Abstract
Background:
Substance use is prevalent among adolescents and young adults (AYAs). Polysubstance use is associated with poorer outcomes than single substance use alone. However, few studies have examined patterns of daily use of substance(s) among AYA females, as well as associated factors.
Objective:
We aimed to 1) identify daily substance use patterns in AYA females, and 2) examine the association with select predisposing factors (i.e. impulsivity and depressive symptoms) and risky sexual behavior.
Method:
A sample of 15–24 year old females with recent binge drinking (n=149), reported daily substance use and sexual activity through weekly Timeline Followback interviews over a 1-month period (n days=4224). Impulsivity, depressive symptoms, and sociodemographics were measured at baseline. Day-level latent class analysis was conducted to characterize patterns of substance use, and Bolck, Croon, and Hagenaars three-step approach was utilized to examine the association of predisposing factors and other health behaviors with latent class membership.
Results:
Four day-level classes of substance use were identified: 1) little-to-no-use (54.1% of days); 2) vaping-only (28.2% of days); 3) cannabis-only (10.5% of days); and 4) heavy-alcohol-and-cannabis (some smoking and vaping; 7.2% of days). More class 4 days were associated with higher impulsivity than class 1 and 3 (M=.42 vs. −.17 and −.13), more depressive symptoms than class 1 and 2 (M=.77 vs. −.19 and .01), and more same day condomless sex than class 1 and 2 (19% vs. 8% and 7%).
Conclusions:
AYA females exhibit varied patterns of daily substance use including polysubstance use. Knowledge of risk factors associated with problematic use as well as co-occurring risky sexual behavior can inform targeted intervention and prevention efforts.
Keywords: polysubstance use, latent class analysis, adolescent, young adult, risky sexual behavior
1. Introduction
Substance use is highly prevalent among adolescents and young adults (AYAs) in the United States (US), with substance use typically initiated in adolescence and peaking in young adulthood (Alcover & Thompson, 2020). In 2022, 18.5% of AYAs (12–20 year olds) reported alcohol use, 15.9% use of tobacco products, 11.5% cannabis, and 3.2% other illicit drugs in the past month (National Survey on Drug Use and Health, 2023). AYA substance use can be particularly disruptive given ongoing brain development (Steinfeld & Torregrossa, 2023). Polysubstance use the co-occurring use of multiple substances, is inconsistently defined in the literature. However, recent conceptualizations distinguish the use of multiple substances over extended periods (e.g. past month, past year, or lifetime) from the short-term behavior of using two or more substances with overlapping effects (e.g. in the same day, same episode, or through co-ingestion), the latter hereafter will be referred to as polysubstance use (Bunting et al., 2023). While the use of multiple substances in AYAs is monitored by nationally representative surveys, such as the National Survey on Drug Use and Health (NSDUH), patterns of polysubstance use are less well characterized. Relative to single substance use, poly-substance use has been linked to greater substance-related consequences, as well as heavier substance use (Bailey et al., 2019; Fernández-Calderón et al., 2020; Karamouzian et al., 2024). Given these deleterious effects, understanding nuanced patterns of polysubstance use, as well as associated factors among AYAs is of great concern.
A broad range of factors have been associated with substance use, but less research has examined these factors in association with polysubstance use. Trait impulsivity, a predisposition toward rapid reactions without regard for consequences (Moeller et al., 2001), is suspected to be an important risk factor for a range of substance use disorders, as well as a possible consequence of substance use (Kozak et al., 2019; Verdejo-García et al., 2008). Existing research, suggests polysubstance use may be associated with greater impulsivity than single substance use (Bailey et al., 2019; Moody et al., 2016). Depression is another well-established factor associated with substance use (Esmaeelzadeh et al., 2018): the relationship between depressive symptoms and substance use is thought to be reciprocal and cascading, with elevations in one promoting the other (Felton et al., 2015). In adolescents, depressive symptoms are associated with an increased likelihood of multiple substance use (Jongenelis et al., 2019; Maslowsky et al., 2014), and this relationship extends into young adulthood (Crane et al., 2021; Schuler et al., 2015).
Substance use has also been linked to risky sexual behavior. In a meta-analysis of studies among AYAs, substance use was determined to have a small to moderate positive association with risky sexual behavior (e.g. sex without a condom, multiple sexual partners, sex with a person that uses intravenous drugs) (Ritchwood et al., 2015). While a large body of research has connected individual-level substance use and risky sexual behavior, evidence of the day-level association and the impact of polysubstance use more scare. However, studies of same-day alcohol and cannabis use, suggest greater condomless sex on days with dual use than those with use of either substance alone (Hayaki et al., 2018; Metrik et al., 2016). Increased understanding can be gained by examining these factors in the context of complex daily substance use patterns, through the application of novel statistical methods.
Examination of polysubstance use presents statistical and methodological challenges (Connor et al., 2014). To combat these challenges, a growing body of research has applied flexible person-centered approaches such as latent class analysis (LCA) to characterize substance use in AYAs (Halladay et al., 2020; Tomczyk et al., 2016). LCA allows for the examination of multiple dimensions simultaneously, using shared responses to observed variables to categorize individuals (Collins & Lanza, 2009). LCA studies have identified common substance use subgroups including low use, single or dual use (most commonly alcohol, alcohol and cannabis, or alcohol and tobacco), moderate multi-use, and high multi-use (Halladay et al., 2020). Polysubstance use in particular has been considered in only a few LCA studies (Bailey et al., 2019; Fernández-Calderón et al., 2020). Select combinations of co-use, such as simultaneous alcohol and cannabis use have also been examined using LCA (Lanza et al., 2022; Patrick et al., 2018). While LCA has traditionally been applied at the person level, this method has recently been extended to identify patterns of behavior in daily life, where the unit of analysis is occasion (e.g. day) rather than person (Linden-Carmichael et al., 2022). Using day-level LCA, longitudinal measures of substance use such as Timeline Followback or daily diary can be leveraged to identify patterns of polysubstance use within a given day.
In recent years rates of substance use have converged by gender, particularly among youth (Goodwin et al., 2022). Contrary to historical patterns, more high school females report alcohol, cigarette, vape, cannabis and prescription opioid use in the past month than their male peers (Birdsey et al., 2023; Centers for Disease Control and Prevention, 2024). Females but not males with dual polysubstance are shown to have an increased likelihood of progressing to same-day use of a third substance (Roche et al., 2019). Thus, exploration of daily patterns of substance use among AYA females is particularly salient.
The present study aimed to 1) identify daily substance use patterns in AYA females reporting recent binge drinking, and 2) examine the extent to which person-level (sociodemographic characteristics, impulsivity, and depressive symptoms) and day-level factors (day of the week and condomless sex) are associated with daily latent class membership. Classification of substance use within a given day in this risk population, can inform prevention and intervention efforts, by clarifying what factors are associated with day to day use patterns.
2. Methods
2.1. Participants and procedures
Participants were recruited from a metropolitan area in the south-central US between 2017 and 2018 (before the legalization of medicinal cannabis in the state), as part of a study of AYA female substance use and nutriture. Recruitment included community-based and participant-driven sampling approaches. Inclusion criteria were female biological sex, 14 to 24 years old, and reporting at least one binge drinking episode (≥ 4 standard alcoholic drinks on a single occasion) within the past two weeks at screening. Eligible participants were scheduled for an initial study visit, wherein they were consented to participate, or parental consent with youth assent was obtained depending on the age of the participant. After consent, participants completed baseline questionnaires, followed by weekly follow-up visits over a one-month period. All methods were approved by Oklahoma State University’s Institutional Review Board and a certificate of confidentiality was in place.
2.2. Measures
2.2.1. Baseline measures
Person-level variables.
Impulsivity was measured using the Barrett Impulsiveness Scale Version 11 (BIS-11) (Patton et al., 1995), one of the most widely used self-report measures of impulsivity (Stanford et al., 2009). Participants indicated how often each statement (e.g. “I act on the spur of the moment.”) applied to them on a 4-point Likert scale ranging from 1 (rarely/never) to 4 (almost always/always). Responses were summed (possible scores ranged from 0 to 120), with higher scores reflecting greater impulsivity (Cronbach’s α=0.82). Depressive symptoms were assessed using the 20-item Center for Epidemiologic Studies Depression Scale Revised (CESD-R), a frequently used measure of depressive symptomology, shown to have good psychometric properties (Eaton et al., 2004; Van Dam & Earleywine, 2011). Responses were on a 5-point Likert scale ranging from not at all or less than 1 day (0) to nearly every day for 2 weeks (4). Responses were summed (possible scores ranged from 0 to 80), with higher scores indicating greater depressive symptoms (Cronbach’s α=0.93) (Bean et al., 2024). BIS-11 and CESD-R sum scores were z-scored to assist with interpretation (BIS-11: M=63.29, SD=10.18; CESD-R: M=13.42, SD=12.62).
Sociodemographic variables.
Participant sociodemographic characteristics of age, race and ethnicity (categorized as: racial and ethnic minority groups; non-Hispanic White), relationship status (partnered (boyfriend, girlfriend); un-partnered); sexual attraction (only or mostly attracted to males; attracted to females and males; only or mostly attracted to females), school enrollment (no; high school; college) and employment (not employed; part-time employment (1–29 hours a week); full-time employment (≥30 hours a week)) were self-reported at baseline. Participants also reported substance use in the past month including how many days they drank alcohol, consumed ≥4 drinks, smoked cigarettes and/or vaped (reported as a weekly average, multiplied by 4 to approximate other monthly estimates), as well as the frequency with which they used cannabis and other medical and non-medical drugs. This question did not distinguish between prescription and non-prescription drug use.
2.2.2. Daily measures
Indicator variables.
During weekly visits, daily substance use was reported using a Timeline Followback (TLFB) approach (Sobell & Sobell, 1992). Participants reported use of alcohol, cigarettes, nicotine vaping, cannabis, ADHD medication, opiates, heroin, MDMA, cocaine, methamphetamine or other amphetamines, inhalants, hallucinogens, acid, Xanax, psychedelic mushrooms, or other pills on each day of the previous week. If participants reported drinking alcohol, they were asked the quantity of drinks consumed of each type (beer, wine, liquor, or mixed drinks). All types were summed to create a daily count, categorized as: no use (<1 drink), moderate use (1–3 drinks), and heavy use (4 or more drinks). Daily cigarette, nicotine vape, and cannabis use was categorized as no use (0) or use (>0). Due to lower prevalence, responses for other substances were combined into an “Other medical/nonmedical drugs” category and coded as no use (0) or use (>0).
Day-level variables.
Participants also reported whether they engaged in vaginal sex and if they used a condom. Vaginal sex without a condom was categorized as condomless sex (1=yes, 0=no). Day of the week was categorized as weekday (Monday-Thursday) or weekend (Friday-Sunday).
2.3. Statistical analysis
Statistical analyses were conducted using Stata Statistical Software: Release 18, StataCorp LLC. To identify daily patterns of substance use, Day-level LCA was conducted using the LCA Stata plugin (University Park: The Methodology Center, Penn State, 2015). Classes were estimated using five daily indicators: four binary indicators of cannabis, cigarette, nicotine vape, and other drug use (1=no use, 2=use) and one 3-level indicator of alcohol use (1=no use; 2=moderate use; 3=heavy use). Within-person clustering of days was included to account for the nesting of days within participants and models estimated using a pseudo-maximum likelihood approach (Skinner et al., 1989; Vermunt & Magidson, 2005). To establish the best fitting model, solutions between 2 and 6 classes were considered. Model selection was guided by information criterion (Akaike information criterion, AIC; Bayesian information criterion, BIC; sample size adjusted Bayesian information criterion, SABIC), as well as stability, interpretability, and parsimony (Collins & Lanza, 2009).
The association of day-level latent class membership with person and day level factors was estimated using the Bolck, Croon, and Hagenarrs (BCH) 3-step approach Stata function (Haung et al., 2017), applied separately for each variable. The BCH 3-step function provides Wald tests comparing the expected values of covariates between each latent class, as well as an Omnibus test comparing all expected values. Covariates examined included age, race and ethnicity, relationship status, employment status, impulsivity and depressive symptoms, day of the week and condomless sex. Cases with missing data on covariates were removed. The multiple number of comparisons was managed using Bonferroni correction.
A post hoc examination of the association of latent classes with eBAC was conducted, as further validation of class findings. Daily estimated blood alcohol concentration (eBAC) was calculated using Mathews and Miller’s 1979 formula (BAC=((c/2)*(GC/w))-(β60*t)), where c=number of drinks, GC=a gender constant (9.0 for females), w=weight, and β60=the metabolism rate of alcohol per hour (.017 g/dl), and t=the number of hours spent drinking (Matthews & Miller, 1979).
3. Results
3.1. Descriptive statistics
Table 1 provides a summary of participant characteristics at baseline. Of the 149 participants, most were under the legal drinking age (<21; 67.1%), partnered (60.4%; 55.0%, male partner, 5.4% female partner), and only or mostly attracted to males (73%). Over 60% of the sample was of non-Hispanic White race and ethnicity, while 38% were of minority race and ethnicity. Approximately 43% were enrolled in high school, 26% in college and fewer than 39% worked full-time. At baseline, 99.3% of participants reported days with alcohol use, 93.3% heavy alcohol use, 33.6% cigarette use, 43.6% vape use, 56.4% cannabis and 25.5% other drugs.
Table 1.
Participant sociodemographic and baseline characteristics (n=149)
| Sociodemographic characteristics | N, % |
|---|---|
| Age (M±SD) | (19.09±2.65) |
| ≥ 21 years | 49, 32.89% |
| Race and ethnicity | |
| Non-Hispanic White | 92, 61.74% |
| Person of minority race or ethnicity | 57, 38.26% |
| Native American | 25, 16.78% |
| Latino/Hispanic | 24, 16.11% |
| Black | 7, 4.70% |
| Asian | 1, 0.67% |
| Relationship status | |
| Unpartnered | 59, 39.60% |
| Partnered | 90, 60.40% |
| Boyfriend | 82, 55.03% |
| Girlfriend | 8, 5.37% |
| Sexual Attraction | |
| Only or mostly attracted to males | 107, 73.29% |
| Attracted to females and males | 32, 21.92% |
| Only or mostly attracted to females | 7, 4.79% |
| School enrollment | |
| No | 45, 31.03% |
| High school | 62, 42.76% |
| College | 38, 26.21% |
| Working | |
| No | 42, 28.19% |
| Part-time (<30 hours) | 50, 33.56% |
| Full-time (≥30 hours) | 57, 38.26% |
| Substance use in the previous month | |
| Alcohol use | 148, 99.33% |
| Days (M±SD) | (7.34±4.64) |
| Heavy alcohol use (≥4 drinks) | 139, 93.29% |
| Days (M±SD) | (4.70±3.78) |
| Cigarettes | 50, 33.56% |
| Days (M±SD) | (2.96±6.84) |
| Vape | 65, 43.62% |
| Days (M±SD) | (9.54±12.12) |
| Cannabis use | 84, 56.38% |
| Everyday | 12, 8.05% |
| 1–3 times a week | 27, 18.12% |
| Less than once a week | 45, 30.20% |
| Other medical/nonmedical drug use | 38, 25.50% |
| ADHD medication | 28, 18.92% |
| Opiates | 17, 11.42% |
| MDMA/ecstasy | 7, 4.70% |
| Cocaine | 9, 6.04% |
| Methamphetamines | 0, 0% |
| Other Amphetamines | 4, 2.68% |
| Inhalants | 2, 1.34% |
Over the study period, Participants reported an average of 3.68 days with moderate alcohol use, 2.84 days with heavy alcohol use, 3.74 with cannabis use, 1.79 days with cigarette use, 8.36 days with vape use, and 0.77 days with other drug use (see Table 2). Participants reported engaging in condomless sex on an average of 2.59 days.
Table 2.
Number of days with substance use and condomless sex during the 1-month study period (n=149)
| Substance Use | N, % |
|---|---|
| No. days with moderate alcohol use (M±SD) | (3.68±3.49) |
| ≥ 1 day with moderate alcohol use | 125, 83.89% |
| No. days with heavy alcohol use (M±SD) | (2.84±2.66) |
| ≥ 1 day with heavy alcohol use | 120, 80.54% |
| No, days with cannabis use (M±SD) | (3.74±7.41) |
| ≥ 1 day with cannabis use | 72, 48.32% |
| No, days with cigarette use (M±SD) | (1.79±5.63) |
| ≥ 1 day with cigarette use | 31, 20.81% |
| No. days with vape use (M±SD) | (8.36±11.29) |
| ≥ 1 day with vape use | 79, 53.02% |
| No. days with other medical/nonmedical drug use (M±SD) | (0.77±2.93) |
| ≥ 1 day with other drug use | 24, 16.11% |
| Risky Sexual behavior | |
| No. Days with condomless sex (M±SD) | (2.59±3.76) |
| ≥ 1 day with condomless sex | 76, 51.00% |
Other medical/nonmedical drugs include opiates, MDMA, cocaine, methamphetamine, amphetamine, heroin, inhalants, hallucinogens, acid, Xanax, shrooms or other pills.
3.2. Daily substance use
Latent class analysis was based on 4224 person days and 149 participants. Models with 2 through 6 latent classes were considered (see Table 3). The BIC and SABIC favored a 3-class solution, and AIC favored a 6-class solution. However, the smallest class size in 5 and 6 class models were below 5%. In the 3-class model the smallest class was judged to be less well defined and harder to interpret (see Supplement Table 1). Therefore, a 4-class solution was selected based on the model fit, interpretability and the meaningfulness of classes.
Table 3.
Summary of information for comparing models with 2 through 6 latent classes of daily substance use
| # of Classes | Log Likelihood | df | AIC | BIC | SABIC | Entropy |
|---|---|---|---|---|---|---|
| 2 | −8423.79 | 34 | 173.47 | 256.00 | 214.69 | 64.16% |
| 3 | −8383.09 | 27 | 106.08 | 233.05 | 169.49 | 68.97% |
| 4 | −8372.12 | 20 | 98.13 | 269.54 | 183.74 | 74.43% |
| 5 | −8364.45 | 13 | 96.79 | 312.63 | 204.59 | 69.07% |
| 6 | −8356.95 | 6 | 95.76 | 356.07 | 225.78 | 86.92% |
Latent class models based on 4224 person days from 149 participants. Abbreviations: df, degrees of freedom; AIC, Akaike information criteria; BIC, Bayesian information criterion; SABIC, sample size adjusted Bayesian information criterion.
Parameter estimates for the 4-class model are presented in Table 4, in order of class prevalence. Class 1, the largest class (54.1% of days), was defined by small probabilities of any use, thus we labeled this class “Little-to-no-use days.” Class 2 (28.2% of days) was labeled “Vaping-only days,” defined by a high probability of vaping and a low probability of other use. Class 3 (10.5% of days), labeled “Cannabis-only days” was characterized by a high probability of cannabis use and lower probabilities of other substance use. Class 4 (7.2% of days) involved a high probability of heavy alcohol use and cannabis use, as well as a moderate probability of vaping and smoking, but a small probability of other drug use, thus we labeled this class “Heavy-alcohol-and-cannabis days (some smoking and vaping).” The distribution of day-level classes by participant are displayed in Figure 1.
Table 4.
Parameter estimates for the 4-class model of daily substance use behavior
| Daily Substance Use | ||||
|---|---|---|---|---|
| Little-to-no-use days | Vaping-only days | Cannabis-only days | Heavy-alcohol-and-cannabis days (some smoking and vaping) | |
| Proportion of days | 54.1% | 28.2% | 10.5% | 7.2% |
| Item response probabilities | ||||
| Alcohol | ||||
| Heavy use | .07 | .07 | .00 | .57 |
| Moderate use | .13 | .06 | .24 | .27 |
| No Use | .80 | .87 | .76 | .17 |
| Cannabis | ||||
| Use | .00 | .08 | .67 | .56 |
| No Use | .99 | .92 | .33 | .44 |
| Cigarettes | ||||
| Use | .00 | .05 | .26 | .32 |
| No Use | .99 | .95 | .74 | .68 |
| Vape | ||||
| Use | .01 | .96 | .07 | .31 |
| No Use | .99 | .04 | .93 | .69 |
| Other medical/nonmedical drugs | ||||
| Use | .02 | .01 | .03 | .13 |
| No Use | .98 | .99 | .97 | .87 |
Parameter estimates greater than 0.50 are bolded. Moderate alcohol use is defined as 1–3 drinks, heavy alcohol use is defined as 4 or more drinks. Other medical/nonmedical drugs include opiates, MDMA, cocaine, methamphetamine, amphetamine, heroin, inhalants, hallucinogens, acid, Xanax, shrooms or other pills.
Figure 1.
Day-level substance use latent classes by participant over study period
Each vertical bar holds data for an individual participant. Bars are color-coded to represent day-level latent classes according to maximum probability assignment. The height of colored sections represents the number of days characterized by each latent class.
3.3. Associations with latent classes
Class comparisons after correcting for multiple tests are presented in Table 5. Bonferroni correction determined that comparisons with p<0.00083 were significant. Vaping-only days were associated with younger age and less likelihood of working full time compared to all other types of days. Whereas little-to-no-use days were associated with a greater likelihood of being of a minority race and ethnicity relative to vaping-only days.
Table 5.
Associations with latent classes of daily substance use
| Daily Substance Use Classes | ||||||
|---|---|---|---|---|---|---|
| Overall (Mean, SD or %) | Little-to-no-use days | Vaping-only days | Cannabis-only days | Heavy-alcohol-and-cannabis days (some smoking and vaping) | Bonferroni Corrected P value | |
| Person-level characteristics | ||||||
| Age (M, SE) | 19.09 (2.65) | 20.06 (.06)V | 16.88 (.05)L,C,H | 19.95 (.17)V | 19.37 (.23)V | <.0001 |
| Race and ethnicity (log odds probability of “minority race and ethnicity”, SE) |
38.26% | .44 (.01)V | .32 (.01)L | .33 (.03) | .31 (.04) | <.0001 |
| Relationship status (log odds probability of “partnered”, SE) |
60.40% | .59 (.01) | .59 (.02) | .73 (.03) | .58 (.05) | 0.0632 |
| Employment status (log odds probability of “full time”, SE) |
38.26% | .48 (.01)V | .11 (.01)L,C.H | .59 (.03)V | .46 (.05)V | <.0001 |
| Impulsivity (M,SE) | 0 (1) | −.17 (.02)V,H | .28 (.03)L,C | −.13 (.07)V, H | .42 (.08)L,C | <.0001 |
| Depression symptoms (M, SE) | 0 (1) | −.19 (.02)V,C,H | .01 (.03)L,C,H | .44 (.09) L,V | .77 (.16) L,V | <.0001 |
| Day-level characteristics | ||||||
| Day of the week (log odds probability of “weekday”, SE) |
57.72% | .59 (.01) H | .57 (.02) H | .60 (.03) | .38 (.05)L,V | .0231 |
| Condomless sex (log odds probability of “condomless sex”, SE |
9.22% | .08 (.01)H | .07 (.01)H | .13 (.02) | .19 (.03)L,V | <.0001 |
Superscript letters denote where pairwise comparisons (Wald chi square) are significant: L=significantly differs from little-to-no-use days; V=significantly differs from vaping-only days; C=significantly differs from cannabis-only days; H=significantly differs from heavy-alcohol-and-cannabis days (some smoking and vaping)
Heavy-alcohol-and-cannabis days (some smoking and vaping) and vaping-only days were associated with higher impulsivity compared to little-to-no-use days and cannabis-only days. Little-to-no-use days were associated with lower depressive symptoms relative to all other types of days. Similarly, vaping-only days were associated with lower depressive symptoms compared to cannabis-only days and heavy-alcohol-and-cannabis days (some smoking and vaping).
Heavy-alcohol-and-cannabis days (some smoking and vaping) were less likely to occur on a weekday relative to little-to-no-use days or vaping-only days. Condomless sex was more likely to occur on heavy-alcohol-and-cannabis days (some smoking and vaping) than both little-to-no-use days and vaping-only days.
In Post hoc analysis (Supplement Table 2), eBAC was higher on heavy-alcohol-and-cannabis days (some smoking and vaping) than all other types of days. Further, EBAC was lower on cannabis-only days compared to little-to-no-use days and vaping-only days.
4. Discussion
We explored patterns of daily substance use in a sample of AYA females reporting binge drinking in the previous 2 weeks, identifying four distinct patterns of daily use: little-to-no-use days, vaping-only days, cannabis-only days, and heavy-alcohol-and-cannabis days (some smoking and vaping). Participants demonstrated a range of daily substance use patterns over the study period (e.g. a single, two, three, or all four types of days over the study period). Daily substance use patterns were differentially associated with person-level factors of impulsivity, depressive symptoms, sociodemographic characteristics, and day-level factors of condomless sex and day of the week.
Latent class findings are consistent with subgroups of multi-substance use identified in other studies of AYAs (e.g. low substance use, use of one or two substances, moderate multi-use, and high multi-use) (De Jonge et al., 2022; Halladay et al., 2020). However, the present study expands previous work by classifying polysubstance and substance use within a given day rather than over a more extended period (e.g. past month, past year, or lifetime). This distinction is important as the timeframe in which polysubstance use is defined informs interpretation, with acute timeframes illuminating overlapping effects and accompanying dangers (Bunting et al., 2023). The smallest class, heavy-alcohol-and-cannabis days (some smoking and vaping) represents days with the highest probability of same-day polysubstance use. Examination of eBAC confirmed higher risk patterns of drinking on these days. Because heavy episodic (binge) drinking was an inclusion criterion, it is particularly notable that heavy alcohol use occurs on a small percentage of days, suggesting that while participants engage in this behavior, they do so infrequently. It is unlikely that these findings occur from an overinflation of estimates at screening, because more than 80% of participants reported one or more days with heavy alcohol use during the study period. Heavy-alcohol-and-cannabis days (some smoking and vaping) were less likely than other day-use-types to occur on a weekday, suggesting potential contextual differences associated with this pattern of use. Weekend patterns of use are particularly relevant for younger participants because drinking occurs primarily in social contexts; minors may lack a place to drink; and those under the minimum legal drinking age tend to lack steady access to alcohol (Cox et al., 2019).
Identified classes differed somewhat from those found in Linden-Carmichael et al., (2022)’s application of day-level LCA in a sample of young adults (Linden-Carmichael et al., 2022). Wherein 5 classes were identified including cigarette-only days, alcohol-only days, cannabis days (some alcohol), vaping days (some alcohol and cannabis), and stimulant and alcohol days (some cannabis and vaping). Our study identified fewer classes overall, fewer polysubstance use classes, and did not identify a class with a high probability of drug use other than cannabis. However, these discrepancies might be explained by sample as well as methodological differences. Linden-Carmichael et al., (2022)’s study included only days with substance use rather than all measured days. Further, they included both males and females, the sample was slightly older, composed primarily of college students, and may have been at higher risk for polysubstance use given inclusion criteria of past month alcohol and cannabis use.
Consistent with previous evidence that impulsivity is associated with substance use (Smith & Cyders, 2016), in our study, little-to-no use days were more likely to be experienced by participants with lower impulsivity compared to vaping-only and heavy-alcohol-and-cannabis days (some smoking and vaping). Findings were partially in keeping with the hypothesis that polysubstance use is associated with greater impulsivity than single substance use (Bailey et al., 2019; Stamates et al., 2023), such that heavy-alcohol-and-cannabis days (some smoking and vaping) were associated with higher baseline impulsivity relative to cannabis-only days. However, the same was not seen when compared to vaping-only days. Age related differences in impulsivity might have contributed to these findings (Inuggi et al., 2014), given that vaping-only days were associated with younger age compared to other day-use-types.
Little-to-no-use days were more likely among participants with lower depressive symptoms relative to all other types of days. This finding is consistent with the considerable literature linking substance use with depressive symptoms in AYAs (Esmaeelzadeh et al., 2018). Further, heavy-alcohol-and-cannabis days (some smoking and vaping) and cannabis-only days were associated with higher baseline depressive symptoms than vaping-only days. This finding builds on previous studies of multi-substance use (Maslowsky et al., 2014; Williams et al., 2021), suggesting polysubstance use may be associated with greater depressive symptomology than single substance use. Depressive symptoms did not vary significantly between cannabis-only days and heavy-alcohol-and-cannabis days (some smoking and vaping).
A substantial body of research has found a relationship between alcohol use and to a lesser extent cannabis use and risky sexual behavior, however, event level analyses are limited (Cho & Yang, 2023; Schumacher et al., 2018). Merik et al., (2015) found alcohol use, cannabis use, and dual alcohol and cannabis use were associated with increased odds of condomless sex with a main partner (Metrik et al., 2016). Similarly, in a sample of young adults, Hayaki et al., (2018) found risk of condomless sex increased on days with alcohol use and cannabis use, but was greatest on days with dual alcohol and cannabis use (Hayaki et al., 2018). Consistent with these studies, heavy-alcohol-and-cannabis days (some smoking and vaping) were associated greater odds condomless sex than little-to-no-use days and vaping-only days.
Given evidence of sex/gender differences in substance use and escalating trends of use among females, the focus on AYA females contributes significantly to the current body of literature. LCA is an ideal approach for the characterization of substance use as it enables the reduction of complex patterns into homogeneous subgroups (Connor et al., 2014). Through the application of day-level LCA which accounts for nested data structure, increased specificity can be achieved in the identification of daily patterns of substance use (Linden-Carmichael et al., 2022). Study limitations should also be considered. First, our study consisted of AYA females who reported recent heavy alcohol use. Thus, findings may not necessarily represent all types of polysubstance use behavior or generalize to all AYA females. Secondly, a distinction was not made between prescription and non-prescription drug use. While participants’ self-report of medical diagnoses does not indicate the majority of use was likely to be prescribed, future studies would benefit from measurement. Most (55%) participants reported having a boyfriend at baseline. As this will influence safe sex practices it should be incorporated in future studies of risky sexual behaviors. Further, data was collected from 2017 to 2018, prior to medicinal cannabis legalization in Oklahoma. Findings may be influenced by the rapidly changing cannabis legal landscape. Finally, findings demonstrate association rather than causation, with a number of the factors examined known to have bidirectional relationships.
Findings elucidate patterns of daily substance use among a sample of AYA females who report recent binge drinking. AYA substance use patterns are typically described over a lifetime, past year, or past month use. Application of day-level LCA to intensive longitudinal data presents a more nuanced perspective of types of use-days. Analysis uncovered a polysubstance use day (heavy-alcohol-and-cannabis days) wherein substance use occurs with potentially overlapping effects, on about 7% of days in the month. Polysubstance use days were associated with higher baseline impulsivity and depressive symptoms, as well as same day condomless sex, increasing risk of sexually transmitted infections. Continued examination of nuanced patterns of substance use with person level and day level factors, inducing contextual factors is needed to extend findings and inform intervention. This specificity may be particularly applicable to development of just-in-time interventions designed to delivery support in moments of need.
Supplementary Material
Highlights.
Emerging adult females demonstrate varied patterns of daily substance use.
Days with heavy alcohol and cannabis use were more likely to be associated with impulsivity.
Days with heavy alcohol and cannabis use were more likely to be associated with depressive symptoms.
Condomless sex was more likely to occur on days with heavy alcohol and cannabis use.
Funding
This research was supported in part by the National Institutes of Health (U01DA055349; P20GM109097). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Health Resources Services Administration.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflicts of Interest
The authors report no relevant financial or non-financial conflicts of interest.
Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References
- Alcover KC, & Thompson CL (2020). Patterns of Mean Age at Drug Use Initiation Among Adolescents and Emerging Adults, 2004–2017. JAMA Pediatrics, 174(7), 725. 10.1001/jamapediatrics.2019.6235 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bailey AJ, Farmer EJ, & Finn PR (2019). Patterns of polysubstance use and simultaneous co-use in high risk young adults. Drug and Alcohol Dependence, 205, 107656. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bean CAL, Mueller SB, Abitante G, Ciesla JA, Cho S-J, & Cole DA (2024). Improved Scoring of the Center for Epidemiologic Studies Depression Scale – Revised: An Item Response Theory Analysis. Journal of Psychopathology and Behavioral Assessment. 10.1007/s10862-024-10155-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- Birdsey J, Cornelius M, Jamal A, Park-Lee E, Cooper MR, Wang J, Sawdey MD, Cullen KA, & Neff L (2023). Tobacco Product Use Among U.S. Middle and High School Students—National Youth Tobacco Survey, 2023. MMWR. Morbidity and Mortality Weekly Report, 72(44), 1173–1182. 10.15585/mmwr.mm7244a1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bunting AM, Shearer R, Linden-Carmichael AN, Williams AR, Comer SD, Cerdá M, & Lorvick J (2023). Are you thinking what I’m thinking? Defining what we mean by “polysubstance use.” The American Journal of Drug and Alcohol Abuse, 1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Centers for Disease Control and Prevention. (2024). Youth Risk Behavior Survey Data Summary & Trends Report: 2013–2023. U.S. Department of Health and Human Services. [Google Scholar]
- Cho H-S, & Yang Y (2023). Relationship Between Alcohol Consumption and Risky Sexual Behaviors Among Adolescents and Young Adults: A Meta-Analysis. International Journal of Public Health, 68, 1605669. 10.3389/ijph.2023.1605669 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Collins LM, & Lanza ST (2009). Latent Class and Latent Transition Analysis. John Wiley & Sons, Inc. 10.1002/9780470567333 [DOI] [Google Scholar]
- Connor JP, Gullo MJ, White A, & Kelly AB (2014). Polysubstance use: Diagnostic challenges, patterns of use and health. Current Opinion in Psychiatry, 27(4), 269–275. 10.1097/YCO.0000000000000069 [DOI] [PubMed] [Google Scholar]
- Cox MJ, Sewell K, Egan KL, Baird S, Eby C, Ellis K, & Kuteh J (2019). A systematic review of high-risk environmental circumstances for adolescent drinking. Journal of Substance Use, 24(5), 465–474. 10.1080/14659891.2019.1620890 [DOI] [Google Scholar]
- Crane NA, Langenecker SA, & Mermelstein RJ (2021). Risk factors for alcohol, marijuana, and cigarette polysubstance use during adolescence and young adulthood: A 7-year longitudinal study of youth at high risk for smoking escalation. Addictive Behaviors, 119, 106944. 10.1016/j.addbeh.2021.106944 [DOI] [PMC free article] [PubMed] [Google Scholar]
- De Jonge MC, Bukman AJ, Van Leeuwen L, Onrust SA, & Kleinjan M (2022). Latent Classes of Substance Use in Young Adults – A Systematic Review. Substance Use & Misuse, 57(5), 769–785. 10.1080/10826084.2022.2040029 [DOI] [PubMed] [Google Scholar]
- Eaton WW, Smith C, Ybarra M, Muntaner C, & Tien A (2004). Center for Epidemiologic Studies Depression Scale: Review and Revision (CESD and CESD-R). The use of psychological testing for treatment planning and outcomes assessment: Instruments for adults. (3rd ed.). Lawrence Erlbaum Associates Publishers. [Google Scholar]
- Esmaeelzadeh S, Moraros J, Thorpe L, & Bird Y (2018). Examining the Association and Directionality between Mental Health Disorders and Substance Use among Adolescents and Young Adults in the U.S. and Canada—A Systematic Review and Meta-Analysis. Journal of Clinical Medicine, 7(12), 543. 10.3390/jcm7120543 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Felton JW, Kofler MJ, Lopez CM, Saunders BE, & Kilpatrick DG (2015). The emergence of co-occurring adolescent polysubstance use and depressive symptoms: A latent growth modeling approach. Development and Psychopathology, 27(4pt1), 1367–1383. 10.1017/S0954579414001473 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fernández-Calderón F, Vidal-Giné C, Rojas-Tejada AJ, & Lozano-Rojas ÓM (2020). Patterns of Simultaneous Polysubstance Use among Partygoers: Correlates and Differences in Adverse Acute Effects Experienced. Journal of Psychoactive Drugs, 52(4), 344–356. 10.1080/02791072.2020.1752959 [DOI] [PubMed] [Google Scholar]
- Goodwin SR, Moskal D, Marks RM, Clark AE, Squeglia LM, & Roche DJO (2022). A Scoping Review of Gender, Sex and Sexuality Differences in Polysubstance Use in Adolescents and Adults. Alcohol and Alcoholism, 57(3), 292–321. 10.1093/alcalc/agac006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Halladay J, Woock R, El-Khechen H, Munn C, MacKillop J, Amlung M, Ogrodnik M, Favotto L, Aryal K, Noori A, Kiflen M, & Georgiades K (2020). Patterns of substance use among adolescents: A systematic review. Drug and Alcohol Dependence, 216, 108222. 10.1016/j.drugalcdep.2020.108222 [DOI] [PubMed] [Google Scholar]
- Haung L, Dziak JJ, Bray BC, & Wagner AT (2017). LCA_Distal_BCH Stata function users’ guide (Version 1.1) [Computer software]. Retrieved from http://methodology.psu.edu
- Hayaki J, Anderson BJ, & Stein MD (2018). Dual use of alcohol and marijuana and condomless sex in young adult men and women: A within-subject day-level analysis. The American Journal on Addictions, 27(5), 413–418. 10.1111/ajad.12738 [DOI] [PubMed] [Google Scholar]
- Inuggi A, Sanz-Arigita E, González-Salinas C, Valero-GarcÃ-a AV, GarcÃ-a-Santos JM, & Fuentes LJ (2014). Brain functional connectivity changes in children that differ in impulsivity temperamental trait. Frontiers in Behavioral Neuroscience, 8. 10.3389/fnbeh.2014.00156 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Jongenelis M, Pettigrew S, Lawrence D, & Rikkers W (2019). Factors Associated with Poly Drug Use in Adolescents. Prevention Science, 20(5), 695–704. 10.1007/s11121-019-00993-8 [DOI] [PubMed] [Google Scholar]
- Karamouzian M, Cui Z, Hayashi K, DeBeck K, Reddon H, Buxton JA, & Kerr T (2024). Longitudinal polysubstance use patterns and non-fatal overdose: A repeated measures latent class analysis. International Journal of Drug Policy, 104301. 10.1016/j.drugpo.2023.104301 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kozak K, Lucatch AM, Lowe DJE, Balodis IM, MacKillop J, & George TP (2019). The neurobiology of impulsivity and substance use disorders: Implications for treatment. Annals of the New York Academy of Sciences, 1451(1), 71–91. 10.1111/nyas.13977 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lanza ST, Sokolovsky AW, Linden-Carmichael AN, White HR, & Jackson KM (2022). Understanding Heterogeneity Among Simultaneous Alcohol and Marijuana Users: Latent Classes Derived From Daily Diary Data. Journal of Studies on Alcohol and Drugs, 83(3), 358–363. 10.15288/jsad.2022.83.358 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Linden-Carmichael AN, Van Doren N, Bray BC, Jackson KM, & Lanza ST (2022). Stress and Affect as Daily Risk Factors for Substance Use Patterns: An Application of Latent Class Analysis for Daily Diary Data. Prevention Science, 23(4), 598–607. 10.1007/s11121-021-01305-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Maslowsky J, Schulenberg JE, O’Malley PM, & Kloska DD (2014). Depressive symptoms, conduct problems, and risk for polysubstance use among adolescents: Results from US national surveys. Mental Health and Substance Use, 7(2), 157–169. 10.1080/17523281.2013.786750 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Matthews DB, & Miller WR (1979). Estimating blood alcohol concentration: Two computer programs and their applications in therapy and research. Addictive Behaviors, 4(1), 55–60. 10.1016/0306-4603(79)90021-2 [DOI] [PubMed] [Google Scholar]
- Metrik J, Caswell AJ, Magill M, Monti PM, & Kahler CW (2016). Sexual Risk Behavior and Heavy Drinking Among Weekly Marijuana Users. Journal of Studies on Alcohol and Drugs, 77(1), 104–112. 10.15288/jsad.2016.77.104 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Moeller FG, Barratt ES, Dougherty DM, Schmitz JM, & Swann AC (2001). Psychiatric Aspects of Impulsivity. American Journal of Psychiatry, 158(11), 1783–1793. 10.1176/appi.ajp.158.11.1783 [DOI] [PubMed] [Google Scholar]
- Moody L, Franck C, Hatz L, & Bickel WK (2016). Impulsivity and polysubstance use: A systematic comparison of delay discounting in mono-, dual-, and trisubstance use. Experimental and Clinical Psychopharmacology, 24(1), 30–37. 10.1037/pha0000059 [DOI] [PMC free article] [PubMed] [Google Scholar]
- National Survey on Drug Use and Health. (2023). 2022 NSDUH Detailed Tables [Data Table]. https://www.samhsa.gov/data/report/2019-nsduh-detailed-tables
- Patrick ME, Kloska DD, Terry-McElrath YM, Lee CM, O’Malley PM, & Johnston LD (2018). Patterns of simultaneous and concurrent alcohol and marijuana use among adolescents. The American Journal of Drug and Alcohol Abuse, 44(4), 441–451. 10.1080/00952990.2017.1402335 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Patton JH, Stanford MS, & Barratt ES (1995). Factor structure of the barratt impulsiveness scale. Journal of Clinical Psychology, 51(6), 768–774. 10.1002/1097-4679(199511)51:6<768::AID-JCLP2270510607>3.0.CO;2-1 [DOI] [PubMed] [Google Scholar]
- Ritchwood TD, Ford H, DeCoster J, Sutton M, & Lochman JE (2015). Risky sexual behavior and substance use among adolescents: A meta-analysis. Children and Youth Services Review, 52, 74–88. 10.1016/j.childyouth.2015.03.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Roche DJO, Bujarski S, Green R, Hartwell EE, Leventhal AM, & Ray LA (2019). Alcohol, tobacco, and marijuana consumption is associated with increased odds of same-day substance co- and tri-use. Drug and Alcohol Dependence, 200, 40–49. 10.1016/j.drugalcdep.2019.02.035 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schuler MS, Vasilenko SA, & Lanza ST (2015). Age-varying associations between substance use behaviors and depressive symptoms during adolescence and young adulthood. Drug and Alcohol Dependence, 157, 75–82. 10.1016/j.drugalcdep.2015.10.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schumacher A, Marzell M, Toepp AJ, & Schweizer ML (2018). Association Between Marijuana Use and Condom Use: A Meta-Analysis of Between-Subject Event-Based Studies. Journal of Studies on Alcohol and Drugs, 79(3), 361–369. 10.15288/jsad.2018.79.361 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Skinner CJ, Holt D, & Smith TMF (Eds.). (1989). Analysis of complex surveys. Wiley. [Google Scholar]
- Smith GT, & Cyders MA (2016). Integrating affect and impulsivity: The role of positive and negative urgency in substance use risk. Drug and Alcohol Dependence, 163, S3–S12. 10.1016/j.drugalcdep.2015.08.038 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sobell L, & Sobell M (1992). Timeline follow-back: A technique for assessing self-reported alcohol consumption. Measuring Alcohol Consumption: Psychosocial and Biochemical Methods, 41–72. [Google Scholar]
- Stamates AL, Linden-Carmichael AN, Miller SE, & Feldstein Ewing SW (2023). Impulsivity typologies and simultaneous alcohol and cannabis use. Experimental and Clinical Psychopharmacology, 31(3), 599–604. 10.1037/pha0000608 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Stanford MS, Mathias CW, Dougherty DM, Lake SL, Anderson NE, & Patton JH (2009). Fifty years of the Barratt Impulsiveness Scale: An update and review. Personality and Individual Differences, 47(5), 385–395. 10.1016/j.paid.2009.04.008 [DOI] [Google Scholar]
- Steinfeld MR, & Torregrossa MM (2023). Consequences of adolescent drug use. Translational Psychiatry, 13(1), 313. 10.1038/s41398-023-02590-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tomczyk S, Isensee B, & Hanewinkel R (2016). Latent classes of polysubstance use among adolescents—A systematic review. Drug and Alcohol Dependence, 160, 12–29. 10.1016/j.drugalcdep.2015.11.035 [DOI] [PubMed] [Google Scholar]
- University Park: The Methodology Center, Penn state. (2015). LCA Stata Plugin (Version 1.2) [Software] [Computer software].
- Van Dam NT, & Earleywine M (2011). Validation of the Center for Epidemiologic Studies Depression Scale—Revised (CESD-R): Pragmatic depression assessment in the general population. Psychiatry Research, 186(1), 128–132. 10.1016/j.psychres.2010.08.018 [DOI] [PubMed] [Google Scholar]
- Verdejo-García A, Lawrence AJ, & Clark L (2008). Impulsivity as a vulnerability marker for substance-use disorders: Review of findings from high-risk research, problem gamblers and genetic association studies. Neuroscience & Biobehavioral Reviews, 32(4), 777–810. 10.1016/j.neubiorev.2007.11.003 [DOI] [PubMed] [Google Scholar]
- Vermunt JK, & Magidson J (2005). Technical guide for Latent GOLD 4.0: Basic and advanced Statistical Innovations Inc. [Google Scholar]
- Williams GC, Patte KA, Ferro MA, & Leatherdale ST (2021). Substance use classes and symptoms of anxiety and depression among Canadian secondary school students. Health Promotion and Chronic Disease Prevention in Canada, 41(5), 153–164. 10.24095/hpcdp.41.5.02 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.