Examining the relationship between cannabis use and drinking levels on co-use days

Abstract

Background:

Co-use of alcohol and cannabis among young adults is often associated with more alcohol consumption than alcohol-only use, yet little work has examined cannabis use specifically during heavy drinking. Much of the work examining heavy drinking has focused on heavy episodic drinking (HED, 4+/5+ drinks/occasion for females/males), however young adults report drinking levels that far exceed this, frequently reporting high intensity drinking (HID; 8+/10+ drinks/occasion for females/males), double the HED threshold, which confers greater risk of acute negative consequences. This study examined whether day-level co-use (vs. alcohol-only use) was associated with greater odds of heavier drinking levels: HID vs. HED vs. moderate drinking (1-3/1-4 drinks/occasion for females/males). We explored whether within-person variations in cannabis use characteristics (frequency, form, quantity) differentiated moderate drinking, HED, and HID on co-use days.

Methods:

Young adults who reported simultaneous alcohol and cannabis use (N=318) from three U.S. universities completed five repeated daily surveys for 54 days reporting number of drinks and number of cannabis uses (i.e., frequency/events), forms, and quantity.

Results:

Participants reported increased likelihood of HID and HED versus moderate drinking on co-use versus alcohol-only days. HID versus HED did not differ between co-use and alcohol-only days. On co-use days, heavier drinking was more likely on days with more cannabis use. The form of cannabis used and the use of multiple versus single forms of cannabis were not associated with drinking level. Yet, HID was more likely than HED and moderate drinking when more grams of flower were used and HID was also more likely than moderate drinking when more hits of concentrates were used.

Conclusions:

Findings show heavy drinking (HID, HED) was more likely than moderate drinking on co-use days and cannabis use characteristics may influence drinking levels. Co-use interventions may benefit from a focus on heavy drinking and cannabis use characteristics.

GRAPHICAL ABSTRACT TEXT:

This study examined whether cannabis use differed on days involving high intensity (HID; 8+/10+ drinks for females/males), heavy episodic (HED; 4-7/5-9 drinks), or moderate (1-3 drinks/1-4 drinks) drinking. HID and HED were more likely than moderate drinking on co-use versus alcohol-only days. On co-use days, heavier drinking was associated with more cannabis use. HID was more likely than (a) HED when more flower (grams) was used and (b) moderate drinking when more flower (grams) or concentrates (hits) were used.

Introduction

Heavy drinking among youth is a public health concern with young adults having higher rates of heavy drinking and alcohol use disorder (AUD) than any other age group (Bohm et al., 2021; Kanny et al., 2020). Much of the work on young adult drinking has examined heavy episodic drinking (HED, 4+/5+ drinks for females/males per occasion) with over one fourth of young adults endorsing this level of drinking in the past two weeks (Patrick et al., 2023). However, a substantial portion of young adults report drinking well above this threshold, with 10% reporting high intensity drinking (HID; 8+/10+ drinks for females/males per occasion) in the past two weeks (Patrick et al., 2023). Furthermore, multiple event-level studies of young adults have reported 10-17% of all drinking days were HID (Cox et al., 2022; Evans-Polce et al., 2022; Terry-McElrath et al., 2023). As such, HID is not only prevalent but frequent and should be examined as a unique marker of risk when considering young adult drinking patterns.

Young adults who drink heavily are especially likely to report co-use of alcohol and cannabis including concurrent use (i.e., using alcohol and cannabis but effects do not overlap) and simultaneous use (i.e., using alcohol and cannabis so effects overlap) (Linden-Carmichael et al., 2019; Patrick et al., 2019). In one study, young adults who reported any past year simultaneous use were more likely to report HID than those who only reported drinking (25% versus 6% respectively) and more frequent simultaneous use was associated with increased odds of engaging in any HID (Linden-Carmichael et al., 2019). Those who engage in co-use report increased adverse consequences compared to young adults who only use alcohol (Jackson et al., 2020; Yurasek et al., 2017). Since heavy drinking, and HID in particular, is associated with a range of serious harms (Cox et al., 2019; Hingson et al., 2017; Linden-Carmichael et al., 2017) and harms may be even more severe for individuals who combine alcohol use with cannabis, it is important to also understand cannabis use in the context of heavy drinking occasions.

Day-level studies have also found co-use days to be associated with more alcohol consumption than alcohol-only days (Boyle et al., 2024; Lee et al., 2020). Yet, while co-use has strong associations with heavy drinking, little research has examined co-use on excessive drinking days specifically. One of our previous daily diary studies among young adults found HED to be more likely than moderate drinking on simultaneous alcohol and cannabis use days versus alcohol-only days (Boyle et al., 2024). On the other hand, a study of college students found that there was no difference in HED on co-use versus alcohol-only days, when focused on high-risk drinking weekends (e.g., home football days, holidays) (Mallett et al., 2019). No event-level research has explored cannabis use on HID days.

Competing hypotheses have suggested alcohol and cannabis may be used as substitutes (i.e., one substance can pharmacologically replace the other; resulting in a substance being used less when the other substance is also used) or as complements (i.e., effects of one substance being enhanced by use of the other substance; resulting in a substance being used more when the other substance is also being used) (Gunn et al., 2022; Risso et al., 2020; Subbaraman, 2016). Recent literature reviews show evidence for both substitution and complementarity (Gunn et al., 2022; Risso et al., 2020; Subbaraman, 2016), yet studies specific to young adults including many of the studies examined above show more support for complementary use of alcohol when cannabis is also used (Risso et al., 2020). However, more research is needed to examine cannabis use in the context of very heavy drinking occasions.

In co-use research, cannabis use is often operationalized as any use and, therefore studies often only examine characteristics of alcohol use (e.g., alcohol consumption) during co-use (Boyle et al., 2024; Lee et al., 2020; Linden-Carmichael et al., 2019), limiting our understanding of patterns of cannabis use during co-use occasions. Studies that have examined level of cannabis use during co-use versus cannabis-only occasions have been mixed. One study found number of hits did not differ on simultaneous use versus cannabis-only days (Linden-Carmichael et al., 2020), while another study found college students report a higher level (number of uses) of cannabis use on simultaneous use days compared to cannabis-only days (Jackson et al., 2021). The different measures used to assess level of cannabis use across studies may explain mixed findings.

Unlike alcohol consumption that can be assessed using an established measure of standard drinks (e.g., 12oz of beer, 5oz of wine, 1.5oz of liquor), the heterogeneity of cannabis products in terms of forms (e.g., flower, concentrate, edibles), mode of administration (e.g., smoke, vape, oral ingestion), and cannabinoid composition (i.e., percent THC/CBD) makes cannabis consumption difficult to assess via self-report (Linden-Carmichael & Wardell, 2021), yet these specific characteristics of cannabis use may influence the experience of co-use events. One study found alcohol consumption did not differ on days when multiple alcohol products and flower was used compared to days when multiple alcohol products and concentrates were used, suggesting forms of cannabis may not influence alcohol consumption at least on days when multiple alcohol products were used (Stevens et al., 2021). A survey study found those who use cannabis products with lower THC/CBD ratio report drinking and engaging in co-use less than those who use higher THC/CBD products (Karoly et al., 2020). In a naturalistic observational study where adults who reported prior alcohol and cannabis use were assigned to one of three legal-market cannabis strains (THC dominant, CBD dominant, and THC+CBD), the individuals assigned to the CBD group drank fewer drinks on drinking days and reported fewer co-use days compared to the other groups (Karoly et al., 2021). Overall, our understanding of patterns of cannabis use during co-use occasions and how cannabis use characteristics may influence alcohol use is limited.

Even less work has examined cannabis use characteristics during heavy drinking occasions. A recent study of youth (age 16-24) found quantity of cannabis use was highest on heavy drinking (4+/5+ drinks for females/males) days followed by moderate drinking (1-3 drinks/1-4 drinks for females/males) days and lowest on cannabis-only days (Coughlin et al., 2022). This study examined only grams of flower used and did not examine other forms of cannabis and corresponding quantity. Additionally, heavy drinking occasions examined in this study encompassed both HED and HID days, not allowing for comparison between these different risky drinking levels.

Present Study

The goal of the present study was to examine cannabis use during heavy drinking (HID and HED) versus moderate drinking days. Aim 1 (a) descriptively examined rates of moderate drinking, HED, and HID on co-use and alcohol-only days and (b) tested the hypothesis that using cannabis and alcohol (vs. alcohol-only use) on a given day would be associated with more excessive alcohol consumption (i.e., greater likelihood of HID versus HED, HID versus moderate drinking, and HED versus moderate drinking). Aim 2 examined whether frequency, form, and/or quantity of cannabis were different on co-use days involving HID, HED, or moderate drinking. Given the lack of literature examining cannabis use characteristics during heavy drinking, no formal hypotheses regarding the association between cannabis form, frequency, or quantity and heavy drinking were made.

Materials and Methods

Participants

All procedures were reviewed and approved by the coordinating university’s Institutional Review Board. College students were recruited from three U.S. universities in states where recreational cannabis use laws varied at the time (School A – recreational cannabis use illegal, School B – recreational cannabis use was decriminalized, and School C – recreational cannabis use legal) for a large study examining co-use of alcohol and cannabis. Of 24,000 students (8,000 from each university) randomly invited to participate, a total of 7,000 completed the screener of which 2,874 (41.1%) were eligible to participate. Eligibility criteria included: (1) age 18-24, (2) current full-time enrollment in one of three specific U.S. universities, (3) used alcohol in the past year, and (4) used cannabis in the past year. For more information on screening procedures for the study see White et al. (2019).

Baseline Survey

Of those eligible, a random sample of 2,501 students stratified by university and over sampled for alcohol and cannabis use in the past month were invited to participate in a baseline survey. The final baseline sample included 1,390 students who each received a $25 gift card for completing the survey.

Daily Survey Study

A daily survey study was completed by a subset of students who completed the baseline survey and reported “using alcohol and cannabis at the same time so their effects overlapped” (simultaneous alcohol and cannabis use) at least once in the past month. We oversampled for those who engaged in frequent simultaneous use (3+ times in the past month) and those who were male at birth. Of the 693 students who reported simultaneous alcohol and cannabis use on the baseline survey, 596 were invited to participate in the daily survey study. Enrollment occurred on a rolling basis until quotas were filled resulting in 379 students invited to participate, of which 343 (90.5%) were enrolled and completed the daily survey study.

Daily Survey Procedures

Participants completed two 28-day bursts of repeated daily surveys, on average three months apart. Due to technical difficulties, the first two days of data collection were discarded resulting in a total of 54 days of surveys. Participants were asked to complete 5 surveys daily that were sent at 9am, 2pm, 5pm, 8pm, and 11pm. The first survey of the day was available from 9am-2pm to accommodate the variable sleep patterns of college students, while all other reports were available for 2 hours. The 9am report assessed (a) prior day behavior that occurred between the last survey taken and when the participant went to bed and (b) any negative substance use related consequences that occurred on the prior day. The 2pm report assessed behavior that occurred between when the participant woke up (as reported on the 9am survey) and the current time and the subsequent surveys (5pm, 8pm, and 11pm surveys) assessed behavior that occurred between the submission of the last report and the current time. If one survey was missed, the next survey would ask about behavior between the time of the prior completed survey and the current time. If more than one survey had been missed, the next survey would ask about behavior only from the prior scheduled report time to the current time. Text message reminders were sent 15 minutes prior to the expiration of the survey (see Jackson et al. (2021) and Stevens et al. (2020) supplemental materials for greater detail on study procedures).

For the main analyses only alcohol days with full coverage (i.e., where behavior across the full days was reported) were used, resulting in a final analytic sample of n= 318 and 3111 alcohol days (20% of all days reported). Participants were compensated $1 for each completed survey (up to $5 per day/$35 per week) with bonuses of $10 each week if they completed at least 85% of surveys for that week and $20 at the end of 4 weeks if they completed at least 90% of all surveys, totaling $200 maximum (Amazon gift cards) per burst.

Baseline Measures

Demographics.

Participants reported demographic information including age, sex assigned at birth, gender, college year, race, and ethnicity.

Baseline Substance Use.

As part of the baseline survey, participants reported their recent alcohol, cannabis, and co-use of alcohol and cannabis. Alcohol Use. Participants reported the number of drinking days and number of HED days in the past 30 days. They also reported the peak number of drinks they had on a single day in the past 30 days and the typical number of drinks they had in a week. Participants also completed the Alcohol Use Disorder Identification Test (AUDIT) (Allen et al., 1997; Saunders et al., 1993). Cannabis Use. Participants reported the number of cannabis use days in the past 30 days. Data on typical cannabis use per week was summarized as number of hours high and number of cannabis use sessions. They also completed the Cannabis Use Disorder Test (CUDIT) (Adamson & Sellman, 2003). Co-Use. Participants were asked to report the number of days they used alcohol and cannabis together “so that their effects overlapped” (i.e., simultaneous use) in the past 30 days.

Daily Survey Measures

Alcohol Use.

At each repeated daily survey, participants were asked what substances they used between their previous reported survey and the current time including (a) both alcohol and marijuana, (b) alcohol, (c) marijuana, and (d) neither. Days where young adults reported alcohol use, but not cannabis use were considered alcohol-only days. When participants reported alcohol use, they were asked to “Tap your finger in the blue box each time you had a drink at the corresponding time.” They were presented with a graphical interface where they could tap at each specific time when they consumed a drink (see Jackson et al. (2021) for images of this interface). The total number of drinks reported across the repeated surveys were summed to determine the total number of drinks consumed each day. Indicators of whether each day was a moderate drinking day (1-3 drinks/1-4 drinks for females/males), a heavy episodic drinking day (HED; 4-7/5-9 drinks for females/males) or a high-intensity drinking day (HID; 8+/10+ drinks for females/males) were created.

Cannabis Use.

Participants were provided with the same graphical interface where they could tap at each specific time when they engaged in cannabis use. Cannabis use occasions were self-defined by participants. The total number of cannabis uses reported across the repeated surveys were summed to determine the total number of uses each day (frequency of use). Participants were also asked to report what forms of cannabis they used including flower, concentrate, and/or edibles (see Stevens et al. (2021) for additional details on forms of cannabis used). A dichotomous indicator of whether single or multiple forms of cannabis were used each day was created. Furthermore, participants reported on the quantity of flower and cannabis concentrates they used. Participants reported the number of grams (<1/4 - 5) of flower and the number of hits of concentrates (1-15+).

Co-use.

Any days where participants reported one or more drinks, and one or more cannabis uses via the graphical interface were considered co-use days. Recent work using this dataset (Sokolovsky et al., 2020) found almost all co-use days (94%) included co-use occurring within four hours, suggesting most co-use days included overlapping effects of alcohol and cannabis. An ordering variable was created indicating which substance (alcohol or cannabis) was used first on co-use days.

Negative Consequences.

On days when participants reported prior day alcohol and/or cannabis use participants were asked to report on seven consequences (i.e., blackout, hangover, nausea/vomiting, hurt self, drove car high/drunk, being rude/aggressive, unwanted sex) in the report triggered at 9am (Johnson & White, 1995, Kahler et al., 2005, Read et al., 2006, White & Labouvie, 1989). Sum scores were created for total number of negative consequences each day and reported as descriptives.

Analyses

Descriptive statistics on drinking levels for alcohol-only and co-use days were examined including percentage of days that were moderate drinking, HED, and HID. Also, descriptives of cannabis use characteristics on co-use days were reported. We used linear mixed-effects models to address our primary research aims. For Aim 1, one multinomial model was run where we compared the influence of day type (co-use day (coded 1) vs. alcohol-only day (coded 0)) on the likelihood of drinking to HID versus moderate drinking levels, as well as HED to moderate drinking levels. In this model, moderate drinking served as the reference group. A second multinomial model was run with HED as the reference group to examine the likelihood of drinking to HID versus HED drinking levels. Together these two multilevel models obtained pairwise comparisons between all three drinking levels.

For Aim 2, models were conducted among the subset of co-use days. First, we examined the association between number of cannabis uses and drinking level (moderate/HED/HID) on co-use days. Next, we examined differences in drinking level during days when multiple forms of cannabis were used (i.e., flower, concentrates, and/or edibles) versus days where only one form of cannabis was used. We also examined whether the form of cannabis used on co-use days (limited to days when only a single form was used) was associated with drinking level. For co-use days of flower use only, we examined whether grams of cannabis used was associated with drinking level. Also, for co-use days of concentrate use only, we examined whether number of hits of concentrates was associated with drinking level. As with Aim 1, for Aim 2, two separate multinomial models were run for each analysis to obtain pairwise comparisons between all three levels of alcohol use.

For all models, we covaried (a) whether each day was a weekend (Friday and Saturday; coded (1)) or weekday (Sunday-Thursday; coded (0)) given increases in substance use on Fridays and Saturdays and (b) day in study (the number of days the participant had been enrolled in the study, corresponding to 3-28 and 85-112 with slight variations across participants) to account for potential changes in drinking behavior over time. Additionally, at Level 2, we controlled for sex (1 = male, 0 = female) and age differences in outcomes. We included school as a covariate to control for differences in cannabis legalization status at the site of recruitment (reference group: School A - recreational cannabis use illegal). For Aim 2 models conducted among the subset of co-use days, we also controlled for order of substance use defined as which substance was used first (1=alcohol use first, 0=cannabis use first) since prior research has found use of alcohol first on co-use days is associated with increased alcohol consumption (Gunn et al., 2021; Karoly et al., 2023).

We also controlled for person-level aggregates of our predictors of interest for each model (i.e., Aim 1: proportion of co-use days; Aim 2: average cannabis uses across all days, the proportion of all co-use days where multiple forms of cannabis were used, average grams used, average number of hits) to isolate the effects at the day level. Intercepts were specified as random to account for mean differences in the outcome across participants. Random slopes were tested and excluded when non-significant. All models were tested in HLM 7.02, using full information maximum likelihood estimation and we relied on robust standard errors when interpreting results.

Results

Descriptives

Sample descriptives that were derived from both the baseline and daily surveys are presented in Table 1. Across the analytic sample of 318 participants, there were 1581 (51%) co-use days and 1530 (49%) alcohol-only days. On alcohol-only days, 259 (17%) were HID days, 455 (30%) HED days, and 816 (53%) were moderate drinking days. On co-use days, 308 (20%) were HID days, 512 (32%) were HED days, and 761 (48%) were moderate drinking days. A total of 258 participants (81% of the analytic sample) reported any co-use during the EMA study (range from 1-35 of 54 days). See Table 2 for descriptives of cannabis use characteristics by co-use days of all drinking levels, co-use HID days, co-use HED days, co-use moderate drinking days, and cannabis-only days.

Table 1.

Participant Characteristics (N= 318)

	N/% or Mean (SD)
Age	19.83 (1.31)
Biological Sex: Female	168 (53%)
Gender (check all that apply)
Female	165 (52%)
Male	151 (48%)
Trans man	2 (1%)
Genderqueer/Gender non-conforming	2 (1%)
Different Identity	1 (<1%)
Race
African American or Black	9 (3%)
Asian	32 (10%)
Hawaiian/Pacific Islander	1 (<1%)
Native American/Alaskan	2 (1%)
White	241 (76%)
Mixed Race	26 (8%)
Other	7 (2%)
Ethnicity: Hispanic/Latino	33 (10%)
College Year
First year	61 (19%)
Sophomore	87 (27%)
Junior	78 (24%)
Senior	79 (25%)
Five or more year senior	13 (4%)
Cannabis Legalization Status in State of School Enrolled
Recreational use legal	107 (34%)
Recreational use decriminalized	111 (35%)
Recreational use illegal	100 (32%)
Baseline Substance Use Behavior
# of drinking days in the past 30 days	8.41 (5.59)
# of HED days in the past 30 days	4.49 (4.04)
# of cannabis use days in the past 30 days	13.38 (10.61)
# of simultaneous use days in the past 30 days	4.11 (4.21)
# of maximum drinks consumed in a day in the past 30 days	7.65 (3.51)
# of drinks per a typical week	13.28 (9.75)
# of cannabis use session per a typical week	7.56 (8.63)
# of hour high from cannabis on a typical week	13.38 (13.40)
AUDIT	9.70 (5.17)
CUDIT	9.49 (6.15)
Daily Survey Descriptives
# of alcohol-only days	4.81 (5.42)
# of cannabis-only days	9.06 (10.61)
# of co-use days	4.97 (6.36)
# of drinks per drinking day	5.12 (2.83)
# of negative consequences on alcohol-only days	0.31 (0.64)
# of negative consequences on cannabis-only days	0.10 (0.31)
# of negative consequences on co-use days	0.36 (0.64)

Table 2.

Descriptives of Cannabis Use Characteristics

Cannabis Use Characteristics	N/% or Mean (SD)
	Co-Use Days				Cannabis-only Days
	All Drinking Level Days	HID Days	HED Days	Moderate Drinking Days
Cannabis Uses1	5.63 (7.26)	8.60 (11.15)	5.33 (5.60)	4.62 (5.81)	5.12 (5.95)
Multiple Forms1	210 (13.3%)	42 (13.6%)	63 (12.3%)	105 (13.8%)	310 (10.8%)
Flower2	1088 (79.4%)	203 (76.3%)	356 (79.3%)	529 (80.6%)	2070 (80.5%)
Concentrates2	247 (18.0%)	57 (21.4%)	81 (18.0%)	109 (16.6%)	438 (17%)
Edibles2	36 (2.6%)	6 (2.3%)	12 (2.7%)	18 (2.7%)	64 (2.5%)
Grams of Flower3	0.91 (1.36)	1.06 (1.32)	0.95 (1.31)	0.81 (1.40)	0.87 (1.13)
Hits of Concentrates4	7.92 (9.26)	12.54 (11.65)	7.41 (8.27)	5.33 (7.34)	7.52 (7.72)

¹Co-use days N=1581 Cannabis-only days N=2882

²Based on days when only a single form of cannabis was used: Co-use days N=1371 Cannabis-only days N=2572

³Based on days when only flower was used: Co-use days N=1088, Cannabis-only days N=2070

⁴Based on days when only concentrate was used: Co-use days N=247, Cannabis-only days N=438

Co-use and Drinking Levels

Results from Aim 1 indicated that HID versus moderate drinking and HED versus moderate drinking were more likely to occur on co-use use days than on alcohol-only days. This finding showed heavy drinking (HID, HED) was more likely than moderate drinking on days when any cannabis was used versus days of only drinking. HID versus HED did not differ between co-use days and alcohol-only days (see Table 3).

Table 3.

Associations between co-use and drinking level: HID, HED, and moderate drinking (N=318, all drinking days 3111)

	Likelihood of HID vs. HED			Likelihood of HID vs. moderate drinking			Likelihood of HED vs. moderate drinking
	Odds Ratio	95% Confidence Interval	p-value	Odds Ratio	95% Confidence Interval	p-value	Odds Ratio	95% Confidence Interval	p-value
Intercept	0.40	0.28, 0.57	<0.001	0.19	0.12, 0.30	<0.001	0.47	0.35, 0.64	0.011
Level 1
Co-use	1.14	0.86, 1.52	0.367	1.52	1.14, 2.04	0.005	1.33	1.04, 1.72	0.026
Weekend	1.80	1.44, 2.24	<0.001	3.61	2.86, 4.54	<0.005	2.01	1.66, 2.43	<0.001
Study Day	1.00	0.99, 0.99	0.002	0.99	0.99, 0.99	<0.001	1.00	0.99, 1.00	<0.430
Level 2
Sex	0.94	0.69, 1.30	0.715	0.79	0.52, 1.20	0.264	0.84	0.66, 1.07	0.151
Age	0.87	0.77, 0.98	0.025	0.74	0.63, 0.88	<0.001	0.86	0.78, 0.94	<0.001
School B	0.97	0.68, 1.40	0.883	1.06	0.65, 1.75	0.813	1.09	0.81, 1.47	0.569
School C	0.92	0.65, 1.32	0.662	0.83	0.50, 1.38	0.479	0.90	0.66, 1.24	0.522
Proportion of co-use days	1.01	0.60, 1.73	0.958	1.09	0.56, 2.11	0.798	1.07	0.70, 1.66	0.744

Note: HID = High-intensity drinking: 8+/10+ drinks for females/males; HED = Heavy episodic drinking: 4-7/5-9 drinks for females/males; Moderate drinking =1-3 drinks/1-4 drinks for females/males. Bold indicates the effects of interest when significant.

Co-use is coded 1=Co-use, 0=alcohol-only use; Sex is coded 0=female, 1=male; Weekend is coded 0=Sunday-Thursday and 1=Friday or Saturday. The referent group for School is School A located where recreational cannabis use was illegal at the time of data collection.

Characteristics of Cannabis Use and Drinking Levels on Co-use Days

Number of cannabis uses.

For Aim 2, models examining number of cannabis uses found that heavier drinking (HID versus HED, HID versus moderate drinking, and HED versus moderate drinking) was more likely when more cannabis uses occurred during a co-use day. Participants who reported more cannabis uses on average in a day were more likely to report HID versus HED and HED versus moderate drinking (See Table 4).

Table 4.

Associations between number of cannabis uses and drinking levels on co-use days (N=258, co-use days 1581)

	Likelihood of HID vs. HED			Likelihood of HID vs. moderate drinking			Likelihood of HED vs. moderate drinking
	Odds Ratio	95% Confidence Interval	p-value	Odds Ratio	95% Confidence Interval	p-value	Odds Ratio	95% Confidence Interval	p-value
Intercept	0.34	0.21, 0.53	<0.001	0.22	0.13, 0.38	<0.001	0.65	0.43, 0.99	0.047
Level 1
Number of cannabis uses	1.05	1.02, 1.08	<0.001	1.08	1.05, 1.17	<0.001	1.06	1.01, 1.11	0.016
Weekend	1.86	1.37, 2.52	<0.001	3.52	2.46, 5.04	<0.001	1.89	1.43, 2.50	<0.001
Study Day	1.00	0.99, 0.99	0.015	0.99	0.99, 0.99	0.001	1.00	0.99, 1.00	0.206
Order	1.81	1.23, 2.65	0.002	2.85	1.91, 4.26	<0.001	1.58	1.21, 2.06	<0.001
Level 2
Sex	0.79	0.54, 1.14	0.203	0.61	0.38, 0.98	0.041	0.78	0.58, 1.05	0.098
Age	0.87	0.75, 1.01	0.065	0.76	0.63, 0.91	0.003	0.87	0.78, 0.97	0.016
School B	0.90	0.55, 1.46	0.660	0.87	0.46, 1.67	0.679	0.97	0.65, 1.46	0.898
School C	1.20	0.77, 1.88	0.432	0.99	0.53, 1.86	0.985	0.83	0.56, 1.24	0.363
Mean number of cannabis uses	1.08	1.02, 1.15	0.012	1.11	1.02, 1.22	0.021	1.03	0.99, 1.07	0.132

Note: HID = High-intensity drinking: 8+/10+ drinks for females/males; HED = Heavy episodic drinking: 4-7/5-9 drinks for females/males; Moderate drinking =1-3 drinks/1-4 drinks for females/males. Bold indicates the effects of interest when significant.

Order is coded 0=cannabis use first, 1=alcohol use first; Sex is coded 0=female, 1=male; Weekend is coded 0=Sunday-Thursday and 1=Friday or Saturday. The referent group for School is School A located where recreational cannabis use was illegal at the time of data collection.

Forms of cannabis used.

As for forms of cannabis used on co-use days, drinking levels (HID vs. HED, HID versus moderate drinking, and HED versus moderate drinking) did not differ between days when multiple forms of cannabis were used versus days when only one form of cannabis was used (See Table 5). Additionally, on co-use days where only one form of cannabis was used, drinking level did not differ depending on the form of cannabis that was used (flower vs. concentrates vs. edibles) (See Table 6).

Table 5.

Associations between multiple vs. a single form of cannabis use and drinking levels on co-use days (N=258, co-use days 1581)

	Likelihood of HID vs. HED			Likelihood of reporting HID vs. moderate drinking			Likelihood of HED vs. moderate drinking
	Odds Ratio	95% Confidence Interval	p-value	Odds Ratio	95% Confidence Interval	p-value	Odds Ratio	95% Confidence Interval	p-value
Intercept	0.40	0.25, 0.64	<0.001	0.27	0.16, 0.46	<0.001	0.68	0.45, 1.03	0.071
Level 1
Multiple forms used	0.99	0.62, 1.58	0.974	0.99	0.65, 1.51	0.973	1.00	0.70, 1.43	0.997
Weekend	1.79	1.32, 2.42	<0.001	3.39	2.40, 4.78	<0.001	1.90	1.43, 2.52	<0.001
Study Day	0.99	0.99, 0.99	0.010	0.99	0.99, 0.99	<0.001	1.00	0.99, 1.00	0.215
Order	1.41	0.98, 2.03	0.066	2.03	1.42, 2.90	<0.001	1.44	1.12, 1.85	0.004
Level 2
Sex	0.79	0.53, 1.17	0.239	0.63	0.39, 1.03	0.065	0.80	0.59, 1.08	0.143
Age	0.84	0.72, 0.99	0.038	0.73	0.60, 0.89	0.002	0.86	0.77, 0.97	0.011
School B	0.97	0.61, 1.54	0.904	0.96	0.53, 1.73	0.893	0.99	0.67, 1.47	0.952
School C	1.05	0.68, 1.62	0.822	0.86	0.46, 1.60	0.629	0.82	0.55, 1.21	0.312
Proportion of multiple forms days	2.40	0.88, 6.53	0.086	2.07	0.62, 6.92	0.239	0.86	0.34, 2.20	0.752

Note: HID = High-intensity drinking: 8+/10+ drinks for females/males; HED = Heavy episodic drinking: 4-7/5-9 drinks for females/males; Moderate drinking =1-3 drinks/1-4 drinks for females/males. Bold indicates the effects of interest when significant.

Multiple forms used is coded 1=Yes, multiple forms, 0=No, Single form used; Order is coded 0=cannabis use first, 1=alcohol use first; Sex is coded 0=female, 1=male; Weekend is coded 0=Sunday-Thursday and 1=Friday or Saturday. The referent group for School is School A located where recreational cannabis use was illegal at the time of data collection.

Table 6.

Associations between specific forms of cannabis use and drinking levels on co-use days (N=254, co-use days where only a single form was used 1371)

	Likelihood of HID vs. HED			Likelihood of HID vs. moderate drinking			Likelihood of HED vs. moderate drinking
	Odds Ratio	95% Confidence Interval	p-value	Odds Ratio	95% Confidence Interval	p-value	Odds Ratio	95% Confidence Interval	p-value
Intercept	0.40	0.25, 0.64	<0.001	0.27	0.16, 0.46	<0.001	0.67	0.44, 1.03	0.068
Level 1
Concentrates	1.15	0.68, 1.93	0.605	1.31	0.77, 2.24	0.316	1.15	0.78, 1.68	0.485
Edibles	0.86	0.24, 3.12	0.816	0.71	0.19, 2.71	0.616	0.83	0.42, 1.65	0.592
Weekend	1.80	1.32, 2.46	<0.001	3.68	2.63, 5.15	<0.001	2.04	1.53, 2.73	<0.001
Study Day	0.99	0.99, 0.99	0.005	0.99	0.99, 0.99	<0.001	1.00	0.99, 1.00	0.117
Order	1.41	0.97, 2.05	0.069	1.99	1.37, 2.90	<0.001	1.41	1.09, 1.82	0.009
Level 2
Sex	0.87	0.57, 1.31	0.497	0.70	0.42, 1.16	0.161	0.80	0.59, 1.09	0.163
Age	0.84	0.71, 0.99	0.049	0.73	0.59, 0.89	0.002	0.86	0.76, 0.97	0.013
School B	1.01	0.61, 1.68	0.960	0.96	0.52, 1.77	0.886	0.94	0.63, 1.41	0.777
School C	1.15	0.72, 1.83	0.556	1.01	0.50, 1.80	0.879	0.83	0.56, 1.22	0.338

Note: High-intensity drinking: 8+/10+ drinks for females/males; HED = Heavy episodic drinking: 4-7/5-9 drinks for females/males; Moderate drinking =1-3 drinks/1-4 drinks for females/males. Bold indicates the effects of interest when significant.

The referent group for form is flower. Order is coded 0=cannabis use first, 1=alcohol use first; Sex is coded 0=female, 1=male; Weekend is coded 0=Sunday-Thursday and 1=Friday or Saturday. The referent group for School is School A located where recreational cannabis use was illegal at the time of data collection.

Quantity of cannabis used.

On co-use days when only flower was used, HID was more likely than HED and HID was more likely than moderate drinking when more grams of cannabis were used. The likelihood of HED versus moderate drinking was not associated with grams of cannabis used (see Table 7). On co-use days when only concentrates were used, HID was more likely than moderate drinking when more hits of cannabis were used. The likelihood of HID versus HED and HED versus moderate drinking was not associated with hits of concentrates used (see Table 8).

Table 7.

Associations between grams of flower used and drinking levels (N=216, co-use days where only a single form of cannabis was used: flower 1088)

	Likelihood HID vs. HED			Likelihood of HID vs. moderate drinking			Likelihood of HED vs. moderate drinking
	Odds Ratio	95% Confidence Interval	p-value	Odds Ratio	95% Confidence Interval	p-value	Odds Ratio	95% Confidence Interval	p-value
Intercept	0.31	0.18, 0.53	<0.001	0.23	0.13, 0.43	<0.001	0.75	0.48, 1.20	0.233
Level 1
Grams of cannabis used	1.30	1.03, 1.65	0.030	1.51	1.12, 2.05	0.008	1.16	0.83, 1.63	0.382
Weekend	1.99	1.40, 2.83	<0.001	3.72	2.51, 5.51	<0.001	1.87	1.33, 2.64	<0.001
Study Day	0.99	0.99, 0.99	0.002	0.99	0.99, 0.99	<0.001	1.00	0.99, 1.00	0.306
Order	2.06	1.42, 2.99	<0.001	2.44	1.59, 3.76	<0.001	1.18	0.89, 1.58	0.248
Level 2
Sex	1.05	0.68, 1.62	0.819	0.78	0.46, 1.33	0.352	0.74	0.52, 1.05	0.090
Age	0.77	0.65, 0.91	0.002	0.69	0.56, 0.84	<0.001	0.90	0.79, 1.02	0.109
School B	1.04	0.64, 1.69	0.873	0.98	0.52, 1.83	0.947	0.94	0.61, 1.46	0.786
School C	1.04	0.60, 1.78	0.898	0.81	0.40, 1.67	0.570	0.78	0.49, 1.26	0.316
Mean number of grams used	0.88	0.70, 1.21	0.308	0.90	0.68, 1.19	0.454	1.02	0.91, 1.14	0.779

Note: High-intensity drinking: 8+/10+ drinks for females/males; HED = Heavy episodic drinking: 4-7/5-9 drinks for females/males; Moderate drinking =1-3 drinks/1-4 drinks for females/males. Bold indicates the effects of interest when significant.

Order is coded 0=cannabis use first, 1=alcohol use first; Sex is coded 0=female, 1=male; Weekend is coded 0=Sunday-Thursday and 1=Friday or Saturday. The referent group for School is School A located where recreational cannabis use was illegal at the time of data collection.

Table 8.

Associations between number of hits of concentrates and drinking levels (N=90, co-use days where only a single form of cannabis was used: concentrates 247)

	Likelihood of HID vs. HED			Likelihood of HID vs. moderate drinking			Likelihood of HED vs. moderate drinking
	Odds Ratio	95% Confidence Interval	p-value	Odds Ratio	95% Confidence Interval	p-value	Odds Ratio	95% Confidence Interval	p-value
Intercept	0.42	0.10, 1.78	0.233	0.11	0.04, 0.35	<0.001	0.27	0.06, 1.23	0.09
Level 1
Number of hits	1.05	0.93, 1.20	0.381	1.09	1.03, 1.15	0.005	1.03	0.89, 1.19	0.965
Weekend	1.58	0.78, 3.21	0.199	3.32	1.64, 6.72	0.001	2.10	1.12, 3.91	0.013
Study Day	1.00	0.99, 1.02	0.654	1.00	0.98, 1.02	0.865	1.00	0.99, 1.01	0.965
Order	0.59	0.17, 2.02	0.394	2.01	0.81, 5.00	0.131	3.40	1.46, 7.93	0.005
Level 2
Sex	0.62	0.26, 1.48	0.276	0.59	0.19, 1.81	0.351	0.96	0.44, 2.05	0.907
Age	1.08	0.77, 1.52	0.651	0.77	0.49, 1.20	0.242	0.71	0.52, 0.97	0.032
School B	0.73	0.23, 2.28	0.581	1.11	0.30, 4.15	0.878	1.52	0.60, 3.85	0.369
School C	1.53	0.55, 4.24	0.413	1.82	0.54, 7.46	0.290	1.32	0.54, 3.22	0.536
Mean number of hits	1.02	0.91, 1.14	0.655	1.02	0.85, 1.07	0.429	0.93	0.81, 1.08	0.689

Note: High-intensity drinking: 8+/10+ drinks for females/males; HED = Heavy episodic drinking: 4-7/5-9 drinks for females/males; Moderate drinking =1-3 drinks/1-4 drinks for females/males. Bold indicates the effects of interest when significant.

Order is coded 0=cannabis use first, 1=alcohol use first; Sex is coded 0=female, 1=male; Weekend is coded 0=Sunday-Thursday and 1=Friday or Saturday. The referent group for School is School A located where recreational cannabis use was illegal at the time of data collection.

Discussion

The current study examined the relationship between cannabis use and drinking levels on co-use days among young adults. Participants reported heavy drinking on almost half of all drinking days with drinking at extremely high levels (10 or more drinks for males, 8 or more drinks for females) reported on 17% of alcohol-only days and 20% of co-use days, highlighting the prevalence of heavy drinking among young adults who engage in co-use.

The results of the study support complementary effects of cannabis on alcohol use. We found participants reported heavier drinking, increased likelihood of HID and HED versus moderate drinking, on co-use days versus alcohol-only days. These findings correspond to previous between (Linden-Carmichael et al., 2019; Patrick et al., 2019) and within-person (Boyle et al., 2024) studies that have highlighted associations between co-use and heavy drinking. Reductions in inhibitory control that have been found to occur following use of cannabis may be one factor that leads to heavier drinking levels on co-use days (Ramaekers et al., 2016).

The relationship between heavier (vs. moderate) drinking levels and co-use may also be explained by a motive to increase positive effects experienced during substance use. Positive effect motives for engaging in co-use such as “cross-fading” (i.e., to enhance the effect of alcohol or cannabis) have been found to be common among young adults and suggest that this population may often engage in co-use, particularly complementary use of alcohol and cannabis, to increase intoxicated effects (Patrick et al., 2020). Young adults may drink to heavy (vs. moderate) drinking levels on co-use days to further facilitate intoxication goals. While young adults may be more likely to drink to heavy levels on co-use days, there was no difference in the likelihood of HID versus HED on co-use versus alcohol-only days. At these heavier drinking levels, individuals may often achieve their intoxication goals without also using cannabis.

Research has also shown increased alcohol and cannabis consumption on planned versus unplanned co-use occasions (Fairlie et al., 2021, Stevens et al., 2022). During individual interviews with young adults who reported simultaneous use, they described consuming more alcohol and/or cannabis on planned occasions due to increased availability of substances and/or because they had set aside time to use substances and be able to recover from possible negative outcomes. They mentioned that consuming more during planned events did not necessarily equate to worse outcomes. During planned occasions they reported often monitoring their use of substances and space out their substance use, while still feeling overlapping effects, to avoid negative consequences (Boyle et al., 2023). Together, this research suggest substance use patterns and effects, motives, cognitions, and context may all influence consumption levels on co-use days.

While much of the co-use research has operationalized cannabis use as any use, we were interested in understanding cannabis use characteristics during co-use occasions. We examined whether frequency (i.e., number of uses in a day), form (i.e., flower, concentrates, edibles), and/or quantity (i.e., grams, hits) of cannabis differentiated drinking levels on co-use days. Prior event-level research has found any cannabis use is associated with heavier alcohol consumption (Boyle et al., 2024; Lee et al., 2020). When we examined frequency of cannabis use on co-use days, heavier drinking levels were more likely on days with more cannabis uses. That is, not only were co-use days characterized by high volume drinking, but also accompanied by high quantities of cannabis. Jackson and colleagues (2021) using the same data set showed that college students reported more cannabis uses on co-use days compared to cannabis-only days. Together these findings suggest increased cannabis use when also drinking particularly during heavy drinking occasions.

In examining how cannabis impacts alcohol use on drinking days specifically, our findings suggest complementary use of alcohol and cannabis on co-use days. Yet, participants in this sample did report more cannabis-only days on average than co-use or alcohol-only days. It is possible that some individuals may choose to use cannabis on some days instead of drinking alcohol (i.e., full substitution of cannabis for alcohol). Future research could directly assess when and under what conditions young adults choose to substitute cannabis for alcohol or substitute alcohol for cannabis.

Reactive or situational reasons may explain more frequent cannabis use occurring on days with heavier drinking. In prior qualitative data we collected, college students who engaged in heavy drinking reported that simultaneous use often occurred in response to one’s environment including using cannabis during a drinking event because cannabis was available and offered (Boyle et al., 2021). Heavier drinking events may occur in contexts where cannabis use is readily available (e.g., parties or celebrations) allowing for more frequent use of cannabis throughout the drinking event. Also, heavier drinking events may occur over longer periods of time than lighter drinking events allowing for more times of cannabis use. In this prior qualitative study, college students stated that they often decided to use cannabis because they were not able to think through this decision while intoxicated (Boyle et al., 2021). Alcohol myopia theory suggests that excessive alcohol use reduces attentional resources and inhibition (Steele & Josephs, 1990). Thus, at heavier drinking levels young adults may allocate their attention on salient, proximal stimuli such as using cannabis more often (than during lighter drinking occasions) with people around them who are also using cannabis.

Further examination of cannabis used during co-use, led us to find that the likelihood of using specific forms of cannabis or using multiple forms versus a single form did not differ at different drinking levels. Yet, when examining quantities of specific forms of cannabis on co-use days, we found drinking to extreme levels (i.e., HID) compared to HED or moderate drinking was associated with more grams of flower and drinking to HID levels compared to moderate drinking was associated with more hits of concentrates. This suggests complementary use of flower and concentrates at HID (vs. lower drinking) levels. Increased quantities of flower and concentrates on HID days suggest that HID co-use days may be particularly risky. Further research is needed to explore how substance use risks are enhanced or decreased when different quantities of cannabis and alcohol are both used.

Cannabis in the form of flower remains the most common form of cannabis used (Russell et al., 2018) and in this study, flower was used on 79% of co-use days when only a single form was used. Yet, concentrates (e.g., shatter, wax, dabs) and edibles (e.g., oral cannabis products) are rapidly increasing in popularity as these novel forms have become more readily available as more cannabis dispensaries open (Goodman et al., 2020). Concentrate use alone has been associated with negative consequences, cannabis use disorder symptomology, and anxiety among college students (Meier, 2017). Edibles can produce unpredictable and potentially extreme subjective effects from over-intoxication/overuse due to the delayed onset of drug effects (Barrus et al., 2016; Ghosh et al., 2015; Hudak et al., 2015; Huestis, 2007). Such adverse outcomes make edibles and concentrates risky cannabis use forms that are important to understand. Notably, increased access has led to individuals rarely using a sole formulation/mode of administration (Streck et al., 2019), and individuals who use multiple formulations report more negative consequences (Gunn et al., 2020), high-risk behaviors such as driving after cannabis use (Krauss et al., 2017), and greater cannabis dependence severity (Baggio et al., 2014; Craft et al., 2019).

At the time of data collection for this study, only 34% of our sample were residing in a state where recreational cannabis use was legal, and college students could access cannabis dispensaries. This may be why we saw low rates of concentrate and edible use, as well as infrequent use of multiple forms of cannabis while drinking. With the increased access and elevated risks associated with concentrates and edibles, it is important for future work to further examine these novel forms in relation to co-use. Research comparing states with and without legalized recreational cannabis use and examining this over time could provide insight into how the changing legal and social landscape influence co-use patterns.

In exploring how patterns of cannabis use may influence drinking levels, we included which substance was used first as a covariate. Similar to prior research (Gunn et al., 2021; Karoly et al., 2023), use of alcohol first was often associated with heavier drinking levels in our models. For example, in the model examining whether number of cannabis uses was associated with drinking level, use of alcohol first was associated with heavier drinking across all three pairwise comparisons. Laboratory research has found alcohol use can increase disinhibition (Källmén & Gustafson, 1998) and alcohol-induced disinhibition has predicted increased alcohol consumption (Jones & Field, 2013; Weafer & Fillmore, 2008). On co-use days when young adults begin by drinking, they may drink more heavily across the day due to alcohol-related behavioral disinhibition.

Yet, other aspects of the topography of co-use, other than which substance was used first, may also influence substance use outcomes. Future work should consider using ecological momentary assessment methods to examine how co-use of alcohol and cannabis unfolds across a substance use occasion in real-time. Protocols could include asking participants to self-initiate each time they use alcohol or cannabis and collect follow-up assessments across the co-use day. This would allow for a detailed characterization of the topography of the co-use day in real-time including when each substance was used, the amount and form that was used at each time point, and the order and spacing of use across the day. Such fine-grained analysis could help to identify patterns of co-use that are of greatest risk.

Clinical Implications

Although brief interventions have been developed for alcohol and cannabis use separately (Gex et al., 2024, Tanner-Smith & Lipsey, 2015), the current findings suggest days of heavy consumption may include use of both alcohol and cannabis. Targeting both alcohol and cannabis use via the promotion of protective behavioral strategies (PBS) such as setting limits on the amount of alcohol and cannabis use in a day and/or limiting the frequency in which alcohol is consumed when also using cannabis may help to reduce negative consequences and prevent the development of substance use disorders. Additionally, psychoeducational strategies could be used to inform young adults that using cannabis may not be an effective strategy to reduce alcohol consumption on drinking days, despite common endorsement of such harm reduction motives (i.e., using cannabis to reduce alcohol consumption and/or related alcohol consequences, Boyle et al., 2021; Patrick et al., 2018). Furthermore, using just-in-time adaptive interventions to target cognitions and behavioral skills (e.g., motives for use, use of PBS) during the substance use event when they are most salient and opportune for modification may improve outcomes.

Limitations

Overall, this study contributes to the growing body of research exploring co-use by examining patterns of cannabis use on heavy drinking days. Yet, this study is not without limitations to consider. Participants were all college students, the majority of which identified as White. Therefore, findings may not generalize to more diverse young adult populations. Our analyses were restricted to alcohol-only days with full coverage (i.e., where behavior across the full day was reported) to confidently (a) categorize each alcohol day into the correct drinking level and (b) examine cannabis use patterns across the day. While our analytic sample included a high number of HID and HED days, it is important to acknowledge that prior research has found noncompliance with daily survey studies to be associated with substance use (Serre et al., 2012, Stevens et al., 2020). Therefore, it is possible some of the heavier drinking days in the study were not captured in our analyses because they were days of incomplete coverage. To avoid potential bias, future daily survey studies examining heavy substance use should consider objective measures of substance use when possible. For example, wearable alcohol sensors provide an accurate and noninvasive method for tracking alcohol intoxication that can supplement retrospective or real-time reports of drinking (Richards et al., 2023, Russell et al., 2022). Also, while we were interested in understanding how forms of cannabis used may differentiate drinking levels on co-use days, most of the cannabis used by this sample was flower with very little use of edibles and we did not collect information on quantity of edibles used. Future research should examine co-use among those who use various forms of cannabis including multiple forms in a day to further understand cannabis use patterns during co-use occasions and related outcomes. In this study participants defined what a cannabis use occasion was. Nonetheless, because we used within-person analyses, their subjective definition should not affect the results. Also, while we were interested in examining characteristics of cannabis use during heavy drinking occasions, we did not collect day-level information on the cannabinoid composition of the cannabis used. Note, however, that in the baseline survey of this study 67% of students reported they did not know the typical potency of the flower they used and 26% reported they did not know the typical potency of concentrates they used. This may be less likely with current college students as young adults have increasing access to cannabis from dispensaries that include packaging with cannabinoid composition information. Two prior studies have found associations between use of cannabis products with lower levels of THC and higher levels of CBD and less alcohol consumption (Karoly et al., 2021; Karoly et al., 2020). This work underscores the importance of considering cannabinoid content in future work examining the relationship between cannabis and alcohol co-use.

Conclusions

This co-use study examined whether alcohol and cannabis were used as substitutes or as complements. The research provided support for the complement hypothesis, finding that heavier drinking levels are associated with more frequent cannabis use in a given day -- that is, more cannabis product (amount used) in a day. Additionally, both more grams of cannabis and hits of concentrates were reported on HID days compared to moderate drinking days and more grams of cannabis were reported on HID days compared to HED days. More research is needed to examine cannabis use characteristics during drinking occasions including examination of how cannabinoid composition is related to co-use of alcohol. Further research should explore whether alcohol and cannabis are complements across different populations (e.g., older adults), and examine how the relationship between cannabis use characteristics and alcohol consumption may differ across contexts and types of occasions (e.g., planned versus unplanned occasions). Next steps in this research should include further examining cannabis use characteristics during drinking to determine what patterns of use are associated with the greatest harm (e.g., driving while intoxicated, feeling nauseous/vomiting, memory problems). Identifying the riskiest patterns of co-use will provide important information for developing ecological momentary interventions (EMI) such as just-in-time adaptive interventions targeting patterns of high risk, adapted to the individual.