Cannabis eCHECKUP TO GO

Study 1

This study referred to the program as Marijuana ECHECKUP TO GO. Since the time of the publication, the program has been rebranded as Cannabis ECHECKUP TO GO with this terminology now used on the intervention website. Thus, this study review uses the term "cannabis."

Summary

Riggs et al. (2018), Prince et al. (2021), and Fetterling et al. (2021) used a randomized controlled trial to examine 301 undergraduate college students from one large public university in Colorado. Students screened for heavy cannabis use were eligible to participate in the study and were randomly assigned to the intervention condition or a healthy stress management control condition. Primary outcomes measured at a six-week posttest were cannabis use and consequences of cannabis use.

Riggs et al. (2018) found that, compared to control group participants, intervention group participants reported significantly greater reductions in:

• cannabis use

Evaluation Methodology

Design:

Recruitment: Undergraduate college students from one large public university in Colorado were recruited online in the fall of 2016 to participate in the study. Eligibility criteria were that participants were 18 years of age or older, a university student, a recreational cannabis user (i.e., non-medicinal), and reported typical cannabis use at least twice per week. Of the 918 completed screeners, 527 (57%) met eligibility requirements. A total of 301 students agreed to participate in the study and completed a baseline assessment.

Assignment: The 301 students were randomly assigned to the adapted Cannabis eCHECKUP TO GO intervention condition (n=146) or healthy stress management (HSM) control condition (n=155). However, because three participants were ineligible after random assignment, the CONSORT diagram and text stated that 144 students received the intervention and 154 students received the HSM comparison program.

Attrition: Assessments occurred at baseline and six weeks post-intervention. A total of 227 students out of 301 randomized (75%) completed the six-week follow-up assessment.

Sample:

The sample of undergraduate college students was 51% male and had a mean age of 19.97 years. Most students were Caucasian (86%) and non-Hispanic/Latino (89%).

Measures:

All measures came from computer-based self-reports. At baseline and posttest, participants completed a 203-item survey asking about their personal substance use, use consequences, perceived cannabis use norms, and protective behavioral strategies. Three categories of dependent variables were measured: cannabis use and cannabis use consequences as the main outcomes, and program targets/risk and protective factors (descriptive norms, injunctive norms, and protective behavioral strategies).

The five indicators of cannabis use were: hours high per week, hours high per using day, days high per week, weeks high per month, and periods high per week. The number of endorsed time periods was summed across days of the week. Cannabis use consequences were assessed by summing the total number of consequences experienced in the last month and the average severity of the endorsed consequences, measured on a 5-point scale from "never" to "always" experienced. Other than citing an established measure for protective behavioral strategies (a protective factor), the authors did not report any reliability or validity information for the outcome measures, despite noting some concerns about social desirability.

Prince et al. (2021) examined four mediator variables: the proportion of time spent high while partying/socializing, exercising/playing sports, studying, and in class. The proportion of time high while participating in each activity was computed from estimates of the number of hours spent high in each activity and the total number of hours spent on each activity. At baseline, the proportion of time participants reported being high while socializing, exercising, studying, and in class was 0.65, 0.21, 0.17, and 0.11, respectively. The measures, developed by the study authors, lacked established psychometric properties.

Analysis:

General linear models were used to examine posttest intervention effects on cannabis use and consequences outcome variables with controls for baseline outcomes and participant biological sex. Additionally, because of differential attrition on two participant characteristics (males and heavier cannabis users), the authors used standard inverse probability weights to create estimated probabilities of dropout variable, which was also included as a covariate in analyses.

Exploratory moderation analyses examining whether there were program differences by participant sex were conducted by adding a sex-by-intervention interaction term to general linear models.

In the mediation analysis, Prince et al. (2021) used beta regression for the proportional measures of time spent high in various activities. The models included baseline outcomes.

Intent-to-Treat: In Riggs et al. (2018), all students who completed the posttest assessment were included in analyses, and the weighting helped adjust the estimates for missing data by better representing the randomized sample. Prince et al. (2021) referred to using a per-protocol analysis, which involves using "data from those who complete all aspects of the study." Other than those dropped at baseline for non-eligibility, all participants with complete data were included in the analysis.

Outcomes

Implementation Fidelity:

There was no information on implementation fidelity, although the intervention delivery is a computer-based standardized program. Riggs et al. (2018) also mentioned that a manipulation check was performed to confirm all participants received the intended information.

Baseline Equivalence:

Riggs et al. (2018) stated that there were no significant differences between the two conditions on sex, racial/ethnic background, or age (Table 1), and Table 3 presents tests showing no significant differences across conditions on the baseline outcomes. In addition, Prince et al. (2021) noted that the baseline measures of the proportion of time spent high during four types of activities did not differ across conditions.

Differential Attrition:

Participants who completed the study reported significantly fewer hours high per week, hours high per use day, days high per week, time periods per week, and were less likely to be male than female than those participants who dropped out of the study.

Riggs et al. (2018) reported that there were no statistically significant differences in the number of completers vs. non-completers for heavy or male users across conditions. It is unclear whether tests were performed for differential attrition by condition for race/ethnicity and all cannabis outcomes.

Posttest:

Riggs et al. (2018). There were significant intervention effects on four of seven cannabis use and cannabis use consequences outcomes, with the intervention group showing significantly greater reductions in hours high per week, days high per week, weeks high per month, and periods high per week compared to the control group. Effect sizes (partial eta-squared) for intervention effects ranged from small to medium effects (.02-.07).

There was also a significant intervention condition effect for one of three risk and protective factors, such that students in the intervention condition reported reduced descriptive norms (perceived use prevalence), compared to students in the control condition.

Moderation analyses showed that participant biological sex did not moderate intervention effects on outcomes. For moderation analyses examining risk and protective factors as outcomes, females in the intervention condition used significantly more protective behavioral strategies at posttest than males in the intervention condition.

Prince et al. (2021). The direct effect analysis found that the intervention group reported a significantly lower proportion of time spent high during studying than the control group, but there were no effects on the proportion of time spent high in class, while exercising, or while partying/socializing. The mediation tests found a significant indirect effect of the intervention on overall marijuana use via decreased time spent high while studying. Controlling for time spent high while studying partially explained the total effect of the intervention on marijuana use.

Fetterling et al. (2021). In examining mediation models separately for males and females, the results demonstrated that the intervention had a significant indirect effect on the recency of marijuana use through the intervention effect on descriptive norms but only for males.

Long-Term: Not examined.

Study 2

The authors referred to the program as e-TOKE as well as eCHECKUP TO GO.

Summary

Elliott et al. (2014) and Elliott (2012) used a randomized controlled trial to examine 317 college students at one university who were past-month marijuana users. After random assignment to an intervention condition or a no-treatment control condition (each with baseline assessment and no-baseline assessment subgroups), a one-month posttest included measures of marijuana use.

Elliott et al. (2014) and Elliott (2012) found no main effects of the intervention on measures of marijuana use and related problems, but relative to the control group, the intervention group reported significantly:

• more accurate descriptive marijuana use norms

Evaluation Methodology

Design:

Recruitment: The study recruited past-month marijuana users from psychology courses at a large private northeastern university. Of the 320 marijuana users recruited, three non-traditional students were excluded, leaving a sample of 317.

Assignment: The study randomly assigned participants within genders to four conditions: the intervention group without baseline assessment (n = 84), the intervention group with baseline assessment (n = 77), the control group without baseline assessment (n = 71), and the control group with baseline assessment (n = 85). The baseline assessment versus no assessment subgroups were intended to check for assessment reactivity, but much of the analysis combined the sample into the pooled control and pooled intervention groups.

Assessments/Attrition: The posttest assessment came one month after baseline. Of the 317 baseline participants, 98.4% completed the follow-up.

Sample:

The sample, 52% female and 78% White, ranged in age from 18 to 23 (mean = 19.3), with most being freshmen (42%) or sophomores (26%). The students reported smoking marijuana an average of 11 days in the month prior to baseline.

Measures:

All measures came from participant self-reports. The four behavioral outcomes measured marijuana use, marijuana problems, marijuana abuse symptoms (alphas = .45-.52), and marijuana dependence symptoms (alphas = .73-.79). Five risk and protective factors measured descriptive norms corresponding to those addressed in the intervention (e.g., the percent of college students who use marijuana more and less than themselves, and the percent of college students who do not use at all in a typical month). Elliott (2012) also reported on measures of the perceived pros and cons of marijuana use.

Analysis:

The main analyses used individual-level analysis of variance (ANOVA) models with controls for gender but not for the baseline outcomes, which were not assessed for half the sample. Additional analyses for the 50% subsample controlled for the baseline outcomes. The authors used non-linear transformations of the continuous outcomes to adjust for non-normality (see Elliott, 2012, Table 1 for details).

Missing Data Methods: The tables list a sample size of 317, indicating the use of the full randomized sample but without providing information on imputation for the small amount of missing data.

Intent-to-Treat: The study used all participants or all participants with complete data in their originally assigned condition.

Outcomes

Implementation Fidelity:

Of the 149 intervention participants who responded to the question on program participation, only 56% remembered completing it, but computer records confirmed completion for 98.1% of the participants assigned to the program.

Baseline Equivalence:

Table 3 in Elliott (2012) shows one difference at p = .05 in 26 tests (year in school). Elliott et al. (2014) summarized these results by stating that "Conditions did not differ by demographics, social desirability, or marijuana use at baseline, indicating successful randomization (ps < 0.05)."

Differential Attrition:

With only 1.6% attrition and the loss of three in the intervention condition and two in the control condition, tests found that completers and non-completers did not differ on any baseline variables (see Table 2 in Elliott, 2012).

Posttest:

Elliott et al. (2014) found no intervention main effects on the four behavioral outcomes. The intervention significantly improved all five of the risk and protective factors related to perceived accuracy of descriptive norms (d ranged from .33 to .56). As reported in Elliott (2012), the intervention did not significantly affect perceived pros and cons of marijuana use. Assessment reactivity (i.e., completing the baseline assessment) did not influence the outcomes, and similar effects emerged when examining only those who completed the baseline assessment.

Moderation tests by gender found inconsistent differences but generally more beneficial effects for women.

Long-Term:

Not examined.

Study 3

The authors referred to the program as e-TOKE as well as eCHECKUP TO GO.

Summary

Elliott and Carey (2012) used a randomized controlled trial to examine 245 college students at one university who had not used marijuana in the past month. After random assignment to an intervention condition or a no-treatment control condition, a one-month posttest included a measure of marijuana initiation.

Elliott and Carey (2012) found no effects of the intervention on a measure of marijuana initiation, but relative to the control group, the intervention group reported significantly:

• more accurate descriptive marijuana use norms
• more positive perceptions of approval of marijuana abstention

Evaluation Methodology

Design:

Recruitment: The sample included 245 college students at a large northeastern private university who in 2011 were recruited from psychology courses. The sample was recruited as part of Study 2 but was restricted to those who had not used marijuana in the past month rather than to recent users. A total of 245 eligible participants joined the study by completing the baseline survey. Although no participants were using marijuana at the time of recruitment, approximately half (46%) had tried marijuana at some point in their lives.

Assignment: Of the 245 students, 111 were randomly assigned to receive the intervention, and 134 were randomly assigned to the assessment-only control condition.

Assessments/Attrition: The posttest occurred one month after the baseline survey (or upon program completion, if later). The results included 241 students (98%).

Sample:

Participants were on average 20.5 years of age, with the majority in their sophomore (34%) or junior (32%) year. Most were women (73%) who self-identified as White (57%).

Measures:

All measures came from self-reports. The one behavioral outcome measured whether or not marijuana had been used in the last month. Five risk and protective factors included four measures of descriptive norms in which participants were asked to estimate marijuana usage rates among college students and one measure of injunctive norms in which participants were asked about their perception of their close friends' approval or disapproval of their abstention from marijuana. The study provided no information on reliability or validity, although the measures appear straightforward.

Analysis:

The individual-level analysis used ANOVAs for continuous outcomes, a chi-square test for the multicategory outcome, and logistic regression for the binary outcome. The models included a baseline control for an outcome only if it differed across conditions at baseline.

Missing Data Strategy: The study used complete case analysis without imputation or FIML for the 2% missing posttest data.

Intent-to-Treat: The study used all participants with complete data in their originally assigned conditions.

Outcomes

Implementation Fidelity: All but one intervention participant completed the online program.

Baseline Equivalence:

Tests found one difference in nine tests. Estimates regarding the percentage of U.S. college students who had used marijuana in the last month were significantly higher among the control group than the intervention group.

Differential Attrition:

Attrition was only 2%.

Posttest:

Tests found significant intervention effects in favor of the treatment group for all four measures of descriptive norms and the one measure of injunctive norms, but the conditions did not differ significantly on the one behavioral outcome of marijuana use initiation.

Long-Term:

Not examined.

Study 4

Summary

Palfai et al. (2014, 2016) used a randomized controlled trial to examine 123 undergraduate students who came to the university Student Health Services and reported using marijuana monthly. After random assignment to an intervention condition or a minimal health feedback control condition, three- and six-month follow-ups measured marijuana use and the negative consequences of marijuana use.

Palfai et al. (2014) found no effects of the intervention on measures of marijuana use or the negative consequences of marijuana use, but relative to the control group, the intervention group reported significantly:

• more accurate marijuana norms

Evaluation Methodology

Design:

Recruitment: The sample included 123 undergraduates at a private university who 1) presented to Student Health Services in 2012 and 2) reported using marijuana at least monthly over the past 90 days. The sample excluded students with high scores for marijuana dependence on a screening instrument; these students were provided with information about their score and encouraged to seek clinical services at the university. Of the 1,080 undergraduate students undergoing screening, 209 indicated that they had smoked marijuana at least monthly and 123 (59%) joined the study and completed the baseline survey.

Assignment: The study randomized the 123 students to the personalized feedback intervention condition (n = 61) or to the control condition (n = 62) that received minimal general health feedback regarding recommended guidelines for sleep, exercise, and nutrition. In addition, students were randomized to complete assessments either on-site (i.e., at Student Health Services) or off-site (i.e., at a place of the student's own choosing).

Assessments/Attrition: Assessments occurred at baseline and at three months and six months after baseline. Completion rates were 89% at three months and 84% at six months. The non-completers included six students who were withdrawn from the trial following the three-month assessment because their scores indicated a high probability of dependence and need for clinical services.

Sample:

About 87% of the sample participants identified their race as White, with 2.4% Black, 1.6% American Indian/Alaskan Native, and 5.7% Asian. About 17% identified as Hispanic, 42.5% were female, and the average age was 19.7.

Measures:

All four outcome measures came from student self-reports. The two behavioral outcomes included reports of the number of days using marijuana in the past 90 days and the number of marijuana-related negative consequences in the past 90 days. The two risk and protective factors included a measure of readiness-to-change and a measure of perceptions of marijuana use norms at the university. Palfai et al. (2016) divided the readiness-to-change measure into two subscales, problem recognition and action. Nearly all the scales showed good reliabilities.

Analysis:

The analysis used conditional latent growth models to regress the time slope of the outcome (including the baseline outcome) on the intervention measure. The robust maximum likelihood estimator adjusted for the non-normal distributions of the outcomes.

Missing Data Methods: The maximum likelihood estimator accommodated missing data in the models and allowed for inclusion of all 123 randomized participants in the analysis.

Intent-to-Treat: The study dropped six participants after three months because of indications of marijuana dependence but appeared to include them in the analysis via maximum likelihood estimation.

Outcomes

Implementation Fidelity:

Not examined.

Baseline Equivalence:

The authors reported that "There were no significant differences between intervention groups on baseline categorical variables as assessed by chi-squared analyses and t-tests for continuous variables." It appears from Table 1 that the tests included two sociodemographic measures (gender and age), four outcome measures, and one screening measure.

Differential Attrition:

The authors reported that "There were no significant differences between intervention groups in terms of follow-up completion," and that the six participants dropped after three months because of high dependence did not differ significantly across conditions. However, the study performed no other tests.

Posttest:

The models found no condition differences in the time slopes (i.e., changes over time in the outcomes) for the two behavioral outcomes, days of marijuana use and negative consequences of marijuana use, or for one of the risk and protective factors, readiness to change. The one significant effect indicated that the intervention group lowered their estimation of peer marijuana use relative to the control group.

Moderation: Palfai et al. (2014) noted that the intervention had a marginally significant effect on negative consequences of marijuana use for the on-site sample but not for the off-site sample. Palfai et al. (2016) found a significant interaction of the intervention with the action subscale of the readiness-to-change measure. Among those with high scores on the action scale, participants in the intervention group reported significantly fewer days of use than those in the control condition at follow-up.

Long-Term:

Not examined.

Study 5

This study added a booster eCHECKUP at three months to the regular program.

Summary

Goodness and Palfai (2020) used a randomized controlled trial to examine 49 graduate students who came to the university Student Health Center and reported using marijuana monthly. After random assignment to an intervention condition plus booster session or a no-feedback control condition, three- and six-month follow-ups measured marijuana use and the negative consequences of marijuana use.

Goodness and Palfai (2020) found that the intervention group relative to the control group reported significantly:

• fewer days using cannabis at three months (before the booster session) but not at six months (after the booster session)

Evaluation Methodology

Design:

Recruitment: The sample included 49 graduate students at a private university who 1) presented to the Student Health Center and 2) reported using marijuana at least monthly over the past 90 days. The sample excluded students with high scores for marijuana use on a screening instrument; these students were provided with information about their score and encouraged to seek clinical services at the university. Of the 701 graduate students who underwent screening, 72 were eligible and 49 joined the study and completed the baseline survey (68%).

Assignment: The students were randomized to either a control condition (n = 25) that answered questions on cannabis use and health-related behaviors but without feedback on cannabis use or the intervention condition (n = 24) with feedback at baseline plus an additional booster feedback intervention at three months.

Assessments/Attrition: Assessments occurred at baseline, three months later, and six months later. Since the intervention was repeated after three months, the first assessment is most relevant to the main program and the six-month assessment is most relevant to the booster program. The completion rate was 92% at both three months and six months. One student was dropped at three months because the results indicated a need for clinical services.

Sample:

The racial composition of the sample was 78% White, 10% Asian, 4% Black/African American, 4% Native Hawaiian or Other Pacific Islander, and 4% Multiracial or Other.  Participants averaged 25.5 years of age and were 49% male.

Measures:

The two behavioral outcomes came from self-reports and included measures of the number of days using marijuana in the past 90 days and the number of marijuana-related negative consequences in the past 90 days (alpha = .77).

Analysis:

The analysis used conditional latent growth models to regress the time slope of the outcome (including the baseline outcome) on the intervention measure. The robust maximum likelihood estimator adjusted for the non-normal distributions of the outcomes.

Missing Data Methods: The maximum likelihood estimator accommodated missing data in the models and allowed for the inclusion of all 49 randomized participants in the analysis. For the one participant dropped after three months, the three-month outcomes were carried forward to six months.

Intent-to-Treat: The study dropped one participant after three months because of an indication of marijuana dependence but used all available data via maximum likelihood estimation and the last observation carried forward method.

Outcomes

Implementation Fidelity:

Not examined.

Baseline Equivalence:

The authors stated that "There were no significant differences between groups" on three demographic variables and 10 cannabis-related baseline variables.

Differential Attrition:

Not examined, although attrition was only 8%.

Posttest:

The intervention group reported significantly fewer days using cannabis than the control group at three months (before the booster intervention) but not at six months (after the booster intervention). The intervention did not significantly affect negative cannabis consequences at either time point. Cohen's f2 value of .09 indicated a small-to-medium effect for days of cannabis use at three months.

Long-Term:

Not examined.