The American Journal on Addictions, 23: 7–14, 2014 Copyright © American Academy of Addiction Psychiatry ISSN: 1055-0496 print / 1521-0391 online DOI: 10.1111/j.1521-0391.2013.12054.x
Cannabis Use among Treatment‐Seeking Smokers: Motives and the Moderating Effects of Anxiety Sensitivity Melissa M. Norberg, PhD,1 Jake Olivier, PhD,2 Norman B. Schmidt, PhD,3 Michael J. Zvolensky, PhD4 1
National Cannabis Prevention and Information Centre, University of New South Wales, Australia School of Mathematics and Statistics, University of New South Wales, Australia 3 Department of Psychology, Florida State University, Tallahassee, Florida 4 Department of Psychology, University of Houston, Houston, Texas 2
Background and Objectives: Concurrent use of cannabis and tobacco is associated with poor tobacco cessation outcomes. As little research has examined why treatment‐seeking tobacco users engage in cannabis use, the objective of this study was to discover if emotional vulnerability and cannabis use motives are associated with concurrent users’ cannabis use. Methods: One hundred thirty‐eight (n ¼ 138) daily cigarette smokers seeking smoking cessation treatment completed measures of anxiety sensitivity, cannabis use motives, and cannabis use. Results: Stronger coping, enhancement, social, and expansion motives were associated with using greater amounts of cannabis per use occasion. In a model accounting for all these motives, anxiety sensitivity moderated the relationship between enhancement motives and cannabis use. Discussion and Conclusions: Clinical interventions for concurrent tobacco–cannabis users may be advanced by targeting low anxiety sensitive individuals’ use of cannabis to increase excitement and fun. Such an approach may consist of having clients identify and engage in healthier pleasurable activities and by teaching clients to accept the trade‐off between perceived less pleasurable, but healthier activities and cannabis use. (Am J Addict 2014;23:7–14)
BACKGROUND AND OBJECTIVES Data from the National Epidemiological Survey on Alcohol and Related Conditions demonstrates that 28% of adults in the US who smoke tobacco concurrently use cannabis.1 Given that concurrent tobacco–cannabis users believe that they are more physically and mentally addicted to nicotine2 and are less likely to quit smoking cigarettes than nonusers of cannabis,3,4 tobacco cessation interventions for concurrent tobacco– cannabis users may be enhanced by targeting mechanisms Received July 30, 2012; revised September 20, 2012; accepted December 7, 2012. Address correspondence to: Dr. Norberg, PO Box 684, Randwick, NSW 2031, Australia. E‐mail: [email protected].
underlying cannabis use. Unfortunately, scarce research describes why tobacco users may use cannabis.5 To date, most research on the underlying mechanisms of cannabis use has neglected cannabis’ link to tobacco. The majority of research into cannabis use motives has relied upon collecting data from mixed samples (ie, some use tobacco while other participants do not) and controlling statistically for concurrent tobacco use. This research has found that stronger endorsement of coping, enhancement, social, and expansion motives is associated with greater frequency of cannabis use.6 Further, studies have shown that using cannabis to cope is related to an increased risk for negative emotional symptoms and disorders, while using cannabis for enhancement purposes is related to greater positive affectivity and lower anxious arousal.7,8 Individual differences in emotional vulnerability may contribute to these observed relations, such that those who are very sensitive to anxiety may be more apt to use cannabis to cope with their negative affect, while those who are less sensitive to anxiety may be more likely to use cannabis to increase their positive affect. Partially consistent with this speculation, positive associations between anxiety sensitivity and coping motives for cannabis use have been reported when controlling statistically for concurrent tobacco use6,9; however, enhancement motives have not been related to anxiety sensitivity.6 In order to identify methodological changes that will improve research and further our knowledge about the relationship between concurrent cannabis and tobacco use, we believe it is necessary to examine the limitations of the above‐mentioned research. First, and foremost, previous work has relied upon statistically controlling for tobacco use. Although useful for establishing how anxiety sensitivity and motives relate to cannabis use independently of tobacco use, failure to examine these risk factors explicitly among concurrent tobacco–cannabis users has prevented an understanding of why cigarette smokers use cannabis, and therefore has limited researchers’ ability to tailor smoking cessation 7
treatment to target concurrent cannabis use, a behavior that contributes to cigarette smoking. Second, most research has focused on the individual effects of these variables in relation to cannabis use, which has not allowed researchers to understand how the combination of these variables influences cannabis use (ie, examination of associations in one overarching model). This issue is clinically important, as cannabis use motives are highly related to one another.10,11 Similarly, past work has predominately targeted the main effects of these variables. Such an approach does not take into account how anxiety sensitivity may qualify cannabis use motives effects and may lead to false conclusions that anxiety sensitivity is not related to use motives. As an example, research with tobacco smokers found that while indulgent smoking motives were not related to anxiety sensitivity, the interaction between indulgent smoking motives and anxiety sensitivity significantly predicted cigarette use. Level of cigarette use was related to indulgent motives only for individuals low in anxiety sensitivity.12 Lastly, previous research has focused on frequency of use. Recent research demonstrates that quantity of cannabis use is a stronger predictor of cannabis problems than frequency of use, and thus, may be a more clinically meaningful variable.13 Importantly, quantity estimates for cannabis use have been found to be reliable.13 The purpose of this paper, therefore, was to address key limitations of past work by evaluating both the main and interactional effects of anxiety sensitivity and cannabis use motives among a treatment‐seeking sample of cigarette smokers. We studied a treatment‐seeking sample as we were interested in obtaining data relevant to the refinement of tobacco cessation interventions for concurrent cannabis use. In line with past research, we expected coping, enhancement, social, and expansion motives to exhibit positive relationships with quantity of cannabis use when examining zero‐order correlations. Further, we hypothesized that anxiety sensitivity would moderate the relationships between enhancement motives and quantity of cannabis use, such that enhancement motives would be associated positively with quantity of cannabis use only for low anxiety sensitive individuals. Conversely, we also hypothesized that anxiety sensitivity would moderate the relationship between coping motives and quantity of cannabis use, such that coping motives would be associated positively with quantity of cannabis use for high anxiety sensitive individuals only. Although our main aim of this paper was to explore these relations for quantity of cannabis use, we conducted an exploratory analysis to discover whether the moderation model would be similar for frequency of cannabis use.
METHODS Participants Participants were a subset of daily cigarette smokers, who had used cannabis in the past month (N ¼ 138; Nfemale ¼ 56; aged 18–60 years old; Mage ¼ 29.66; SD ¼ 11.99) and 8
responded to advertisements for an ongoing study examining a smoking cessation intervention that focused on vulnerability to panic in comparison to a standard smoking cessation program. On average, participants had smoked cigarettes for 11.81 years (SD ¼ 11.15) and smoked approximately 15.43 cigarettes per day (SD ¼ 10.53). In addition, participants had smoked cannabis for an average of 10.23 years (SD ¼ 9.69). Ninety‐four percent of the sample (n ¼ 129) had used cannabis in the week prior to being assessed for the current study. Of the 125 individuals who reported on their drinking behavior, 68 (54%) indicated that they drank alcohol more than once a week. Approximately half of participants reported attending part of college, but having no degree (47%), and most were unmarried (82%) and Caucasian (84%). Based on the Fagerström Test for Nicotine Dependence,14 approximately half of the sample was classified with medium to very high nicotine dependence (52%). Sixty‐nine individuals (50%) met criteria for an Axis I disorder based on the Structured Clinical Interview for DSM‐IV Axis I Disorders‐Non‐Patient Edition (SCID‐I/NP).15 Twenty individuals (15%) met criteria for two disorders and 11 individuals (8%) met criteria for three or more disorders. The distribution of disorders were as follows: 21 individuals (15%) met criteria for a cannabis use disorder, 22 individuals (16%) met criteria for a substance use disorder other than cannabis (most commonly an alcohol use disorder, n ¼ 20), 42 individuals (30%) met criteria for an anxiety disorder (most commonly social phobia, n ¼ 19), 18 individuals (13%) met criteria for a mood disorder (most commonly major depression, n ¼ 10), and 2 (1%) were diagnosed with another disorder (anorexia nervosa or body dysmorphic disorder). See Table 1 for descriptive information on predictor and outcome variables. See papers by Johnson and colleagues,16 Capron and colleagues,17 and Peasley‐Miklus and colleagues18 for details of other findings obtained from the treatment‐seeking database that was used for this study. Measures Diagnostic Information The SCID‐I/NP was used to determine clinical diagnoses. Diagnoses obtained with the SCID‐I/NP have been shown to have fair to excellent reliability.19 All SCID‐I/NP interviews were administered by trained research assistants or doctoral level staff, blind to study hypotheses, and supervised by independent doctoral‐level professionals. A random selection of 13% of the SCID‐I/NP interviews were checked by the last author for accuracy. There was 100% agreement between these judgments. Cigarette and Cannabis Use The six‐item Fagerström Test for Nicotine Dependence14 was used as a measure of nicotine dependence severity. The Fagerström Test for Nicotine Dependence exhibits good internal consistency, positive relations with key smoking variables, and high test–retest reliability.14,20 The Smoking
Cannabis Use among Treatment‐Seeking Smokers
January–February 2014
TABLE 1. Means and standard deviations for primary study variables
Variables Frequency of cannabis use Quantity of cannabis use ASI‐3 MMM‐enhancement MMM‐social MMM‐conformity MMM‐coping MMM‐expansion
Anxiety sensitivity status
Total sample (N ¼ 138)
Low (n ¼ 96)
High (n ¼ 42)
M (SD)
M (SD)
M (SD)
4.84 3.26 16.75 17.49 13.45 7.04 11.10 10.84
(2.55) (2.59) (12.01) (4.89) (5.71) (3.64) (5.60) (6.05)
4.81 3.36 10.22 17.20 12.98 6.71 9.95 10.27
(2.59) (2.78) (5.44) (5.04) (5.43) (3.35) (4.90) (5.80)
4.90 3.00 31.67 18.14 14.52 7.79 13.74 12.14
(2.87) (2.15) (9.26) (4.52) (6.25) (4.18) (6.06) (6.46)
t‐Test t(136) ¼ 20, p ¼ .85 t(136) ¼ .14, p ¼ .89 t(53.80) ¼ 13.96, p < .001 t(136) ¼ 1.03, p ¼ .30 t(136) ¼ 1.47, p ¼ .15 t(65.08) ¼ 1.48, p ¼ .15 t(136) ¼ 3.84, p < .001 t(136) ¼ 1.68, p ¼ .09
ASI‐3 ¼ anxiety sensitivity index‐3; MMM ¼ marijuana motives measure.
History Questionnaire21 was used to capture information about cigarette use. Items used for the current study included: “For how many years, altogether, have you been a regular daily smoker” and “Think about your smoking during the last week, how many cigarettes did you smoke in an average day.” The Marijuana Smoking History Questionnaire (MSHQ)22 was used to assess participants’ history and patterns of cannabis use. MSHQ items used in study analyses included, “How many years have you smoked marijuana” and “Think about your smoking during the last week, how much marijuana did you smoke per occasion in an average day.” This latter item was rated on an 8‐point Likert scale (1–8). Scores correspond to pictures depicting increasing sizes of marijuana joints, with 1 indicating the smallest joint and 8 indicating the largest joint. The MSHQ has been used successfully as an indicator of cannabis use in many previous studies.8,22
general anxiety sensitivity.24 Each item is rated on a 5‐point Likert scale ranging from 0 (not at all) to 4 (very much); thus, total scores range from 0 to 72. The Cronbach alpha for the total score in the current study was .92.
Motives The marijuana motives measure (MMM)10 contains 25‐ items and five subscales that assess positive and negative reinforcement motives, each with five items. Response options range from 1 (almost never/never) to 5 (almost always/always) scale. The five subscales and representative items are as follows: Enhancement (because I like the feeling), Coping (to forget my worries), Social (because it helps me enjoy a party), Conformity (to be liked), and Expansion (to understand things differently). The subscales have demonstrated fair to excellent internal consistency and support has been found for the multidimensional nature of measure.11 In the current study, Cronbach alphas for the subscales ranged from .82 (Enhancement) to .94 (Expansion).
Analytic Approach Zero‐order correlations were conducted to examine which motives would be individually related to cannabis use, as well as to discover which theoretically relevant variables would need to be controlled for in the regression analysis. As previous research has consistently found lower levels of educational achievement, being unmarried, psychological disorder, race, gender, age, and years of use to predict cannabis and tobacco use,25–29 these variables were examined for appropriateness of model inclusion before serving as covariates in the study analyses. Educational achievement (no college degree vs. at least a 2‐year college degree), relationship status (married vs. single), race (Caucasian vs. non‐Caucasian), and psychiatric diagnosis (absent vs. present) were dummy coded for appropriate inclusion. Given that the cannabis use variables were overdispersed (the conditional variance exceeded the conditional mean), negative binominal regressions were conducted. As anxiety sensitivity may have a categorical latent structure,30,31 ASI‐3 scores were contrast coded (low vs. high) prior to entry into the model. Following Aiken and West’s32 recommendation to use
Anxiety Sensitivity The Anxiety Sensitivity Index‐3 (ASI‐3)23 is an 18‐item multidimensional instrument defined by three highly related subscales: Physical Concerns, Cognitive Concerns, and Social Concerns. Bifactor analysis has demonstrated that it is appropriate to use the total score to obtain an indication of Norberg et al.
Procedure Participants were recruited at two sites (University of Vermont, Burlington, VT, USA and Florida State University, Tallahassee, FL, USA) for an ongoing treatment study at which identical procedures were executed. The current report is based on secondary analyses of data collected during the study’s baseline assessment session, which took place prior to randomization and the commencement of the intervention. All participants provided informed consent after study procedures were explained. Both universities’ institutional review boards approved the study protocol.
January–February 2014
9
West for following up a significant interaction with linear regression. Lastly, to support our method for dichotomizing individuals into high and low anxiety sensitive groups, three additional negative binomial regressions were conducted for quantity of cannabis use. These models examined ASI‐3 scores as a continuous variable and ASI‐3 scores dichotomized using median and tertile splits.
clinically meaningful rather than standard deviation cut‐offs, individuals were classified as experiencing high anxiety sensitivity if their ASI‐3 scores were classified as being in the dysfunctional range using Jacobson and Truax’s33 formula for calculating clinical significance, c ¼ s 0 M1 þ s 1
M0 s0 þ s1
where M0 and M1 denote the average scores of the nonclinical and clinical samples, respectively, and where s0 and s1 indicate the standard deviations of those means. Normative data provided by Taylor et al.23 (M0 nonclinical M ¼ 12.80, s0 ¼ nonclinical SD ¼ 10.60; M1 ¼ clinical M ¼ 32.60, s1 ¼ clinical SD ¼ 14.3) were used to calculate the cut‐off score for the dysfunctional range. Accordingly, individuals were deemed to be low in anxiety sensitivity if they had ASI‐3 scores 21 and lower, whereas individuals were deemed to be high in anxiety sensitivity if they had ASI‐3 scores 22 and higher. It is important to note that this cut‐off was used since no established categorization for anxiety sensitivity exists in the literature. Taxometric analysis was not used in this study to model groups as sample sizes larger than 300 are needed for this procedure.34 In order to improve interpretation of interaction effects and to reduce multicollinearity, continuous explanatory variables were centered prior to creating interaction terms.35 Since interactive effects are less powerful than main effects and because loss of power occurs when dichotomizing scores,36–38 marginally insignificant interaction terms (p .10) for anxiety sensitivity and motives were kept in the model; p values ranging from .05 to .10 were assessed by their unstandardized effect sizes.39–42 Simple slopes were plotted to allow for visual inspection of the interaction.32 For significant interactions, the significance of the simple slopes was assessed by reference coding categorical variables and reestimating model parameters. This is equivalent to the method presented by Aiken and
RESULTS See Table 2 for zero‐order correlations among studied variables. As expected, enhancement, social, coping, and expansion motives were correlated positively with quantity of cannabis use per occasion in the previous week. Education, age, and marital status were related to quantity of cannabis used and were included as covariates in the regression analysis. Conformity motives were not included in the model because they were not related statistically to quantity of cannabis use. The interaction between anxiety sensitivity level and coping motives was not statistically significant (p ¼ .99, B ¼ .0001, SEB ¼ .01), and it was removed from the model (data not displayed in Table 3). Table 3 demonstrates that when all relevant motives and covariates were included in one model, the interaction between anxiety sensitivity and enhancement motives was related marginally to quantity of cannabis use (p ¼ .10). Examination of the simple slopes demonstrated a disordinal interaction (lines that intersect; see Fig. 1). Enhancement motives were not related to cannabis use for individuals high in anxiety sensitivity (Wald x2 ¼ .06, p ¼ .81, B ¼ .01, SEB ¼ .03), but were related to cannabis use for those low in anxiety sensitivity (Wald x2 ¼ 11.06, p < .001, B ¼ .05, SEB ¼ .02). When the model was repeated for frequency of use, the interactions between anxiety sensitivity level and coping (p ¼ .74, B ¼ .003, SEB
TABLE 2. Zero‐order correlations among theoretically relevant variables Variable
2
3
4
5
1. Age .15 .09 .30 2. Gender .04 .07 3. Race .05 4. Education 5. Marital status 6. Psychiatric disorder 7. Years of cannabis use 8. Frequency of cannabis use 9. Quantity of cannabis use 10. ASI‐3 11. MMM‐enhancement 12. MMM‐social 13. MMM‐conformity 14. MMM‐coping 15. MMM‐expansion
6
.372 .07 .05 .05
7
8
9
.17 .74 .30 .19 .22 .03 .11 .02 .004 .13 .34 .18 .01 .27 .20 .11 .19 .003
10
11
.24 .04 .23 .03 .16 .08 .05 .10 .15 .18 .10 .12 .20 .07 .07 .11 .17 .11 .01 .03 .03 .30 .02 .48 .07 .40 .09
12 .04 .003 .19 .03 .01 .13 .02 .20 .30 .13 .51
13
14
.01 .12 .10 .09 .14 .13 .14 .15 .17 .03 .04 .21 .15 .01 .18 .41 .04 .27 .14 .31 .02 .33 .39 .52 .24
15 .18 .14s .18 .02 .01 .12 .10 .26 .27 .14 .46 .56 .22 .44
N reduced to 133 for years of cannabis use due to missing data. N reduced to 129 for quantity of cannabis use as seven people did not report smoking cannabis in the last week. ASI‐3 ¼ anxiety sensitivity index‐3 (low: 0–21, high: 22þ); MMM ¼ marijuana motives measure. p < .05; p < .01; p < .001.
10
Cannabis Use among Treatment‐Seeking Smokers
January–February 2014
TABLE 3. Negative binomial regression analysis for anxiety sensitivity, motives, and cannabis use
Model Quantity of cannabis use Age Marital status Education ASI‐3 MMM‐enhancement MMM‐social MMM‐coping MMM‐expansion MMM‐enhancement by ASI‐3 Frequency of cannabis use Age Marital status Education Psychiatric disorder ASI‐3 MMM‐enhancement MMM‐social MMM‐coping MMM‐expansion MMM‐conformity
B
SEB
95% CI
Wald x2
p‐Value
.003 .28 .25 .21 .006 .01 .02 .004 .05
.006 .17 .16 .14 .03 .01 .01 .01 .03
.02, .008 .06, .61 .56, .07 .05, .48 .05, .06 .02, .04 .003, .05 .02, .03 .009, .10
.35 2.58 2.40 2.50 .06 .79 3.12 .11 2.73
.56 .11 .12 .11 .81 .38 .08 .74 .10
.004 .25 .10 .04 .15 .04 .01 .04 .002 .04
.004 .12 .11 .04 .09 .01 .01 .009 .008 .01
.01, .004 .008, .49 .31, .11 .04, .12 .03, .33 .02, .06 .03, .008 .02, .05 .01, .02 .07, .01
1.02 4.11 .85 .84 2.79 14.52 1.50 18.30 .09 9.67
.31 .04 .36 .36 .09 .0001 .22
Cannabis Use among Treatment‐Seeking Smokers: Motives and the Moderating Effects of Anxiety Sensitivity Melissa M. Norberg, PhD,1 Jake Olivier, PhD,2 Norman B. Schmidt, PhD,3 Michael J. Zvolensky, PhD4 1
National Cannabis Prevention and Information Centre, University of New South Wales, Australia School of Mathematics and Statistics, University of New South Wales, Australia 3 Department of Psychology, Florida State University, Tallahassee, Florida 4 Department of Psychology, University of Houston, Houston, Texas 2
Background and Objectives: Concurrent use of cannabis and tobacco is associated with poor tobacco cessation outcomes. As little research has examined why treatment‐seeking tobacco users engage in cannabis use, the objective of this study was to discover if emotional vulnerability and cannabis use motives are associated with concurrent users’ cannabis use. Methods: One hundred thirty‐eight (n ¼ 138) daily cigarette smokers seeking smoking cessation treatment completed measures of anxiety sensitivity, cannabis use motives, and cannabis use. Results: Stronger coping, enhancement, social, and expansion motives were associated with using greater amounts of cannabis per use occasion. In a model accounting for all these motives, anxiety sensitivity moderated the relationship between enhancement motives and cannabis use. Discussion and Conclusions: Clinical interventions for concurrent tobacco–cannabis users may be advanced by targeting low anxiety sensitive individuals’ use of cannabis to increase excitement and fun. Such an approach may consist of having clients identify and engage in healthier pleasurable activities and by teaching clients to accept the trade‐off between perceived less pleasurable, but healthier activities and cannabis use. (Am J Addict 2014;23:7–14)
BACKGROUND AND OBJECTIVES Data from the National Epidemiological Survey on Alcohol and Related Conditions demonstrates that 28% of adults in the US who smoke tobacco concurrently use cannabis.1 Given that concurrent tobacco–cannabis users believe that they are more physically and mentally addicted to nicotine2 and are less likely to quit smoking cigarettes than nonusers of cannabis,3,4 tobacco cessation interventions for concurrent tobacco– cannabis users may be enhanced by targeting mechanisms Received July 30, 2012; revised September 20, 2012; accepted December 7, 2012. Address correspondence to: Dr. Norberg, PO Box 684, Randwick, NSW 2031, Australia. E‐mail: [email protected].
underlying cannabis use. Unfortunately, scarce research describes why tobacco users may use cannabis.5 To date, most research on the underlying mechanisms of cannabis use has neglected cannabis’ link to tobacco. The majority of research into cannabis use motives has relied upon collecting data from mixed samples (ie, some use tobacco while other participants do not) and controlling statistically for concurrent tobacco use. This research has found that stronger endorsement of coping, enhancement, social, and expansion motives is associated with greater frequency of cannabis use.6 Further, studies have shown that using cannabis to cope is related to an increased risk for negative emotional symptoms and disorders, while using cannabis for enhancement purposes is related to greater positive affectivity and lower anxious arousal.7,8 Individual differences in emotional vulnerability may contribute to these observed relations, such that those who are very sensitive to anxiety may be more apt to use cannabis to cope with their negative affect, while those who are less sensitive to anxiety may be more likely to use cannabis to increase their positive affect. Partially consistent with this speculation, positive associations between anxiety sensitivity and coping motives for cannabis use have been reported when controlling statistically for concurrent tobacco use6,9; however, enhancement motives have not been related to anxiety sensitivity.6 In order to identify methodological changes that will improve research and further our knowledge about the relationship between concurrent cannabis and tobacco use, we believe it is necessary to examine the limitations of the above‐mentioned research. First, and foremost, previous work has relied upon statistically controlling for tobacco use. Although useful for establishing how anxiety sensitivity and motives relate to cannabis use independently of tobacco use, failure to examine these risk factors explicitly among concurrent tobacco–cannabis users has prevented an understanding of why cigarette smokers use cannabis, and therefore has limited researchers’ ability to tailor smoking cessation 7
treatment to target concurrent cannabis use, a behavior that contributes to cigarette smoking. Second, most research has focused on the individual effects of these variables in relation to cannabis use, which has not allowed researchers to understand how the combination of these variables influences cannabis use (ie, examination of associations in one overarching model). This issue is clinically important, as cannabis use motives are highly related to one another.10,11 Similarly, past work has predominately targeted the main effects of these variables. Such an approach does not take into account how anxiety sensitivity may qualify cannabis use motives effects and may lead to false conclusions that anxiety sensitivity is not related to use motives. As an example, research with tobacco smokers found that while indulgent smoking motives were not related to anxiety sensitivity, the interaction between indulgent smoking motives and anxiety sensitivity significantly predicted cigarette use. Level of cigarette use was related to indulgent motives only for individuals low in anxiety sensitivity.12 Lastly, previous research has focused on frequency of use. Recent research demonstrates that quantity of cannabis use is a stronger predictor of cannabis problems than frequency of use, and thus, may be a more clinically meaningful variable.13 Importantly, quantity estimates for cannabis use have been found to be reliable.13 The purpose of this paper, therefore, was to address key limitations of past work by evaluating both the main and interactional effects of anxiety sensitivity and cannabis use motives among a treatment‐seeking sample of cigarette smokers. We studied a treatment‐seeking sample as we were interested in obtaining data relevant to the refinement of tobacco cessation interventions for concurrent cannabis use. In line with past research, we expected coping, enhancement, social, and expansion motives to exhibit positive relationships with quantity of cannabis use when examining zero‐order correlations. Further, we hypothesized that anxiety sensitivity would moderate the relationships between enhancement motives and quantity of cannabis use, such that enhancement motives would be associated positively with quantity of cannabis use only for low anxiety sensitive individuals. Conversely, we also hypothesized that anxiety sensitivity would moderate the relationship between coping motives and quantity of cannabis use, such that coping motives would be associated positively with quantity of cannabis use for high anxiety sensitive individuals only. Although our main aim of this paper was to explore these relations for quantity of cannabis use, we conducted an exploratory analysis to discover whether the moderation model would be similar for frequency of cannabis use.
METHODS Participants Participants were a subset of daily cigarette smokers, who had used cannabis in the past month (N ¼ 138; Nfemale ¼ 56; aged 18–60 years old; Mage ¼ 29.66; SD ¼ 11.99) and 8
responded to advertisements for an ongoing study examining a smoking cessation intervention that focused on vulnerability to panic in comparison to a standard smoking cessation program. On average, participants had smoked cigarettes for 11.81 years (SD ¼ 11.15) and smoked approximately 15.43 cigarettes per day (SD ¼ 10.53). In addition, participants had smoked cannabis for an average of 10.23 years (SD ¼ 9.69). Ninety‐four percent of the sample (n ¼ 129) had used cannabis in the week prior to being assessed for the current study. Of the 125 individuals who reported on their drinking behavior, 68 (54%) indicated that they drank alcohol more than once a week. Approximately half of participants reported attending part of college, but having no degree (47%), and most were unmarried (82%) and Caucasian (84%). Based on the Fagerström Test for Nicotine Dependence,14 approximately half of the sample was classified with medium to very high nicotine dependence (52%). Sixty‐nine individuals (50%) met criteria for an Axis I disorder based on the Structured Clinical Interview for DSM‐IV Axis I Disorders‐Non‐Patient Edition (SCID‐I/NP).15 Twenty individuals (15%) met criteria for two disorders and 11 individuals (8%) met criteria for three or more disorders. The distribution of disorders were as follows: 21 individuals (15%) met criteria for a cannabis use disorder, 22 individuals (16%) met criteria for a substance use disorder other than cannabis (most commonly an alcohol use disorder, n ¼ 20), 42 individuals (30%) met criteria for an anxiety disorder (most commonly social phobia, n ¼ 19), 18 individuals (13%) met criteria for a mood disorder (most commonly major depression, n ¼ 10), and 2 (1%) were diagnosed with another disorder (anorexia nervosa or body dysmorphic disorder). See Table 1 for descriptive information on predictor and outcome variables. See papers by Johnson and colleagues,16 Capron and colleagues,17 and Peasley‐Miklus and colleagues18 for details of other findings obtained from the treatment‐seeking database that was used for this study. Measures Diagnostic Information The SCID‐I/NP was used to determine clinical diagnoses. Diagnoses obtained with the SCID‐I/NP have been shown to have fair to excellent reliability.19 All SCID‐I/NP interviews were administered by trained research assistants or doctoral level staff, blind to study hypotheses, and supervised by independent doctoral‐level professionals. A random selection of 13% of the SCID‐I/NP interviews were checked by the last author for accuracy. There was 100% agreement between these judgments. Cigarette and Cannabis Use The six‐item Fagerström Test for Nicotine Dependence14 was used as a measure of nicotine dependence severity. The Fagerström Test for Nicotine Dependence exhibits good internal consistency, positive relations with key smoking variables, and high test–retest reliability.14,20 The Smoking
Cannabis Use among Treatment‐Seeking Smokers
January–February 2014
TABLE 1. Means and standard deviations for primary study variables
Variables Frequency of cannabis use Quantity of cannabis use ASI‐3 MMM‐enhancement MMM‐social MMM‐conformity MMM‐coping MMM‐expansion
Anxiety sensitivity status
Total sample (N ¼ 138)
Low (n ¼ 96)
High (n ¼ 42)
M (SD)
M (SD)
M (SD)
4.84 3.26 16.75 17.49 13.45 7.04 11.10 10.84
(2.55) (2.59) (12.01) (4.89) (5.71) (3.64) (5.60) (6.05)
4.81 3.36 10.22 17.20 12.98 6.71 9.95 10.27
(2.59) (2.78) (5.44) (5.04) (5.43) (3.35) (4.90) (5.80)
4.90 3.00 31.67 18.14 14.52 7.79 13.74 12.14
(2.87) (2.15) (9.26) (4.52) (6.25) (4.18) (6.06) (6.46)
t‐Test t(136) ¼ 20, p ¼ .85 t(136) ¼ .14, p ¼ .89 t(53.80) ¼ 13.96, p < .001 t(136) ¼ 1.03, p ¼ .30 t(136) ¼ 1.47, p ¼ .15 t(65.08) ¼ 1.48, p ¼ .15 t(136) ¼ 3.84, p < .001 t(136) ¼ 1.68, p ¼ .09
ASI‐3 ¼ anxiety sensitivity index‐3; MMM ¼ marijuana motives measure.
History Questionnaire21 was used to capture information about cigarette use. Items used for the current study included: “For how many years, altogether, have you been a regular daily smoker” and “Think about your smoking during the last week, how many cigarettes did you smoke in an average day.” The Marijuana Smoking History Questionnaire (MSHQ)22 was used to assess participants’ history and patterns of cannabis use. MSHQ items used in study analyses included, “How many years have you smoked marijuana” and “Think about your smoking during the last week, how much marijuana did you smoke per occasion in an average day.” This latter item was rated on an 8‐point Likert scale (1–8). Scores correspond to pictures depicting increasing sizes of marijuana joints, with 1 indicating the smallest joint and 8 indicating the largest joint. The MSHQ has been used successfully as an indicator of cannabis use in many previous studies.8,22
general anxiety sensitivity.24 Each item is rated on a 5‐point Likert scale ranging from 0 (not at all) to 4 (very much); thus, total scores range from 0 to 72. The Cronbach alpha for the total score in the current study was .92.
Motives The marijuana motives measure (MMM)10 contains 25‐ items and five subscales that assess positive and negative reinforcement motives, each with five items. Response options range from 1 (almost never/never) to 5 (almost always/always) scale. The five subscales and representative items are as follows: Enhancement (because I like the feeling), Coping (to forget my worries), Social (because it helps me enjoy a party), Conformity (to be liked), and Expansion (to understand things differently). The subscales have demonstrated fair to excellent internal consistency and support has been found for the multidimensional nature of measure.11 In the current study, Cronbach alphas for the subscales ranged from .82 (Enhancement) to .94 (Expansion).
Analytic Approach Zero‐order correlations were conducted to examine which motives would be individually related to cannabis use, as well as to discover which theoretically relevant variables would need to be controlled for in the regression analysis. As previous research has consistently found lower levels of educational achievement, being unmarried, psychological disorder, race, gender, age, and years of use to predict cannabis and tobacco use,25–29 these variables were examined for appropriateness of model inclusion before serving as covariates in the study analyses. Educational achievement (no college degree vs. at least a 2‐year college degree), relationship status (married vs. single), race (Caucasian vs. non‐Caucasian), and psychiatric diagnosis (absent vs. present) were dummy coded for appropriate inclusion. Given that the cannabis use variables were overdispersed (the conditional variance exceeded the conditional mean), negative binominal regressions were conducted. As anxiety sensitivity may have a categorical latent structure,30,31 ASI‐3 scores were contrast coded (low vs. high) prior to entry into the model. Following Aiken and West’s32 recommendation to use
Anxiety Sensitivity The Anxiety Sensitivity Index‐3 (ASI‐3)23 is an 18‐item multidimensional instrument defined by three highly related subscales: Physical Concerns, Cognitive Concerns, and Social Concerns. Bifactor analysis has demonstrated that it is appropriate to use the total score to obtain an indication of Norberg et al.
Procedure Participants were recruited at two sites (University of Vermont, Burlington, VT, USA and Florida State University, Tallahassee, FL, USA) for an ongoing treatment study at which identical procedures were executed. The current report is based on secondary analyses of data collected during the study’s baseline assessment session, which took place prior to randomization and the commencement of the intervention. All participants provided informed consent after study procedures were explained. Both universities’ institutional review boards approved the study protocol.
January–February 2014
9
West for following up a significant interaction with linear regression. Lastly, to support our method for dichotomizing individuals into high and low anxiety sensitive groups, three additional negative binomial regressions were conducted for quantity of cannabis use. These models examined ASI‐3 scores as a continuous variable and ASI‐3 scores dichotomized using median and tertile splits.
clinically meaningful rather than standard deviation cut‐offs, individuals were classified as experiencing high anxiety sensitivity if their ASI‐3 scores were classified as being in the dysfunctional range using Jacobson and Truax’s33 formula for calculating clinical significance, c ¼ s 0 M1 þ s 1
M0 s0 þ s1
where M0 and M1 denote the average scores of the nonclinical and clinical samples, respectively, and where s0 and s1 indicate the standard deviations of those means. Normative data provided by Taylor et al.23 (M0 nonclinical M ¼ 12.80, s0 ¼ nonclinical SD ¼ 10.60; M1 ¼ clinical M ¼ 32.60, s1 ¼ clinical SD ¼ 14.3) were used to calculate the cut‐off score for the dysfunctional range. Accordingly, individuals were deemed to be low in anxiety sensitivity if they had ASI‐3 scores 21 and lower, whereas individuals were deemed to be high in anxiety sensitivity if they had ASI‐3 scores 22 and higher. It is important to note that this cut‐off was used since no established categorization for anxiety sensitivity exists in the literature. Taxometric analysis was not used in this study to model groups as sample sizes larger than 300 are needed for this procedure.34 In order to improve interpretation of interaction effects and to reduce multicollinearity, continuous explanatory variables were centered prior to creating interaction terms.35 Since interactive effects are less powerful than main effects and because loss of power occurs when dichotomizing scores,36–38 marginally insignificant interaction terms (p .10) for anxiety sensitivity and motives were kept in the model; p values ranging from .05 to .10 were assessed by their unstandardized effect sizes.39–42 Simple slopes were plotted to allow for visual inspection of the interaction.32 For significant interactions, the significance of the simple slopes was assessed by reference coding categorical variables and reestimating model parameters. This is equivalent to the method presented by Aiken and
RESULTS See Table 2 for zero‐order correlations among studied variables. As expected, enhancement, social, coping, and expansion motives were correlated positively with quantity of cannabis use per occasion in the previous week. Education, age, and marital status were related to quantity of cannabis used and were included as covariates in the regression analysis. Conformity motives were not included in the model because they were not related statistically to quantity of cannabis use. The interaction between anxiety sensitivity level and coping motives was not statistically significant (p ¼ .99, B ¼ .0001, SEB ¼ .01), and it was removed from the model (data not displayed in Table 3). Table 3 demonstrates that when all relevant motives and covariates were included in one model, the interaction between anxiety sensitivity and enhancement motives was related marginally to quantity of cannabis use (p ¼ .10). Examination of the simple slopes demonstrated a disordinal interaction (lines that intersect; see Fig. 1). Enhancement motives were not related to cannabis use for individuals high in anxiety sensitivity (Wald x2 ¼ .06, p ¼ .81, B ¼ .01, SEB ¼ .03), but were related to cannabis use for those low in anxiety sensitivity (Wald x2 ¼ 11.06, p < .001, B ¼ .05, SEB ¼ .02). When the model was repeated for frequency of use, the interactions between anxiety sensitivity level and coping (p ¼ .74, B ¼ .003, SEB
TABLE 2. Zero‐order correlations among theoretically relevant variables Variable
2
3
4
5
1. Age .15 .09 .30 2. Gender .04 .07 3. Race .05 4. Education 5. Marital status 6. Psychiatric disorder 7. Years of cannabis use 8. Frequency of cannabis use 9. Quantity of cannabis use 10. ASI‐3 11. MMM‐enhancement 12. MMM‐social 13. MMM‐conformity 14. MMM‐coping 15. MMM‐expansion
6
.372 .07 .05 .05
7
8
9
.17 .74 .30 .19 .22 .03 .11 .02 .004 .13 .34 .18 .01 .27 .20 .11 .19 .003
10
11
.24 .04 .23 .03 .16 .08 .05 .10 .15 .18 .10 .12 .20 .07 .07 .11 .17 .11 .01 .03 .03 .30 .02 .48 .07 .40 .09
12 .04 .003 .19 .03 .01 .13 .02 .20 .30 .13 .51
13
14
.01 .12 .10 .09 .14 .13 .14 .15 .17 .03 .04 .21 .15 .01 .18 .41 .04 .27 .14 .31 .02 .33 .39 .52 .24
15 .18 .14s .18 .02 .01 .12 .10 .26 .27 .14 .46 .56 .22 .44
N reduced to 133 for years of cannabis use due to missing data. N reduced to 129 for quantity of cannabis use as seven people did not report smoking cannabis in the last week. ASI‐3 ¼ anxiety sensitivity index‐3 (low: 0–21, high: 22þ); MMM ¼ marijuana motives measure. p < .05; p < .01; p < .001.
10
Cannabis Use among Treatment‐Seeking Smokers
January–February 2014
TABLE 3. Negative binomial regression analysis for anxiety sensitivity, motives, and cannabis use
Model Quantity of cannabis use Age Marital status Education ASI‐3 MMM‐enhancement MMM‐social MMM‐coping MMM‐expansion MMM‐enhancement by ASI‐3 Frequency of cannabis use Age Marital status Education Psychiatric disorder ASI‐3 MMM‐enhancement MMM‐social MMM‐coping MMM‐expansion MMM‐conformity
B
SEB
95% CI
Wald x2
p‐Value
.003 .28 .25 .21 .006 .01 .02 .004 .05
.006 .17 .16 .14 .03 .01 .01 .01 .03
.02, .008 .06, .61 .56, .07 .05, .48 .05, .06 .02, .04 .003, .05 .02, .03 .009, .10
.35 2.58 2.40 2.50 .06 .79 3.12 .11 2.73
.56 .11 .12 .11 .81 .38 .08 .74 .10
.004 .25 .10 .04 .15 .04 .01 .04 .002 .04
.004 .12 .11 .04 .09 .01 .01 .009 .008 .01
.01, .004 .008, .49 .31, .11 .04, .12 .03, .33 .02, .06 .03, .008 .02, .05 .01, .02 .07, .01
1.02 4.11 .85 .84 2.79 14.52 1.50 18.30 .09 9.67
.31 .04 .36 .36 .09 .0001 .22
