Prenatal Substance Exposure: What Predicts Behavioral Resilience by Early Adolescence?

Abstract

Understanding behavioral resilience among at-risk adolescents may guide public policy decisions and future programs. We examined factors predicting behavioral resilience following intrauterine substance exposure (IUSE) in a prospective longitudinal birth-cohort study of 136 early adolescents (age 12.4–15.9) at-risk for poor behavioral outcomes. We defined behavioral resilience as a composite measure of lack of early substance use initiation (before age 14), lack of risky sexual behavior, or lack of delinquency. IUSEs included in this analysis were cocaine (IUCE), tobacco (IUTE), alcohol (IUAE), and marijuana (IUME). We recruited participants from Boston Medical Center as mother-infant dyads between 1990 and 1993. The majority of the sample was African-American/Caribbean (88%) and 49% female. In bivariate analyses, none and lower IUCE level predicted resilience compared to higher IUCE, but this effect was not found in an adjusted model. Instead, strict caregiver supervision (adjusted odds ratio (AOR)=6.02, 95% confidence interval (CI)=1.90–19.00, p=0.002), lower violence exposure (AOR=4.07, 95% CI=1.77–9.38, p<0.001), and absence of intrauterine tobacco exposure (AOR=3.71, 95% CI= 1.28–10.74, p=0.02) predicted behavioral resilience. In conclusion, caregiver supervision in early adolescence, lower violence exposure in childhood, and lack of intrauterine tobacco exposure predict behavioral resilience among a cohort of early adolescents with significant social and environmental risk. Future interventions should work to enhance parental supervision as a way to mitigate the effects of adversity on high-risk groups of adolescents.

Introduction

Resilience is a dynamic process, influenced by multiple factors encompassing genetics, biology, environment, psychology, and exposure to adversity (Rutter, 2006). Resilience can be defined as positive adaptation or recovery in the context of adversity. An individual may display resilience in one functional domain at one period of development, but not necessarily in multiple domains or over different developmental epochs (Luthar & Brown, 2007). Of particular interest for policy and scientific inquiry is understanding behavioral resilience among children, adolescents, and young adults who grow up with the most extreme adversity, such as witnessed parental violence, parental addiction, and poverty. In this study, we focus specifically on a cluster of multiple problem behaviors in a high risk cohort, as problem behaviors tend to co- occur and are also negatively related to prosocial behaviors (e.g. school attendance) and positive health behaviors (e.g. exercise, diet) (Jessor, Donovan, & Costa, 1991).

There are known links between intrauterine substance exposure (IUSE) and indicators of poorer behavioral and health outcomes in adolescence. In this study we focus on intrauterine cocaine exposure (IUCE), intrauterine tobacco exposure (IUTE), intrauterine alcohol exposure (IUAE), and intrauterine marijuana exposure (IUME) because there is substantial literature linking these exposures to negative developmental and health outcomes in children and adolescents. IUCE is linked to adolescent substance use (Delaney-Black, Chiodo, Hannigan, Greenwald, Janisse, Patterson, Huestis, Patridge, et al., 2011; Frank et al., 2011; Richardson, Larkby, Goldschmidt, & Day, 2013), childhood externalizing behavior problems (Bennett, Marini, Berzenski, Carmody, & Lewis, 2013), inattention, and impulsivity (Richardson, Goldschmidt, Leech, & Willford, 2011). IUTE is associated with childhood and adolescent conduct disorder, externalizing behavior (Cornelius, Goldschmidt, De Genna, & Larkby, 2012; Piper, Gray, & Birkett, 2012; Stene-Larsen, Borge, & Vollrath, 2009), adolescent delinquent behaviors, and adult criminal behavior (Paradis, Fitzmaurice, Koenen, & Buka, 2011; Rantakallio, Laara, Isohanni, & Moilanen, 1992). IUAE correlates with attention difficulties (Mattson, Crocker, & Nguyen, 2011; Underbjerg et al., 2012), delinquent behaviors (Schonfeld, Mattson, & Riley, 2005), and higher rates of Attention Deficit Hyperactivity Disorder (ADHD) (Mattson, et al., 2011). IUME in the context of environmental disadvantage has been linked to increased attention problems and aggression in 18-month old girls (El Marroun et al., 2011), future substance use (Frank, et al., 2011), delinquency in late childhood (Goldschmidt, Day, & Richardson, 2000) and adolescence (Day, Leech, & Goldschmidt, 2011), and poor academic performance (Goldschmidt, Richardson, Willford, Severtson, & Day, 2012). Due to this well- established literature on intrauterine exposure to individual substances and adolescent outcomes, we examine multiple forms of IUSE simultaneously to see if some exposures are more detrimental than others, and to avoid mistakenly attributing the effects of one substance on outcomes to another.

A prospective study design and biological markers to confirm IUSE are of particular importance when studying resilience due to the temporal relationship of the predictors of resilience and the markers of resilience over the lifetime of children and adolescents, and to avoid ascertainment bias. For this reason it is inappropriate to examine predictors of resilience and markers of resilience in a cross-sectional survey. Like many other research teams who examine resilience (Bennett, et al., 2013; Delaney-Black, Chiodo, Hannigan, Greenwald, Janisse, Patterson, Huestis, Partridge, et al., 2011; Frank, et al., 2011), we employ a prospective design. We also confirm IUSE with at least one biological marker from each mother-infant dyad, including either maternal or neonatal urine drug tests, or meconium radioimmunoassays.

In order to gain a clearer understanding of what contributes to resilience, it is essential to examine the effects of IUSE in the context of other components that influence resilience. Other factors that influence adolescent resilience that we examine in this study include parental supervision, exposure to violence, and sex. Population based studies in the US have found social support, including parental support and monitoring, increases healthy resilient behaviors in adolescents (Goldstein, Faulkner, & Wekerle, 2013; Mistry, McCarthy, Yancey, Lu, & Patel, 2009; Tiet, Huizinga, & Byrnes, 2010). In previous work, we found that parental incarceration correlated with depressive symptoms and externalizing behaviors (Wilbur et al., 2007), and others have found that it appears to increase not only anti-social behaviors but also mental health problems, drug use or educational underperformance (Murray, Farrington, & Sekol, 2012). Thus, it is not clear how it impacts long term resilience. In violent neighborhoods and among high-risk populations, close parental supervision and monitoring increases adolescent resilience (Burlew et al., 2009; X. Li, Feigelman, & Stanton, 2000; Stanton et al., 2002). Macrosystem community influences, particularly exposure to violence and neighborhood safety, can strongly impact coping skills (Benzies & Mychasiuk, 2009; S. T. Li, Nussbaum, & Richards, 2007). Li found that neighborhood hassles and violence exposure increased internalizing and externalizing symptoms in adolescents (S. T. Li, et al., 2007). Neighborhood cohesion may lead to positive behavioral outcomes for adolescents, while neighborhood disorganization may be related to delinquency (Cantillon, 2006; Chung & Steinberg, 2006). Additionally, exposure to violence, either as victim or witness, correlates with increased suicidal ideation in 9–10 year olds and delinquent behavior in early adolescence, irrespective of intrauterine cocaine exposure or parental distress (Gerteis et al., 2011; O'Leary et al., 2006).

The relationship between sex and resilience is complex and not consistent across studies. Females show increased likelihood of behavioral resilience in young adulthood in some samples (Ackerman, Riggins, & Black, 2010; Skinner, Haggerty, Fleming, & Catalano, 2009). This may be the result of increased parental monitoring among female adolescents (X. Li, et al., 2000). However, other samples have shown that female sex correlates with less resilience (Tusaie, Puskar, & Sereika, 2007). The interactions between sex and intrauterine substance exposure effects are also inconsistent (Bennett, et al., 2013; Bridgett & Mayes, 2011; Dennis, Bendersky, Ramsay, & Lewis, 2006; Dixon, Kurtz, & Chin, 2008; El Marroun, et al., 2011). Non-white race has been associated with decreased resilience (Dumont, Widom, & Czaja, 2007; Fantuzzo, LeBoeuf, Rouse, & Chen, 2012; Mistry, et al., 2009), not as a biologic factor but as a marker for discrimination and material deprivation, low socioeconomic status or lack of supportive social networks (Brown, 2008; S. T. Li, et al., 2007; Tusaie, et al., 2007). In contrast, a sample of 489 rural African-American youth developed psychosocial competence under conditions of high risk, even as they displayed lower health resilience outcomes (Brody et al., 2013).

An ecological model, which takes into account the individual, family, community, environment and larger social context, can facilitate understanding of contributing factors toward behavioral resilience in adolescents who have experienced violence and IUSE. Using an ecological model, this exploratory study examines selected predictors of components of resilience among high risk adolescents. We hypothesized that both intrauterine substance exposure and the family and social environment during childhood and adolescence would predict presence or absence of behavioral resilience in adolescence. Specifically, we hypothesized that lower levels of intrauterine exposure to substances, lower household substance use, lower exposure to violence, lower rates of parental incarceration, and greater parental supervision, and greater neighborhood cohesion would be associated with greater adolescent behavioral resilience.

Methods

Study Design

This is a masked prospective longitudinal cohort study of adolescents recruited at birth to examine the effects of levels of IUCE on behavior and development. Participants and their caregivers were repeatedly assessed from birth, using interviews and urine assays, as well as neuropsychological and behavioral assessments that have been reported elsewhere (Frank, et al., 2011; Gerteis, et al., 2011).

Sample Selection

Sample recruitment took place at the post-partum unit of Boston City Hospital (now Boston Medical Center) from 1990 to 1993. Mother-infant dyads met the following eligibility criteria: maternal age ≥18 years; infant gestational age ≥ 36 weeks; no need for neonatal intensive care; no diagnosis of fetal alcohol syndrome; and no indication (either by neonatal or maternal urine toxic screen or meconium assay or by history in medical record) of intrauterine exposure to illegal opiates, methadone, amphetamines, phencyclidine, barbiturates, or hallucinogens; and no history of HIV seropositivity in the infant or mother. Further information about recruitment procedures and sample characteristics has been previously described (Tronick, Frank, Cabral, Mirochnick, & Zuckerman, 1996). Boston Medical Center Institutional Review Board approval was obtained yearly. Mothers or primary caregivers also provided ongoing informed consent. Beginning at 8 years, study participants provided assent.

Of the original 252 cohort members, we analyzed data from the 136 participants examined at early adolescence, targeted for ages 12.5–14.5. Due to challenges in getting participants for interviews in target range, the actual age range was 12.4–15.9. These participants did not differ significantly from the non-participants in terms of sex, ethnicity, maternal age at birth, or intrauterine exposure to alcohol, tobacco or marijuana or cocaine.

Data Collection

This paper uses data collected at birth and ages 8.5, 9.5, 11 years and early adolescence (12.4–15.9 years). The primary outcomes were obtained during early adolescence. Primary caregivers were interviewed in parallel with the participants at each time point. Children’s evaluators were masked to participants’ IUCE status and to all information furnished by their caregivers. Questions about delinquent behavior, substance use, and sexual activity were asked as part of an audio computer assisted self-interview (ACASI) in which the participant read and listened via head phones to written and audio text of the questions and answers. The participant answered questions by clicking the computer mouse. Using the ACASI is thought to promote more truthful answers to questions on potentially sensitive subjects than would be obtained via face to face interview (Riley et al., 2001). ACASI questions included items from the Hooked on Nicotine Checklist (DiFranza et al., 2002), parts of the CDC's 2005 Youth Risk Behavior Surveillance System (YRBSS) (Eaton et al., 2006) the Wisconsin YRBS Middle School Questionnaire and the Wisconsin YRBS High School Questionnaire. In addition, urine samples were tested for cotinine, marijuana, and other illicit drugs (Frank, et al., 2011).

Dependent Variable

Resilience was defined as the absence of three outcomes: HIV risk behavior, early initiation of substance use, and delinquency. Each of these were measured at the early adolescent interview at ages 12.4–15.9. The latter two outcomes have been examined individually in previous publications from this study (Frank, et al., 2011; Gerteis, et al., 2011). Any of these three behaviors was considered an indicator that the participant was not showing resilience. Absence of these risk factors were analyzed cumulatively as a better indicator of resilience (Jessor, 1987). HIV risk behavior was defined as endorsement of 1 or more of the following behaviors: lack of condom use during 1^st intercourse or most recent intercourse, injection drug use, or pregnancy (self or partner). Early initiation of substances was defined as use of substances (tobacco, alcohol, marijuana or other illicit substances) before age 14. Specific questions included: “How old were you when you smoked a whole cigarette for the first time?” and “How old were you when you had your 1st drink of alcohol other than a few sips?”. It was specified that, “A drink of alcohol is equal to having a can of beer (the same size as a soda can), a glass of wine, a wine cooler, or a shot of liquor such as rum, gin, vodka, or whiskey.” For misuse of prescription medications (e.g. amphetamines, steroids, oxycodone and other pain killers, or benzodiazepines), the question was “How old were you when you first tried taking (substance of interest) without a doctor or nurse telling you to take them?” In the case of illicit substances (marijuana, heroin, cocaine, “club drugs”) a quantity was not specified, with the question framed, “How old were you when you first tried (substance) for the first time?” Delinquency was defined as self-report of at least three delinquent activities in response to seven questions on minor criminal behavior from the National Longitudinal Study of Adolescent Health (Gerteis, et al., 2011; Udry, 2003).

Independent Variables

We identified a priori a set of caregiver and adolescent variables to be tested as predictors of behavioral resilience in the context of an ecological model: 1) No intrauterine substance exposure, 2) Lack of household substance use during participants’ early adolescence, 3) Lower exposure to violence, 4) Higher neighborhood cohesion, 5) Strict supervision during adolescence, 6) No history of parental incarceration, 7) Female sex, 8) Birth mother’s race/ethnicity (African-American/African-Caribbean vs. other).

Intrauterine Substance Exposure (IUSE)

Levels of IUSE was determined by infant urine and meconium assays as well as urine assays and post-partum interviews of the mothers using an adaptation of the Addiction Severity Index, fifth edition (McLellan et al., 1992). IUCE was classified heavier, lighter or none where heavier is defined as the top quartile of self-reported days of maternal cocaine use during index pregnancy and/or the top quartile of cocaine metabolites in infant’s meconium. All other intrauterine cocaine use was defined as lighter. Self-reported intrauterine tobacco exposure was coded as yes/ no and as none, < ½ pack per day, ≥ ½ pack of cigarettes per day. Intrauterine tobacco use was initially coded in a three level variable, but both levels of tobacco exposure had similar results. To conserve degrees of freedom in the analysis, we chose to combine them into a single variable (exposed or not) for the multivariable analyses. Self-reported intrauterine alcohol exposure was coded as none vs. any drinking by mother during last 30 days of index pregnancy. Intrauterine marijuana exposure was determined by a positive result on any one of the following: mothers’ self-report and urine and meconium assays (any detection was considered positive) obtained from mothers and newborns. A third of the marijuana users in this cohort who denied marijuana use were identified solely on the basis of meconium or urine assay.

Household Substance Use

At each study visit, household substance use was determined by the caregivers’ responses to questions asking whether any member of the household where the child was living or spent considerable time used individual substances (marijuana, cocaine, tobacco, heroin, prescription medications not taken as prescribed, methadone) or had a drinking problem. Household tobacco use (yes/no) was analyzed separately from other household substance use because it was thought that it might have unique effects separate from other substances (Weitzman, Gortmaker, & Sobol, 1992). Of note, because caregiver’s own substance use was too highly correlated with pre-natal exposure to give an independent effect, substance use by members other than the caregiver were used for this variable.

Exposure to Violence

Children’s self-reported exposure to violence (either as a witness or a victim) was measured using the Revised Violence Exposure Scale for Children (VEX-R) (Fox & Leavitt, 1995). This measure was used in its original cartoon format accompanying the questionnaire at ages 8.5, 9.5 and 11, and then as a modified questionnaire without cartoons in early adolescence. Scores at each age were grouped in quartiles with the 4^th quartile being the highest level of exposure and the 1^st quartile being the lowest level of exposure. Quartiles were chosen because the VEX total score is not weighted for severity (Frank, et al., 2011; Gerteis, et al., 2011). Others have used rank order in quartiles as a mode of analysis (Shahinfar, Fox, & Leavitt, 2000). Never being in the highest quartile of self-reported violence exposure at any study assessment was used as a predictor of resilience.

Neighborhood Cohesion

Caregiver reported neighborhood cohesion during early adolescence was a composite variable of caregivers’ responses to three questions drawn from the National Survey of Children’s Health (Data Resource Center for & Adolescent, 2003) on perceived aspects of the neighborhood (e.g. “we watch out for each other’s children”, “people in my neighborhood help each other out” and “there are people I can count on in this neighborhood”). The instrument uses a 7 point Likert scale that ranges from very strongly disagree to very strongly agree with a composite score ranging from 3 to 21 (low to high cohesion).

Supervision during Adolescence

Participants reported on their perceptions of caregivers’ parenting during adolescence with relation to strictness/supervision (Eaton, et al., 2006). The supervision scale consists of questions that ask specifically about parental strictness about the adolescent’s activities (e.g., “In a typical week, what is the latest you can stay out on school nights?”) and parental knowledge about adolescent’s activities (e.g., “My parents know exactly where I am most afternoons after school” (Yes/No), “How much do your parents really know what you do with your free time?” (Don’t know/Know a little/Know a lot)). For each of these scale scores, we divided the continuous scores into quartiles and created a dichotomous variable of the highest quartile versus others in our sample.

Parental Incarceration

Parental incarceration at any time was determined by any positive caregiver response during childhood or adolescent interviews to the following questions: “In (child’s) lifetime, has his/her father been in jail or prison?” and “How many times since (the child) was born, have you been in jail or prison (either because you were serving a sentence or because you were detained before a trial)?”

Caregiver Type

Caregiver type in early adolescence was defined as birth mother, kin, or unrelated as identified by the adult accompanying the participant who gave informed consent at each study visit. This was measured at each study visit.

Analysis

Bivariate analyses using logistic regression models unadjusted for covariates were performed to determine associations with resilience for the seven theoretical predictor variables. A multivariable logistic regression analysis included all independent variables that were associated with resilience at p<0.10, age, and all intrauterine substance exposures, which were a main focus of the construction of the source sample. Multicollinearity was not found in the multivariable regression. Odds ratios (OR) and 95% confidence intervals (CI) were computed from the logistic regression models. Statistically significant results had two tailed p-values less than 0.05. All analyses were conducted using SAS, version 9.3.

Results

Of the 136 participants mean age=14.2 (standard deviation = 0.7, age range 12.4–15.9), 72 (53%) were classified as behaviorally resilient. Specifically, 94% of the cohort exhibited no HIV risk behaviors, 76% reported two or fewer delinquent acts, and 59% had not initiated alcohol, tobacco or other substance use (Table 1).

Table 1.

Behavioral Resilience by Maternal and Adolescent Characteristics, bivariate analyses (n=136)

	N (%) of sample or mean (s.d.)	% Resilient	Odds ratio (95% CIa)	p-value
Intrauterine cocaine exposure (global p=0.12)
None	61 (45)	61%	2.74 (1.04, 7.19)	0.04
Lighter	50 (37)	52%	1.93 (0.72, 5.17)	0.19
Heavier	25 (18)	36%	Referent
Intrauterine marijuana exposure
No	103 (76)	53%	1.08 (0.49, 2.36)	0.85
Yes	33 (24)	52%	referent
Intrauterine tobacco exposure (global p=0.001)
None	66 (49)	68%	2.65 (1.16, 6.03)	0.02
<1/2 pack per day	32 (24)	31%	0.56 (0.21, 1.50)	0.25
>=1/2 pack per day	38 (28)	45%	referent
Intrauterine alcohol exposure
No	97 (71)	60%	2.66 (1.23, 5.74)	0.01
Yes	39 (29)	36%	referent
Maternal
Age at delivery (years)	26.6 (5.2)		0.99 (0.93, 1.06)	0.74
Education (years)	11.5 (1.3)		0.97 (0.75, 1.26)	0.83
Ethnicity
African- American/Caribbean	120 (88)	52%	1.56 (0.53, 4.56)	0.42
Other	16 (12)	63%	referent
Adolescent
Sex
Male	70 (51)	49%	referent	0.29
Female	66 (49)	58%	1.44 (0.73, 2.83)
Violence exposure
Quartiles 1–3	79 (58)	66%	3.56 (1.74, 7.29)	0.0005
Top quartile	57 (42)	35%	referent
Parental strictness/supervision
Quartiles 1–3	105 (77)	44%	referent	0.0003
Top quartile	31 (23)	84%	6.67 (2.38, 18.72)
Household Tobacco Use
No	104 (76)	57%	1.92 (0.86, 4.29)	0.11
Yes	32 (24)	41%	referent

^aCI: confidence interval

In bivariate analyses, a number of variables were found to be significantly (p<0.05) associated with resilience, including lower levels of violence exposure between ages 8.5–11.5 years, strict parental supervision, lack of substance use in the home during adolescence, no IUCE compared to heavy IUCE, and lack of intrauterine alcohol and tobacco exposures. Sex and race/ethnicity (see Table 1) were not correlated with behavioral resilience. Scores on acceptance/involvement, psychological autonomy, parental incarceration, neighborhood cohesion, and caregiver type during early adolescence were not related to resilience and are not presented here.

In multivariable logistic regression analysis, strictest supervision (adjusted odds ratio (AOR)=6.02, 95% CI=1.90–19.00, p=0.002), lower violence exposure (AOR=4.07, 95% CI=1.77–9.38, p=0.001), and lack of intrauterine tobacco exposure (AOR=3.71, 95% CI=1.28–10.74, p=0.02) were statistically significant protective factors associated with behavioral resilience (Table 2). Older age appeared protective with results just above statistical significance, (AOR=0.52, 95 CI=0.27–1.02, p=0.06). No statistically significant interactions were found with IUCE and other salient independent variables.

Table 2.

Predictors of Resilience, Multivariable logistic regression

	AORa	95% CIb	p-value
Intrauterine Cocaine Exposure
None	0.67	(0.13, 3.39)	0.97
Lighter	1.02	(0.27, 3.81)	0.63
Heavier	referent
Intrauterine Marijuana Exposure
No	0.65	(0.24, 1.76)	0.39
Yes	referent
Intrauterine Tobacco Exposure
No	4.04	(1.36, 12.04)	0.01
Yes	referent
Intrauterine Alcohol Exposure
No	1.88	(0.59, 5.96)	0.29
Yes	referent
Parental Strictness/Supervision
Top Quartile	5.63	(1.74, 18.25)	0.004
Quartiles 1–3	referent
Violence Exposure
Top Quartile	referent
Quartiles 1–3	3.49	(1.49, 8.20)	0.004
Household Tobacco Use at Early Adolescence
No	2.06	(0.76, 5.55)	0.15
Yes	referent
Age, for each year older	0.52	(0.27, 1.02)	0.06

^aAOR: adjusted odds ratio

^bCI: confidence interval

To further explore the tobacco results, we conducted a (post-hoc) multivariable analysis limited to those with both lighter and heavier IUCE to test whether the intrauterine tobacco effects found were due to high prevalence of intrauterine tobacco exposure in participants with heavy cocaine exposure. The results showed no difference by low vs. high IUCE (results not shown).

Discussion

Strict caregiver supervision in early adolescence, lower violence exposure from ages 8–14 and lack of intrauterine tobacco exposure were predictors of behavioral resilience in this exploratory analysis among a cohort of early adolescents with significant social and environmental risk regardless of intrauterine cocaine exposure status. In contrast to our hypothesis, although heavy intrauterine cocaine exposure was associated with decreased odds of resilience when compared to no cocaine exposure in bivariate analyses, this effect was not identified after controlling for other factors. Viewed from an ecological stance, the biologic factor of intrauterine tobacco exposure, the microsystem factor of parental supervision, and the combined micro/macrosystem factors of lower violence exposure each contributed significantly to behavioral resilience.

The findings from this current study also support growing understanding of the potent effect of violence on development and behavior. Adverse childhood experiences, including exposure to violence, have been associated in retrospective studies with a number of life-long problems, including depression, substance abuse and high risk health behaviors (Felitti et al., 1998). The proposed mechanism for this relationship is that violence exposure disrupts the normal stress response of the hypothalamic-pituitary-adrenal axis (Neigh, Gillespie, & Nemeroff, 2009), so that cortisol does not appropriately increase in response to stress. This blunted cortisol response is highly associated with depression and other mental health problems (Neigh, et al., 2009). Lester and colleagues found an exaggerated blunting of the cortisol response in 11 year olds who had both childhood exposure to domestic violence and IUCE, compared to those without IUCE, without violence, or without either (Lester & Padbury, 2009). In our study, the impact of not experiencing high levels of violence on behavioral resilience, points to an urgent need to prevent violence exposure.

Our findings on parental supervision are consistent with work conducted in other studies, which did not account for documented IUSE (Steinberg, Lamborn, Darling, Mounts, & Dornbusch, 1994). Chilcoat and Anthony reported that lower parental supervision increased the risk for early initiation of substances in a sample of largely minority children from an urban setting (Chilcoat & Anthony, 1996). Burlew and colleagues noted that parental supervision buffered the impact of increased neighborhood risk for early substance initiation in a sample of African-American youth living in low income neighborhoods (Burlew, et al., 2009). In intergenerational longitudinal studies, decreased parental monitoring was associated with externalizing behaviors (e.g. precursors of delinquency) (Bailey, Hill, Oesterle, & Hawkins, 2009). Lahey and colleagues reported that parental knowledge of children’s peers and limit setting influenced risk of adolescent delinquency particularly in adolescents in high risk neighborhoods (Lahey, Van Hulle, D'Onofrio, Rodgers, & Waldman, 2008). Similarly, parental monitoring and supervision has shown to decrease early sexual activity among high risk adolescents (Boislard & Poulin, 2011; Browning, Leventhanl, & Brooks-Gunn, 2005).

The continued impact of intrauterine tobacco exposure into adolescence confirms other studies that have shown that such exposure has been associated with conduct disorder and behavioral problems in childhood and adolescence in samples without IUCE (Brook, Zhang, Rosenberg, & Brook, 2006; Desrosiers et al., 2013; Gaysina et al., 2013; Rantakallio, et al., 1992). Furthermore, the findings on intrauterine tobacco expose misconceptions about relative impact of intrauterine exposure to legal compared to illegal substances on long term behavioral outcomes. The fact that tobacco is legal and cocaine is not, is often misinterpreted to suggest that intrauterine tobacco exposure is less harmful than intrauterine exposure to illicit substances. However, as our study has demonstrated, intrauterine tobacco exposure can have long lasting behavioral consequences identifiable even when such use co-occurs with IUCE. Smoking cessation programs should focus on women of childbearing age (Chamberlain et al., 2013; Lumley et al., 2009; Valanis et al., 2001).

Strengths and Limitations

The strengths of this study include the prospective longitudinal cohort design and the detailed biological information on intrauterine exposures, as well as frequent prospective data collection on predictors and outcomes of interest. We acknowledge that this study has several limitations. First, the sample size, which may affect the statistical power to detect a significant association of IUCE and resilience or lack of resilience. While we may have failed to identify all factors associated with resilience because of lack of power we did find three factors, one at each level predicted by the ecological model which significantly influenced resilience. Furthermore, we constructed multiple models to test our predetermined hypothesis which used a limited number of variables in order to diminish the impact of sample size on statistical power. Second, use of a dichotomous outcome (i.e. resilient vs. not resilient) may obscure more subtle findings that relate to this complex developmental process. Third, the model in this study predicting resilience in early adolescence may not predict resilience in other developmental periods. In particular, this sample had a low rate of risky sexual behavior, which may have been due to the age itself. Prevalence of sexual behavior itself is low in this age group as compared to older age groups. The risky sexual behavior measure of resiliency likely played a minor role in this study, but in older age groups, resiliency to risky sexual behaviors may be more common. Lastly, our findings may be generalized only to urban, low-income, predominantly African- American/African-Caribbean populations. Further research needs to be conducted in other samples, including cohorts who are of higher socio-economic status, rural, or of other ethnicities.

Public Health Implications

Family practices and environmental factors, particularly stricter caregiver supervision and less exposure to violence, may buffer the negative behavioral impact of intrauterine substance exposure for at-risk urban youth. While intrauterine tobacco exposure remains a risk for negative behavioral outcomes, this study points also to potential post-natal points of modifiable environmental experiences that can moderate early life disadvantages. Because this is an observational study, it is not known whether an intervention to reduce violence and increase parental supervision will enhance behavioral resilience.

Conclusion

Lower exposure to violence in childhood, close parental supervision in adolescence, and lack of intrauterine tobacco exposure predicted increased behavioral resilience in high-risk urban adolescents, half of whom had IUCE. Despite the presumed increased risk for adolescent maladaptive behaviors associated with IUCE, level of IUCE was not related to lower odds of behavioral resilience after covariate control. Interventions to enhance parental supervision in adolescence should be tested as a method to mitigate the effects of harmful exposures for high risk youth.