from discrimination has often been suggested as a central factor contributing to the pronounced disparity in health status that exists between Blacks and Whites in the US [43,44,45]. Also, children from low-income families have serious concerns in terms of conduct and attention functioning [28,29]. As per our study findings, lower median household income (<25th percentile) was associated with the higher prevalence of conduct disorder hospitalization (36.4%) and it decreases as the household income increases. D’Onofrio et al. [46] concluded that there is a causal association between family income and conduct disorder, and low family income is a critical risk factor for the development of early-onset conduct disorder.

Disney et al. [8] studied the relationships between ADHD, conduct disorder, and gender to SUD in a large population-based sample of adolescent twins. In this study, they concluded that a diagnosis of conduct disorder is an important predictor of SUD especially for cannabis, alcohol and tobacco abuse [8]. These findings coincide with our study that among all SUD, cannabis use disorder (23.7%) was most prevalent in conduct disorder inpatients followed by tobacco and alcohol use disorders (10.1% each). A study by Henry B et al. [47] shows that a strong association between conduct disorder and SUD emerged at age 15 for both males and females. Findings of Morse et al. [48] indicated that depressive symptoms mediated the relation between cannabis use and conduct disorder diagnosis.

Nearly 42% of the men diagnosed with conduct disorder met the diagnostic criteria for alcohol dependence [49]. In addition, nearly 23% percent of the women diagnosed with conduct disorder met the diagnostic criteria for alcohol dependence, high comorbidities exist between alcohol abuse/dependence and conduct disorder, particularly for the adolescent-onset subtype of conduct disorder [49]. There also exists an increased risk for a young person diagnosed with conduct disorder to develop an alcohol use disorder if one or both parents also had alcohol abuse [49].

A past history of conduct disorder increases the chances of substance use and abuse at age 16 [50]. In a study by Hopfer et al. [7] conduct disorder was associated with elevated adjusted hazards for initiation of all substances, with comparatively greater hazard ratios at age 15. By age 18, the adjusted hazard ratios remained significant except for alcohol. At age 21, the adjusted hazard ratios were significant only for cocaine, amphetamines, inhalants and club drugs [7]. A similar pattern was seen in our study for ages 16 to 18 years’ cocaine/amphetamine use disorders were prevalent followed by opioid, tobacco, alcohol, and cannabis whereas cannabis-use disorder was seen majorly in adolescents 12 to 15 years. Pacek et al. [51] showed that cannabis-use disorder was greatest among African Americans while cocaine-use disorder was more prevalent in the Whites, which is similar to our findings, as about half of African American adolescents abused cannabis (45.6%). However, in Whites, substances significantly abused were opioids (73.7%), tobacco (64.2%), and cocaine/amphetamine (61.7%). Another study states that among cannabis users, the odds of cannabis abuse and dependence were greater among African American, Native-Americans, and Hispanics than Whites [52].

A study by Patrick et al. [53] found that smoking in young adulthood was associated with lower childhood family socioeconomic status, while alcohol use and cannabis use in young adulthood was associated with higher childhood family socioeconomic status. By contrast, our study findings showed that adolescents from low-income families (below 25th percentile) abused majorly cannabis (42.9%) and alcohol (31.3%), middle-income families (51st–75th percentile) abused opioid (44.1%), and those from high-income families (above 75th percentile) equally abused alcohol, tobacco, cannabis, and cocaine/amphetamine. This variation could be due to the fact that we took into account only household income while the study by Patrick et al. [53] included education, wealth and income altogether. We focused on the demographic characteristics of adolescents with conduct disorder by SUD in order to delineate the predictors for the rising risk of SUD in these at-risk populations. Fergusson and Horwood [54] found that after controlling on a wide range of predisposing factors, including gender, family socioeconomic status, IQ, and childhood conduct disorder, a direct association remained between early marijuana use and poorer educational attainment and unemployment. Because adequate education is needed for successful functioning in a variety of adult roles, especially employment, and studies have shown that early marijuana use may lead to poor education [55] early marijuana use’s association with later poor outcomes may be an indirect one through education. There is mounting evidence that gender plays such an essential role in drug use. Men and women have different risk factors, opportunities to use, and rates of use [56,57]. Men and women also experience different social issues when it comes to drug use. For example, though substance use, in general, is stigmatized, women who use drugs encounter more social disapproval than men [58] a study [59] concluded that drugs are more likely to affect a man’s job or career path, whereas women are more likely to experience family problems, as evidenced by, for example, higher rates of divorce in substance-using women than men [60]. This selection and the consequences of early drug use may vary for males and females. A study [61] found that family factors influence the onset of drug use, especially among females.

Interestingly, for the illegal substances of cannabis, cocaine, and heroin, we found a consistent association with family factors for females only and no influence of family factors for males. Social implications of substance use include employment, marriage, and parenting. Also, success in employment and family relations have been linked to better long-term health, and high school graduation is a potential target for drug intervention programs [62].

Similarly, programs that focus on preventing or reducing teen substance use may increase the likelihood of high school graduation, diminish adult drug use, and have benefits about the transition to adult roles. Therefore, it seems necessary that those who design school dropout programs and drug prevention programs consider focusing on all the risk factors. Wealthier individuals have easier access to substances, both because of their financial means and their social activities [63].

On the other hand, low-income individuals have higher levels of stress, which leads to higher rates of drug and alcohol use. It is possible that the factors causing stress among wealthy and middle-class Americans—such as financial concerns, relationship conflicts, or parenting issues—are as influential and harmful as the factors that cause anxiety or depression among the poor. The lack of health insurance and financial resources is cited as a barrier to treatment in national surveys [63].

Research has indicated that there are several common pathways through which children and adolescents develop conduct disorder, each with different risk factors and each with different underlying developmental mechanisms leading to the child’s aggressive and antisocial behavior. Future research should consider whether other pathways could also explain the problems experienced by a significant number of youths with conduct disorder. It is important to note that this way of conceptualizing the development of conduct disorder has significant implications for how such research is conducted. Specifically, research should no longer focus simply on documenting what risk factors are associated with conduct disorder or which risk factors account for the most or the unique variance in measures of antisocial behavior, aggression, or delinquency. Such methods assume that conduct disorder is a unitary outcome. For future interventions, one key implication is the importance of prevention. As noted previously, the most aggressive youth and the youth most likely to continue their antisocial behavior into adulthood tend to show a childhood onset to their antisocial behavior [64].

Furthermore, several interventions have proven effective in treating early emerging conduct problems, with a significant decrease in their effectiveness in older children and adolescents [65]. Thus, intervening early in the developmental trajectory of childhood-onset conduct problems is an essential goal for preventing later severe aggression and antisocial behavior. Secondly, interventions need to be comprehensive and target multiple risk factors. Also, interventions need to be not only comprehensive but also individualized. That is, given that the causal processes leading to antisocial behavior appear to be different across subgroups of youths with conduct disorder [55]. Unfortunately, there is only very minimal research testing the utility of this matching of individuals with conduct disorder to different types of treatment depending on their unique characteristics, and this is a clear need for future research.

The key strength of this study is the national representation provided by the NIS dataset, with a uniform collection of data using ICD-9 diagnoses codes, and its large sample size of 800,614 adolescent inpatients. This study includes children with a psychiatric diagnosis of conduct disorder versus those without conduct disorder. The identification of patients using the ICD-9 diagnosis code may be affected by some external factors such as insurance and billing. Also, due to the cross-sectional nature of study we are not able to delineate a causal relationship between conduct disorder and SUD. Another major limitation is lack of information about the SUD patterns of consumption (e.g., weekly consumption, years of consumption, and polysubstance use). However, we included patients with a discharge diagnoses of SUD, and the clinical and non-clinical information in this dataset are coded independently of the individual practitioner and thus it is subjected to minimal reporting bias. Nevertheless, our study does have a few limitations as it is an administrative dataset, including lack of information of criminal history and other comorbidities including traumatic brain injury. Re-hospitalizations, which add to the total inpatient burden, was not accounted for in our study. However, despite these limitations, NIS is still an excellent population-based representation of disease associations with comorbidities.