Abstract
Background:
Measuring marijuana use quantity in survey research is complicated due to wide variation in the types (e.g., flower, edibles) and potency of marijuana products and in the modes (e.g., smoking, dabbing) used to consume products. There is currently no gold standard marijuana use quantity measure for survey research. This study examined whether number of hours high can be used as a proxy for marijuana use quantity in survey research, particularly in intensive longitudinal designs.
Methods:
Participants came from a community sample of young adults participating in a longitudinal study on simultaneous alcohol and marijuana use that used a longitudinal measurement-burst design in which participants completed surveys on up to 14 consecutive days in up to five bursts across nearly two calendar years. Those who reported using marijuana on at least one sampled day were included in present analyses (N = 379; Mage = 21.6; 50.7 % female). Hypotheses were tested using Poisson multilevel models and a logistic regression.
Results:
Within persons, mode-specific marijuana use quantity variables predicted same-day number of hours high indicating evidence of initial criterion validity. In turn, hours high predicted same-day negative marijuana-related consequences indicating evidence of proximal predictive validity. Between persons, participants’ average number of hours high was positively associated with their odds of possible cannabis use disorder following the last burst demonstrating distal predictive validity.
Conclusions:
Number of hours high may be a parsimonious proxy for measuring marijuana use quantity (regardless of mode of use) in survey research, particularly in intensive longitudinal designs.
Keywords: Marijuana, Cannabis, Measurement of marijuana use, Young adults, Intensive longitudinal data
1. Introduction
With recent changes in the legalization of recreational marijuana in the United States and historical increases in its use among young adults (Schulenberg et al., 2021), being able to measure how much marijuana young adults are using is more important than ever. However, measuring marijuana use quantity remains a complex and challenging task. Whereas survey researchers have developed standardized methods of measuring alcohol use quantity via self-report (i.e., a standard drink, defined as 12 ounces of beer, 5 ounces of wine, or 1.5 ounces of distilled spirits; NIAAA, n.d.), there is currently no standard method for measuring marijuana use quantity via self-report (Lorenzetti et al., 2021). Many researchers studying adolescent and young adult marijuana use have opted for using measures of marijuana use frequency, but not quantity (Lorenzetti et al., 2021; Prince et al., 2018). However, marijuana use quantity is likely more closely related to impairment and other acute and longer-term consequences (Callaghan et al., 2020; Zeisser et al., 2012).
1.1. Why is measuring marijuana use quantity challenging?
Measuring marijuana use quantity is complicated by several factors. First, there are different types of marijuana products, and their properties vary. Products refer to the actual substance or substance-containing material people use. Examples include marijuana flower, vaping liquids, and edibles (i.e., marijuana-infused food products). Although the primary psychoactive ingredient of interest (i.e., THC, or Δ-9-tetrahydrocannabinol) is contained in all marijuana products, properties such as potency can vary substantially between and within product types (Chandra et al., 2019; Smart et al., 2017). Second, there are several different routes of administration, or general physiological processes through which THC enters the bloodstream (Ehrler et al., 2015; Russell et al., 2018). Examples include inhalation through the lungs and absorption through the digestive system. The speed, duration, and level of intoxication varies across routes of administration (Spindle et al., 2018; Vandrey et al., 2017). Third, there are numerous modes of marijuana use, or the actual devices or techniques used to consume marijuana products (e.g., Streck et al., 2019). For instance, marijuana flower can be smoked using joints, blunts, pipes, or bongs, among other modes. Other examples of modes include vaping and dabbing (i.e., inhaling vapor from waxy cannabis preparations consisting of extremely concentrated THC). Modes can vary in the average amount of marijuana product used and THC consumed per hit as well as in the resulting levels of intoxication (Cloutier et al., 2022). This is further complicated by new products and modes continually being produced and made available. Lastly, marijuana use may involve use of multiple products and modes, sometimes shared with others, making it difficult to easily compare use quantity between people or within people across days or occasions. Furthermore, assessing products and modes for each use occasion can be time consuming and burdensome for participants.
Researchers who have measured marijuana use quantity have tried operationalizing this construct in various ways, including number of hits (e.g., Cloutier et al., 2022; Lankenau et al., 2019; Linden-Carmichael et al., 2020), number of joints used (Zeisser et al., 2012), and grams (Prince et al., 2018). However, these operationalizations have clear limitations. The size of an average hit varies across modes, and there may be between- and within-person variability in the size of hits within modes. Also, some modes do not involve taking hits (e.g., edible use). Similarly, number of joints is overly specific to one type of mode and relies on the assumption that joints are a standardized size. Grams is perhaps the most standardized operationalization, but some modes do not typically involve the use of marijuana flower (e.g., vaping, edibles), in which case the amount of product used cannot be easily measured in grams. Further, some research suggests participants have limited ability to accurately report their quantity of use in grams (Prince et al., 2018). Ultimately, none of the previously used operationalizations are satisfactory, and marijuana measurement challenges are compounded by use of multiple products and modes within the same reference period.
As a result of this complexity, there is currently no widely accepted, standardized method for measuring marijuana use quantity (Lorenzetti et al., 2021; Prince et al., 2018), though some researchers have proposed standardized methods, such as standardized THC units (e.g., Freeman and Lorenzetti, 2021). In fact, the National Institute on Drug Abuse recently established standardized THC units for applicable research it funds (Volkow, 2021). Although using THC units may have utility for lab-based research (e.g., THC administration protocols), it is not practical in survey research, particularly research using intensive longitudinal designs. Since marijuana products are not currently packaged in standardized amounts based on THC units, many survey items would still be needed to accurately quantify use across different modes and products.
1.2. Could hours high simplify measuring marijuana use quantity?
In the absence of concise standardized marijuana quantity measures, number of hours high may be a reasonable proxy in survey research, particularly research using intensive longitudinal designs. Theoretically, given the presumed dose-response relationship between marijuana use and intoxication, use of greater quantities should result in being intoxicated (or high) for longer periods of time (McCrady and Epstein, 2013). There are two primary potential benefits of using hours high as a proxy for marijuana use quantity. First, hours high may capture marijuana use quantity across various modes of use in a given time period. Rather than asking about specific product types, the potency of those products, and the modes through which those products were used, hours high can be assessed with a single item for each reference period (e.g., day). This concise proxy measure would likely reduce participant burden in intensive longitudinal designs, especially if marijuana use is not the primary focus of a study. An hours high item would also allow researchers to account for marijuana use quantity in analyses more easily. Second, variance due to other aspects of marijuana use that affect intoxication (e.g., product potency) are presumably captured by hours high. For instance, number of hours high should be greater on days when individuals use marijuana with greater THC concentrations, when holding all other factors constant.
1.3. Current study
The purpose of the present study was to assess whether daily measurements of hours high were a reasonable proxy for daily-level marijuana use quantity in a longitudinal measurement-burst design study of young adults who engage in simultaneous alcohol and marijuana (SAM) use. In Aim 1, we tested for initial criterion validity by assessing whether daily-level, mode-specific marijuana quantity measures (i.e., grams for smoking and vaping, hits for dabbing, milligrams of THC for edible use) predicted same-day number of hours high. We hypothesized that each mode-specific quantity variable would be positively associated with hours high within persons. In Aim 2, we tested for proximal predictive validity by examining whether hours high on marijuana use days predicted same-day number of negative marijuana consequences. We hypothesized that hours high would be positively associated with negative consequences within persons. Third, we tested distal predictive validity by assessing whether participants’ average number of hours high on marijuana use days throughout the study period predicted their likelihood of possible cannabis use disorder (CUD) following the completion of all daily surveys. We hypothesized that young adults who reported a higher average number of hours high would be at greater risk for CUD.
2. Materials and methods
2.1. Participants and procedures
Data came from a longitudinal study of health behaviors and experiences in 409 young adults originally recruited in the greater Seattle metropolitan area in Washington State (Lee et al., 2020; Patrick et al., 2020). In a longitudinal measurement-burst design, participants completed daily web-based surveys for up to 14 consecutive days in each of up to six bursts across two years in addition to a baseline survey and three yearly follow-up surveys. The onset of the COVID-19 pandemic occurred during Burst 6. Since sampled days in Burst 6 occurred prior to the pandemic for some participants, but not others, daily surveys from Burst 6 were excluded from analyses. Therefore, analyses were limited to the 379 participants who used marijuana on at least one sampled day in Bursts 1–5 (December 2017 to January 2020). The study was approved by the institutional review board of the first author’s university.
Participants were recruited using a multimethod sampling approach that included advertising online, in print, and on social media; outreach at community colleges and local events and to community agencies involved with young adults; flyers; and friend referral. Eligible participants were 18–25 years old, reported alcohol use three or more times in the past month, reported past-month SAM use, and lived within 60 miles of the study office. Additional details of the study and procedures are described elsewhere (Lee at al, 2020; Patrick et al., 2020). Across Bursts 1–5, participants completed surveys on 88.4 % of sampled days and on an average of 12.37 (SD = 3.55) out of 14 days per burst. Retention of participants across bursts was high with 89.0 % responding to daily surveys in each of the first five bursts.
In the analytic sample, 47.8 % of participants identified as White Non-Hispanic/Latinx (NHL), 20.3 % Other NHL (e.g., Black or African American, more than one race), 16.9 % Hispanic/Latinx, and 15.0 % Asian NHL. Approximately half (50.7 %) of participants reported their biological sex as female. At baseline, the average age of participants was 21.63 years (SD = 2.16), 61.5 % were enrolled in post-secondary education, and 26.4 % were employed full-time.
2.2. Daily-level measures
2.2.1. Number of hours high
Each day, participants were asked “Did you use marijuana yesterday?” On days participants reported using marijuana, they were asked “How many total hours were you high yesterday?” Response options ranged from “Less than 1 h” (0) to “23–24 h” (23).
2.2.2. Marijuana use modes
When participants reported using marijuana, they were also asked “How did you use marijuana?” and were able to select from a total of five different modes in a check-all-that-apply format. Available response options were “smoking (e.g., joint, bong, pipe, blunt, hookah),” “edibles (e.g., brownies, cakes, cookies, candy),” “vaping with an electronic device (e.g., Volcano, vape pen, or e-cig),” “dabbing (e.g., oil, wax, shatter, butane hash oil, dabs),” and “some other way.”
2.2.3. Mode-specific marijuana quantity
Grams: When participants reported smoking or vaping marijuana, they were asked “When you smoked or vaped yesterday, how many grams of marijuana did you personally use?” Participants were instructed not to include other forms of marijuana they may have used, such as concentrates or edibles. In a training session prior to enrollment, participants were shown pictures of various amounts of marijuana flower to help calibrate their self-reports. Response options ranged from “Up to 1/8 of a gram” to “More than 28 g (1 ounce).” Responses were recoded into numeric values representing the number of grams at the midpoint of each category range (e.g., 1.5 g for “More than 1–2 g”). Thus, the scale of the recoded variable was grams. Hits: When participants reported dabbing, they were asked “How many hits of marijuana concentrates did you personally take yesterday?” with response options consisting of integers ranging from 1 to 100. Milligrams: When participants reported consuming edibles, they were asked “How many milligrams of THC did you personally ingest yesterday?” as quantity of edible use is often measured in milligrams of THC (e.g., Lorenzetti et al., 2021). Response options consisted of integers ranging from 1 to 100. Milligrams of THC was divided by 5 so that each one-unit change represented 5 milligrams of THC, which is the standardized unit some have proposed and NIH has adopted (Freeman and Lorenzetti, 2021; Volkow, 2021).
2.2.4. Mode-specific marijuana potency
When participants smoked or vaped marijuana, they were asked “What was the average THC content of the marijuana used yesterday?” There were seven response options ranging from “Up to 4 %” to “30 % or greater.” On days participants reported dabbing, they were asked “What is the THC content of the concentrates you used yesterday?” with ten response options ranging from “Up to 9 %” to “90 % or greater.” For both items, values were recoded to reflect the THC percentage at the midpoint of each response option (e.g., 22 % for “20–24 %” THC content of marijuana smoked or vaped).
2.2.5. Negative marijuana consequences
When marijuana use was reported, participants were asked “Did any of the following happen to you as a result of your marijuana use yesterday?” Participants were presented with a list of 10 negative marijuana consequences (e.g., had low motivation, felt anxious or worried, felt dizzy) and selected which they experienced. Consequence items were adapted from multiple sources, including a validated daily alcohol consequences measure (Lee et al., 2017) and a validated past-month marijuana consequences measure (Lee et al., 2021). Items were selected based on relevance to marijuana and their appropriateness for daily assessment. Sum scores were created indicating the number of negative marijuana consequences experienced each day. Past work using this measure has shown that number of negative marijuana consequences was positively associated with coping and conformity general substance use motives at the daily and person levels and with social motives at the person level (Patrick et al., 2020).
2.2.6. Number of drinks
Each day, participants were asked “Did you drink alcohol yesterday?” On days participants reported drinking alcohol, they were asked “How many total drinks did you have yesterday?” Response options ranged from “1 drink” (1) to “25 or more drinks” (25). On days participants reported not drinking, number of drinks was recoded as 0.
2.2.7. Any tobacco/nicotine use
Each day, participants were asked if they had used tobacco/nicotine the previous day. Days they reported not using tobacco/nicotine were coded as 0, and days they reported using tobacco/nicotine were coded as 1.
2.3. Person-level measures
2.3.1. Cannabis use disorder identification test – revised (CUDIT-R)
Participants completed the CUDIT-R (Adamson et al., 2010), regarding their marijuana use in the past six months, at the Year 2 follow-up after the sixth and final burst. This measure was completed during the COVID-19 pandemic for some, but not all, participants. The CUDIT-R contains nine items with varying response options. Responses for the last eight items were summed, with scores of 13 or more indicating possible CUD. A binary indicator variable was created to represent whether participants had scores at or above 13, thereby indicating possible CUD.
2.4. Data analysis
To test initial criterion validity (Aim 1), Poisson multilevel models were used to assess whether daily-level, mode-specific marijuana quantity variables (i.e., grams on smoking and vaping days, hits on dabbing days, and milligrams of THC on edible use days) predicted number of hours high that day. Since marijuana quantity variables were mode-specific, three separate models were estimated that were limited to smoking and vaping, dabbing, and edible use days, respectively. To test proximal predictive validity (Aim 2), one Poisson multilevel model was used to examine whether hours high predicted number of same-day negative marijuana consequences across all marijuana use days. All multilevel models were estimated using maximum likelihood estimation based on the Laplace approximation in the glmmTMB package (Brooks et al., 2017) in R 4.1.2 (R Core Team, 2021). A daily-level random effect was included to account for overdispersion (Harrison, 2014). To fully disentangle within- and between-person associations at Levels 1 and 2, respectively, burst number (coded as 0–4) and day number within burst (coded as 0–13) variables were included at Level 1 to account for any potential trends over time in predictor or outcome variables and all other Level 1 variables were person-mean-centered (Hamaker and Muthén, 2020; Wang and Maxwell, 2015). All Level 2 variables were grand-mean-centered. To examine distal predictive validity (Aim 3), the person-level association between participants’ average number of hours high on marijuana use days and their likelihood of possible CUD after the last measurement burst was tested using a logistic regression estimated in R and with all variables grand-mean-centered. All models controlled for sex, age, and race/ethnicity, as reported at baseline.
3. Results
3.1. Descriptive statistics
The analytic sample consisted of 8410 marijuana use days (29.5 % of the 28,486 sampled days in Bursts 1–5) nested within 379 individuals who reported marijuana use on at least one day (92.7 % of the 409 participants). Descriptive statistics are shown in Table 1. The number of hours participants reported being high on marijuana use days ranged from 0 to 21 (M=3.14, SD=2.45). Participants smoked or vaped marijuana on 87.6 % of use days, dabbed on 15.0 %, and consumed edibles on 10.0 %. Multiple modes were endorsed on 21.1 % of use days (M=1.24, SD=0.51). Negative marijuana consequences were reported on 31.4 % of use days. The average number of consequences reported per day was 1.21 (SD=1.53, range=0–10). The average CUDIT-R score was 9.67 (SD=6.82), and 28.5 % of participants had scores at or above 13, indicating possible CUD.
Table 1.
Descriptive Statistics.
| Day-Level (N = 8410) | n (%) | M (SD) | Min. | Max. |
|---|---|---|---|---|
| Number of hours high | – | 3.14 (2.45) | 0 | 21 |
| Marijuana use mode (Check all that apply) | ||||
| Smoking and/or vaping | 7365 (87.6) | – | 0 | 1 |
| Smoking only | 6041 (71.8) | – | 0 | 1 |
| Vaping only | 2307 (27.4) | – | 0 | 1 |
| Dabbing | 1260 (15.0) | – | 0 | 1 |
| Edibles | 838 (10.0) | – | 0 | 1 |
| Number of modes | – | 1.24 (0.51) | 0 | 4 |
| Mode-specific marijuana use quantity | ||||
| Grams (smoking and/or vaping days) | – | 0.54 (1.12) | 0.06 | 42 |
| Hits of concentrates (dabbing days) | – | 5.22 (4.51) | 1 | 40 |
| Milligrams of THC (edible use days) | 18.27 (21.95) | 1 | 100 | |
| Marijuana potency (% THC) | ||||
| Smoking and/or vaping days | – | 21.8 (6.65) | 2 | 32 |
| Dabbing days | – | 73.3 (18.10) | 4.5 | 95 |
| Negative marijuana consequences | – | 1.21 (1.53) | 0 | 10 |
| Number of drinks | 1.52 (2.55) | 0 | 25 | |
| Days with any tobacco/nicotine use | 2325 (27.7) | – | 0 | 1 |
| Person-Level (N = 379) | N (%) | M (SD) | ||
| CUDIT-R score at Year 2 | – | 9.67 (6.82) | 0 | 30 |
| Possible CUD (CUDIT-R ≥ 13) at Year 2 | 108 (28.5) | – | 0 | 1 |
| Male sex | 189 (49.2) | 0 | 1 | |
| Age at baseline | – | 21.63 (2.16) | 18 | 26 |
| Race/ethnicity | ||||
| Asian NHL | 57 (15.0) | – | 0 | 1 |
| Hispanic/Latinx | 64 (16.9) | – | 0 | 1 |
| Other NHL | 77 (20.3) | – | 0 | 1 |
| White NHL | 181 (47.8) | – | 0 | 1 |
| Four-year college student at baseline | 182 (48.0) | – | 0 | 1 |
| Employed part-time at baseline | 149 (39.3) | – | 0 | 1 |
| Employed full-time at baseline | 100 (26.4) | – | 0 | 1 |
Note. THC = Δ-9-tetrahydrocannabinol; CUDIT-R = Cannabis Use Disorder Identification Test – Revised (Adamson et al., 2010); CUD = Cannabis Use Disorder; NHL = Non-Hispanic/Latinx.
Intraclass correlation coefficients (ICCs) for daily-level variables included in the models are shown in Supplemental Table A. The ICC of 0.41 for hours high indicates that daily-level reports of hours high varied more within-person across days than between-person (i.e., 59 % of the variability was within-person). Within- and between-person correlations between hours high and mode-specific marijuana quantity and potency variables are shown in Supplemental Table B. Within-person correlations between hours high and the three mode-specific quantity variables were relatively small (i.e., rs ranged 0.23–0.27, all p-values <0.05). However, it is important to note that hours high should capture the amount of marijuana used across all modes, whereas each specific quantity variable (e.g., grams on smoking and/or vaping days) is only representative of amounts used by a single mode. Further, hours high accounts for variation in potency, whereas the mode-specific quantity variables do not. This demonstrates the challenges of accurately measuring marijuana use quantity in intensive longitudinal designs and highlights the value of using hours high as a proxy for daily-level marijuana use due to its simplicity and comprehensive consideration of both quantity and potency (which are often not accurately estimated by young adults).
3.2. Aim 1: Did marijuana use quantity predict hours high?
Results of Poisson multilevel models testing whether marijuana use quantity predicted hours high are presented separately by mode in Table 2.
Table 2.
Poisson Multilevel Models Testing Whether Mode-Specific Marijuana Use Quantity Predicted Number of Hours High.
| Outcome: Number of Hours High | ||||||
|---|---|---|---|---|---|---|
|
| ||||||
| Model 1: Smoking/Vaping Days | Model 2: Dabbing Days | Model 3: Edible Days | ||||
|
NDays = 7102 NPersons = 363 |
NDays = 1233 NPersons = 143 |
NDays = 790 NPersons = 194 |
||||
| Fixed Effects | RR | 95 % CI | RR | 95 % CI | RR | 95 % CI |
| Level 2: Person Level | ||||||
| Intercept | 2.16 *** | 2.04, 2.30 | 3.32 *** | 2.99, 3.68 | 3.14 *** | 2.80, 3.52 |
| Mode-specific quantity | ||||||
| Mean grams | 1.20 *** | 1.09, 1.31 | – | – | – | – |
| Mean hits of concentrates | – | – | 1.01 | 0.99, 1.03 | – | – |
| Milligrams THC | – | – | – | – | 1.04 *** | 1.02, 1.06 |
| Mean potency (% THC) | 1.03 *** | 1.02, 1.04 | 1.00 | 1.00, 1.01 | – | – |
| Mean no. of drinks on marijuana use days | 0.96 * | 0.93, 0.99 | 1.01 | 0.97, 1.05 | 1.03 | 1.00, 1.06 |
| Proportion of marijuana use days tobacco/nicotine was used | 1.04 | 0.89, 1.21 | 0.91 | 0.73, 1.14 | 1.07 | 0.85, 1.34 |
| Male sex | 1.19 ** | 1.07, 1.33 | 1.26 * | 1.05, 1.52 | 1.09 | 0.94, 1.27 |
| Age | 1.00 | 0.97, 1.02 | 1.01 | 0.97, 1.05 | 1.01 | 0.97, 1.04 |
| Race/ethnicity (Ref.: White NH) | ||||||
| Asian NH | 0.84 * | 0.72, 0.99 | 0.65 ** | 0.48, 0.89 | 0.88 | 0.71, 1.09 |
| Other NH | 1.02 | 0.89, 1.17 | 1.19 | 0.95, 1.48 | 1.08 | 0.89, 1.30 |
| Hispanic | 0.94 | 0.81, 1.08 | 0.92 | 0.74, 1.15 | 0.90 | 0.73, 1.12 |
| Level 1: Day Level | ||||||
| Mode-specific quantity | ||||||
| Grams | 1.61 *** | 1.48, 1.74 | – | – | – | – |
| Hits of concentrates | – | – | 1.07 *** | 1.04, 1.09 | – | – |
| Milligrams THC | – | – | – | – | 1.06 *** | 1.03, 1.09 |
| Potency (% THC) | 1.01 *** | 1.01, 1.01 | 1.00 ** | 1.00, 1.01 | – | – |
| Used other modes | 1.24 *** | 1.19, 1.30 | 1.36 *** | 1.25, 1.47 | 1.36 *** | 1.20, 1.55 |
| Number of drinks | 1.00 | 1.00, 1.01 | 1.00 | 0.99, 1.01 | 1.00 | 0.98, 1.03 |
| Any tobacco/nicotine use | 1.03 | 0.98, 1.08 | 0.90 | 0.80, 1.02 | 1.02 | 0.81, 1.30 |
| Weekend | 1.10 *** | 1.07, 1.13 | 1.03 | 0.97, 1.09 | 1.04 | 0.95, 1.14 |
| Burst number | 1.02 ** | 1.01, 1.03 | 1.03 * | 1.01, 1.05 | 1.06 *** | 1.03, 1.10 |
| Day number within burst | 1.00 ** | 0.99, 1.00 | 0.99 * | 0.98, 1.00 | 0.99 | 0.98, 1.00 |
Note. RR = Rate ratio; THC = Δ-9-tetrahydrocannabinol; NHL = Non-Hispanic/Latinx. Weekend: 0 = Monday-Friday, 1 = Saturday-Sunday.
p < .05,
p < .01,
p < .001.
3.2.1. Smoking/Vaping days
At the daily level, the number of grams used was positively associated with hours high such that each gram of marijuana used was associated with being high for 61 % more hours, on average (Model 1). Marijuana potency and use of other modes were also positively associated with hours high at the daily level. At the person level, the average number of grams used on smoking/vaping days was positively associated with the average number of hours high. Average potency was also positively associated with the average number of hours high on smoking/vaping days.
3.2.2. Dabbing days
At the daily level, hits of marijuana concentrates was positively associated with hours high such that each hit was associated with being high for 7 % more hours, on average (Model 2). Marijuana potency and use of other modes were also positively associated with hours high at the daily level. At the person level, neither the average number of hits of marijuana concentrates taken on dabbing days nor average potency was associated with the average number of hours high.
3.2.3. Edible days
At the daily level, milligrams of THC was positively associated with hours high such that each additional 5 milligrams of THC consumed was associated with being high for 6 % more hours, on average (Model 3). Use of other modes was also positively associated with hours high at the daily level. At the person level, participants who consumed a greater average number of milligrams of THC via edibles tended to be high for more hours on edible use days.
3.3. Aim 2: Did hours high predict negative marijuana consequences?
Results of a Poisson multilevel model testing whether hours high predicted negative marijuana consequences across all marijuana use days is presented in Table 3. At the daily level, hours high was positively associated with the number of negative marijuana consequences such that each additional hour high was associated with experiencing 14 % more negative consequences, on average (Model 4). At the person level, the average number of hours high on marijuana use days was not associated with the average number of negative marijuana consequences experienced. Sensitivity analyses in which the association between hours high and negative marijuana consequences was tested separately for days participants smoked and/or vaped, dabbed, and used edibles (as was done for Aim 1) produced similar findings (Supplemental Table C). That is, these stratified models also showed that hours high was positively associated with negative marijuana consequences at the daily level, but not at the person level. A second set of supplemental analyses (Supplemental Table D) showed that the mode-specific marijuana quantity variables used in Aim 1 were also positively associated with negative marijuana consequences at the daily level.
Table 3.
Poisson Multilevel Model Testing Whether Number of Hours High Predicted Number of Negative Marijuana Consequences.
| Model 4 Outcome: Number of Negative Marijuana Consequences |
||
|---|---|---|
|
| ||
| Fixed Effects | Rate Ratio | 95 % CI |
| Level 2: Person Level | ||
| Intercept | 1.10 | 1.00, 1.21 |
| Mean number of hours high | 0.99 | 0.93, 1.06 |
| Mean no. of drinks on marijuana use days | 0.99 | 0.93, 1.04 |
| Proportion of marijuana use days tobacco/nicotine was used | 0.92 | 0.72, 1.17 |
| Male sex | 0.84 | 0.71, 1.01 |
| Age | 0.99 | 0.95, 1.03 |
| Race/ethnicity (Ref.: White NHL) | ||
| Asian NHL | 1.24 | 0.96, 1.59 |
| Other NHL | 0.82 | 0.66, 1.03 |
| Hispanic/Latinx | 0.90 | 0.71, 1.14 |
| Level 1: Day Level | ||
| Number of hours high | 1.14 * ** | 1.12, 1.17 |
| Number of drinks | 0.97 * ** | 0.96, 0.98 |
| Any tobacco/nicotine use | 1.09 | 1.00, 1.19 |
| Weekend | 0.94 * | 0.90, 0.99 |
| Burst number | 0.97 *** | 0.95, 0.99 |
| Day number within burst | 0.98 *** | 0.97, 0.98 |
Note. NDays = 8218, NPersons = 379. NHL = Non-Hispanic/Latinx. Weekend: 0 = Monday-Friday, 1 = Saturday-Sunday.
p < .05,
p < .01,
p < .001
3.4. Aim 3: Did average number of hours high predict odds of possible CUD?
Results of a logistic regression testing whether participants’ average number of hours high on marijuana use days across the five bursts predicted their likelihood of possible CUD are presented in Table 4. The average number of hours high was positively associated with the likelihood of possible CUD such that each additional average hour high was associated with 1.84 times greater odds of possible CUD (Model 5).
Table 4.
Logistic Regression Testing Whether Participants’ Average Number of Hours High Predicted Symptoms of CUD at Two Year Follow-up.
| Model 5 Outcome: CUDIT-R Scores (0 = 0–12, 1 = 13 +) |
||
|---|---|---|
|
| ||
| Predictors | Odds Ratio | 95 % CI |
| Intercept | 0.40 *** | 0.31, 0.52 |
| Average number of hours high | 1.84 *** | 1.51, 2.25 |
| Average number of drinks | 0.93 | 0.73, 1.20 |
| Proportion of days tobacco/nicotine was used | 1.99 | 0.90, 4.39 |
| Male sex | 1.14 | 0.68, 1.90 |
| Age | 0.98 | 0.87, 1.11 |
| Race/ethnicity (Ref.: White NHL) | ||
| Asian NHL | 0.69 | 0.32, 1.48 |
| Other NHL | 0.80 | 0.41, 1.57 |
| Hispanic/Latinx | 1.09 | 0.55, 2.18 |
Note. N = 354 participants. CUDIT-R = Cannabis Use Disorder Identification Test – Revised (Adamson et al., 2010); NHL = Non-Hispanic/Latinx.
p < .05,
p < .01,
p < .001.
4. Discussion
This paper examined whether number of hours high can be used as a proxy for marijuana use quantity in survey research. Findings demonstrated initial criterion validity of hours high as a proxy for marijuana use quantity as evidenced by positive daily-level associations between mode-specific marijuana use quantity and number of hours high. Findings also provided evidence of proximal predictive validity as hours high was positively associated with the number of same-day negative marijuana consequences. Lastly, findings provided evidence of distal predictive validity as participants’ average number of hours high was positively associated with their likelihood of possible CUD at the end of the study period. These findings provide initial evidence that hours high may be a reasonable and parsimonious proxy for marijuana use quantity in survey research.
Measuring marijuana use quantity is complicated by variation in the use of marijuana products and modes between people and within people across days or occasions. Comprehensively assessing marijuana use quantity with product- and mode-specific measures is burdensome for participants, especially those using a larger variety of products or modes, and can use a large amount of total survey time, something that is a premium in intensive longitudinal designs. For instance, in our study, eight marijuana quantity items were asked depending on which modes were reported each day. Further, comparing across different metrics and across days different modes were used is not straightforward. In contrast, hours high was assessed using a single item, is generally applicable across modes, and can be easily included into statistical models. Lorenzetti et al. (2021) recommend biological measurements of THC concentration as a gold standard for accurately measuring marijuana use quantity under certain conditions. However, in addition to being invasive and costly, such techniques are not well suited for survey research. Therefore, given the theoretical dose-response relationship between marijuana use quantity and duration of intoxication (McCrady and Epstein, 2013) and the evidence presented here, hours high may be a useful way for researchers to assess marijuana use quantity in survey research.
The associations of hours high with THC concentration and use of other modes highlight the additional utility of this measure. Although models testing associations between mode-specific marijuana use quantity and hours high did not account for amounts of marijuana used via other modes, positive associations between the indicator representing use of other modes and hours high suggests that the hours high variable captures these additional amounts of marijuana to some degree. This supports the notion that hours high is applicable across modes. Similarly, even within modes, participants may have used multiple marijuana products that differed in potency across days. Positive daily-level associations between potency and hours high suggest the hours high measure also captures some within-person variability related to potency. Thus, the use of an hours high measure may also lessen the need to account for within-person sources of variation in use, such as use of multiple modes or potency.
4.1. Strengths and limitations
This study had several strengths. First, the longitudinal measurement-burst design allowed for a modeling approach that tested both within- and between-person associations. Most research assessing the psychometric properties and utility of marijuana use quantity measures have not used intensive longitudinal data and have been limited to between-person analyses. Second, stratifying the analytic sample by mode allowed for three different marijuana use quantity variables to be compared with hours high. The general consistency of findings across different types of marijuana use days strengthens the validity of the findings. Third, analyses testing the association between marijuana use quantity and hours high controlled for THC concentration and use of other modes, which are often unaccounted for in intensive longitudinal designs.
There were also several limitations. First, analyses did not account for potential tolerance effects. When using the same amount of marijuana on a given day, individuals who use marijuana in greater quantities, on average, may become less high (or be high for fewer hours) than those who typically use in lesser quantities. However, the daily-level associations tested here were within-person associations that controlled for stable, between-person differences (Hox et al., 2017). Thus, to the extent that tolerance levels were fairly stable across the study period, tolerance should not have largely affected the within-person associations found here. Second, the daily-level negative consequences measure used here had not been previously validated; however, the models in Supplemental Table D showing positive daily-level associations between mode-specific quantity and potency variables and negative consequences provides initial evidence of the predictive validity of this measure. Third, the sample was selected for past-month SAM use. This was a higher-risk sample than the general young adult population as indicated by high levels of cannabis use disorder. Fifth, the CUDIT-R was asked following the completion of the final measurement burst. For some participants, but not others, this occurred during the early stages of the COVID-19 pandemic.
5. Conclusions
This paper provides evidence supporting the use of number of hours high as a proxy for daily-level marijuana use quantity in survey research using a daily design. Initial findings suggest that assessing hours high could provide researchers with a clean, simple method of measuring marijuana use quantity that is generally applicable across modes of use and is less burdensome for participants to complete and less cumbersome for researchers to use in analyses. The general approach of prioritizing items about being high over actual quantity of use items may be as or more helpful in ecological momentary assessment designs, though further research is needed to provide evidence of this assertion and to determine whether a related item (e.g., “how high are you right now?”) may be more appropriate than number of hours high for assessing within-day variability in marijuana use. Future work that more rigorously evaluates the validity of the hours high measure is warranted.
Supplementary Material
Funding source
Funding for this study was provided by NIAAA Grant R01AA025037. NIAAA did have any role in the study design, collection, analysis or interpretation of the data, writing the manuscript, or the decision to submit the paper for publication.
This research was supported by a grant from the National Institute on Alcohol Abuse and Alcoholism (R01AA025037, MPIs Lee and Patrick). The content of this manuscript is solely the responsibility of the author(s) and does not necessarily represent the official views of the National Institute on Alcohol Abuse and Alcoholism or the National Institutes of Health. The authors have no conflicts of interest to report.
Footnotes
CRediT authorship contribution statement
Dr. Calhoun performed the statistical analyses and wrote the first draft of the manuscript. Drs. Patrick and Lee designed the study, wrote the protocol, curated the data, helped conceptualize the paper idea and research questions, and edited all sections of the paper. Drs. Fairlie, Graupensperger, and Walukevich-Dienst helped plan analyses and helped edit all sections of the paper. All authors contributed to and have approved the final manuscript.
Conflict of Interest
All authors declare that they have no conflicts of interest.
Appendix A. Supporting information
Supplementary data associated with this article can be found in the online version at doi:10.1016/j.drugalcdep.2022.109628.
References
- Adamson SJ, Kay-Lambkin FJ, Baker AL, Lewin TJ, Thornton L, Kelly BJ, Sellman JD, 2010. An improved brief measure of cannabis misuse: the Cannabis Use Disorders Identification Test-Revised (CUDIT-R). Drug Alcohol Depend. 110, 137–143. 10.1016/j.drugalcdep.2010.02.017. [DOI] [PubMed] [Google Scholar]
- Brooks ME, Kristensen K, van Benthem KJ, Magnusson A, Berg CW, Nielsen A, Skaug HJ, Machler M, Bolker BM, 2017. glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. R. J 9, 378–400. 10.3929/ethz-b-000240890. [DOI] [Google Scholar]
- Callaghan RC, Sanches M, Kish SJ, 2020. Quantity and frequency of cannabis use in relation to cannabis-use disorder and cannabis-related problems. Drug Alcohol Depend. 217, 108271 10.1016/j.drugalcdep.2020.108271. [DOI] [PubMed] [Google Scholar]
- Chandra S, Radwan MM, Majumdar CG, Church JC, Freeman TP, ElSohly MA, 2019. New trends in cannabis potency in USA and Europe during the last decade (2008–2017). Eur. Arch. Psychiatry Clin. Neurosci 269, 5–15. 10.1007/s00406-019-00983-5. [DOI] [PubMed] [Google Scholar]
- Cloutier RM, Calhoun BH, Linden-Carmichael AN, 2022. Associations of mode of administration on cannabis consumption and subjective intoxication in daily life. Psychol. Addict. Behav 10.1037/adb0000726. [DOI] [PMC free article] [PubMed] [Google Scholar]
- R. Core Team, 2021. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/. [Google Scholar]
- Ehrler MR, McGlade EC, Yurgelun-Todd DA, 2015. Subjective and cognitive effects of cannabinoids in marijuana smokers. In: Campolongo P, Fattore L (Eds.), Cannabinoid Modulation of Emotion, Memory, and Motivation. Springer, New York, pp. 159–181. 10.1007/978-1-4939-2294-9_7. [DOI] [Google Scholar]
- Freeman TP, Lorenzetti V, 2021. A standard THC unit for reporting of health research on cannabis and cannabinoids. Lancet Psychiatry 8, 944–946. 10.1016/S2215-0366(21)00355-2. [DOI] [PubMed] [Google Scholar]
- Hamaker EL, Muthén B, 2020. The fixed versus random effects debate and how it relates to centering in multilevel modeling. Psychol. Methods 25, 365–379. 10.1037/met0000239. [DOI] [PubMed] [Google Scholar]
- Harrison XA, 2014. Using observation-level random effects to model overdispersion in count data in ecology and evolution. PeerJ 2. 10.7717/peerj.616. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hox JJ, Moerbeek M, Van de Schoot R, 2017. Multilevel analysis: Techniques and applications, Third ed. Routledge, New York. 10.4324/9781315650982. [DOI] [Google Scholar]
- Lankenau SE, Tabb LP, Kioumarsi A, Ataiants J, Iverson E, Wong CF, 2019. Density of medical marijuana dispensaries and current marijuana use among young adult marijuana users in Los Angeles. Subst. Use Misuse 54, 1862–1874. 10.1080/10826084.2019.1618332. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lee CM, Cronce JM, Baldwin SA, Fairlie AM, Atkins DC, Patrick ME, Leigh BC, 2017. Psychometric analysis and validity of the daily alcohol-related consequences and evaluations measure for young adults. Psychol. Assess 29, 253–263. 10.1037/pas0000320. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lee CM, Patrick ME, Fleming CB, Cadigan JM, Abdallah DA, Fairlie AM, Larimer ME, 2020. A daily study comparing alcohol-related positive and negative consequences for days with only alcohol use versus days with simultaneous alcohol and marijuana use in a community sample of young adults. Alcohol. Clin. Exp. Res 44, 689–696. 10.1111/acer.14279. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lee CM, Kilmer JR, Neighbors C, Cadigan JM, Fairlie AM, Patrick ME, White HR, 2021. A marijuana consequences checklist for young adults with implications for brief motivational intervention research. Prev. Sci 22, 758–768. 10.1007/s11121-020-01171-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Linden-Carmichael AN, Van Doren N, Masters LD, Lanza ST, 2020. Simultaneous alcohol and marijuana use in daily life: Implications for level of use, subjective intoxication, and positive and negative consequences. Psychol. Addict. Behav 34, 447–453 10.1037/adb0000556. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lorenzetti V, Hindocha C, Petrilli K, Griffiths P, Brown J, Castillo-Carniglia Á, Freeman TP, 2021. The International cannabis toolkit (iCannToolkit): a multidisciplinary expert consensus on minimum standards for measuring cannabis use. Addiction 1–8. 10.1111/add.15702. [DOI] [PMC free article] [PubMed] [Google Scholar]
- McCrady BS, Epstein EE, 2013. Addictions: A Comprehensive Guidebook, Second ed. Oxford University Press, Oxford. [Google Scholar]
- National Institute on Alcohol Abuse and Alcoholism (NIAAA), no date. What is a standard drink? Retrieved March 23, 2022, from https://www.niaaa.nih.gov/alcohols-effects-health/overview-alcohol-consumption/what-standard-drink.
- Patrick ME, Fleming CB, Fairlie AM, Lee CM, 2020. Cross-fading motives for simultaneous alcohol and marijuana use: associations with young adults’ use and consequences across days. Drug Alcohol Depend. 213, 108077 10.1016/j.drugalcdep.2020.108077. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Prince MA, Conner BT, Pearson MR, 2018. Quantifying cannabis: a field study of marijuana quantity estimation. Psychol. Addict. Behav 32, 426–433 10.1037/adb0000370. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Russell C, Rueda S, Room R, Tyndall M, Fischer B, 2018. Routes of administration for cannabis use–basic prevalence and related health outcomes: a scoping review and synthesis. Int. J. Drug Policy 52, 87–96. 10.1016/j.drugpo.2017.11.008. [DOI] [PubMed] [Google Scholar]
- Schulenberg JE, Patrick ME, Johnston LD, O’Malley PM, Bachman JG, Miech RA, 2021. Monitoring the Future national survey results on drug use, 1975–2020: Volume II, college students and adults ages. Institute for Social Research. The University of Michigan, Ann Arbor, MI, pp. 19–60 (Available at). 〈http://monitoringthefuture.org/pubs.html#monographs〉. [Google Scholar]
- Smart R, Caulkins JP, Kilmer B, Davenport S, Midgette G, 2017. Variation in cannabis potency and prices in a newly legal market: evidence from 30 million cannabis sales in Washington state. Addiction 112, 2167–2177. 10.1111/add.13886. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Spindle TR, Cone EJ, Schlienz NJ, Mitchell JM, Bigelow GE, Flegel R, Hayes E, Vandrey R, 2018. Acute effects of smoked and vaporized cannabis in healthy adults who infrequently use cannabis: a crossover trial. JAMA Netw. Open 1, e184841. 10.1001/jamanetworkopen.2018.4841. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Streck JM, Hughes JR, Klemperer EM, Howard AB, Budney AJ, 2019. Modes of cannabis use: a secondary analysis of an intensive longitudinal natural history study. Addict. Behav 98, 106033 10.1016/j.addbeh.2019.106033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Vandrey R, Herrmann ES, Mitchell JM, Bigelow GE, Flegel R, LoDico C, Cone EJ, 2017. Pharmacokinetic profile of oral cannabis in humans: blood and oral fluid disposition and relation to pharmacodynamic outcomes. J. Anal. Toxicol 41, 83–99. 10.1093/jat/bkx012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Volkow ND, 2021, May 10. Establishing 5mg of THC as the standard unit for research. National Institute on Drug Abuse. https://nida.nih.gov/about-nida/noras-blog/2021/05/establishing-5mg-thc-standard-unit-research. [Google Scholar]
- Wang LP, Maxwell SE, 2015. On disaggregating between-person and within-person effects with longitudinal data using multilevel models. Psychol. Methods 20, 63–83. 10.1037/met0000030. [DOI] [PubMed] [Google Scholar]
- Zeisser C, Thompson K, Stockwell T, Duff C, Chow C, Vallance K, Ivsins A, Michelow W, Marsh D, Lucas P, 2012. A ‘standard joint’? The role of quantity in predicting cannabis-related problems. Addict. Res. Theory 20, 82–92. 10.3109/16066359.2011.569101. [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.