Social and environmental predictors of youth alcohol and cannabis initiation risk: The moderating role of family history of substance use disorders

Abstract

The social and local-home environment impacts youth’s likelihood of early substance use initiation (SUI). Yet, it is unknown whether protective or risk factors are salient for all forms of youth SUI, or may be specific to certain substances (e.g., alcohol, cannabis). Additionally, certain family factors – such as having a family history of SUD (FH+)—may exacerbate risk for and mitigate protection from alcohol (AUI) and cannabis (CUI) use initiation. Using a longitudinal cohort of 387 adolescents, we aimed to understand how the link between social and local-home factors on AUI and CUI was moderated by FH status. Results indicated that affiliating with risky peers significantly increased the risk of AUI and CUI, but that this relationship was not moderated by FH status. On the other hand, the link between mother-youth relationship quality and CUI, was potentially moderated by FH status such that being FH+ attenuated the protective effect of a positive relationship with parents on SUI. This research could provide evidence that youth who are at risk of early SUI, particularly those who are FH+, may benefit from programs designed to reduce affiliation with risky peers.

Introduction

Risky behaviors in adolescence can have many long-term consequences on later health and developmental outcomes. One risky behavior that becomes more prevalent in adolescence and has been linked to negative health outcomes is substance use (Bonnie et al., 2004; Marshall, 2014; McCambridge et al., 2011). Youth who have early substance use initiation (SUI; initiation of substance use before age 15; Trujillo et al., 2019) are at increased risk of developing a substance use disorder (SUD; Chassin et al., 2013; Hussong et al., 2012), and are at risk for other social and health problems in adulthood (Raposa et al., 2014; Taylor et al., 2014). Past research has found that youth who have positive relationships with their parents have a lower risk of early SUI, but youth who affiliate with risky peers and those who have poor local-home environments (e.g., neighborhood safety) have a higher risk (Nash et al., 2005; Schuler et al., 2019; Van Ryzin et al., 2012). Yet, the overall strength of these protective or risk factors may have on youth early SUI risk may be underscored by having a family history of SUD (FH+; Kuppens et al., 2020). Past research has found that FH+ status attenuates protective factors while also exacerbating risks (Gorka et al., 2013; Joyner et al., 2018; Zhou et al., 2006). Much of this research focuses on alcohol use and more knowledge is needed to understand how these factors may also predict the initiation of other substances such as cannabis. The current paper aims to explore how the link between social and local-home factors and the risk of youth alcohol (AUI) and cannabis (CUI) use initiation may be moderated by FH+ status.

Social relationships and local-home factors as predictors of SUI

Parent and peer relationships, along with the quality of the environment in which youth live, have been found to play a large role in determining SUI risk (Anderson et al., 2007; Knyazev, 2004; Moore et al., 2010). For example, the quality of the relationship with parents can be a critical factor in adolescent development; positive relationships can decrease the risk for early SUI, while negative relationship quality can increase risk (Velleman & Templeton, 2016). Having more positive relationships with parents is associated with greater parent-youth involvement, communication, and parental monitoring (Criss et al., 2015; Kim et al., 2015; Yoder et al., 2016). Parents who engage in more monitoring may be more aware of what is going on in their child’s life (Crouter & Head, 2002; Lippold et al., 2014; Racz & McMahon, 2011), allowing them to be better equipped to dissuade their children from early SUI (Abar et al., 2015; Mills et al., 2021; Rusby et al., 2018; Trucco, 2020). Additionally, if the parent and child have positive, open communication, they may be more likely to discuss topics such as the use of alcohol or other substances. In turn, these parents may be better able to inform their children of the potential risks of early SUI—ultimately decreasing their risk of engaging in early substance use (Ohannessian, 2013; Ryan et al., 2015). However, it is important to note that while a large amount of research has examined the association between parent-youth relationship and SUI risk, the majority of developmental research has focused on the mother-youth relationship (Dishion & Loeber, 1985; Rusby et al., 2018), leaving the role of the father largely underexplored (Yoder et al., 2016). Yet, there is increasing evidence that indicates that mothers and fathers may play unique roles in predicting youth outcomes (Erickson, 2015; Jeynes, 2016). These findings may extend to youth SUI such that while youth may have positive (or negative) relationships with both of their parents, one relationship may be more salient at predicting risk SUI than the other. Therefore, when trying to assess the impact of the parent-youth relationship on SUI, it is important to simultaneously examine the relationships with both the mother and father.

In addition to the role that parental relationships play in SUI risk, the types of peers youth affiliate with have also been shown to play a key role in determining SUI risk (Trucco, 2020). During adolescence, youth begin to spend more time with their peers than with their parents, often while unsupervised (Haase et al., 2008; Spillane et al., 2020; Steinberg & Morris, 2001). Spending more time in unstructured and unsupervised environments is associated with increased risk for substance use (Greene & Banerjee, 2009; Spillane et al., 2020). Associating with peers who engage in risky behavior increases the likelihood that youth will engage in deviant behavior and substance use themselves (Kobus, 2003; Levey et al., 2019; Van Ryzin et al., 2012). Conversely, association with less risky peers can reduce their own risk of deviant behavior and substance use (Van den Bos et al., 2014; van Hoorn et al., 2016).

Outside of the effect of social relationships on SUI risk, local-home environment factors, such as neighborhood safety, are also related to adolescent SUI. One factor that predicts SUI is family socioeconomic status (SES). In many societies, individuals with less financial means are marginalized and systematically excluded from advancement opportunities (Atkinson & Da Voudi, 2000; Smith, 2015; Walker & Smith, 2018). For example, youth raised in families with higher SES typically live in safer neighborhoods and have greater opportunities for supervised extracurricular activities, which may decrease the likelihood of SUI (Leventhal & Brooks-Gunn, 2000; Veitch et al., 2006). Yet, youth from low SES families may live in more dangerous neighborhoods and may have less supervision, which may increase the risk of SUI (Buu et al., 2009; Owens, 2017; Pettit et al., 1999; Veitch et al., 2006). Additionally, youth from low SES families may be socially excluded or victimized by their higher SES peers (Hong et al., 2014; Karlsson et al., 2014), placing youth at increased risk for engaging in SUI if they have difficulties coping (Brady et al., 2009).

FH status as a moderator

Family history of SUD (FH+) marks the risk of engaging in early SUI for a variety of reasons. One reason is that FH+ youth may inherit genetic risk from family members who have a SUD (Carmelli et al., 1990; Kuppens et al., 2020). Youth with an inherited predisposition may be more sensitive to the effects of substances which may ultimately increase their risk for developing a later SUD (Carmelli et al., 1990; McGue et al., 2000; Pasman et al., 2019). Alternatively, their risk may be increased by inheriting conditions with a high comorbidity with substance misuse (e.g., attention deficit hyperactivity disorder, depression, anxiety; Clark et al., 2004; De Sanctis et al., 2008; Hines et al., 2015). These genes impact youth risk of SUI because they may simultaneously exacerbate pre-existing risk factors while suppressing protective factors (Kuppens et al., 2020). For example, youth who have difficulties with inhibitory control may be more susceptible to peer pressure (Stautz & Cooper, 2014), and more likely to impulsively initiate substance use without considering long-term consequences (Fosco et al., 2019). Furthermore, youth who have internalizing and externalizing problems may have more controlling and less warm relationships with their parents (Lansford et al., 2018) —placing youth at risk for early substance initiation as well.

Notwithstanding potential genetic predispositions, past research examining the moderating role of FH+ status on predictors of alcohol problems found engaging in more substance-free activities is associated with decreased alcohol problems and this link was moderated by FH+ status: FH+ youth reported significantly more alcohol problems than FH− youth, even when reporting the same number of substance-free activities (Joyner et al., 2018). Furthermore, when examining how FH+ status moderates the link between social relationships and substance-related outcomes, research has found that FH+ mitigates the protective effects that family harmony has on alcohol and drug dependence. When experiencing the same level of family harmony, FH+ youth have a greater likelihood of developing alcohol and drug dependence than FH− youth (Zhou et al., 2006). In addition to providing evidence that FH+ status mitigates protective factors, additional research has found that FH+ status can also exacerbate risk factors. When examining the association between depression and substance use, research has also found that FH+ status exacerbates risk such that depressed youth who were FH+ developed substance use problems significantly faster than FH− youth (Gorka et al., 2013).

In sum, although research has indicated that FH+ status may mitigate protective factors while exacerbating risk factors, there are currently gaps in the literature that may limit our understanding of risk and protective factors for early substance use for FH+ youth. For example, much of this research has been conducted on youth who are already participating in substance use. Additionally, past research examining the role of parents on substance use outcomes has typically combined reports of maternal and paternal attributes to create one composite variable rather than examining the roles of mothers and fathers separately (Zhou et al., 2006). By using composite variables, this research does not allow us to understand the unique and distinct role of mothers and fathers in conferring SUI risk. Lastly, although much of the literature designed to understand the predictors and moderators of substance use are longitudinal in nature, many of these studies have truncated collection windows, or rely on few data collection timepoints. Therefore, while these studies provide evidence of the predictors and moderators of SUI risk, it is unclear at which point in development these factors may be most salient. The current study is designed to address the aforementioned gaps in the present literature to provide a more comprehensive understanding of the factors that may protect or place youth at risk for substance use, particularly for FH+ youth.

Current study

Past research has shown that social relationships and local-home factors play a significant role in determining SUI risk. However, there is evidence that FH+ may attenuate protective factors and exacerbate risk factors for SUI (Gorka et al., 2013; Joyner et al., 2018; Zhou et al., 2006). While these factors have mainly been explored in research examining the initiation of alcohol or general substance use, less is known about the discrete effects that parents, peers, and local-home factors have on the initiation of specific substances.

Past research has found that individuals – both teens and adults – often have different motivations for engaging in different substances. Many teens report being motivated to initiate alcohol use to enhance social experiences, to cope with difficult situations, or to look “cool” amongst their friends (Comasco et al., 2010). On the other hand, teens report being motivated to initiate cannabis use not only to enhance social experiences or improve mood but to also expand cognitive awareness of surroundings (Arterberry et al., 2021; Jackson et al., 2021). Furthermore, due to the discrepancies in the social acceptance and accessibility of alcohol and cannabis use (Cappelli et al., 2021; Krohn et al., 1982), it may also be true that risk and protective factors may have different effects when examined separately. Because differing social motives and community acceptance of alcohol or cannabis use can impact youth’s likelihood of initiating use, it may be that protective and risk factors may be more or less salient when examined in alcohol or cannabis separately. By extension, FH+ status may have a unique moderating role when considering the unique predictive role protective/risk factors have on youth alcohol and cannabis initiation.

The present study aimed to determine how relationships with parents, peer risk behaviors, and local-home factors predicted the risk for alcohol or cannabis use initiation (AUI/CUI). Further, we aimed to test how the strength of the protective and risk factors may be moderated by FH+ status. We hypothesized that when controlling for behavioral phenotypes associated with increased risk for AUI/CUI, more positive relationships with mothers and fathers, more neighborhood safety, and higher SES would decrease the risk for AUI/CUI (Hypothesis 1a), but that affiliating with risky peers would increase the risk for AUI/CUI (Hypothesis 1b). Additionally, we also hypothesized that the link between social and local-home factors and youth AUI/CUI would be moderated by FH status such that being FH+ would mitigate the protective nature of positive relationships with parents, SES, and neighborhood safety on AUI/CUI (Hypothesis 2a) but would exacerbate the risk of affiliating with risky peers on AUI/CUI (Hypothesis 2b).

Method

Participants & Recruitment Procedure

Participants were 387 adolescents who were part of a longitudinal study tracking risks for substance use disorder in adolescents with a family history of SUD (see Charles et al., 2015; Ryan et al., 2016 for more information on the cohort). Of the 387 adolescents, 307 (79.33%) were identified as FH+, and 80 were identified as FH−. Based on the goals of the initial longitudinal project (Charles et al., 2015; Ryan et al., 2016), all participants in the FH+ group had a biological father who met the criteria for an alcohol use disorder. The present study had almost equal representation for males (48.3%) and females (51.7%). Most participants were White (86.6%) and of Hispanic/Latinx descent (78.3%). See Table 1 for detailed study demographics.

Table 1.

Demographic and baseline information

	FH+ (n = 307)	FH- (n = 80)	Total (n = 387)	P
Sex, n (%)				0.36
Male	152(49.5)	35(43.8%)	187(48.3%)
Female	155(50.5)	45(56.3%)	200(51.7%)
Age at Baseline, mean (SD)				0.42
Age	11.24(0.88)	11.33(0.90)	11.25(0.88)
Ethnicity, n(%)				0.04
Non-Hispanic/Latinx	60(19.5)	24(30.0)	84(21.7)
Hispanic/Latinx	247(80.5)	56(70.0)	303(78.3)
Race, n (%)				0.50
White	262(85.3)	73(91.3)	335(86.6)
Black	38(12.4)	5(6.3)	43(11.1)
American Indian/Alaskan Native	5(1.6)	0(0.0)	5(1.3)
Multiracial	1(0.3)	2(2.5)	3(0.8)
Unknown	1(0.3)	0(0.0)	1(0.3)
Social Relationships, mean (SD)
Mother-youth relationship quality	0.71(0.16)	0.77(0.15)	0.72(0.16)	0.01
Father-youth relationship quality	0.64(0.21)	0.78(0.16)	0.67(0.16)	<0.001
Peer risky behaviors	0.05(0.1)	0.03(0.05)	0.05(0.08)	0.04
Local-Home Factors, mean (SD)
Neighborhood safety	0.91(0.24)	0.99(0.06)	0.93(0.22)	0.002
SES	0.40(0.20)	0.60(0.19)	0.44(0.23)	<0.001
Heritable Risk Factor, mean (SD)
Behavioral phenotypes	0.71(0.32)	0.47(0.29)	0.66(0.31)	<0.001

Adolescents and their parents were recruited from the general community when the adolescent was 10–12 years of age by media advertisement from a large southwestern city in the USA between April 2010 and February 2013. The adolescent proband and a biological parent completed six-hour baseline assessments and DSM-IV diagnostic interviews at study intake. Adolescents and their parents were followed every six months thereafter for 4-hour follow-up assessments - though parents stopped participating once the adolescent turned 18 years of age. All participants completed self-reported questionnaires, structured interviews, and behavioral tasks at both baseline and follow-up assessments. Longitudinal follow-up visits were conducted from 2010 through December 2021. Adolescents’ age range across the study period began as early as 10 years of age and concluded with the eldest of the cohort being 24.5 years of age.

To ensure that participants were substance naïve at study entry, participants were excluded if they reported SUI before the baseline assessment or tested positive for alcohol or substance use (via breath alcohol monitor and urinalysis, respectively). Additionally, participants were excluded at baseline if they had a positive pregnancy test (girls only) at study entry or a developmental disability that would diminish their cognitive capacity to assent to study procedures. Also excluded was any psychiatric disorder other than externalizing disorders commonly co-morbid with substance use, such as conduct disorder, oppositional defiant disorder, and attention deficit hyperactivity disorder. Both adolescents and their parents were compensated $120 each for completing the baseline assessment; in follow-up assessments, adolescents received $120 compensation, and parents received $75. Breaks, lunch, and transportation were offered during all visits.

All study procedures were approved by the institution’s IRB and informed consent and assent were acquired from parents and youth, respectively. Data were protected through a Department of Health and Human Services Certificate of Confidentiality.

Materials

Family History of Substance Use Disorder.

At the initial baseline assessment, paternal SUD was determined by completing the Family History Assessment Module (Janca et al., 1992; Rice et al., 1995). The Family History Assessment Module assesses the family history of psychiatric and substance use disorders for both parents and maternal and paternal grandparents. For each family member, parents were asked to report how family members of the participating youth (e.g., another parent, maternal/paternal grandparents) engaged in problematic substance use (sample item: “drinking ever caused any of [their] relatives to have problems with health, family, job, or police?”) on a dichotomous scale (1=yes, 0=no). Parents then completed a 19-item yes/no individual assessment module for each family member that was indicated to have issues related to alcohol consumption to determine if family members met the criteria for alcohol use disorder (sample item: “because of drinking did your partner have problems such as being able to stop or cut down on drinking”). If the respondent indicated three or more activities within the past year, the target relative would be considered to meet the criteria for a substance use disorder. Due to the nature of the initial study, all participants who had a father who met criteria for alcohol use disorder were identified as “FH+” and those who did not have a family history of SUD were identified as “FH−.”

Social and Local-Home Risk Factors.

To assess the social and environmental factors that place youth at increased risk for substance use initiation, youth completed at baseline the Non-Transmissible Liability Index (NTLI; Kirisci et al., 2009). The NLTI is an 86-item self-report measure that is comprised of four subscales, including: mother-youth relationship quality (24 items on a varying 3- to 4-point scale: sample item: “When I have a problem, my mom helps me solve it”), father-youth relationship quality (46 items on a varying 3- to 4-point scale: sample item: “My dad is reasonable when I make a mistake”), peer risky behavior (11 items on a 2-point scale: sample item: “During the past six months, how many of your friends have used cannabis or hashish?”), and local-home environment (2 items on a 2-point scale: sample item: “Do you know where people in our neighborhood or school get cannabis or other drugs?”). Three items measured youth risky behaviors (e.g., “Have you ever driven a car by yourself”); however, these were not included in the present analyses. For mother- and father-youth relationship quality factors, higher scores indicated more positive parent-youth relationship quality. For peer risky behavior and neighborhood safety, higher scores indicated greater affiliation with deviant peers and greater neighborhood danger, respectively. Parent-youth relationship quality subscales had strong reliability: mother-youth relationship quality α = 0.89, father-youth relationship quality α =0.95. Peer and local-home factors were found to have acceptable reliability: peer attributes α = 0.788, neighborhood safety α =0.66.

Socio-Economic Status (SES).

To assess the effect of socio-economic status on risk for substance use initiation, parents completed the Four Factor Index of Social Status (FFI; Hollingshead, 1975). Parents reported employment and education for both them and their live-in partner or spouse, if applicable. For employment, occupations were categorized into nine various categories, with “1” including positions such as “farm labor and service worker positions” and “9” including occupations such as “higher executive, major professional, large business owner.” Similarly, for education level, education was categorized into seven various categories, with “1” indicating “less than 7th-grade education” and “7” indicating “graduate professional training.” Participant ratings were then calculated using the factor scores for each of the employment and educational modules to come up with a final marker of socio-economic status. Scores on the FFI could range from 5 to 64, in which higher scores indicated greater socio-economic status.

Behavioral Phenotypes.

To control for the heritable behavioral factors that place youth at increased risk for substance use initiation, parents completed the Transmissible Liability Index (TLI; Vanyukov et al., 2009). The TLI is a 17-item parent-report scale measuring youth behaviors that are highly heritable and place them at increased risk for externalizing and substance use behaviors (sample item: “prior to age 13, did s/he get into trouble because s/he would do things without thinking about them first, for example running into the street without looking”). Scale items were on either a 2- or 4-point scale such that higher scores on the TLI indicate greater heritable risk for SUI. Reports were averaged to create one factor. The scale had acceptable reliability α = 0.739.

All the above composite scores were rescaled to 0 to 1 using the formula (score – min)/(max-min) to allow for an easier interpretation in regression analysis. For a predictor ranging from 0 to 1, the regression coefficient represents the maximum effect this predictor could make on the outcome (e.g., the regression coefficient quantified the maximum predicted change in adolescent substance use initiation risk based on moderating factors).

Drug History Questionnaire.

To track patterns of adolescent alcohol and cannabis use initiation, adolescents completed a modified Drug History Questionnaire that was adapted from Sobell & Sobell (1992) at each follow-up assessment visit. The DHQ asked participants to endorse whether they have “ever in their life” used one of fifteen categories of substances (i.e., alcohol, marijuana/cannabis, cocaine, amphetamines). Positive affirmations of substance use on the DHQ were followed by additional questions regarding frequency and use patterns; however, the current analysis used the first follow-up occasion of endorsing ever used in a lifetime for alcohol or cannabis as the substance use initiation endpoint for the present analyses. Three time-to-event outcomes were analyzed: age to initiation of first alcohol or cannabis use, age to alcohol use initiation, and age to cannabis use initiation.

Analytical Plan

Demographic and baseline information were summarized using descriptive statistics and comparisons between FH+ and FH− were conducted using a t-test or Mann-Whitney U test for continuous variables and a Chi-square test of Fisher’s exact test for categorical variables as appropriate. For each type of substance use initiation, the incidence rate was computed for each FH group using the number of events observed divided by the total number of person-years of observation (e.g., the sum number of years of data collected for all participants in the study in each FH group) and compared between the FH+/− groups using a Chi-square test. For each FH group, the proportion of not engaging in each event of interest over age was presented using Kaplan-Meier (KM) survival curves with the age of 10 years old as the time origin (time 0) and allowing participants to enter the study at different times (i.e., their age at screening). Log-rank test was used to compare the survival curves between the FH+ and FH− groups; and, when applicable, median survival time and corresponding 95% CIs were estimated for each FH group separately.

For each time-to-event outcome of interest, Cox proportional hazards (PH) regression with robust standard errors was used to examine the effect of FH group membership and other risks/protective factors (i.e., SES, mother-youth relationship quality, father-youth relationship quality, peer risky behavior, neighborhood safety, and TLI) while adjusting for demographic information (age at screening, and sex). Two sets of Cox PH models were fitted. Model 1 fitted a model with the main effects only. Model 2 started by fitting a full model with all main effects and all first-order interactions between FH and each other risk/protective factors and demographic variables to examine the potential moderating effect of FH. Then an automatic backward model selection process was implemented to identify predictors including the interaction terms that were significant at a significance level of 0.1 (Harrell, 2015). The final reduced Model 2 included all predictors (including the interaction terms) from the automatic model selection process. If an interaction term was present, the corresponding main effects were always included in the model. The PH assumption was examined using Schoenfeld residuals after fitting each Cox PH model and the variance inflation factor was used to verify that there was no multicollinearity issue. In response to the misuses and misconceptions surrounding p-values, we reported both p-values and confidence intervals (CIs), as recommended by the American Statistical Association (Wasserstein & Lazar, 2016). This approach allows us to assess both the strength and precision of the evidence. The strength of the evidence is indicated by p-values, where smaller values suggest stronger evidence against the null hypothesis. Precision is conveyed through 95% CIs, with narrower intervals reflecting greater precision. By considering both p-values and 95% CIs, we can provide a more nuanced interpretation of the findings. All analyses were performed using Stata/SE (version 17).

Results

On average, participants were active in study follow-up visits for 7.90 years. For the present study, 82 participants voluntarily exited the study prior to engaging in either alcohol or cannabis use. There were no significant differences found between those who left the study and those who remained in the study until after they engaged in substance use, and the Cox models account for censoring. Table 1 provides descriptive information obtained at baseline for the study sample as a whole and by FH+ status. There were no significant differences between FH+ and FH− groups in sex, age at baseline, or race, but there were significant differences based on ethnicity. Additionally, there were significant differences across all six risk or protective factors such that FH− youth reported more positive parent-youth relationship quality, less neighborhood safety, and higher SES; FH+ youth reported more frequent affiliation with peers who engage in risky behaviors and had significantly higher behavioral phenotypic risk factors than FH− youth.

Risks of Alcohol Use Initiation

Over 1994.23 person-years of follow-up, 266 (68.7%) participants initiated alcohol use. Of these, 214 were FH+ (69.7% of the FH+ sample) and 52 were FH− (65.0% of the FH− sample). FH+ adolescents had a 36% higher rate of alcohol use initiation than FH− participants (Model 1 in Table 2). Furthermore, FH+ adolescents had an earlier age of alcohol use onset than adolescents who were FH− (log-rank test p = 0.013; Figure 1). By 17 years (95% CI: 16 to 17.5), 50% of FH+ participants had initiated alcohol use, while the median age was 18.5 years (95% CI: 18 to 19) for FH− participants.

Table 2.

Predictors of Alcohol Use Initiation

Predictors	Hazard Ratio	95% CI	p
Model 1: Main effects only
FH+ vs. FH−	1.36	[1.01,1.84]	.045
Sex (Male vs. Female)	0.94	[0.73,1.20]	.603
Age at baseline, year	1.14	[0.99,1.31]	.076
Mother-youth relationship quality	0.91	[0.41,2.00]	.811
Father-youth relationship quality	0.57	[0.31,1.06]	.077
Peer risky behavior	34.26	[8.64,135.81]	<.001
Neighborhood safety	0.89	[0.48,1.68]	.733
SES	1.18	[0.61,2.26]	.627
Behavioral phenotypes	1.04	[0.48,2.28]	.924
Model 2: Final model with three interaction terms
Peer risky behavior	38.61	[10.47,142.35]	<.001
Among FH -
Age at baseline	1.05	[0.84,1.33]	.641
Father-youth relationship quality	0.53	[0.13,2.16]	.375
Behavioral phenotypes	4.29	[0.58,31.91]	.155
Among FH +
Age at baseline	1.16	[0.98,1.36]	.083
Father-youth relationship quality	0.57	[0.30,1.09]	.089
Behavioral phenotypes	0.90	[0.40,2.02]	.805

FIGURE 1:

Survival trajectories of Alcohol Use Initiation for FH+ and FH− youth

The final Cox FH model (Model 2 in Table 2) included the large main effect of peer risky behavior and three interactions with FH status. Behavioral phenotype risk factors increased the risk of alcohol use for FH− youth but not for FH+ youth. Although the confidence of these estimates was low and non-significant, the effect size was more than four times larger among FH− youth compared with FH+ youth. Additionally, for FH+ youth, youth who were older at study enrollment were more likely to initiate alcohol use. The final model indicates that the protective effect of positive father-youth relationship quality on AUI was similar for FH− and FH+ youth, although the precision for FH+ youth is much higher than it is for FH− youth (HR[95% CI] = 0.57 [0.30, 1.09], p=0.089 for FH+; HR[95% CI] = 0.53 [0.13, 2.16], p=0.375 for FH−).

Risk of Cannabis Use Initiation

Over 2,116.63 person-years of follow-up, 220 (56.85%) participants engaged in cannabis initiation. Of these, 192 were FH+ (62.54% of the FH+ sample) and 28 were FH− (31.25% of the FH− sample). FH+ adolescents were 2.36 times more likely to initiate cannabis use than those who were FH− (Model 1 in Table 3). Furthermore, FH+ adolescents had an earlier age of CUI than adolescents who were FH− (log-rank test p <0.001; Figure 2). By age 17.5 years (95% CI: 16.5 to 18), the median age of initiation was 22 years for the FH− group (95% CI: 19.5 to > 24 unobserved in the study period).

Table 3.

Predictors of Cannabis Use Initiation

Predictors	Hazard Ratio	95% CI	p
Model 1: Main effects only
FH+ vs. FH−	2.36	[1.58,3.54]	<.001
Sex (Male vs. Female)	1.17	[0.89,1.52]	.259
Age at baseline, year	1.04	[0.89,1.22]	.602
Mother-youth relationship quality	0.54	[0.22,1.30]	.169
Father-youth relationship quality	0.59	[0.29,1.20]	.146
Peer risky behavior	15.73	[2.87,86.04]	.001
Neighborhood safety	0.74	[0.43,1.30]	.300
SES	0.73	[0.37,1.43]	.359
Behavioral phenotypes	1.22	[0.56,2.67]	.614
Model 2: Final model with one interaction term
Peer risky behavior	13.99	[2.50,78.45]	.003
Among FH -
Mother-youth relationship quality	0.11	[0.01,1.22]	.072
Among FH +
Mother-youth relationship quality	0.45	[0.19,1.09]	.078

FIGURE 2:

Survival trajectories of Cannabis Use Initiation for FH+ and FH− youth

The final Cox FH model (Model 2 in Table 3) included the large main effect of peer risky behavior and one interaction with FH status. The final model indicates the protective effect of positive mother-youth relationship quality on CUI for both FH− and FH+ youth. Positive mother-youth relationship quality had a greater protective effect for FH− youth than FH+ youth (a one-unit increase in mother-youth relationship quality reduced the hazard of CUI by 89% for FH− youth but only 55% for FH+ youth), although not statistically significant at 5% level (HR=0.45 for FH+ vs. HR=0.11 for FH−, a ratio of HRs = 4.20, p=0.268 ) —indicating that FH+ status may attenuate the association between positive mother-youth relationship quality and CUI.

Discussion

The present study was designed to understand how the predictive association between social and local-home factors and SUI was moderated by FH+ status, and how these associations may differ for alcohol (AUI) and cannabis (CUI) use initiation. Across the observation period, FH+ youth were at greater risk for SUI than FH− youth. However, when controlling for potential genetic heritability via behavioral phenotypes (though the final model did not find strong evidence of behavioral phenotypes predicting SUI, despite the AUI model indicating a non-statistically significant but clinically large risk for FH− youth with large variation), there was evidence that social relationships, rather than local-home factors, served as salient predictors of SUI risk. Broadly, the results found that, regardless of FH+ status, youth who affiliated with risky peers were significantly more likely to engage in SUI. In contrast, youth who reported better relationship quality with their parents were less likely to engage in SUI for both FH− and FH+ groups, although this finding was not statistically significant. Yet, as evidenced in the CUI model, the protective nature of positive mother-youth relationship quality on CUI may be attenuated for FH+ youth, although this difference between FH+ and FH− was also not statistically significant.

The distinct findings for mother- and father-youth relationship quality when examined simultaneously, we were able to provide a more nuanced understanding of the unique impact that mothers and fathers make to predict the overall risk of SUI. The present study adds to a growing literature that mothers and fathers play unique roles in socializing behavior toward their children (Erickson, 2015; Jeynes, 2016). For both FH+ and FH− youth, positive father-youth relationships marginally decreased the risk of alcohol initiation. This decreased risk may be attributable to the greater stability and open discussions about substance use which may decrease youth likelihood of early alcohol initiation—even among FH+ youth (Seilhamer et al., 1993). In contrast, whereas relationship quality with fathers may deter AUI, relationship quality with mothers may not predict AUI because, regardless of FH status, mothers typically spend more time in contact with their children, and have more opportunities to socialize expectations about alcohol use. On the other hand, when considering CUI, regardless of FH status, having a positive relationship with mothers may dissuade youth from seeking out other forms of social support and decreases the opportunity to engage in cannabis use (Hundleby & Mercer, 1987). It is important to note that, in the final model assessing for CUI, the difference in hazard ratios for mother-youth relationship quality between FH+ and FH− is sizeable, but the substantial overlap in the confidence intervals suggests that this difference is not statistically significant.

Although the present study found differences in the impact of mothers and fathers on SUI risk, it is important to discuss these findings within the context of the present sample, where all of the FH+ participants (79.33% of the sample; 100% of the FH+ group) had a biological father with a SUD. Our data indicated that, although being FH+ may attenuate the protective nature of parent-youth relationships on SUI risk, particularly when examining CUI, positive relationships with mothers and fathers are still important deterrents of risk. When examining the results in the AUI model, positive father-youth relationship quality was found to decrease AUI risk for both FH+ and FH− youth. This may indicate that – particularly for FH+ youth – fostering positive family relations may still be important for deterring youth from SUI.

Additionally, whereas there was evidence for moderation when examining the association between parent-youth relationship quality and SUI risk, all youth – regardless of FH+ status – were at increased risk for both AUI and CUI if they affiliated with risky peers. This data supports prior findings that youth behavior mirrors that of their peers. However, in the present study, we are unable to tell whether the association between peer risky behavior and SUI was due to socialization or selection effects. Some researchers would argue that substance naïve youth (like the current sample was at baseline) may be encouraged by risky peers to partake in alcohol and cannabis use, and therefore, if they have more risky peers, they are at greater risk of SUI (i.e., socialization effects; Ennett et al., 2006; Monahan et al., 2009). Others may argue that even if youth are substance naïve, they may already be engaging in risky behaviors and may purposely seek out peers who engage in similar levels of risky behavior (i.e., selection effects; Bryant & Zimmerman, 2002; Ennett et al., 2006). Therefore, it may be that youth inclined to early SUI may also be likely to affiliate with risky peers. Yet, exploration of the motivations behind these friendships is outside the scope of the present study.

Interestingly, while we found support for the hypotheses regarding relationships with parents and peers, we did not find evidence that local-home factors can predict SUI risk. Despite previous work showing adolescents who are raised in lower SES families and in more dangerous neighborhoods are at increased risk of substance use initiation (Leventhal & Brooks-Gunn, 2000; Owens, 2017), local-home factors were not retained in our hazard models when placed in a model that examined FH+ status, social relationships, and local-home factors simultaneously. The lack of significant findings for local-home factors as a predictor of SUI risk may potentially be due to “spillover effects.” Spillover effects occur when the emotions or behaviors that arise from stressful environments and/or relationship “spillover” to impact other relationships and associations (Gerard et al., 2006; Liu et al., 2023). Although stressors in the local-home environment, such as financial insecurity or neighborhood danger, may place youth at risk for SUI, it may also be true that stressors in the local-home environment may also impact more proximal predictors of SUI, such as social relationships with parents or peers (Brody et al., 2001; Winslow & Shaw, 2007). We may find that youth who have negative relationships with their parents and engage more frequently with risky peers have significantly more local-home stressors than youth who have positive relationships with their parents and engage less frequently with risky peers. Therefore, in the present analyses, there is a chance that when examined simultaneously with these social relationships, local-home factors were not as strong of predictors because they enhanced the strength of the social predictors of SUI.

Limitations

Although the present study supports and extends the current adolescent substance use literature, there are also limitations that must be considered. Primarily, while this was a longitudinal study, the social and local-home predictors used in the present analyses were only collected at baseline, usually pre-adolescence and always before initiation of substance use. As we know, adolescence is a period of rapid social change and development. Although the present analyses showed the effects of parents and peers, the current analysis does not capture the dynamic nature of these factors and how those may change over time. Therefore, we are unable to understand the specific periods during which these relationships are most impactful at impacting SUI risk, nor are we able to determine if there are periods in which one relationship may be more salient than another. For example, research has found that peers overall influence on youth early SUI risk may be stronger in early adolescence than later adolescence (Beardslee et al., 2018). Future research should utilize repeated measure assessments of the family, social, and environmental factors surrounding the adolescent to create a more comprehensive understanding of the exacerbating and attenuating factors affecting risk for early substance use initiation. Furthermore, for the present sample, we oversampled FH+ youth, particularly those who had a father with SUD, therefore, our sample may have been biased toward earlier SUI. Additionally, the original study was not powered to detect moderator effects, therefore lack of statistical significance in the present analyses does not necessarily indicate the absence of any effect. Additionally, similar to the measures of social and local-home predictors, FH status was only assessed at baseline. Therefore, there is a possibility that youth who were FH− at study entry may have transitioned to being FH+ across the observation window. Furthermore, adolescent substance use initiation was measured via adolescent self-reports. Although it seems unlikely that participants would have reported the use of alcohol or cannabis when they had not used them, adolescents might not have accurately reported when they first began using alcohol and cannabis due to parental involvement and expectancy effects. Lastly, it is important to note that, by using automatic backward model selection, the specification of the final reported models was data-driven. To account for this, all results in the final model were carefully reviewed to ensure that, when interaction terms were included, the corresponding main effects were retained, regardless of their p-values, but, still, the limitation needs to be acknowledged.

Conclusion

In summary, the present study was designed to understand the risk and protective factors of adolescent substance use for FH+ and FH− youth. Regardless of FH status, affiliating with risky peers significantly increased risk for SUI, but positive parent-youth relationship quality decreased this risk. Additionally, finding separate patterns for AUI and CUI suggests that mothers and fathers play separate and unique roles in predicting SUI risk. However, because the protective nature of positive parent-youth relationship quality was sometimes attenuated in FH+ youth, these findings may underscore the importance of peer-focused interventions (e.g., cooperative learning; (Ryzin et al., 2020) designed to reduce the impact of negative experiences (e.g., having a family member with a SUD) on youth outcomes.

Data availability statement:

The raw data, analysis code, and materials used in this study are not openly available but are available upon request to the corresponding author.