To characterize the relationship between marijuana and tobacco use among young adults, the present study used a national online survey to examine whether heavier marijuana use was associated with greater cigarette use . We also sought to examine whether marijuana use was associated with differences in the likelihood of a past-year tobacco quit attempt. Furthermore, while controlling for smoking severity, we sought to examine whether marijuana use was associated with differences in smoking related cognitions, including tobacco-related expectancies, desire and self-efficacy for quitting, abstinence goals, pros and cons for quitting , and readiness to quit .Participants were young adults between ages 18 and 25 years who were English literate and reported having smoked at least one cigarette in the past 30 days. Marijuana use was not an inclusion criterion. This cross-sectional survey used three Internet-based recruitment methods described previously : a paid advertisement campaign on Facebook, a free campaign on Craigslist,vertical farming racks and a paid email advertising campaign through a survey sampling company. Advertisements on the three Internet-based channels invited young adults to participate in a 20-minute online survey with a chance to win a prize in a drawing . Advertisements were targeted to tobacco smokers and/or both tobacco and marijuana use . The campaign ran for 18 consecutive months, between April 1, 2009, and December 31, 2010.
Advertisements contained a hyperlink that directed potential participants to the study’s University of California Institutional Review Board–approved consent form that included verification questions to determine understanding of the consent process and a screener for determining initial eligibility, including English literacy. Screening questions assessed age and past-month use of tobacco and other substances, including alcohol and marijuana . Participants who consented and were deemed eligible were asked to complete a demographic questionnaire and measures of tobacco, marijuana, and other substance use and thoughts about tobacco use. The survey was anonymous, and data were encrypted for added security protection. Participants were required to answer all questions before they could continue to the next page of the survey but could quit and return to the survey at any time. Computer Internet Protocol addresses were tracked, and only one entry was allowed from a single computer to prevent duplicate entries from the same person; however, multiple entries were allowed from the same Internet connection . Eligibility checks excluded respondents who had discrepant data on similar demographic questions or grossly discrepant data on substance use measures , reported the same email address across multiple survey entries, and had clearly invalid data . Respondents found to be ineligible based on initial screening questions or the above criteria were considered invalid, and data were not analyzed.Chi-squares and t tests were used to compare tobacco smokers who did and did not use marijuana on demographic, tobacco use, alcohol use, and other drug use characteristics. Non-parametric Mann–Whitney U tests were used to test for differences in variables that were skewed. Analyses of study aims consisted of estimating and testing multivariate regression and logistic regression models.
Because this was an observational study comparing self-selected groups, propensity scoring was used to help account for differences between marijuana users and nonusers that could lead to biased estimates of smoking patterns and cognitions. This method, widely used in observational studies, incorporates a scalar summary of covariate information to be included in the study design . For each participant, sociodemographic, alcohol, and other drug use variables thought to be related to marijuana use were entered in a logistic regression analysis to generate a propensity score. Propensity scores were then entered as independent variables in all analyses testing the effect of marijuana use on tobacco use and thoughts about use. In addition to propensity scores, the total number of cigarettes smoked in the past month was included as a control variable in all analyses of thoughts about tobacco use. Given the possibility that equating groups on alcohol and other substance use may overly correct for risk taking or substance problems in the marijuana use groups, all analyses were rerun with a modified version of propensity scores that did not include alcohol or other drug use variables. Given that propensity scoring is typically used when making causal inferences , we also reran any models in which propensity scores were significant, using only those individual covariates that were significant in analyses comparing marijuana users with nonusers . All findings were consistent with those described below.The online survey received more than 7,567 hits, and 7,260 people gave online consent to determine eligibility to complete the survey. Of those, 4,242 met criteria and 494 were deemed invalid, leaving 3,748 eligible and valid cases.
Of those, 3,379 completed the demographic items, and 1,987 completed the entire survey. The completion rate was consistent with other online smoking studies . Survey completers differed significantly from those who completed demographic information only on several variables, but the differences were of small magnitude . Because significant differences were more likely attributable to the large sample size than to meaningful group differences, we used only the completed cases in analyses for the present study. More than half of the sample reported marijuana use in the past month. The proportion of marijuana users was identical to that in a sub-sample of respondents who were recruited from advertisements targeted to tobacco use only. Table 1 presents comparisons between marijuana users and nonusers on sociodemographic, tobacco, alcohol, and other drug use variables. Compared with those who smoked only cigarettes, those who also used marijuana were slightly younger, were more likely to be male, were more likely to be multi-ethnic, and had higher household incomes. Marijuana users were more likely to have used alcohol or illicit drugs in the past month, and those who did so had used them more often and in greater amounts. Without controlling for demographic differences, we found that marijuana users had fewer total years of smoking, were less likely to be daily smokers, reported fewer pros of smoking,vertical rack system and were less likely to endorse a goal of abstinence from smoking.Demographic, alcohol, and other drug use variables were included in propensity score computations. Regression analyses tested whether marijuana use frequency was associated with cigarette use. Use was examined as the number of days smoking in the past 30 days, the total number of cigarettes smoked in the past 30 days, the average number of cigarettes smoked per day, daily smoking status [yes/no; logistic regression], nicotine dependence , and the total number of years smoking. Model fit statistics are summarized for multiple regression models and logistic regression models testing the relationships between marijuana use and tobacco use variables. Models with propensity scores added first indicated that propensity scores were associated with the number of days smoking in the past 30 days , the total number of cigarettes smoked in the past 30 days , the average number of cigarettes smoked per day , nicotine dependence , the total number of years smoking , and daily smoking status . Controlling for propensity scores, we found that there were significant associations between the number of days using marijuana and the number of days smoking in the past 30 days , the number of cigarettes smoked in the past 30 days , and the average number of cigarettes per smoking day . There were no significant relationships between marijuana use and nicotine dependence , the total number of years smoking , or daily smoking status .Logistic regression was used to examine whether marijuana use was associated with a past-year quit attempt and, in addition to propensity scores, controlled for cigarettes smoked in the past 30 days. Tobacco use significantly predicted the likelihood of making a past-year tobacco quit attempt, but marijuana use did not .
Regression analysis tested whether, when we controlled for propensity scores and the number of cigarettes smoked in the past 30 days, past-30-day marijuana use was associated with smoking-related outcome expectancies , the desire to quit smoking, self-efficacy for smoking cessation, expected difficulty with staying quit, pros and cons for smoking, the stage of change for tobacco use or having a goal of abstinence . Propensity scores were associated with the desire to quit smoking and the pros of tobacco use . A greater number of cigarettes smoked in the past 30 days was associated with more positive tobacco outcome expectancies , a lower desire to quit , lower self-efficacy for quitting , greater perceived difficulty with staying quit , more pros of smoking , and an earlier stage of change for tobacco . Compared with marijuana nonusers, marijuana users were three fourths as likely to endorse a goal of complete, sustained abstinence from tobacco use . Given the large number of comparisons made for this study , if we controlled for type I error using a Šidák correction, the adjusted p value threshold would be .003, and the p value for marijuana use in this model would not be significant either. Marijuana use was not significantly associated with tobacco outcome expectancies , thoughts about tobacco abstinence , or the pros or cons of smoking . Marijuana use did not differentiate stages of change for quitting tobacco use .The present study tested whether marijuana use had an effect on tobacco use and associated cognitions among young adult smokers. A high proportion of the sample reported marijuana use in the past 30 days . Although this study did use some recruitment advertisements targeted to both tobacco and marijuana users, the proportion of marijuana users was identical regardless of recruitment strategy. The anonymity of the online environment may have provided for reduced bias in reporting of illegal or stigmatized activity. Consistent with much work on the relationship between tobacco and marijuana use, we found that heavier marijuana use was associated with heavier tobacco use when groups were equated on demographic and substance use characteristics. This is consistent with the large body of work demonstrating that tobacco use is associated with the use of marijuana among young people and that prior cannabis use increases the risk for later smoking and developing nicotine dependence among adolescents . The present study stands out in that it used a continuous measure of marijuana use, making it possible to assess the relationship between tobacco and marijuana use at all levels of marijuana use severity. A recent review of clinical outcomes of tobacco and marijuana co-use found that, relative to tobacco use only, co-occurring use was not associated with a greater likelihood of tobacco use disorder, psychosocial problems, or poorer tobacco-cessation outcomes . However, most of the studies reviewed measured marijuana use as a dichotomous variable . The present study suggests that contributions to tobacco related use and problems are likely seen at higher levels of marijuana use. Thoughts about tobacco use and quitting were not associated with past-month marijuana use. This speaks to the importance of addressing tobacco cessation with a similar effort in those who use marijuana compared with those who do not . A notable exception was that marijuana users were less likely to select a complete abstinence goal for tobacco use, yet associations with the desire to quit smoking and the stage of change for smoking were not significant. Future research should examine co-users’ thoughts about tobacco in relation to their thoughts about marijuana to clarify this issue. This national online survey was a convenience rather than a representative sample of young adult smokers, and thus findings may differ from previous work because of differences in survey methodology. Marijuana users and nonusers were naturally occurring groups and differed on a number of characteristics, which were statistically controlled for through the use of propensity scores and by controlling for tobacco use in all analyses of the effect of marijuana use on tobacco-related cognitions. Statistically adjusting for group differences may have discounted true differences in the population. Effect sizes for most of the models tested were relatively low , suggesting that there are additional variables not controlled for in these analyses that could account for the differences between marijuana users and nonusers. The data were self-reported; however, previous reports have demonstrated good reliability and validity of tobacco and marijuana reports compared with multiple measures of these behaviors and national epidemiological data. Only 52% of the entire eligible sample completed the survey; however, this completion rate is consistent with other online survey studies with young adults , and methods of tracking participants beyond what were used here would have compromised participant anonymity.