Gender matters when examining the association between substance use and mental health outcomes, not only because prevalence rates of depression, anxiety, and substance use differ by gender , but also because what drives individuals to use might differ by gender as well as the ensuing use behaviors. Additionally, although differences in use prevalence continue to differ by gender, the gap in prevalence of use between genders is decreasing . There is also an argument to be made that thus far, most of the work around men who use marijuana, making women marijuana users a minority, understudied population. Finally, prior studies conducted in states where marijuana is illegal may not generalize to states like California, where marijuana is legal. Individuals who use marijuana in the latter contexts may have fewer concerns about social desirability, and thus be more forthcoming about their attitudes, behaviors, and use practices. Data from the Cannabis, Health and Young Adult study were used for the purposes of this dissertation. The Cannabis, Health and Young Adult study is a five-year, mixed method study designed to understand the impact of medical marijuana policies on the physical and psychological health of young adults residing in Los Angeles,indoor vertical garden system as well as the influence of medical marijuana dispensaries on individual and community health. It is the first study funded by the National Institute on Drug Abuse to specifically examine medical marijuana use among a young adult population in the United States.
Data collection for the first wave of the study occurred between February 2014 and April 2015. To be eligible for enrollment, participants had to: 1) be between the ages of 18 and 26; 2) have used marijuana at least four times in the past 30 days; 3) currently reside in the Los Angeles Metro area; and 4) speak and read English. Participants were identified as medical marijuana users or patients if they had a medical marijuana recommendation issued in California within the last three years. Participants were identified as non-patient users if they had never received a recommendation for medical marijuana in any state. Targeted and chain referral sampling were used to recruit young adults, between the ages of 18 and 26, who use marijuana in the Los Angeles Metro area. These two recruitment methods have been proven to be successful to recruit hard to reach populations such as substance using individuals . This sampling methodology allowed control of screening and enrollment so that the sample is stratified to have specified gender, race, and age diversity . The targeted sampling used mapped data of medical marijuana dispensaries in the Los Angeles metro area to target surrounding locations containing the population of interest such as dispensaries, parks, and college campuses. Interviewers at these locations approached potential participants to present the study, and to screen potential participants should they manifest interest in participating. Chain-referral sampling, a non-random sampling approach, utilized currently enrolled participants to refer others within their network to join the study. Chain referral sampling was used in addition to targeted sampling to avoid biasing the sample towards those living in proximity to dispensaries.
Flyers posted in public location across Los Angeles and adds on Craigslist, a classified advertisement website, were also used as recruitment strategies. Individuals screened for the study were compensated with a $3 gift card. Out of 710 individuals 436 screened eligible and 366 were enrolled in the study . Attempts were made to sample from multiple networks, socioeconomic and geographically diverse areas of Los Angeles to increase the diversity of the sample. Although, this is not a representative sample, it is the only sample we know of recruited in a city where medical marijuana is legal, that includes young adults who use marijuana exclusively for recreational reasons, young adults who use marijuana exclusively for medical reasons, and young adults who use marijuana for recreational and medical reasons. Furthermore, it is also the only study we know of that contains information that pertains both to motives of marijuana use as well as to symptoms of depression, symptoms of anxiety, and overall psychological distress. The study instrument was developed using Research Electronic Data Capture , a secure web application for building and managing online surveys and databases. Interviews, lasting between 60 to 90 minutes, were conducted in private or semiprivate locations in the neighborhoods where participants were recruited or lived. Most questions were administered face-to-face except for psychometric scales and sensitive questions involving sexual behavior, which were self-administered. Participants were compensated with a $25 cash incentive for the interview.
Study procedures were approved by the Institutional Review Boards at Children’s Hospital Los Angeles and at Drexel University. Three continuous dependent variables were studied separately: symptoms of depression, symptoms of anxiety, and overall psychological distress. These dependent variables were operationalized by the depression sub-scale, the anxiety sub-scale, and the Global Severity Index of the Brief Symptom Inventory-18. A shortened version of the Brief Symptom Inventory , the BSI-18, is an 18 item self-report symptom checklist designed to measure three dimensions of psychological distress in clinical and non-clinical populations: depression, anxiety and somatization. A Global Severity Index, an indicator of overall psychological distress, can also be derived from the BSI-18. Using a five-point Likert scale that ranges from “Not at all” to “Extremely”, participants were asked to rate how much they were distressed by each symptom listed during the past seven days. Examples of symptoms listed include: faintness or dizziness, feeling blue, feelings of worthlessness, and nausea or upset stomach. Each of the previously mentioned sub-scales, depression, anxiety, and somatization is comprised of six items and the range of possible scores for each is 0 to 24. The Global SeverityIndex is calculated by summing the 18 items. The range of possible scores for the GSI is 0 to 72. Higher scores correspond to higher psychological distress. A prior study of the BSI-18 among drug using individuals ages 18 and over showed high Cronbach alpha values of the sub-scales: 0.84 for somatization, 0.86 for depression, 0.88 for anxiety, and 0.93 for the Global Severity Index . Given its high internal consistency and test-retest reliability , as well as its usefulness for mental health screenings of substance using individuals , the BSI and BSI-18 are common measures of mental health in substance use research . The independent variables of interest for the purpose of this dissertation, are motives of use as operationalized by an amended version of Lee et al. Comprehensive Marijuana Motives Questionnaire . Lee et al. original questionnaire is comprised of 36 items representing 12 sub-scales of motives of marijuana use with high Cronbach alphas ranging from 0.78 to 0.89 . The 12 motive sub-scales and their respective Cronbach alphas are: enjoyment , conformity , coping , experimentation , boredom , alcohol , celebration , altered perception , social anxiety , relative low risk , sleep , and availability . For the purposes of the CHAYA study, 15 items were added to the original 36 for a total of 51 items, to create the five medical use sub-scales. The five medical sub-scales are: natural medicine, pain, nausea, substitution, and attention. Examples of the added items are: to lessen the intensity of my pain, so that I don’t feel sick to my stomach,clone rack and as a natural alternative to prescription or over the counter drugs. Participants were asked to respond to “Thinking of all the times you have used marijuana; how often would you say that you use for each of the following reasons” using a five-point Likert scale ranging from “Almost Never/Never=1” to “Almost always/Always=5.” Examples of reasons listed are: to make you feel more confident, because you were drunk, to help you sleep, because you were experimenting, and because you were depressed. The mean weighted range of possible scores for each sub-scale is 1= Almost never/Never, to 5= Almost always/Always. For this dissertation, the sub-scales were kept continuous. Higher scores indicate a stronger endorsement for any given motive of use.Race and ethnicity. Race and ethnicity were recorded as a categorical variable. Participants were asked what they considered to be their primary racial or ethnic group.
Possible answer choices were: Non-Hispanic Black/African American, Non-Hispanic White/Caucasian, Non-Hispanic Asian/Pacific Islander, Non-Hispanic Native American, Non-Hispanic Multiracial or Hispanic/Latino. Race/ethnicity was dummy coded for analyses and Non-Hispanic White/Caucasian was used as the reference category. Non-Hispanic White/Caucasian was used as the reference category as they represent the majority of participants in much of the research to date on motives of marijuana use . User group. User group was operationalized and controlled for as follows: participants who have never received a recommendation for medical marijuana in any state were categorized as non-medical users , while participants who currently have or ever had a recommendation for medical marijuana were categorized as medical marijuana users . Although traditionally included as a control variable, socioeconomic status was not included as a control variable here due to the lack of variance for this variable in our sample. Table 3.4 indicates the number of missing cases for key variables of interest. Number of missing cases for key variables ranges between 0 and 8. Given that the missing data accounts for less than 10% of our dataset, analyses were performed using listwise deletion for participants with missing data on key variables to maximize sample size for each analysis. This is deemed to be an acceptable strategy to avoid biased statistical analyses because the number of missing cases in our sample is small . Performing multiple imputations to replace missing variables would have not been appropriate here given that it is unlikely that variables were missing at random . Sample size for various analyses therefore range from 346 to 364 depending on the variables being tested in each model. The breakdown of sample sizes is as follows. In Aim 1, n=364. In Aim 2, n=355 for multiple linear regression analyses performed without control variables, and n=350 for multiple linear regression analyses done with control variables. The sample sizes remain the same for mediation analyses performed using past 90 days marijuana use as a mediator. For daily number of marijuana hits as a mediator, n=351 when no control variables are entered in the model, and n=346 with control variables present in the model. In Aim 3, n=355 for moderation analyses and conditional process analyses without control variables and n=350 for moderation analyses and conditional process analyses with control variables. The purpose of an exploratory factor analysis is to explore which observed variables relate to factors to achieve a model that fits the data and has theoretical support . As such, an exploratory factor analysis was performed using wave 1 data to determine a plausible model for the factor structure of motives of marijuana use for young adults who use marijuana for recreational and/or medical reasons in Los Angeles. Using a geomin rotated solution, seventeen alternative models were requested along with a Scree plot. Maximum likelihood estimation was used as it can account for missing data, generates unbiased parameter estimates and standard errors, allows for significant testing, and provides fit estimates . Confirmatory factor analysis, is used to determine how a hypothesized factor model fits a new sample from a different population by examining factor variances and covariances . I thus proceeded with confirmatory factor analyses to evaluate the fit of the most theoretically and conceptually sound models generated by the exploratory factor analysis as well as the fit of the original 17 factors hypothesized model. Separate confirmatory factor analyses were also conducted for the motives from the Comprehensive Marijuana Motive Questionnaire only and medical use motives only. Confirmatory factor analyses were also conducted using wave 2 data for the retained factor structure as well as for the CMMQ items and the MM items to test for factor consistency across waves. Except for two factors in Models 16 and 17, all latent variables were specified with three indicators as it is recommended in the literature . Figures 3.1 to 3.4 depict the models that were confirmed using a confirmatory factor analysis. The purpose of the second aim was to investigate the associations between motives of marijuana use and symptoms of depression and symptoms of anxiety, as well as overall psychological distress in young adults who use marijuana.