Momentary Craving, Craving Variability, and Cannabis Use: Associations With THC Concentrates and Sex

Abstract

Objective:

Ecological momentary assessment (EMA) studies are well positioned to assess the impact of craving on cannabis use in real time and may better capture its time-varying nature. The goal of this exploratory study was to examine whether momentary craving and craving variability predict subsequent use of cannabis and how baseline concentrate use status and male sex might affect these relationships.

Method:

College students residing in a state with legal recreational cannabis use who used cannabis twice a week or more completed a baseline interview and signal-contingent EMA for 2 weeks using a smartphone application. Hierarchical (multi-level) regression was used to examine time-lagged associations between craving, craving variability, and subsequent cannabis use. Baseline concentrate use and male sex were examined as moderators.

Results:

Participants (N = 109) included 59% women, with an average age of 20.2 years, and most using cannabis near-daily or daily. A main effect for craving (within-level effect) on the likelihood of cannabis use at the next EMA instance was found (odds ratio = 1.292, p < .001), although this effect was moderated by concentrate use status. For men, between-level increases in craving led to a greater likelihood of cannabis use at the next instance, but greater craving variability led to a lower likelihood of use. Greater variability in craving was associated with a greater likelihood of cannabis use among those using concentrates.

Conclusions:

The experience of craving may differ based on important participant characteristics. More research examining the fluctuating nature of craving and the role of cannabis potency on craving is warranted.

Approximately 29% of young adults report cannabis use in the past month, with 11% reporting daily use (Patrick et al., 2022). Chronic, heavy cannabis use is associated with a number of negative outcomes, such as mental health conditions and cannabis use disorder (National Academies of Sciences, Engineering & Medicine, 2017). Although men are more likely to use cannabis daily and may experience more negative outcomes, women tend to experience a greater number of and more severe withdrawal symptoms and an accelerated transition to cannabis use disorder (Copersino et al., 2010; Herrmann et al., 2015; Kerridge et al., 2018; Terry-McElrath et al., 2022).

The potency of cannabis—or the amount of tetrahydrocannabinol (THC) in cannabis products—has increased over the past decade (Bidwell et al., 2021). Cannabis concentrates (also known as dabs, shatter, wax, etc.) are known to have higher potency (upward of 90% THC) compared with flower (20% or less; Bidwell et al., 2018, 2020). The form of cannabis and potency, as well as how it is ingested, largely drive the duration and type of effects experienced (Spindle et al., 2019). Several studies report that men are more likely to use high-potency concentrates compared with women (Cuttler et al., 2016; Daniulaityte et al., 2017).

Craving—often defined as an intense urge or desire to use—is common in those using cannabis regularly and has been implicated as a mechanism of drug use maintenance (Sayette, 2016). Because craving tends to fluctuate moment to moment, intensive longitudinal methods such as ecological momentary assessment (EMA) can be useful for capturing its variation (Buckner et al., 2012; Enkema et al., 2020; Phillips et al., 2015). A recent EMA study by Enkema and colleagues (2020) found that an increase in craving above a participant's average level led to more than three times the odds of cannabis use at the next assessment point. Thus, assessing craving variability may be particularly important when trying to understand how craving functions in the moment.

To address gaps in the literature, we conducted a secondary analysis of a larger 2-week EMA study with college students using cannabis in Colorado. The goal of this exploratory analysis was to examine whether momentary craving and craving variability predict subsequent use of cannabis (time lagged) and whether baseline cannabis concentrate use and male sex might affect these relationships. Because rates of cannabis use differ by sex and preliminary evidence suggests that men use concentrates more often, we hypothesized that male sex would moderate the association between greater craving and increased cannabis use. Moderation analyses related to concentrates were exploratory. To capture the fluctuating nature of craving, we conceptualized it based on its variability across EMA reports.

Method

Participants and procedures

Participants included 109 students recruited at a mid-sized Colorado university. To be eligible to participate, young adults had to (a) be age 18 or older, (b) be enrolled at the university for one full semester before participation and not planning to graduate until the following semester or later, (c) report using cannabis within the last week, (d) report using cannabis at least twice per week, (e) have a smartphone capable of downloading the required application (app), and (f) test positive on a single panel cannabis urine toxicology dip test (.50 ng/ml cutoff; Redwood Toxicology Laboratory, Santa Rosa, CA). All procedures were approved by the university's institutional review board at the data collection site.

Participants completed an in-person, 90-minute interview covering prior and current cannabis use, followed by a series of measures. At the conclusion of the appointment, participants downloaded the app used for signal-contingent EMA (RealLife Exp, lifedatacorp.com) and completed a practice session. A research assistant discussed the questions participants would complete, as well as the duration of the EMA protocol. Participants were scheduled for a follow-up session after completion of the 14-day EMA, where they were compensated with a $50 gift card ($40 for EMA response rate < 80%).

EMA prompts began the day after the baseline session and continued for 14 days, three times per day, randomly within predefined strata (morning, afternoon, evening), with a reminder at 30 minutes. Each survey remained open for up to a 1-hour window (Phillips et al., 2014).

Measures

Baseline measures.

(A) DEMOGRAPHICS: We assessed participant sex (assigned at birth), age, and race/ethnicity.

(B) CANNABIS USE AND CONCENTRATES: Cannabis use was assessed during an interview using select items from the psychometrically validated Daily Sessions, Frequency, Age of Onset, and Quantity of Cannabis Use Inventory (DFAQ-CU; Cuttler & Spradlin, 2017). Frequency was assessed as the number of days cannabis was consumed in the past 30 days. Participants were asked to report the forms of cannabis (i.e., flower/bud, concentrates, edibles) they used primarily and secondarily (i.e., other methods used at least 25% of the time). Concentrates were described as “oil, wax, shatter, butane hash oil, or dabs.” For the purposes of the current article, “any” concentrate use was defined as use of concentrates as a primary or secondary form. Cannabis potency was assessed for flower and concentrates separately. Participants were asked to report the average THC content of the product they use, using multiple-choice responses ranging from 0 to more than 90%.

(C) RUTGERS MARIJUANA PROBLEM INDEX (RMPI): The RMPI (White et al., 2005) includes 23 items rated from 0 to 3 that assess negative consequences associated with cannabis use within the last year (Cronbach's α = .86). Total score was used.

EMA data sources.

(A) DAY OF THE WEEK: Day of the week was calculated via the smartphone app with a date and time stamp.

(B) CANNABIS USE: At each random prompt, participants were asked if they had used cannabis since the last prompt (yes/no).

(C) CANNABIS CRAVING: Participants were asked to rate their current level of cannabis craving during all prompts using the following question: “Please rate your current craving or desire to use at this exact moment on a scale of 0–10, with 0 being ‘no cravings’ and 10 being ‘extremely intense cravings’” (Phillips et al., 2015).

Statistical analyses

Descriptive statistics are reported to characterize cannabis use and other variables. An extensive power analysis was conducted a priori for the original study to establish sample size (see Supplemental Appendix A, which appears as an online-only addendum to this article on the journal's website). Because of limited information on cannabis concentrates and assessment of craving variability, the current analysis was exploratory. The primary analysis was a time-lagged model that examined whether craving (and craving variability) at one EMA point predicted cannabis use at the next instance. A random effect was included to account for repeated observations (Hilbe, 2011). A mixed logistic regression model was constructed to investigate the strength and direction of associations between craving and cannabis use at the next instance and testing for moderating effects of “any baseline concentrate use” and male sex. Participants were prompted 42 times over 14 consecutive days. A total of 2,981 responses were available, eliminating concerns about convergence issues because of sample size.

The current model included baseline RMPI and day of the week as controls. Craving included three components in the model. First, the between-level effect of craving included each individual's average craving (0–10) over the 2-week period and is representative of a typical regression term of craving. Second, the within-level effect of craving represents the effect of craving change over time for an individual. Finally, craving variability included the squared deviations between craving values and individual averages (i.e., the square of the within-level effect) and represents the effect of craving variability on the outcome. The within-level effect considers being above or below a mean, whereas the variation considers solely the changing magnitude of deviation from the mean. A component of variability was calculated for each observation and each participant at each time of measurement. Interactions were included between each craving component and both the indicator of baseline concentrate use and male status to assess moderating effects. Estimation and hypothesis testing of all effects used Maximum Likelihood with Laplace Approximation, with Wald z tests (Demidenko, 2004). All analyses were conducted using R Version 4.1.1 (http://www.R-project.org).

Results

EMA prompts and response rate

The overall EMA response rate was 79.9% with day of the week responses ranging from 75.1% to 84.5%. Participants completed the signal-contingent assessments on average 11.31 minutes (SD = 19.34; Mdn = 4) after the first signal occurred.

Background characteristics and cannabis use

Participants included 64 women (58.7%) and 45 men (41.3%), with an average age of 20.2 years (SD = 3.5). The majority of participants were White (n = 69; 63.3%), followed by 21 (19.3%) who were Latinx, 12 (11%) multiracial, and 7 (6.4%) African American. Supplemental Table A includes participant background characteristics and model variables. At baseline, 67.8% of participants reported cannabis use 5 or more days per week (M days used = 22.6, SD = 7.5). More than one fourth of participants reported that concentrates were their primary form of cannabis typically used, and an additional 30.3% reported concentrates as a secondary method (any concentrate use = 62 participants or 56.9%). More than half of participants who used concentrates reported typical potency levels above 70%. Across the 2-week period, craving ranged from 0 to 6.9 (M = 2.9, SD = 1.7). On average, those who reported any concentrate use reported slightly higher craving (M = 2.96, SD = 1.60, 95% CI [2.555, 3.367] compared with those who did not use concentrates (M = 2.72, SD = 1.73, 95% CI [2.211, 3.228]).

Does craving and craving variability predict subsequent cannabis use at the next assessment?

A time-lagged, mixed logistic regression model was applied to address associations between craving, craving variability, and subsequent likelihood of cannabis use. The model (Table 1) showed a significant, positive main effect for the association between craving within level and the likelihood of future use (odds ratio [OR] = 1.292, p < .001), indicating that if craving increased for an individual over time, there would be an expected increase in the odds of future use of about 29.2%. No other main effects were found, although the concentrate variable approached statistical significance (p = .066). There was a significant positive interaction for between-level craving by male status, suggesting that greater typical craving for men is associated with greater likelihood of future use (moderating OR = 1.336, p = .047). This gives an expected increase in the odds of future use of about 33.6% for men with greater average craving. A significant negative interaction was found for craving variability by male status (moderating OR = 0.963, p = .004), suggesting a reduction of about 3.7% in the odds of future use for men with greater craving fluctuation.

Table 1.

Mixed logistic regression model assessing predictors of the likelihood of subsequent cannabis use (yes/no)^a

Variable	Estimate	Odds ratio	p b
Craving (between level)	.043	1.044	.694
Craving (within level)	.257	1.292	<.001
Craving (variability)	-.004	0.996	.720
RMPI	.013	1.013	.388
Male	-.230	0.795	.629
Concentrate use	.866	2.377	.066
Craving (Between Level) × Male	.290	1.336	.047
Craving (Within Level) × Male	-.063	0.939	.143
Craving (Variability) × Male	-.038	0.963	.004
Craving (Between Level) × Concentrate	-.040	0.961	.778
Craving (Within Level) × Concentrate	-.107	0.898	.012
Craving (Variability) × Concentrate	.028	1.028	.035

Notes: RMPI = Rutgers Marijuana Problem Index.

^a Friday is the referent for day of week and was controlled for in the model; no days were significant.

^b p < .05 is bold.

The interaction of craving within level by concentrates was significant and negative (moderating OR = 0.898, p = .012), demonstrating that for those who used concentrates, the increased likelihood of cannabis use at the next instance (within-level effect) was reduced, as compared with those who did not use concentrates. Specifically, the odds of future use for those who used concentrates would be expected to be reduced by 10.2%. When we combined this interaction with the main craving effect, those using concentrates still demonstrated an increased likelihood of use when craving increased (based on the significant main effect) but less of an effect than participants who did not use concentrates. Lastly, there was a significant interaction for craving variability by concentrate status (moderating OR = 1.028, p = .035), indicating an increase in the odds of future use of 2.8%.

Discussion

This study examined whether measures of craving are temporally associated with cannabis use among college students over a 2-week period, while also examining possible moderation effects of male sex and concentrate use. Consistent with past EMA research (Buckner et al., 2012; Enkema et al., 2020; Phillips et al., 2015), our analyses confirm a main effect of greater within-level craving on subsequent cannabis use. This association was moderated by concentrate use. Although there was no main effect for between-level craving or craving variability and cannabis use, complex relations were found among the concentrate and sex moderators. For men, between-level increases in craving led to a 34% greater likelihood of cannabis use at the next instance.

However, unexpectedly for men, greater craving variability led to a lower likelihood of use. Past literature has shown sex differences in cannabis-related impairment and subjective effects, although the experience of craving has not been examined (Nia et al., 2018). Craving may be influenced by physiological processes that differ by sex (e.g., differences because of sex hormones or endocannabinoid brain functioning). Cue reactivity may also offer explanation. Although we cannot address this with our data, men may experience greater cannabis-related cues in their natural environment (e.g., greater exposure to peers who use cannabis) compared with women, thus contributing to a lower sensitivity for cue reactivity (Lundahl & Johanson, 2011; Prashad et al., 2020). Past work has emphasized the potential causal role of cue reactivity in the development of substance use disorder because of increased craving and substance-focused cognitions (Carter & Tiffany, 1999).

Greater variation in craving was more likely to lead to cannabis use for those who use concentrates. Experiencing more frequent disruption in craving may be a function of potency. Concentrates have been shown to produce higher THC blood levels compared with other cannabis products (Bidwell et al., 2020). Concentrate blood levels drop precipitously within an hour after use, thus potentially amplifying the impact of craving variability on cannabis use (Drennan et al., 2020). Future EMA, naturalistic, and/or lab-based studies that assess craving, THC blood levels, and the type of cannabis used in real time could help answer questions related to the experience of higher potency products. Because early work in this area suggests that use of high-potency products may be a marker for greater cannabis dependence, future studies should explore whether concentrates are more problematic for certain populations, such as those with lower levels of tolerance (Arterberry et al., 2019; Meier, 2017; Steeger et al., 2021).

Limitations and implications

The current study includes a number of limitations. We did not assess concentrate use in the moment, only at baseline. Future EMA research should consider assessing the form of cannabis used, method of ingestion, quantity, and potency in the moment. This information could prove useful to understanding dose-dependent effects. Participants included college students who were using cannabis more heavily, thus limiting generalizability. Young adults in the community may experience different cannabis-related cues.

Because craving can prompt cannabis use, assessing craving in real time may be particularly beneficial for intervention. Although it remains untested for singular cannabis use, behavioral and mindfulness interventions targeting alcohol and tobacco craving have shown some success (e.g., Dulin et al., 2017; Elwafi et al., 2013). Because of its fluctuating nature, a just-in-time-adaptive intervention offers an opportunity to target momentary craving and deliver evidence-based coping strategies when they’re needed (Nahum-Shani et al., 2018).

Conclusions

Craving is demanding and disruptive, vacillates over the course of a day, and prompts substance use; as such, this has led to its inclusion as a symptom of substance use disorder (American Psychiatric Association, 2013; Enkema et al., 2020). Our findings are consistent with existing EMA studies that have found associations between craving and cannabis use, while also offering a nuanced examination of moderators that might affect this relationship. Consistent with other work (Cleveland et al., 2021), our research provides justification for the measurement of craving variability. Future studies should continue to assess the influence of a wide range of cannabis-related variables that may affect the experience of craving in order to best intervene with those hoping to stop or reduce their cannabis use.