However, college students are a heterogeneous population, and not all require the same level of intervention . To our knowledge, no one has examined the influence of an alcohol intervention on marijuana use when alcohol interventions are provided sequentially in the context of stepped care, in which individuals who do not respond to an initial, low-intensity level of treatment are provided a more intensive treatment . The purpose of the current study was to examine marijuana use in the context of a stepped care intervention for alcohol use.We conducted a secondary analysis of data from a randomized clinical trial implementing stepped care with mandated college students . In this study, all participants received a brief advice session administered by a peer counselor. Participants who continued to drink in a risky manner six weeks following the BA session were randomly assigned to either BMI or AO conditions . Step 2 participants who completed the BMI as opposed to AO reported greater reductions in alcohol-related consequences at all follow-up assessments . We tested three hypotheses to examine whether interventions that reduce alcohol-related outcomes may also reduce marijuana use. First, because dual marijuana and alcohol users consume higher levels of alcohol use and experience more alcohol-related consequences , we hypothesized that marijuana users would report higher HED frequency, peak blood alcohol content , and alcohol related consequences in the 6 weeks following a BA session, after controlling for their pre-BA drinking behavior. Second, we hypothesized that heavy-drinking marijuana users who did not respond to the BA session and, therefore, were randomized to a Step 2 BMI or AO would report worse alcohol-related outcomes at 3-, 6-, and 9-month follow-ups than non-users. Third,rolling benches canada we examined whether marijuana users changed their marijuana use frequency at any of the three assessment time points following the Step 2 BMI. Examination of marijuana use in this context will improve our understanding of whether marijuana use lessens the efficacy of alcohol interventions, even when delivered sequentially in stepped care.
Furthermore, it will inform future intervention efforts aimed at reducing both alcohol and marijuana use.Participants indicated how many times they used marijuana in the past 30 days at baseline and at each follow-up assessment time point. Because marijuana use was highly zero-inflated , and due to our interest in whether being a marijuana user influenced intervention outcomes, dichotomous variables were created to group individuals into user versus non-user for use in analyses to compare these subgroups.To determine if participants who completed Step 1 of the intervention would also complete Step 2, participants reported the number of times they engaged in heavy episodic drinking , defined as consumption of 5+ drinks for males , in the past month. The maximum number of drinks consumed during their highest drinking event in the past month and the amount of time spent drinking during this episode were used to calculate the students’ estimated peak blood alcohol concentration using the Matthews and Miller equation and an average metabolism rate of 0.017 g/dL per hour.Alcohol-related consequences were assessed using the Brief Young Adult Alcohol Consequences Questionnaire , a 24-item subset of the 48-item Young Adult Alcohol Consequences Questionnaire . Dichotomous items are summed for a total number of consequences experienced in the past month. The B-YAACQ is reliable and sensitive to changes in alcohol use over time and has demonstrated high internal consistency in research with college students . In this study, the B-YYACQ demonstrated good internal consistency at baseline, 6-week and follow-up assessments .First, distributions of outcome variables were examined, and outliers falling three standard deviations above the mean were recoded to the highest non-outlying value plus one , resolving initial non-normality in outcomes. Demographic information and descriptive statistics for the outcome variables were calculated . To examine marijuana users’ drinking behavior following BA for alcohol misuse , multiple regression models were run to predict each alcohol outcome variable at the 6- week assessment from baseline marijuana user status , controlling for gender and the corresponding alcohol outcome assessed at baseline.
To test hypotheses 2 and 3, hierarchical linear models were run in the HLM 7.01 program , using full maximum likelihood estimation. HLM is ideal for data nested within participants across time, for testing between-person effects and within-person effects on outcomes. An additional advantage of HLM is its flexibility in handling missing data at the within-person level, allowing us to retain for analysis any participant that contributed at least one follow-up assessment. We interpreted models that relied on robust standard errors in the determination of effect significance. All intercepts and slopes were specified as random in order to account for individual variation in both mean levels of the outcomes and time-varying associations. Fully unconditional HLM models were run first in order to determine intraclass correlations for each outcome. ICCs provided information on the percentage of variation in each outcome at both the between- and within-person level. Next, three dummy coded time components were created for inclusion at Level 1. The first was coded and therefore allowed examination of the impact of effects on change in the outcome variable from baseline to the first followup, the second was coded to model the impact of effects on change in the outcome variable from baseline to the second follow-up , and the third was coded in order to estimate the impact of effects on change in the outcome variable from the first to the third follow-up . In the context of these three dummy codes, effects on the intercept represent effects when all time effects are equal to 0 . Of note, as all participants received a BA session in the interim between the true baseline and 6-week assessment, marijuana user status at the 6-week assessment was used as the baseline for these analyses .To address hypothesis 2 , Level 2 effects for marijuana user status, treatment condition, and the interaction between marijuana user status and treatment condition were regressed on the three time components. Following recommendations of Aiken and West , prior to forming interactions, marijuana user status and treatment condition were recoded using effects coding ,flood table to remove collinearity with interaction terms so that all main effects of time could be evaluated in the context of models including interactions. To control for potential baseline group differences, we also regressed marijuana user status and treatment condition on the intercept. To address hypothesis 3 [i.e., whether treatment group impacts marijuana use frequency at any of the three follow-up time points, among those who reported marijuana use at 6-week pre-BMI assessment], at Level 2, treatment condition was regressed on the Level 1 intercept and all three time effects of marijuana use frequency. In models for both hypotheses 2 and 3, at Level 2, gender also was included as a covariate.
Descriptive statistics for the full sample of 530 are presented in Tables 1–2. Among participants randomized to BMI or AO in Step 2 , the person-period data set was represented by 392 participants with complete baseline data , each with up to 3 follow-up assessments. Across these participants, we have complete data for a total of 1084 out of 1176 assessments . Specifically, 368 participants completed the 3- month follow-up, 349 completed the 6-month follow-up, and 367 completed the 9-month follow-up. The ICC for alcohol consequences was 0.63 meaning that 63% of the variance in consequences is due to between-person differences, while 37% is due to within-person differences across the follow-ups. The ICCs for HED frequency and pBAC were 0.53 and 0.52, respectively. In the subset of participants who reported marijuana use at the pre-BMI assessment and were therefore included in hypothesis 3 analyses, the ICC of marijuana frequency was 0.59. In all cases, a two-level model was appropriate.Multiple regression models indicated that baseline marijuana user status was not associated with changes in HED frequency, pBAC, or alcohol consequences following the BA session .Results of the HLM models predicting three alcohol outcomes at each follow-up by marijuana user status, treatment condition, and marijuana user status by condition interactions are displayed in Table 4. In the prediction of HED frequency, marijuana user status was associated with higher baseline HED frequency; however, being a marijuana user was not associated with more or less change in HED frequency between the pre-BMI assessment and any of the three follow-ups. There were no interactions between marijuana user status and treatment condition at any follow-up, suggesting that the BMI was not more or less effective for marijuana users. In the prediction of pBAC, marijuana user status was associated with higher pre-BMI pBAC. Additionally, those in the BMI condition had significantly lower pre-BMI pBACs. Controlling for these pre-BMI differences, being a marijuana user, treatment condition, and their interaction were all non-significantly associated with change in pBAC from pre-BMI to each of the follow-ups. In the prediction of alcohol consequences, being a marijuana user was associated with higher pre-BMI levels of consequences. There were no significant effects of marijuana user status, treatment condition, or their interaction on change in consequences between baseline and either the 3- or the 6- month follow-ups. At the 9-month follow-up, those in the BMI reported fewer alcohol consequences1 ; however, this was not moderated by marijuana user status. Overall, these findings suggest that collapsing across treatment condition, marijuana users had heavier alcohol consumption and consequences compared to non-users at the pre-BMI assessment, but they did not increase or decrease their consumption or consequences between pre-BMI and any of the follow-ups.
Additionally, marijuana users responded to the BMI similarly to non-marijuana users at each time point .The purpose of the current study was to examine whether heavy drinking marijuana users demonstrate poorer response to two different alcohol-focused interventions compared to non-users and to examine the efficacy of an alcohol-focused BMI on marijuana use frequency among marijuana users receiving stepped care for alcohol use. Our findings indicated that marijuana users and nonusers evidenced equivalent treatment responses to the alcohol-focused BA session and reported similar alcohol-related outcomes following the BMI. Consistent with prior research , the alcohol-focused BMI did not significantly reduce marijuana use frequency in comparison to the assessment-only group. In our sample, marijuana users did report higher alcohol consumption and problems at baseline/pre-BMI regardless of condition, and these differences between users and nonusers persisted over time. The findings of the current study are somewhat consistent with studies indicating that marijuana use does not decrease the efficacy of alcohol interventions . Although marijuana use did not necessarily lessen the efficacy of the BA and BMI sessions on alcohol use and consequences, regardless of condition, marijuana users reported higher levels of alcohol consumption and consequences at baseline and the pre-BMI assessment. These patterns suggest that heavy drinking marijuana users may still benefit from alcohol use interventions. This is especially noteworthy because dual users typically report increased consequences related to their alcohol use and may have a higher likelihood of being referred to alcohol-focused treatment or mandated to receive intervention for alcohol-related sanctions. Although heavy drinking marijuana users may demonstrate reductions in alcohol consequences following an alcohol-focused intervention , their frequency of marijuana use did not change as a result of receiving a BMI. We can posit several reasons for the participants’ continued use of marijuana, despite a decrease in alcohol-related consequences. First, the parent study found a reduction in alcohol consequences following the alcohol-focused BMI, but not a decrease in alcohol consumption. Prior research examining secondary effects of alcohol BMIs have noted a decrease in marijuana use when there was also a decrease in alcohol consumption . It could be that factors that result in students’ experiencing fewer alcohol-related consequences without changing their drinking differ from ones that would lead to reductions in alcohol or marijuana use. Although our study did not include a measure of marijuana-related consequences, future research should examine changes in marijuana consequences to investigate whether changes in alcohol-related consequences correspond with changes in marijuana consequences following alcohol-focused BMIs. Second, a lack of effects may be due to the fact that our BMI was focused solely on changing alcohol-related behaviors and did not discuss the participant’s marijuana use.