All the one hundred and forty-seven (147) participants were males, which is a direct consequence of the single-sex (only males) Borstal care system being practiced currently in Nigeria [5, 23], unlike what obtains in other parts of the world where there are facilities for both sexes [4, 7, 14]. The mean age of 17.1 ± 1.1 years agrees with what had been previously reported in forensic settings in Nigeria [5, 23] and majority (86.7 %) are from Yoruba ethnic group which is probably a reflection of the location of the facility.
Our study showed that more than half (56.5 %) of the incarcerated participants had CD; a rate comparable to those observed across Europe and America, in systematic review of 25 surveys (56.5 vs 52.8 %) , but slightly lower than what was reported by Adeguloye (64.2 %)  in Ilorin, North Central part of Nigeria. The current rate may either be a consequence of regional differences within the country or a result of the slightly larger sample size in our study, which obviously reduces the proportion of the actual number of in-mates meeting the diagnosis of CD. Nonetheless, it follows the trend that had been shown by earlier studies [4, 5, 7, 12, 14], which indicate that CD is several times higher in residential care settings than in the general population. In the current study for example, the prevalence rate (56.5 %) is three times as high as was reported in school children (15.6 %) in Nigeria .
The specific correlates of CD in this study revealed some interesting but disturbing trends. Although, with multiple regression analysis, older age did not contribute significantly to the prediction of CD (OR 821; p = 0.756; 95 % CI 0.236–2.851), we found a significant relationship between older and CD (χ
2 = 4.164, p = 0.041). On this, our study diverges from the findings of Adegunloye , in a different part of the country and an American study, both of which indicated an association in the opposite direction [5, 24]. It is possible that variables such as duration of stay (54.4 % had spent >12 months) or repeated incarceration might have contributed to an older age of the affected adolescent in this study.
Religiosity (described as frequent participation in religious activities) with frequent teaching of retributive justice may lower the chance of being involved in antisocial practices, whereas, a lower level of religiosity might predispose to vices and criminal behavior, with consequent delinquency and ultimate incarceration [18, 25]. In our study, neither the type of religion (χ
2 = 1.661, p = 0.197) nor religiosity (X = 0.002, p = 0.881) was significantly associated with a lower prevalence of CD. This was unlike the studies conducted in school children [18, 25] but in accord with those conducted among incarcerated juveniles [5, 12] where religious participation was not shown to be protective. This might suggest that religiosity was only associated with CD among non-incarcerated young population rather than in forensic settings wherein participation in religious activities might be an imposition rather than choice.
While our study does not produce evidence that variables such as marital status and ‘who brought the participant up’ were independent predictors of CD, initial association of these variables with CD on bivariate analysis perhaps suggests the need for further investigation of this lack of association. Studies have shown that family dysfunction, parental loss and parental deprivation ultimately produced an attendant loss of a sense of submission to authority figures (primarily, the parents) since they were absent [26, 27]. This is because, family is the foundation of human society and is the strongest socializing force of life. “The family teaches children to eschew unacceptable behavior, to delay gratification and to respect the right of others” . Establishments such as schools, religious bodies, and the law enforcement agency, though acting as surrogates, cannot substitute an intact or healthy family [17, 26]. Furthermore, the absence of an intact family, either as a result of dysfunction or loss, (by death or divorce/separation) may lead to emotional and material deprivation.
In many African cultures, whenever there is divorce or separation, the mother is often the one to leave the house regardless of the children’s interests . These children are sometimes left to cater for themselves and the very absence of material resources may lead them into wandering in order to beg for alms or resorting to stealing as means of surviving.
More importantly, the age at which a parent is lost is very crucial and significantly related to developing conduct behaviors . Indeed this variable demonstrated a trend of significance in the regression model (p = 0.065; OR 3.259; CI 0.927–11.450) and was clearly significant in the bivariate analysis with a significantly higher proportion of these incarcerated adolescents who met the criteria for conduct disorder, reporting losing at least one of their parents before their fifth birthday (χ
2 = 4.362, p = 0.037). This age coincides with the period when a child internalizes the family values and develops a conscience, i.e., the “superego” . It has been proposed to serve as “agency that provides ongoing scrutiny of a person’s behavior, thought, and feelings; makes comparison with expected standards of behavior and offers approval or disapproval” . A defect in the establishment of conscience or in internalizing the family values may be responsible for the development of delinquency in various ways which include; loss of a sense of submission to authority figures including parents and loss of social control, which could eventually give way to frank criminal behavior. Moreover, it had been reported that living with other relatives for up to 1 year before primary school age is associated with delinquency .
Notwithstanding that family setting does not explain any variance in the regression model (OR 0.869; p = 0.852; 95 % CI 0.199–3.793), a signifiant proportion (68 %) of those who met the criteria for conduct disorder were from polygamous homes (χ
2 = 5.276; p = 0.022).
Polygamy, which is a common practice in Africa  and many Islamic states , is often characterized by adverse conditions such as; unhealthy rivalry among co-wives and among half siblings, most often as a result of disproportionate display of attention or unequal distribution of resources. These may eventually lead to family breakdown and resultant departure of children from such homes to the street where they are exposed to bad influences leading to delinquency.
In Nigeria, family setting (monogamous or polygamous) is a major determinant of the family size, with polygamous families being more likely to be larger than monogamous ones. Previous authors have technically defined large sibship as having five or more siblings [12, 29] and this definition was adopted for the purpose of this study. In this setting, our observation reveals that, family size is 4.6 times as important as family setting in predicting CD, hence individuals from homes with large sibship are 4.6 times more like to develop CD than those from homes with small sibship (OR 4. 630; 95 % CI 1.433–14.964).
Large sibship, which may be a function of polygamous family setting, is often uncommon in most developed world, where polygamy is either not encouraged by the law or not a usual practice. This could largely explain the reason why this variable was not explored by most of the studies from Europe and America [4, 6, 7, 30]. Nevertheless, our study agrees with those of some authors [16, 17], who found an association between large family and conduct behaviors.
Among these people, conduct behaviors may have been caused by the dilution of family resources such as funds and accommodation, where finite family resources are allocated thinly across a larger number of children. The inadequate resources may not only be in form of monetary or material possessions, but lack of other nonmaterial resources such as adequate time and attention spent on each child, helping them with schoolwork, providing encouragement and positive reinforcement. These nonmaterial resources are equally crucial for children’s development, because, they are positively related to their moral and social development [17, 28].
Finally, another variable with significant contribution to the prediction of CD was ‘Previous history of incarceration’ (p = 0.043). Juveniles who were repeatedly incarcerated were about five times more likely to develop CD OR 4.99; 95 % CI (1.048–23.846).
This is consistent with previous studies [12, 14, 19] which have indicated that a significant proportion of those with CD demonstrated some level of recidivism. This may further support an earlier argument that a mental disorder of a behavioral type was a major contributor to juvenile re-offending and such antisocial trends could continue except affected adolescent offenders were systematically investigated and properly managed [4, 5, 9, 11]. As opined by Alemika and Chukwuka , in their study among juvenile offenders in Nigeria, more efforts are being channeled towards keeping young offenders/disturbed adolescents off the community rather than rehabilitating them. Most often these efforts only incubate more delinquent behaviors, rather than properly addressing them. Some of these incarcerated but ‘untreated’ juveniles are even introduced to substance use [2, 19, 20, 31] and more dangerous vices while in incarceration . This may explain why they go back into offending or sometimes become more hardened, leading to more serious offences. Moreover, about 97 (66.0 %) of these incarcerated adolescents, who were in for noncriminal reasons, which included, wandering and labelled, ‘beyond parental control’ stand the risk of being exposed to criminal vices. Nevertheless, the meta-analysis of DeSwart et al. had clearly indicated that the implementation of evidence-based treatment in residential care could have a more positive effect on behavioral problems . This is however challenging to the low or middle income countries such as Nigeria, where there are no separate facilities for the juvenile offenders and those merely in need of protection [5, 15].
A modest proportion of the variance (23.6 %) in CD occurrence was explained by the two independent predictors, which perhaps suggest that other factors related to the development of CD were not investigated in this study. These factors include some biological and other psychological factors which may form the subject for further research.
Based on our study findings, there is the need for well-planned psychosocial interventions for adolescent offenders rather than mere incarceration.
On the basis of these, we therefore recommend a collaboration between the mental health providers and the juvenile justice system in adopting a more productive interventional method in handling disturbed children. It is equally important to develop social support strategies as is obtainable elsewhere in the world [4, 32], for children with dysfunctional family. These strategies should involve comprehensive early interventions such as improved parental supervision which will serve potent preventative functions.
Furthermore, more intensive media programs, in creating awareness and training of parents on early identification of this disorder are recommended. This approach has been shown to be an effective way of reducing conduct problems before antisocial behavior crystallizes . Lastly, future research is necessary to investigate how these highlighted risk factors and other psychological/biological factors interact with CD and its severity particularly in Nigeria.
Our study has some important limitations which should be addressed. Since it was not feasible to interview the participants’ parents or care giver, the data obtained were subject to the reliability of participants’ self-report of symptoms. It is therefore not possible to rule out the likelihood of over or under reporting symptoms particularly among those with conduct problems.
The generalizability of our study across gender is clearly limited, owing to the single sex nature of our sample which is a direct result of the legal provisions regarding Borstal Institutions in Nigeria .
The cross-sectional nature of this study clearly limits its ability to establish the direction of causality in the relationship between identified risk factors and CD.
Conversely, the strengths of this study lie in its use of an internationally accepted instrument, which has also been used in this environment with comparable results  and the use of a larger sample size than had been used in the country by previous researchers [5, 23] as well as robust attempt at clearly delineating specific independent predictors of CD from its myriad risk factors.