Abstract
The purpose of the current study was to examine the potential mediating effects of internalizing and externalizing problems at ages 14, 16 and 18 between types of childhood maltreatment and alcohol and marijuana use problems and disorders in young adulthood. Data were from 473 young adults who participated in the Longitudinal Studies of Child Abuse and Neglect (LONGSCAN). Path analysis was conducted to examine pathways between maltreatment type (birth through age 12), internalizing and externalizing problems at three time points during adolescence, and alcohol and marijuana problem use in young adulthood. Findings indicated significant pathways between physical abuse and internalizing problems at 14, which was associated with alcohol-related substance use disorder in adulthood. Externalizing problems and internalizing problems at age 14 mediated the relationship between physical abuse and marijuana-related symptoms in young adulthood. Emotional and sexual abuse were not associated with substance use problems. Implications for practice are discussed.
Child maltreatment (CM) is a major public health problem, with significant costs to individuals, families and society. In 2017, 7.5 million children were reported to Child Protective Services (U.S. Department of Health & Human Services, 2019). The consequences of CM are significant in both the short and long-term. A relatively consistent finding is the association with substance use (SU) problems in adulthood (Dubowitz et al., 2019; Fergusson et al., 2013; Herrenkoh et al., 2013; Lo & Cheng, 2007; Widom et al., 2006). Given the consequences of problem and disordered SU, it is important to identify mechanistic pathways and sensitive periods for intervention, particularly for maltreated children.
Neurobiological models coupled with a developmental psychopathology perspective have been posited to explain the association between early maltreatment and young adult SU, suggesting that the impact of childhood adversity on neurobiology increases vulnerability to psychopathology, contributing to later SU (Andersen, 2019; Cicchetti & Handley, 2019; De Bellis, 2002; Hovdestad et al., 2011; Tarullo & Gunnar, 2006; Teicher et al., 2003). Related data suggests that internalizing and externalizing problems might be important mediators linking CM and SU (Douglas et al., 2010; Jones et al., 2013; T. L. Lewis et al., 2011; Oshri et al., 2011; Proctor et al., 2017). For example, one study found externalizing problems at age 12 mediated the link between child sexual abuse and alcohol use at age 14 for girls (Jones et al., 2013). Another study found externalizing problems at age 8 mediated the link between sexual abuse and neglect and age of initiation of alcohol and marijuana use during adolescence (Proctor et al., 2017). Lewis and colleagues found internalizing problems at age 12 partially mediated the link between CM and smoking at age 16 (T. L. Lewis et al., 2011). Although these studies suggest internalizing and externalizing behaviors may be pathways to adolescent SU, their role in young adult SU is unclear. Few empirical studies have examined the impact of adolescent SU among young adults with a history of CM. Extant literature yields inconsistent findings linking CM type and problem SU, and between internalizing problems and problematic SU. Further, it is unclear whether pathways differ as a function of CM type and whether there are different points during adolescence that are salient.
Child Maltreatment and Young Adult Substance Use (SU)
The bulk of extant literature supports a link between child adversity, including CM, and young adult SU problems (Anda et al., 2006; Elliott et al., 2014; Huang et al., 2011; Shin et al., 2013). However, there is variability in what type(s) of CM is assessed and inconsistency in findings. One study found that emotional abuse and neglect were associated with problem alcohol use, but physical and sexual abuse were not (Kisely et al., 2019). In contrast, Afifi and colleagues (2012) found associations between all forms of abuse and alcohol and marijuana use disorders. Lo and Cheng (2007) found associations between physical abuse and alcohol and marijuana abuse, but not sexual abuse. Wilson and Widom (2010) found that childhood neglect, but not physical or sexual abuse, was was linked to SU.
Internalizing and Externalizing Problems and Substance Use (SU)
There is a robust link between externalizing problems and problem SU (Farmer et al., 2015; Guo et al., 2001; King et al., 2004; Stone et al., 2012). However, the association between internalizing problems and later SU is mixed (Hussong et al., 2017; O’Neil et al., 2011; Scalco et al., 2014; Stone et al., 2012), as is the direction of association (Colder et al., 2017). For example, a review found internalizing problems increased risk for youth SU (O’Neill et al., 2011). Similarly, Woodward and Fergusson (2001) found anxiety disorders in adolescence increased the risk of substance dependence in young adulthood. However, others found no association between internalizing problems and SU in adolescents (Miettunen et al., 2014), and one study found found that internalizing behavior at age 14 was inversely associated with marijuana use among young adults with a history of CM (Mills et al., 2016). Across these studies, the assessment of internalizing problems varies widely. Some studies focused on specific disorders such as anxiety and depression, while others focused on internalizing problems more generally. The time frame of assessment also varied considerably across developmental periods as well as the temporal distance from the outcome. Finally, many studies were cross-sectional or retrospective (Dube et al., 2006; Glover et al., 1995. Johnsen & Harlow, 1996), limiting conclusions about causal pathways.
Adolescence as a Risk Period
Adolescence may be a critical period in which internalizing and externalizing problems heighten risk for managing exposure to and use of substances particularly for maltreated youth (Andersen, 2019; Cicchetti & Handley, 2019; De Bellis, 2002; Stone et al., 2012). Adolescence is a challenging developmental period with major transitions, biological changes, and social challenges including exposure to substances. These challenges change with different influences at different periods (Cicchetti & Rogosch, 2002; Compas et al., 1995). Regarding behavior problems, some studies suggest problems are fairly stable during adolescence (Deković et al., 2004), others suggest variability over time (Bor et al., 2014; Compas et al., 1995). T. Lewis and colleagues (2016) examined internalizing and externalizing problems from age 6 to age 16 in sexually abused, maltreated but not sexually abused, and non-maltreated youth. Overall, externalizing behaviors declined over time in all groups, while internalizing problems increased in the two maltreated groups. In studies to date, the timing of the assessment of internalizing and externalizing behaviors during adolescence varies considerably despite significant changes in stage salient challenges during this period. This limits understanding of whether internalizing and externalizing problems during adolescence are associated with later problematic SU (Mason et al., 2009). Longitudinal studies with assessments of problem behaviors at multiple times during adolescence may elucidate specific periods of vulnerability and inform timing of intervention efforts.
The current study addresses the noted gaps in the extant literature by utilizing prospective data with multiple assessments of internalizing and externalizing behavior throughout childhood and adolescence. We examined potential mediation by internalizing and externalizing behaviors at ages 14, 16 and 18, between CM (birth to age 12) and young adult problem SU. We examined these links as a function of CM type (physical abuse, sexual abuse, emotional abuse, neglect) and substance type (alcohol and marijuana), controlling for demographic factors and adolescent and parental substance use. Given evidence linking externalizing problems and SU, we hypothesized that externalizing problems would mediate the relationship between CM and young adult problem SU. No a priori hypotheses were made regarding type of CM, internalizing problems, and developmental period given the lack of or inconsistent empirical data.
Methods
Sample
The sample was derived from the Longitudinal Studies of Child Abuse and Neglect (LONGSCAN), a consortium of five prospective studies of the antecedents and consequences of CM (Runyan et al., 1998). The original sample of 1,354 children included families where CM had been identified or who were considered high risk for CM. The data are based on a follow-up survey examining participants’ SU as young adults. To be eligible for the follow-up, participants had to have had at least one interview at age 14, 16 or 18 and at least four interviews between ages 4 and 18. This yielded a subsample of 1,053 eligible participants. The data reported here are based on the 473 who completed the follow-up survey (“Completers”). There were a few statistically significant differences between Completers and those who did not complete the survey. There was an association between completion status and site (χ2 [df = 4] = 52.06, p < .001), with fewer Completers in the Southern (29.4%) and Southwestern (28.6%) sites. There was also an association between completion status and gender (χ2 [df = 1] = 29.14, p < .001). More males did not complete than did complete (63.0% vs 37.0%); no significant differences were found for females. CM, race/ethnicity, family income, and caregiver education, employment and marital status were not associated with completion status. The average age of completion of the young adult interview was 23.8 years (SD = 1.89; range = 19–29).
Procedure
LONGCSAN conducted face to face interviews with child participants and their primary carergiver every two years from approximately 4 years of age until age 18, between 1992 and 2015. Beginning at age 12, youth and caregivers completed separate interviews utilizing an audio computer-assisted, self-administered format (Runyan et al., 1998). Child Protective Services (CPS) data were obtained by each site from the local or state agency approximately every two years. A follow-up online survey was then completed by the young adults. Each site employed procedures approved by its Institutional Review Board. Families were paid a nominal amount for their time.
Measures
CPS and self-reports of child maltreatment
History and type of CM from birth to age 12 were assessed utilizing a combination of CPS referrals and youth self-reported history of maltreatment. CPS case narratives were abstracted and coded using the Modified Maltreatment Classification System (MMCS; Barnett et al., 1993; English & the LONGSCAN Investigators, 1997). The MMCS enables the standardization of CM type regardless of definitional differences among agencies and states. Because previous research has demonstrated that substantiated CM is no better at predicting outcomes than alleged CM (e.g., Hussey et al., 2005), any allegation to CPS from birth to age 12 was considered indicative of CM, regardless of substantiation.
Beginning at age 12, LONGSCAN incorporated youth self-report measures of specific CM experiences (Knight et al., 2008). Items from the Self-Report of Physical Abuse, Self-Report of Sexual Abuse and Assault, and Self-Report of Psychological Maltreatment were based on definitions developed by Barnett et al. (1993) and definitions of the American Professional Society on the Abuse of Children (Hart et al., 1996). The physical abuse measure includes 15 stem items (e.g., “Has an adult ever kicked or punched you?”). The sexual abuse measure includes 11 stem items (e.g., “Has anyone ever touched your private parts or bottom in some way?”). The psychological abuse measure includes 18 stem items (e.g., “Have any of your parents ever threatened to hurt you badly?”). Any endorsement was followed by asking if the event occurred before elementary school, since elementary school, and/or in the past year. Endorsement of any item over any of those time periods was considered indicative of experiencing that type of abuse. See Knight et al. (2008) for details of this measure.
Given the potential biases of self-report and CPS records, scholars have recommended the integration of data from both sources to optimally capture maltreatment experiences (Everson et al., 2008; Knight et al., 2000; Shaffer et al., 2008). Accordingly, both self-report and CPS allegation data were used to develop indicators for physical, sexual, and emotional abuse. Specifically, either a CPS allegation or self-report of a type of CM was coded as affirmative (=1), or “0” if neither source indicated a history of that type. Because there was no analogous measure of self-reported neglect, the indicator for neglect was based on CPS history alone. Finally, an overall indicator of CM was based on the presence of any CPS or self-report data for physical, sexual, emotional abuse or neglect.
Substance Use (SU)
Young Adult Problem SU
The National Survey on Drug Use and Health (NSDUH; Substance Abuse and Mental Health Services Administration, 2008) was used to assess young adult use of alcohol and marijuana. The NSDUH gathers information on the use of alcohol, tobacco, illicit and non-medical prescription drugs for determining DSM-V diagnoses of abuse and dependency. Utilizing scoring criteria from the NSDUH, two versions of the alcohol and marijuana criteria were created. First, we summed the endorsed DSM-V substance-related symptoms for each substance (e.g., drink more than intended) for “count outcomes” (range = 0–11). Second, DSM-V Substance Use Disorder categories were created as ordinal variables: No use this year, No disorder, Mild (2–3 criteria), Moderate (4–5 criteria), and Severe (6 or more criteria).
Adolescent SU
At age 14, self-reported use of a wide variety of drugs was assessed using the NIMH Diagnostic Interview Schedule for Children-IV (DISC; Shaffer et al., 1998). Two dichotomous indicators captured any use of alcohol or marijuana in the last year.
Caregiver SU
At child age 8, caregivers completed the Caregiver Substance Use form, a project developed measure for past and current use of legal and illicit drugs. We summed the number of endorsed substances for a count variable (range = 0–8).
Child behavior problems
Child internalizing and externalizing behavior problems were assessed with the Child Behavior Checklist (CBCL; Achenbach, 1991a) at ages 4, 14 and 16, and with the Youth Self-Report at 18 (YSR; Achenbach, 1991b). For the CBCL, caregivers indicate the extent to which each of 113 behaviors is characteristic of the child/youth. The YSR includes 112 analogous items. Scoring includes raw scores for two broad categories of problem behaviors (Internalizing and Externalizing). The CBCL and YSR have good psychometric properties regarding test-retest reliability, inter-rater agreement and validity (Achenbach, 1991a, 1991b).
Young adult demographics
Respondents provided information on employment, relationship status, educational attainment, and current income.
Data-Analytic Procedure
Path analysis for count and ordinal outcomes was used as the primary statistical model. All analyses were conducted with MPlus software (Muthén & Muthén, 2014). MPlus software (ver. 8.4) and MLR estimation with Monte Carlo Integration was used for the count outcomes (symptom criteria) and WLSMV estimation for the ordinal outcomes (substance problem use/abuse). These procedures produce unbiased parameter estimates and standard errors under various missing data conditions (Enders, 2010). All models included study site, child’s sex, child race/ethnicity, caregiver substance use, young adult employment status and relationship status as covariates. The relationship between each covariate and each mediator/outcome was specified with a direct path from a given covariate to a target mediator/outcome. The mediation models were tested individually for each early adult SU outcome. The four CM type variables from ages 0 to 12 (Physical Abuse, Sexual Abuse, Emotional Abuse, Neglect) were the antecedent variables. Variables representing the Internalizing and Externalizing mediators were tested in separate models, given the high degree of collinearity among these variables within each time-point (rs ranged from .62 to .72). Internalizing and Externalizing scores at each time-point (ages 14, 16, 18) were tested as mediators in the same model. The paths from (a) each CM type to Internalizing/Externalizing and (b) Internalizing/Externalizing to each early adult problem SU outcome represented the target mediated effects. The asymmetric confidence interval (CI) was used to formally test for mediation (Tofighi & MacKinnon, 2011); CIs that do not contain the value 0 are considered statistically significant mediated effects. These models also included first-order autoregressive effects for Internalizing/Externalizing to account for the stability of these variables across time. The models controlled for alcohol/marijuana use at age 14 in the prediction of each early adult problem SU by specifying a direct path.
The percentage of missing data varied by variable and time-point, the highest being for Internalizing and Externalizing at age 18 (25.8% missing) followed by age 16 (19.9%) and age 14 (14.4%). Missing data for the early adult problem SU variables was 8.5%. Statistically significant differences (ps < .01) were found for those missing (vs. not missing) the early adult SU symptoms with regard to Internalizing and Externalizing problems. Specifically, those missing the early adult alcohol and marijuana variables had significantly lower Internalizing scores at age 4 (Mmissing = 7.36 vs. Mnot missing = 4.94) and Externalizing scores at age 16 (Mmissing = 9.99 vs. Mnot missing = 6.55). No other statistically significant differences were found (all ps > .01). All variables for which a significant difference was found were controlled for in the models.
Results
Descriptive Statistics and Missing Data Analysis
Sample characteristics are presented in Table 1. Participants were primarily from the Eastern, Southwestern and Northwestern sites. The majority of participants were female, Black/African American, had a high school diploma/GED or education beyond high school, unmarried, and currently working for pay. The sample was evenly split on income (<$20,000, >$20,001). Descriptive statistics for target study variables are described in Table 2. Most participants reported no or minimal symptoms related to alcohol or marijuana use at age 14 and in early adulthood. Over half had a history of neglect (58%), followed by emotional abuse (53%), physical abuse (39%), and sexual abuse (18%). Rates of Internalizing and Externalizing problems are presented in Table 2.
Sample Characteristics.
Descriptive Statistics for Study Variables.
a Maltreatment types are not mutually exclusive. b CBCL internalizing and externalizing scores range from 0 to 62 and 0 to 66 respectively. YSR internalizing and externalizing scores range from 0 to 64 and 0 to 60 respectively.
Modeling Testing
The paths that comprise the mediation effect and the direct effect from the CM variables to the outcomes are presented below by early adult problem SU outcome.
Alcohol-related symptoms
Physical abuse (b = 0.42, p = .033) and neglect (b = −0.48, p = .012) were directly associated with early adult alcohol-related symptoms. Regarding mediation, physical abuse was associated with Internalizing problems at age 14 (b = 1.58, p = .048) and emotional abuse was associated with Internalizing problems at age 18 (b = 2.91, p = .001). In turn, Internalizing problems at age 14 was associated with alcohol-related symptoms in early adulthood (b = 0.04, p = .040), but Internalizing problems at age 18 was not (b = 0.01, p = .445). A formal test of mediation, however, did not confirm that Internalizing problems at age 14 mediated the relationship between physical abuse and early adult alcohol-related symptoms (MacKinnon’s 95% Asymmetric CI: −.002 to .15). Physical abuse was also associated with Externalizing problems at age 14 (b = 3.54, p = .001), but not with early adult alcohol-related symptoms (p > .05). Externalizing problems at age 18, however, was associated with early adulthood alcohol-related symptoms (b = 0.04, p = .008). In sum, mediation for early adult alcohol-related symptoms by Internalizing or Externalizing problems was not established.
Alcohol substance use disorder (SUD)
Neglect was directly associated with less likelihood of early adult alcohol SUD (b=−0.31, p = .020). Regarding mediation, physical abuse was associated with internalizing problems at age 14 (b = 1.98, p = .024), which was associated with early adult alcohol SUD (b = 0.03, p = .024), and mediated the physical abuse—early adult SUD relationship (MacKinnon’s 95% Asymmetric CI: .003 to .119; see Figure 1). Emotional abuse was associated with Internalizing problems at age 18 (b = 2.62, p = .021); the latter was not related to alcohol SUD (p > .05). Physical abuse was also associated with Externalizing problems at age 14 (b = 4.07, p < .001), but the path from Externalizing problems to early adult alcohol SUD was not significant (p > .05). No other significant effects were observed (all ps > .05).
Types of child maltreatment predicting alcohol substance use disorder in early adulthood, mediated by internalizing adolescent behavior problems.
Marijuana-related symptoms
No direct effects from CM to early adult marijuana-related symptoms were found (all ps > .05). For the internalizing problems mediated effect, physical abuse was associated with Internalizing problems at age 14 (b = 1.59, p = .047) and emotional abuse with Internalizing problems at 18 (b = 3.05, p < .001). In turn, Internalizing at age 14 (b = 0.05, p = .025) but not at 18 (B = 0.03, p = .400) was associated with early adult marijuana-related symptoms. A formal test of mediation confirmed that Internalizing at age 14 mediated the physical abuse—early adult marijuana-related symptoms relationship (MacKinnon’s 95% Asymmetric CI: .0004 to .1949). Physical abuse was associated only with Externalizing problems at age 14 (b = 3.33, p = .001). Externalizing problems at age 14 (b = 0.04, p = .009), 16 (b = −0.05, p = .024) and at 18 (b = 0.06, p = .011) were all associated with early adult marijuana-related symptoms. A formal test of mediation confirmed that only the physical abuse—early adult marijuana-related symptoms relationship was mediated by Externalizing problems at age 14 (MacKinnon’s 95% Asymmetric CI: .0285 to .2929; See Figure 2).
Types of child maltreatment predicting marijuana symptom criteria in early adulthood, mediated by externalizing adolescent behavior problems.
Marijuana SUD
No direct effects from CM to early adult marijuana SUD were found (all ps > .05). Regarding mediation, physical abuse was associated with Internalizing problems at age 14 (b = 1.98, p = .014) and emotional abuse with Internalizing problems at age 18 (b = 2.67, p=.011). However, Internalizing problems at these ages were not associated with early adult marijuana SUD (all ps > .05 for all mediators). Physical abuse was associated with Externalizing problems at age 14 (b = 4.33, p < .001), but the path to early adult marijuana SUD was not significant (p > .05). Externalizing problems at age 18 was associated with early marijuana SUD (b = 0.03, p = .020). However, none of the CM types were associated with this potential mediator (all ps > .05). In sum, no mediation for the early adult marijuana SUD outcome was established.
Discussion
The purpose of the present study was to assess the mediating role of internalizing and externalizing behavior problems during adolescence between CM types and early adult substance use problems and disorders. Our findings indicate internalizing problems at age 14 mediated the link between childhood physical abuse and later alcohol SUD and marijuana-related symptoms. Externalizing problems at age 14 mediated the link between physical abuse and marijuana-related symptoms, but not problem alcohol use. Neither internalizing nor externalizing problems at ages 16 and 18 mediated the link between any CM type and SU problems or disorder.
Broadly, our data support research showing associations between CM and adult SU problems (Dubowitz et al., 2019; Fergusson et al., 2013; Lo & Cheng, 2007; Widom et al., 2006), CM and adolescent internalizing and externalizing problems (Bolger & Patterson, 2003; Dvir et al., 2014; Herrenkohl et al., 2013; Lansford et al., 2002; Mills et al., 2013; Nemeroff, 2016), and adolescent internalizing and externalizing problems and adult SU (Englund et al., 2008; Fergusson et al., 2007). However, findings from this prospective study suggest direct effects between CM and young adult SU may be more nuanced depending on the type of CM, type of substance, and degree of problem use. Specifically, we found direct effects for childhood physical abuse and neglect on alcohol-related symptoms, but not marijuana-related symptoms or marijuana use disorder. Physical abuse was not directly associated with alcohol use disorders. Childhood neglect was negatively associated with alcohol-related symptoms and alcohol use disorder, but not related to marijuana-related symptoms or disorder. Neither emotional abuse nor sexual abuse were directly associated with alcohol or marijuana-related symptoms or disorders.
Neurobiogical perspectives suggest that childhood adversity and trauma negatively impact the developing child’s stress response system, subsequently impacting brain structures and systems responsible for emotion regulation and behavioral control (De Bellis, 2002; Tarullo & Gunnar, 2006). This may increase vulnerability to psychopathology including depression, anxiety, anger and post-traumatic stress (Faravelli et al., 2012; McCrory et al., 2012). This is consistent with findings here demonstrating associations between physical and emotional abuse and internalizing and externalizing behaviors— at multiple points during adolescence. Self-medication through SU may be an attempt to cope with negative affective states (Dixon et al., 2009; Hussong et al., 2011; Khantzian, 1987). Indeed, several studies have documented a mediating role of negative affect between early adversity and later SU problems (Asgeirsdottir et al., 2011; T. L. Lewis et al., 2011; White & Widom, 2008).
Few studies, however, have assessed whether affect and behavioral problems are more salient at different times during adolescence. An advantage of the current study is the measurement of internalizing and externalizing problems at three points during adolescence. Our findings indicated that problem behaviors at age 14, but not at 16 and 18, mediated the link between physical abuse and SU problems in early adulthood. This age may represent a particularly vulnerable period as many youth around age 14 have completed pubertal development, are gaining more autonomy, navigating complex peer relationships, and are experiencing increased exposure to and availability of drugs, tobacco and alcohol. Maltreated youth with elevated affective and behavioral dysregulation may have difficulty navigating these developmental challenges subsequently increasing the risk for SU and other health risk behaviors. Even after controlling for SU during adolescence, behavioral problems at age 14 mediated links between early CM and SU problems nearly 10 years later. This period of transition to high school may be a particularly sensitive period for intervening with high-risk adolescents, especially those with a history of physical abuse.
Our findings also suggest there may be different mediating pathways for different substances and degree of problem use. Specifically, internalizing and externalizing problems mediated the link between physical abuse and marijuana-related symptoms (but not marijuana SUD), and only internalizing problems mediated the link to alcohol SUD (but not alcohol-related symptoms). Compared to studies examining an internalizing problem pathway, research on pathways from externalizing problems have been more commonly reported for both alcohol and marijuana (Colder et al., 2013; King et al., 2004). Findings from studies assessing internalizing problem pathways have been inconsistent and are predominantly couched in the Self-Medication model (Khantzian, 1987). Scholarship is growing in this area including the conceptualization of internalizing problems as a pathway to a specific subtype of alcoholism, Negative Affect Alcoholism, particularly among women (Hussong et al., 2011). Others have suggested pathways to SU may be influenced by the co-occurrence of internalizing and externalizing problems. One study found externalizing problems without internalizing problems predicted adolescent alcohol and marijuana use, whereas internalizing problems without externalizing problems served as a protective factor reducing marijuana use (Colder et al., 2013). More research is needed regarding differential pathways by substance type.
We can only speculate reasons neglect was inversely associated with alcohol problem use in the current study. Given that there was no comparable self-report measure, the coding of neglect was based solely on CPS allegations. CPS reports alone, without participants’ perceptions of their early care, may have inadequately captured their experiences. In addition, one study found that neglect experienced in later adolescence versus early childhood, predicted young adult SU (Dubowitz et al., 2019), and Wilson and Widom (2010) found that childhood neglect was associated with a late-drug-use pattern. These findings suggest that neglect may have a more complex role in the emergence of problem substance use, the impact of which may be related to informant, timing, and SU onset.
The lack of findings regarding sexual and emotional abuse may be due a number of factors. First, we controlled for the presence of other CM types. In a study using similar data, T. Lewis and colleagues (2016) reported sexually abused youth were more likely to have experienced other types of CM compared to those who had been maltreated but not sexually abused. Similarly, another study found that emotional abuse alone did not predict adolescent smoking, but the combination of emotional abuse and other abuse types did (T. Lewis et al., 2019). Studies may have failed to control for co-occurring CM thus inflating the contribution of sexual and emotional abuse to young adult SU. Finally, with some exceptions most studies assessing CM and adult SU either rely on retrospective reports from adult respondents or on administrative data only. Our use of prospective data from both CPS and self-report may explain discrepant findings.
This study has a number of important strengths including prospective data from birth to early adulthood with repeated assessment of behavioral problems, SU, and CM over time including careful classification of CM type, integrating where possible, CPS and self-report data. We controlled for other important factors such as caregiver SU, adolescent use of substances, and young adult demographics.
Some study limitations should be noted. First, we integrated both CPS and self-report measures for three of the four CM variables. There are limitations to reliance on either CPS or self-report. Although some scholars advocate for the integration of both to minimize the biases inherent in each (Everson et al., 2008; Knight et al., 2000; Shaffer et al., 2008), others argue the two should not be used interchangeably (Baldwin et al., 2019). Further research is needed to better identify potential differences in findings based on source and how to adequately integrate multi-informant CM data. Second, we did not have a self-report measure of neglect that could be integrated with CPS data. Neglect occurs on a spectrum of care, often without discrete cut-points; this precluded combining it with the CPS data and may have influenced findings and/or comparisons to other types of CM. Third, we did not account for polyvictimization; rather, we chose to assess the unique contribution of each CM type. Mediating pathways for polyvictims may differ from those for specific CM types. In addition, considering type as present or not is admittedly crude given the heterogeneity of circumstances within this label. Fourth, we were limited to youth self-report of internalizing and externalizing problems at age 18; having parental information can offer a more comprehensive understanding of youth’s behavior. Fifth, patterns and reporting of substance use may vary by age. In this study, young adult respondents ranged in age (19–20, Mean = 23.8) Finally, as with any lengthy longitudinal study, attrition is inevitable. However, there were no differences in baseline demographic and CM histories aside from gender.
In summary, our findings suggest that age 14 may be a particularly vulnerable time for youth with a history of CM to develop SU problems in early adulthood, particularly for those who had been physically abused. Interventions to identify and address internalizing and externalizing behavior problems in high-risk youth are needed to help reduce later SU problems. The inclusion and standardized use of screenings for problem behaviors in primary care settings may be one way for early identification of at-risk youth. Multidisciplinary networks including primary healthcare, behavioral health, school, and social services could provide critical components of care and services to prevent a trajectory of problem substance use.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by grants from the Office of Child Abuse and Neglect, Administration on Children and Families, US DHHS (Grants Nos. 90CA1401, 90CA156901, 90CA1681, and 90CA1749), and the National Institute on Drug Abuse (Grant no. 5R01DA031189-04).
