Cross-substance patterns of alcohol, cigarette, and cannabis use initiation in Black and White adolescent girls

Carolyn E. Sartor, Feifei Ye, Patricia Simon, Zu Wei Zhai, Alison E. Hipwell, Tammy Chung

Publication Date: 03/01/2022


Characterizing variations in the timing of alcohol, cigarette, and cannabis use onset both among and between Black and White youth can inform targeted prevention. The current study aimed to capture cross-substance initiation patterns in Black and White girls and characterize these patterns with respect to substance use related socioeconomic, neighborhood, family, community, and individual level factors. Data were drawn from interviews conducted at ages 8 through 17 in an urban sample of girls (n = 2172; 56.86% Black, 43.14% White). Discrete-time multiple event process survival mixture modeling was used to identify patterns (i.e., classes) representing timing of alcohol, cigarette, and cannabis use initiation, separately by race. Class characteristics were compared using multinomial logistic regression. Among both Black and White girls, four classes, including abstainer and cross-substance early onset classes, emerged. Two classes characterized by mid-adolescence onset (Black girls) and variation in onset by substance (White girls) were also observed. Class differences centered around cannabis for Black girls (e.g., preceding or following cigarette use) and alcohol for White girls (e.g., (in)consistency over time in greater likelihood of initiation relative to cigarette and cannabis use). Several factors distinguishing the classes were common across race (e.g., externalizing behaviors, friends’ cannabis use); some were specific to Black girls (e.g., intentions to smoke cigarettes) or White girls (e.g., primary caregiver problem drinking). Findings underscore the need to recognize a more complex picture than a high-risk/low-risk dichotomy for substance use initiation and to attend to nuanced differences in markers of risky onset pathways between Black and White girls.

1. Introduction

By 12th grade, 60.2% of girls have consumed alcohol, 18.8% have smoked cigarettes, and 42.9% have used cannabis (Miech et al., 2020). Whereas Black youth are less likely than White youth to use alcohol or cigarettes, they are more vulnerable to later health and social consequences of alcohol and tobacco use (Heron, 2019Zapolski et al., 2014) and more likely to use cannabis (Miech et al., 2020). There is growing evidence as well that it is more common for Black youth than White youth to initiate cannabis use at the same age or prior to alcohol or tobacco use (Banks et al., 2017Fairman et al., 2019Sartor et al., 2013), counter to the typical, i.e., ‘gateway’ sequence of alcohol and tobacco before cannabis. Establishing a more complete picture of the heterogeneity within and between Black and White youth in the co-occurrence of alcohol, cigarette, and cannabis use onset – which substances, at what age, and in what order – may inform targeted prevention efforts. Capturing simultaneously the timing of initiation of all three substances, which have many common sources of risk (Ramo and Prochaska, 2012Whitesell et al., 2013), can contribute to this goal.

A limited number of studies have applied a statistical approach integrating survival analysis and latent class analysis, discrete-time multiple event process survival mixture (MEPSUM) modeling (Dean et al., 2014), for this purpose. In addition to identifying patterns in the timing of initiation, it captures cumulative risk of use over time. Classes may be differentiated by a combination of the likelihood of ever initiating use, the timing of onset, and the number of substances/products used. In Dean et al.’s (2015) investigation using National Survey on Drug Use and Health data (n = 55,772; ages 12 and older), the onset of nine substances was modeled, revealing six classes: an overall low risk class, two early and two late onset classes (one each for ‘hard’ and ‘soft’ drugs), and a class characterized by cross-substance consistent risk. Richmond-Rakerd and colleagues’ study (Richmond-Rakerd et al., 2016) of alcohol, tobacco, and cannabis use onset based on Add Health data (n = 18,923; adolescent to adult years) revealed four classes, similarly distinguished by overall risk and stability of risk for initiation: a low risk class, two cross-substance early initiation classes with slightly different peaks in risk, and a late adolescent onset class. Likewise, Cho et al. (2021) examined initiation of cannabis and five tobacco products in a longitudinal study of high school students (n = 2272) and identified four classes: a low risk class, two cross-substance early onset classes (one high-risk), and a late onset class specific to cannabis and e-cigarette use.


Read More