Adverse childhood experiences and past 30-day cannabis use among middle and high school students: the protective influence of families and schools

Abstract

Exposure to adverse childhood experiences (ACEs) is a risk factor for adolescent cannabis use (CU). We explored whether family communication and school connectedness can offer direct protection (the compensatory model of resiliency) or moderating protection (the protective factors model of resiliency).

Using cluster random sampling, a Youth Risk Behavior Survey (YRBS) was conducted with 5,341 middle school and 4,980 high school students in 2019. Generalized estimating equations were used to estimate whether family communication and school connectedness offered independent direct protection (multiple regression) or moderating protection (multiplicative interaction) in the relationship between ACEs and past 30-day CU. Adjusted prevalence ratios (APR) and 95% confidence intervals (95% CI) were calculated.

There was a graded relationship between ACEs and past 30-day CU for all students that was particularly strong among middle school students: 1 ACE (APR=2.37, 95% CI=2.16, 2.62), 2 ACEs (APR=2.89, 95% CI=2.60, 3.23), 3ACEs (APR=5.30, 95% CI=4.75, 5.90), 4+ACEs (APR=7.86, 95% CI=7.13, 8.67). Results supported the compensatory model of resiliency with both family communication (middle school APR=0.90, 95% CI=0.88, 0.93; high school APR=0.90, 95% CI=0.87, 0.93) and school connectedness (middle school APR=0.76, 95% CI=0.72, 0.79; high school APR=0.72, 95% CI=0.68, 0.77) demonstrating a direct, independent protective relationship with past 30-day CU. There was no consistent evidence supporting the protective factors resiliency model.

INTRODUCTION

Over the past five years, there has been a significant decrease in past 30-day use of cigarettes, alcohol, and most illicit drugs among adolescents, but 30-day cannabis use (CU) has remained stable.¹ In 2019, 6.6% of 8^th graders, 18.4% of 10^th graders, and 22.5% of 12^th graders reported 30-day CU.¹ Furthermore, using cannabis 20+ times in the past 30-days (frequent use) increased by 85.7% among 8^th graders and 41.2% among 10^th graders between 2018 and 2019.² Among high school seniors with a history of CU, frequent use has also increased significantly since the early 1990s, suggesting that youth who experiment with cannabis at an early age may be at greater risk of progressing to frequent use.³

Initiating CU during adolescence increases the probability of developing a substance use disorder, with greater risk associated with use that is more frequent.⁴ Adolescents who use cannabis regularly are also at higher risk for major depression and suicidality⁵ and have increased rates of poor academic achievement,⁶ justice system involvement,⁷ and negative social outcomes during adulthood.⁸ Increased cannabis potency and access to vaping, a mode of administration that is easier to conceal than smoking, may also increase the risk of developing dependence during this vulnerable developmental period.^9,10 To inform substance use prevention, it is important to identify factors that increase the probability of CU among adolescents as well as modifiable protective factors that can ameliorate the influence of risk factors.

A growing body of research shows that exposure to adverse childhood experiences (ACEs), contributes to CU during adolescence.^11–15 ACEs include traumatic events that often occur in a family setting such as abuse, neglect, or witnessing domestic violence.¹⁶ Children are often exposed to multiple ACEs and there is evidence that cumulative trauma exposure has a greater impact on health outcomes than individual stressors.¹⁷ A cross sectional study with 126,868 students (8–11 grade) found a graded relationship between the number of ACEs students experienced and CU after adjustment for demographics and contextual factors.¹² Furthermore, a national prospective study showed that compared to youth with no ACEs, the adjusted odds of lifetime CU during adolescence increased approximately 60% for every increase in the number of ACEs.¹³

Despite the strong relationship between cumulative ACE exposure and CU, many youth display resilience and have the ability to adapt to chronic adversity. Resiliency can operate within different socio-ecological domains (e.g., individual, interpersonal, community, social) highlighting multiple levels of positive influence.¹⁸ Examples of sources of strength that can protective youth from CU include religiosity,^19,20 social skills,²⁰ having goals and aspirations,²¹ prosocial peer involvement,²¹ peer disapproval of CU,^22,23 and community involvement.¹⁹ Positive family and school environments are particularly influential during adolescent development^24,25 and could play an important role in protecting ACE-exposed youth engaging in CU.

The connection youth have with their parents and other family members, including the ability to communicate openly, can protect youth from a range of behavioral health problems.^26–28 In addition, youth spend a large portion of their lives in school and the acceptance and support they receive from their school community has a protective influence on their risk-taking behaviors.^26,27 Numerous studies have demonstrated that positive family communication and supportive school environments reduce adolescent CU.^{21,26,29–37} Family communication and school connectedness are modifiable protective factors that may decrease the risk of CU among adolescents, but limited research has evaluated such protection in the context of ACE exposure, which is a strong risk factor for CU.³⁸

One study used data from the Longitudinal Studies of Child Abuse and Neglect (LONGSCAN) to assess the moderating influence of quality of relationships with parents on the association between child maltreatment and past year CU.³⁹ The authors did not find that child maltreatment was associated with CU, possibly because most of the adolescents had been reported to child protective services (CPS) by age 12. However, the study did find that for adolescents who experienced child maltreatment, a positive relationship with the father was associated with less CU.³⁹ Another study followed maternal-child dyads for 18 years and assessed whether school engagement moderated the influence of maternal depression on alcohol and/or CU.⁴⁰ There was a strong relationship between maternal depression and adolescent substance use, but school engagement did not moderate this relationship. Both studies used ACE measures with limited scope (e.g., using CPS reports as a proxy for child maltreatment, only investigating one ACE [maternal depression]) and are generalizable to youth at high risk for childhood trauma and poor behavioral health outcomes, not the general U.S youth population.

There is also a need to investigate whether family and school factors can mitigate the influence of ACEs on CU across adolescent developmental periods as the effects of risk and protective factors may differ as youth mature.⁴¹ Early adolescence (7^th grade) and the transition to high school (9^th grade) are periods where risk factors may have greater influence on CU.⁴² Limited research has investigated whether ACEs influence substance use differently in early versus late adolescence, but one study found much stronger correlations between sexual abuse and substance use among students in 6^th grade compared to 12^th grade.⁴³ There is also evidence that different protective factors have a stronger influence across adolescent development periods. A study with students in grades 6–12 found that family protective factors were more influential in earlier grades, whereas school protective factors became more important for older adolescents.⁴¹ Similarly, another study found that talking to parents about personal problems was associated with lower rates of cannabis initiation as youth moved from grade 7 to grade 8, but was no longer protective when youth transitioned to later grades.⁴⁴

To address these limitations, we used data from representative state samples of middle and high school students and tested two common models of resilience: 1) the compensatory model and 2) the protective factors model. The compensatory model of resiliency highlights protective factors that work in the opposite direction of a risk factor. Such protective factors have a direct and independent effect on the outcome and may help youth compensate for the potential harms associated with chronic stressors.⁴⁵ This model suggests that family communication and school connectedness have a protective relationship with CU, separate from the influence of ACEs. The protective factors model posits that protective factors will moderate or buffer the relationship between risk factors and outcomes.⁴⁵ According to the protective factors model, the relationship between ACEs and CU would be reduced among adolescents with higher levels of family communication and school connectedness.⁴⁵ Using these models as a framework, the objectives of our study were to: 1) describe the relationship between ACE exposure and past 30-day CU in representative state samples of middle and high school students; 2) assess whether family communication and school connectedness have a direct protective relationship with past 30-day CU regardless of ACE exposure (compensatory model); and 3) determine whether family communication and school connectedness moderate/buffer the relationship between ACE exposure and past 30-day CU (protective factors model).

MATERIAL AND METHODS

Study design, setting, and participants

The Youth Risk Behavior Survey (YRBS) is a national, school-based surveillance system established by the Centers for Disease Control and Prevention (CDC) in 1991 to monitor the risk behaviors among high school youth through bi-annual cross-sectional surveys. A small number of states also conduct the YRBS in middle schools.⁴⁶ Data for the current study were obtained from a middle and high school YRBS conducted February-May 2019 in Nevada.

A cluster random sampling procedure was used to select classrooms from middle schools (6–8 grade) and high schools (9–12 grade); all regular public, charter, and alternative secondary schools in the state were included. First, the seventeen school districts in the state were grouped into eight regions that align with the state’s substance use prevention coalition structure. Second, a random sample of 2^nd period or required English classrooms were randomly selected from all middle and high schools. The middle and high school samples were drawn separately, using sampling weights appropriate for each. Passive parental consent (10 school districts) or active parental consent (7 school districts) was obtained before survey administration. An anonymous survey was administered to students in randomly selected classrooms. Students could decline participation or skip any question.

In 2019, 5,341 Nevada middle school students from 113 schools completed the survey, resulting in a school response rate of 95.0%, a student response rate of 70.4%, and a combined response rate of 67.2%. The high school survey was completed by 4,980 students from 98 schools, with a school response rate of 98.0%, student response rate of 68.6%, and a combined response rate of 67.3%. The study was approved by a university institutional review board and school district institutional review boards when required.

Measures

Outcome: Past 30-day CU

A standard CDC-YRBS question assessed past 30-day CU: “During the past 30 days, how many times did you use marijuana?” Response categories included 0 times, 1–2 times, 3–9 times, 20–39 times, and 40+ times; responses were dichotomized (use versus no use).⁴⁶

Primary Exposure: Adverse childhood experiences (ACEs)

The CDC-YRBS survey includes a measure of lifetime sexual abuse. In addition, five state-added variables were adapted from the Behavioral Risk Factor Surveillance System (BRFSS) ACE module⁴⁷ to assess lifetime prevalence of physical abuse by an adult, verbal abuse by an adult, witnessing household domestic violence, household mental illness, and household substance abuse. Responses to all ACE questions were dichotomized (yes versus no) except verbal abuse, which was coded as yes if it occurred sometimes, most of the time, or always, which is consistent with recommended scoring from the CDC.⁴⁸ Of the 5,341 middle school students who participated in the survey, 92.9% of participants answered all six ACE questions and 99.8% answered at least four questions. Of the 4,980 high school students who participated in the survey, 91.1% answered all ACE questions and 99.2% answered at least four questions. The ACE questions were summed (range=0–6). Consistent with BRFSS scoring, we categorized the ACE measure as 0, 1, 2, 3, and 4+.⁴⁹ Based on the relationship with the outcome, we collapsed ACE categories to 0, 1–2, 3+ for interaction analyses to ensure adequate sample size.

Moderators

Family communication.

Three items adapted from the Youth Asset Survey (YAS)²⁷ assessed family communication. Example: “How often do you talk to your parents or other adults in your home about your problems?” Responses ranged from never (0) to always (4) on a five-point Likert scale. The item reliability was similar for the middle (Cronbach’s α=0.79) and high school samples (Cronbach’s α=0.81). In the middle school sample, 95% of participants answered all family communication questions and 97.8% answered at least one question. Among high school students, 94% answered all family communication questions; 94.9% answered at least one question. Following recommendations from the survey developers,²⁷ responses from youth who answered at least one of the questions were summed and divided by the number of answered items to create a family communication score, with higher scores reflecting greater family communication (range=0–4). The mean and median scores were similar and consistent in the middle and high school samples.

School connectedness.

Three items adapted from YAS²⁷ assessed school connectedness. Example: “How often do you feel close to people at your school?” Responses ranged from never (0) to always (4) on a five-point Likert scale. Reliability estimates were slightly lower in the middle school sample (Cronbach’s α=0.55) compared to the high school sample (Cronbach’s α=0.64). In the middle school sample, 95.5% of participants answered all school connectedness questions and 97.7% answered at least one question. Among high school students, 93.6% answered all school connectedness questions; 94.9% answered at least one question. The school connectedness score (range=0–4) was computed similarly to the family communication score.²⁷ The mean and median school connectedness scores were similar and consistent across samples.

Covariates

Sociodemographic covariates included sex, age, race/ethnicity, and qualification for free or reduced lunch as a proxy for income. We also controlled for differences school district parental consent (active versus passive).

Statistical Analysis

Due to different weights and clustering in the middle and high school samples, analyses were conducted separately. All analyses accounted for the complex survey design of the YRBS.⁵⁰ For both samples, data were weighted based on sex, grade, and race/ethnicity among youth within each of the eight regions and we accounted for clustering at regional and classroom levels. Overall, the amount of missing data on individual measures was well below 5% for both samples except for the high school family communication and school connectedness scores (5.1% missing). We used Little’s test statistic to assess if the data were missing completely at random (MCAR). The values were significant (p<.05) for middle and high school samples; therefore, we completed multiple imputation rather than complete case sample analyses.⁵¹ Multiple imputation was performed using chained equations through the fully conditional method (FCS)⁵² for past 30-day CU, ACEs, family communication, school connectedness, and all covariates. Twenty imputations were performed.

We used the Spearman rank correlation coefficient to assess the strength of association between ACE exposure and family communication, ACE exposure and school connectedness, and family communication and school connectedness (Table 2). Weighted chi-square analyses and median tests assessed the relationship between sociodemographics, ACE exposure, family communication, and school connectedness with past 30-day CU (Table 3). Next (Table 4), generalized estimating equation models with an unstructured correlation structure and robust variance estimators were used to estimate the relationship between ACE exposure and family communication on the weighted prevalence of past 30-day CU. Model 1a included ACE exposure only, model 2a included family communication, and model 3a included ACE exposure and family communication. The same models were repeated using school connectedness (models 1b-3b). For the interaction analyses (Table 4) the models included ACE exposure, the protective factor (family communication or school connectedness), and a multiplicative interaction term (ACE exposure × family communication or school connectedness). Based on the relationship with the outcome, ACE exposure was collapsed to 0, 1–2, and 3+ and the protective factors were dichotomized at the median to ensure adequate sample size for interaction analyses. Adjusted prevalence ratios (APR) and corresponding 95% confidence intervals (95% CI) were calculated for all models, adjusting for sociodemographic variables and parental permission status. We used SAS version 9.4 (SAS Institute, Cary NC) for all analyses.

RESULTS

Table 1 shows the weighted prevalence of all variables for the middle and high school samples. The most common racial ethnic group was Hispanic (44.9% of middle school and 42.9% of high school students), followed by non-Hispanic white (29.5% of middle school and 32.0% of high school students). About 42% of both samples qualified for a free or reduced free lunch program. Over half (55.5%) of middle school and almost two-thirds (64.1%) of high school students experienced at least one ACE, and high ACE exposure (4+ ACEs) was reported by 6.8% of middle and 9.9% of high school students. The mean and median family communication and school connectedness scores were similar for both samples and past 30-day CU was reported by 7.9% of middle school and 18.5% of high school students.

Table 1.

Weighted characteristics and frequency of missing data for middle school and high school survey participants

	Middle School (N=5,341)		High School (N=4,980)
	N (%)a	Missing (%)	N (%)a	Missing (%)
Sex
Female	2825 (48.7)	38 (0.7)	2607 (48.9)	32 (0.6)
Male	2478 (51.3)		2341 (51.1)
Race/ethnicity
Hispanic	2208 (44.9)	210 (3.9)	1986 (42.9)	125 (2.5)
Non-Hispanic black	254 (11.5)		238 (10.3)
Non-Hispanic other	779 (14.1)		727 (14.7)
Non-Hispanic white	1890 (29.5)		1904 (32.0)
Grade
6th grade	1460 (30.9)	53 (1.0)	---
7th grade	2116 (34.7)		---
8th grade	1712 (34.4)		---
9th grade	---		1317 (26.0)	40 (0.8)
10th grade	---		1341 (25.9)
11th grade	---		1263 (25.0)
12th grade	---		1019 (23.1)
Location of residence
Rural	1716 (8.6)	0 (0.0)	1692 (9.3)	0 (0.0)
Urban	3625 (91.4)		3288 (90.7)
Free or reduced lunch
Yes	2014 (42.6)	48 (0.9)	1832 (42.2)	61 (1.2)
No	3279 (57.4)		3087 (57.8)
ACE score
0 ACEs	2360 (44.5)	13 (0.1)	1774 (35.9)	20 (0.1)
1 ACE	1266 (24.7)		1244 (26.2)
2 ACEs	813 (15.8)		859 (17.2)
3 ACEs	458 (8.2)		534 (10.8)
4+ ACEs	431 (6.8)		549 (9.9)
Family communication (range=0–4)
Mean (SD)	1.8 (1.2)	118 (2.2)	1.8 (1.1)	252 (5.1)
Median	1.7		1.7
School connectedness (range=0–4)
Mean (SD)	2.4 (0.9)	125 (2.3)	2.2 (0.9)	256 (5.1)
Median	2.3		2.3
Past 30-day cannabis use
Yes	420 (7.9)	125 (2.3)	943 (18.5)	407 (3.8)
No	4796 (92.1)		3846 (81.5)

^aWeighted column %.

May not always add to 100 due to rounding.

The correlations between ACE exposure, family communication, and school connectedness are displayed in Table 2. ACEs were not highly correlated with family communication or school connectedness, but associations were in the direction anticipated (as ACE exposure increased, family communication and school connectedness scores decreased).

Table 2.

Correlations between adverse childhood experience (ACEs), family communication, and school connectedness

Middle School
	ACE Score	Family Communication	School Connectedness
ACE Score	---	−0.291***	−0.224***
Family Communication	−0.291***	---	0.314***
School Connectedness	−0.224***	0.314***	---
High School
	ACE Score	Family Communication	School Connectedness
ACE Score	---	−0.276***	−0.211***
Family Communication	−0.276***	---	0.288***
School Connectedness	−0.211***	0.288***	---

Note: Reported estimates are Spearman correlation coefficients.

^***p<.001

Table 3 shows the unadjusted relationships between sociodemographic variables and ACE exposure by past 30-day CU. In the middle school sample, students who reported past 30-day CU were more likely to be female (p=.007), Hispanic (p=.023), in the 8^th grade (p<.001), and qualified for free/reduced lunch (p<.001). Among high school students, past 30-day cannabis users were more likely to be non-Hispanic black (p<.001), in higher grades (p=.047) and qualified for free/reduced lunch (p<.001). In both samples, there was a strong and graded association between ACEs and past 30-day CU (p<.001) and students who reported past 30-day CU had lower family communication (p<.001) and school connectedness scores (p<.001).

Table 3.

Weighted characteristics of middle school and high school participants by past 30-day cannabis use

	Middle School (N=5,341)			High School (N=4,980)
	Past 30-Day Cannabis Use			Past 30-Day Cannabis Use
	Yes	No	p-value	Yes	No	p-value
	n (%)a	n (%)a		n (%)a	n (%)a
Sex
Female	258 (57.1)	2586 (47.9)	0.007	529 (51.2)	2094 (48.4)	0.144
Male	177 (42.9)	2320 (52.1)		464 (48.8)	1891 (51.6)
Race/ethnicity
Hispanic	224 (53.9)	2068 (44.1)	0.023	437 (45.3)	1601 (42.3)	<0.001
black	23 (11.5)	243 (11.4)		70 (14.6)	173 (9.3)
other	61 (11.4)	750 (14.3)		116 (10.2)	630 (15.8)
white	126 (23.1)	1846 (30.1)		372 (29.8)	1581 (32.5)
Grade
6th grade	34 (10.3)	1438 (32.6)	<0.001	---	---	0.047
7th grade	168 (34.9)	1970 (34.7)		---	---
8th grade	232 (54.7)	1499 (32.6)		---	---
9th grade	---	---		230 (23.6)	1097 (26.6)
10th grade	---	---		215 (21.7)	1137 (26.9)
11th grade	---	---		287 (29.8)	986 (24.0)
12th grade	---	---		262 (24.9)	766 (22.7)
Free or reduced lunch
Yes	233 (57.3)	1811 (41.4)	<0.001	443 (49.9)	1413 (40.4)	<0.001
No	211 (42.7)	3096 (58.6)		551 (50.1)	2573 (59.6)
ACE score
0 ACEs	65 (15.9)	2298 (46.9)	<0.001	180 (18.7)	1599 (39.8)	<0.001
1 ACE	94 (23.1)	1175 (24.9)		221 (25.4)	1027 (26.4)
2 ACEs	80 (18.9)	736 (15.6)		202 (19.4)	661 (16.6)
3 ACEs	82 (18.2)	379 (7.3)		161 (16.5)	377 (9.6)
4+ ACEs	114 (24.0)	319 (5.3)		231 (20.0)	322 (7.6)
Family communication (range=0–4)
Mean (SD)	1.3 (0.1)	1.8 (0.1)	<0.001	1.5 (0.1)	1.8 (0.1)	<0.001
Median	1.0	1.7		1.3	2.0
School connectedness (range=0–4)
Mean (SD)	2.0 (0.1)	2.4 (0.1)	<0.001	1.8 (0.1)	2.2 (0.1)	<0.001
Median	2.0	2.3		2.0	2.3

^aWeighted column %.

Does not always add to 100 due to rounding.

Compensatory Model of Resiliency

Table 4 depicts the individual and combined influence of ACE exposure and protective factors (family communication, school connectedness) on past 30-day CU, controlling for covariates. In both samples, there was a graded relationship between ACEs and CU (Models 1a and 1b; p<.001); however, the relationship was stronger among middle school students. Compared to middle school students with no ACEs, those with 1 ACE (APR=2.37, 95%CI=2.16, 2.62), 2 ACEs (APR=2.89, 95%CI=2.60, 3.23), 3ACEs (APR=5.30, 95%CI=4.75, 5.90), 4+ACEs (APR=7.86, 95%CI=7.13, 8.67) had higher prevalence of past 30-day CU. Family communication and school connectedness had a strong protective association with past 30-day CU (Models 2a and 2b; p<.001). When ACE exposure and family communication were included in the same model to evaluate the compensatory model of resiliency the influence of family communication was attenuated, but remained protective (Model 3a; p<.001). For each one-unit increase in the family communication score, there was 10% lower prevalence of past 30-day CU for middle school (APR=0.90, 95%CI=0.88) and high school students (APR=0.90, 95%CI=0.87, 0.93). The analyses including school connectedness and ACEs showed greater protective associations (model 3b; p<.001): for every one-unit increase in the school connectedness score, there was 24%–28% lower prevalence of past 30-day CU for middle school (APR=0.76, 95%CI=0.72, 0.79) and high school students (APR=0.72, 95%CI=0.68).

Table 4.

Compensatory model of resiliency: Direct, independent associations between ACE exposure and protective factors and past 30-day cannabis use

	Middle School	High School
	Past 30-Day Cannabis Use	Past 30-Day Cannabis Use
	APR (95% CI)	APR (95% CI)
Model 1a. ACE score a,b
0 ACEs	Ref	Ref
1 ACE	2.37 (2.16, 2.62)***	1.81 (1.70, 1.93)***
2 ACEs	2.89 (2.60, 3.23)***	2.15 (1.99, 2.33)***
3 ACEs	5.30 (4.75, 5.90)***	2.89 (2.68, 3.11)***
4+ ACEs	7.86 (7.13, 8.67)***	3.71 (3.50, 3.93)***
Model 2a. Family communication a,b,c	0.76 (0.72, 0.77)***	0.81 (0.79, 0.84)***
Models 3a. ACE score + family communication a,b,c
ACE Score
0 ACEs	Ref	Ref
1 ACE	2.28 (2.07, 2.51)***	1.76 (1.65, 1.87)***
2 ACEs	2.71 (2.41, 3.04)***	2.05 (1.90, 2.22)***
3 ACEs	4.92 (4.40, 5.50)***	2.70 (2.50, 2.92)***
4+ ACEs	7.01 (6.33, 7.76)***	3.39 (3.21, 3.58)***
Family communicationc	0.90 (0.88, 0.93)***	0.90 (0.87, 0.93)***
Model 1b. ACE score a,b
0 ACEs	Ref	Ref
1 ACE	2.37 (2.16, 2.62)***	1.81 (1.70, 1.93)***
2 ACEs	2.89 (2.60, 3.23)***	2.15 (1.99, 2.33)***
3 ACEs	5.30 (4.75, 5.90)***	2.89 (2.68, 3.11)***
4+ ACEs	7.86 (7.13, 8.67)***	3.71 (3.50, 3.93)***
Model 2b. School connectedness a,b,c	0.65 (0.63, 0.68)***	0.67 (0.63, 0.71)***
Model 3b. ACE score + school connectedness a,b,c
ACE Score
0 ACEs	Ref	Ref
1 ACE	2.25 (2.04, 2.48)***	1.70 (1.60, 1.81)***
2 ACEs	2.71 (2.42, 3.04)***	1.94 (1.79, 2.10)***
3 ACEs	4.74 (4.26, 5.27)***	2.56 (2.35, 2.79)***
4+ ACEs	6.73 (6.07, 7.45)***	3.28 (3.07, 3.50)***
School connectednessc	0.76 (0.72, 0.79)***	0.72 (0.68, 0.77)***

Note: APR = adjusted prevalence ratio. CI = confidence interval

^aAll models were weighted and accounted for classroom- and regional-level clustering.

^bAll models adjusted for sex, grade, race, free and reduced lunch status, and parental permission status.

^cContinuous measure of family communication and school connectedness.

^***p<.001

Protective Factors Model of Resiliency

In the analyses that assessed the protective model of resiliency (Table 5), there was no consistent evidence that family communication or school connectedness moderated or buffered the relationship between ACEs and past 30-day CU. All levels of ACE exposure continued to be associated with CU (p<.001), regardless of the level of family communication or school connectedness.

Table 5.

Protective factors model of resiliency: Multiplicative interaction between ACE exposure and protective factors on past 30-day cannabis use

	Past 30-Day Cannabis use
	High Family Communicationa	Low Family Communication
	APR (95% CI)	APR (95% CI)
Middle School ACE score × family communication b,c
0 ACEs	Ref	2.79 (1.45, 5.36)**
1 – 2 ACEs	4.10 (2.42, 6.96)***	4.27 (2.47, 7.36)***
3+ ACEs	10.96 (6.15, 19.53)***	10.23 (5.94, 17.60)***
High School ACE score × family communication b,c
0 ACEs	Ref	1.75 (1.09, 2.80)*
1 – 2 ACEs	2.33 (1.75, 3.12)***	2.50 (1.82, 3.42)***
3+ ACEs	3.42 (2.43, 4.80)***	4.47 (3.31 ,6.03)***
	Past 30-Day Cannabis use
	High School Connectednessa	Low School Connectedness
	APR (95% CI)	APR (95% CI)
Middle School ACE score × school connectedness b,c
0 ACEs	Ref	2.85 (1.39, 5.85)**
1 – 2 ACEs	3.69 (2.11, 6.46)***	4.59 (2.69, 7.81)***
3+ ACEs	8.19 (4.72, 14.21)***	11.79 (6.91, 20.10)***
High School ACE score × school connectedness b,c
0 ACEs	Ref	2.60 (1.72, 3.93)***
1 – 2 ACEs	2.50 (1.76, 3.57)***	3.83 (2.35, 5.54)***
3+ ACEs	4.16 (2.86, 6.05)***	6.23 (4.40, 8.82)***

Note. APR = adjusted prevalence ratio. CI = confidence interval.

^aScore at or above the median=high family communication and high school connectedness

^bAll models were weighted and accounted for classroom- and regional- level clustering.

^cAll models adjusted for sex, grade, race, free and reduced lunch status, and parental permission status.

^*p<.05;

^**p<.01;

^***p<.001

DISCUSSION

The purpose of this study was to describe the relationship between exposure to ACEs and past 30-day CU among middle school and high school students, and to determine whether family communication and school connectedness can offer direct, independent protection (compensatory model of resiliency) or moderating protection (protective factors model of resiliency). Consistent with previous research,^12,13 we found a graded relationship between ACE exposure and past 30-day CU among students, which was particularly strong among middle school students. The mechanism by which childhood adversity influences CU during early adolescence could not be ascertained in our study, but youth may initiate substance use at an early age as a coping mechanism to deal with emotional distress caused by their exposure to trauma.⁵³ This finding has important prevention implications, as the earlier CU is initiated, the greater its impact on physical, mental, and social health outcomes across the lifespan.⁵⁴ Early intervention may most effectively mitigate the influence of ACE exposure on future risk of negative cannabis-related outcomes, but our findings suggest that supportive home and school environments will likely benefit adolescents of all ages.

Our analyses supported the compensatory model of resiliency as we found that family communication had a direct protective relationship with past 30-day CU for both middle and high school students, independent of ACE exposure. It is important to note that our family communication measure assessed open and supportive communication, which has been shown to decrease adolescent CU in other studies.^21,31–34 Positive family communication is also a component of parenting programs that have been found to reduce substance use, particularly during early adolescence.⁵⁵ However, positive communication should not be confused with parent child communication specifically regarding CU, which may have no influence or even a harmful influence on adolescent CU.^56,57

The independent protective influence of school connectedness was stronger than family communication. For every unit increase the school connectedness score, there was 24%–28% lower prevalence of past 30-day CU among middle and high school students respectively, compared to 10% lower prevalence of past 30-day CU for family communication. Adolescents spend a large proportion of their formative years in school settings and schools are stable and accessible sites for screening and preventive intervention.⁵⁸ Longitudinal research has shown that greater school connectedness during early adolescence protects youth from CU at later ages.³⁵ However, research has also shown that children and adolescents with high ACE exposure are less likely to engage with school.^59,60 A trauma-informed approach to interventions that promote school connectedness is essential to ensure that the most vulnerable populations are included.

Although family communication and school connectedness were strong protective factors for middle and high school students, we did not find evidence in support of the protective factors model of resiliency, which is similar to previous research.^39,40 Individual preventive domains may not be enough to buffer the strong relationship between ACEs and CU. Future research should explore whether the combined effect of protective factors across multiple domains can offer greater protection. Additionally, Hispanic students were the largest population in our study and in the middle school sample past 30-day CU was highest among Hispanic students. These findings highlight a critical need for research that investigates whether cultural sources of strength could have a protective effect.

Limitations

Our results should be interpreted in the context of study limitations. First, the middle and high school surveys were cross sectional and temporality cannot be established. While the temporal relationship between ACEs and subsequent CU has been demonstrated in previous studies¹³ there is a need for prospective research to differentiate the timing and impact of family and school preventive influences. Second, the CU measure was very broad and students many not have realized that they should include vaporized use or edibles, which could lead to underreporting. Third, we measured lifetime ACEs, whereas family communication focused on present experiences; it is likely that family communication was influenced by prior ACE exposure. Fourth, while our cluster random sampling design and weighted analyses increased representativeness of findings across Nevada, our results cannot be generalized to out-of-school youth, students in private schools, or those in other states that may have different cannabis regulation or school environments.

Conclusions

As more states legalize cannabis for adult recreational use, the prevention of underage CU is increasingly important. Our study contributes to a better understanding of family and school environmental factors that may offer direct protection from the influence of ACE exposure on recent CU among adolescents. Our findings highlight the importance of positive youth development interventions that build youth assets such as family communication and school connectedness during the early middle school years when ACE exposure may have a stronger influence on CU.

There was a strong graded relationship between ACEs and past 30-day CU.

Family communication had an independent relationship with past 30-day CU.

School connectedness had an independent relationship with past 30-day CU.

Findings support the compensatory model of resiliency.