Abstract
Objectives
This study examined associations between drinking and smoking prior to treatment (biochemically measured at baseline), alcohol and tobacco craving, and biochemical alcohol and tobacco use during the analog trial period.
Methods
We conducted a secondary data analysis of a Randomized Clinical Analog Trial where participants with a Diagnostic and Statistical Manual, Fourth Edition Text Revision (DSM-IV-TR) diagnosis of alcohol dependence, abuse or reported heavy drinking, with a co-occurring DSM-IV-TR diagnosis of nicotine dependence, abuse or reported heavy use, who were not seeking treatment were recruited. A generalized estimation equation model for longitudinal binary outcomes was created (N=34) to determine the predictive effects of baseline tobacco use, alcohol craving and tobacco craving on alcohol use over the four weeks of the trial.
Results
Baseline smoking was significantly (*p<0.05) associated with drinking over time (OR=3.09*), while baseline drinking was associated with smoking (OR=4.17*). Baseline alcohol and tobacco craving were positively associated with smoking over time (OR=3.21* and OR=1.92*, respectively) but were negatively associated with alcohol use over time (OR=0.79* and OR=0.57*, respectively).
Conclusions
Heavier use of either tobacco or alcohol preceding treatment may require more intensive interventions in order to reduce tobacco and alcohol use. Future trials designed to address mechanisms of behavior change in the context of novel treatments could promote a better understanding of the cross-rewarding effects related to the co-use of these substances and lead to the development of more integrated and appropriately intense treatments for individuals with concomitant tobacco and alcohol use disorders.
Keywords: Alcohol, tobacco, co-use, contingency management, heavy drinkers
Introduction
Alcohol and tobacco use frequently co-occur, with past-year alcohol consumption predicting a nearly doubled rate of tobacco use, when compared to alcohol abstainers (McKee & Weinberger, 2013). This combined behavior is concerning, as co-use of alcohol and tobacco multiplies several health risks already associated with the use of each substance alone and contributes to significantly increased mortality, illness, and healthcare costs worldwide (Hurt et al., 1996).
While alcohol and tobacco are independently reinforcing, evidence suggests that combined use leads to a synergistic subjective experience (Littleton, Barron, Prendergast, & Nixon, 2007). Notably, the neural mechanisms behind this cross-reinforcing effect may also allow for ‘off-target’ decreases in use of one substance as a result of abstinence from another. We have previously referred to this as the “cascade-up, cascade-down” hypothesis, where the shared liability and cross-sensitivity that leads to co-addiction can be effectively leveraged to reverse the process using combined treatment approaches. A better understanding of the bi-directional nature of alcohol and tobacco craving and their withdrawal may promote more effective integration of their treatments.
We aimed to determine associations between alcohol and tobacco craving and their objectively verified use at baseline, with alcohol and tobacco use during a four-week, Contingency Management (CM) randomized clinical analog trial (RCaT). Based on previous evidence (Orr et al., 2018), we hypothesized that baseline alcohol use and craving would be positively associated with tobacco use over time and that baseline tobacco use and craving would be positively associated with alcohol use over time. While other research has documented similar findings, such has not used biochemical evidence for both tobacco and alcohol verification.
Methods
Sample
We analyzed a previously reported RCaT (Orr et al., 2018). Participants with a Diagnostic and Statistical Manual, Fourth Edition Text Revision (DSM-IV-TR) diagnosis of alcohol dependence, abuse or reported heavy drinking, with a co-occurring DSM-IV-TR diagnosis of nicotine dependence, abuse or reported heavy use, who were not seeking treatment were recruited. Recruiting non-treatment-seeking participants is a common RCaT methodology and has produced important findings leading to future treatment strategies (McDonell et al., 2017). This methodology was chosen to 1) maintain the highest ethical standards of not providing an untested treatment to individuals who are seeking evidence-based care, 2) provide the most rigorous examination of efficacy, and 3) provide homogeneity in participant motivation to change their use of the substances under investigation. We recruited participants through classified ads and Craigslist advertisements in the northwestern United States. Interested individuals contacted study staff who explained the study and screened for tobacco and alcohol use. Eligible participants were scheduled for an in-person visit where they provided written informed consent prior to participation. The Washington State University Institutional Review Board approved this study.
Study design
Data used for this secondary analysis come from a four-week CM trial with a 2×2 factorial design. After screening 500 potential participants, 34 were identified as eligible. After initial baseline assessments, participants were randomly assigned to one of four conditions 1) non-contingent reinforcement equivalent to CM groups (neither substance targeted) (n = 8), 2) CM for alcohol abstinence (n = 8), 3) CM for smoking abstinence (n = 10), 4) CM for both substances (n = 8). Because no participants in condition four completed the study, and only two attended visit one, this group was dropped from the analyses. Therefore, data used for these analyses comprise only the first three conditions, resulting in 26 participants. For a full description of the methods and primary outcomes please see Orr et al. (2018). Consistent with our previous study, participants had to complete their initial baseline and a first follow-up visit (i.e. baseline) in order to be included in our analyses. In our hypotheses above and our analyses reported below, this is the visit we are using to operationally define baseline.
Outcome measures
Alcohol and tobacco use were objectively determined through urine specimens at baseline and thrice weekly during the 4-week trial. Urine samples were analyzed using Diagnostic Reagents Incorporated (DRI) EtG and cotinine semi-quantitative enzyme immunoassay tests by Absolute Drug Testing, a local independent company. Alcohol use was identified through detection of EtG using a cutoff level of 200 ng/mL. This cutoff level accurately detects 80% of drinking for up to the previous five days (McDonell et al., 2015). Tobacco use was detected via cotinine using a threshold of 100 ng/mL, an established cutoff for detecting cigarette smoking in the previous 5–7 days (Cooke et al., 2008).
Alcohol and tobacco “craving” were assessed at baseline using the Questionnaire of Smoking Urges (QSU-Brief)(Toll, Katulak, & McKee, 2006) and Visual Analog Scales (VAS) respectively, with a 10 cm VAS anchored at 0 (no craving) and 10 (most intense craving possible).
Statistical Analysis
A generalized estimation equation (GEE) model for longitudinal binary outcomes was created to determine the predictive effects of baseline tobacco use, alcohol craving and tobacco craving on alcohol use over the four weeks of the trial. We utilized the logit link from the binomial family, and the exchangeable within-subject covariance matrix for all analyses. Our outcome was alcohol use over time (12 repeated measures) and pre-specified covariates were baseline tobacco use, alcohol craving and tobacco craving.
Similarly, a second GEE was created to access the predictive effects of baseline alcohol use, alcohol and tobacco craving on tobacco use during the trial. In this GEE model, our outcome was tobacco use over time and the pre-specified covariates were baseline alcohol use, alcohol and tobacco craving. In both models, randomized group was included as a control variable.
Following the intention to treat paradigm, missing data were managed consistent with GEE procedures (participants contributing two or more data points to the longitudinal analysis were included in the model). All analyses were completed in Stata 14.2 (StataCorp., College Station, Texas, USA) with an alpha threshold of p < 0.05.
Results
Our sample was primarily Caucasian (79%), with 11% multiracial, and 68% male. Participants’ average age was 36 years (range 20–59). At baseline, 97% of the sample was positive for cotinine and 65% for EtG. The mean EtG value for participants in groups 1, 2, and 3 at baseline (n=26) was 840.62 ng/ml (SD=886.19) and the mean value for cotinine at baseline was 952.49 ng/ml (SD=507.98). The mean cigarette craving at baseline was 3.16 (SD=1.63) and the mean alcohol craving at baseline was 1.25 (SD=1.37). See Table 1 for means (SDs) of alcohol and tobacco use biomarkers and ratings of craving for alcohol and cigarettes at baseline and within the study, by study group. Among those randomized, 50% tested positive for cannabis, 9% for methamphetamine, 3% for methadone, and 6% for opioids.
Table 1.
Alcohol and tobacco use and craving at baseline and within the study, by study group
| Group Assignment | Predictors/Outcomes | Baseline values: M (SD) | Within-Study: M (SD) |
|---|---|---|---|
| NC alcohol, NC tobacco (n=8) | Alcohol use UA (EtG) | 1070.43 (987.67) | 822.21 (701.54) |
| Tobacco use UA (Cotinine) | 1166.39 (579.96) | 895.58 (598.13) | |
| Alcohol Craving | 1.05 (1.80) | 1.06 (1.10) | |
| Tobacco Craving | 2.78 (1.45) | 1.31 (.96) | |
| CM alcohol, NC tobacco (n=8) | Alcohol use UA (EtG) | 873.03 (859.33) | 1001.91 (589.66) |
| Tobacco use UA (Cotinine) | 935.54 (510.72) | 784.13 (396.06) | |
| Alcohol Craving | 1.28 (0.84) | 1.44 (1.12) | |
| Tobacco Craving | 3.04 (1.45) | 1.48 (1.33) | |
| NC alcohol, CM tobacco (n=10) | Alcohol use UA (EtG) | 630.84 (866.98) | 770.86 (810.51) |
| Tobacco use UA (Cotinine) | 794.92 (428.71) | 256.92 (257.04) | |
| Alcohol Craving | 1.38 (1.45) | 1.54 (1.60) | |
| Tobacco Craving | 3.56 (1.94) | 2.42 (2.15) | |
As can be seen in Table 2, participants with a cotinine positive UA at their first post-randomization visit had significantly higher odds of submitting an EtG positive UA over time (OR:3.09, p<0.05). Similarly, participants who submitted an EtG positive UA for their first post-randomization visit had significantly higher odds of submitting a cotinine positive UA over time (OR:4.17, p<0.05).
Table 2.
Longitudinal associations between baseline substance use biomarkers, cravings, treatment group and alcohol and tobacco outcomes.
| Odds Ratio | SE | 95% CI | ||
|---|---|---|---|---|
| Outcome: Alcohol use over time | ||||
| Positive Tobacco UA | 3.087* | 1.381 | 1.285 | 7.417 |
| Cigarette Craving | 0.794* | 0.081 | 0.650 | 0.970 |
| Alcohol Craving | 0.568* | 0.066 | 0.452 | 0.714 |
| CM alcohol, NC tobacco | 0.236* | 0.104 | 0.100 | 0.560 |
| NC alcohol, CM tobacco | 0.368* | 0.171 | 0.148 | 0.913 |
| Outcome: Tobacco use over time | ||||
| Positive Alcohol UA | 4.170* | 2.208 | 1.478 | 11.771 |
| Cigarette Craving | 3.207* | 1.109 | 1.628 | 6.317 |
| Alcohol Craving | 1.917* | 0.576 | 1.064 | 3.453 |
| CM alcohol, NC tobacco | 0.059* | 0.064 | 0.007 | 0.495 |
| NC alcohol, CM tobacco | 0.003* | 0.004 | 0.000 | 0.041 |
Note: The intervention group NC alcohol, NC tobacco has no reported parameters because it is the reference group. CM, contingency management; NC, noncontingent control; SE, standard error; CI, 95% confidence interval; UA, urinalysis.
p<0.05
Participant’s tobacco and alcohol craving at their first post-randomization visit were each associated with a significant increase in the odds of using tobacco during the study (tobacco craving OR:3.21; alcohol craving OR:1.92, p<0.05 for both). Inversely, post-randomization baseline tobacco and alcohol craving were each associated with a significant decrease in the odds of using alcohol over time (tobacco craving OR:0.79; alcohol craving OR:0.57, p<0.05 for both). Thus, higher craving for tobacco and alcohol at baseline indicated that the participant was more likely to smoke and less likely to drink alcohol during the study.
Discussion
Our objective was to examine possible crossover associations between alcohol and tobacco co-use by exploring the relationship between baseline alcohol use and craving on tobacco use, and baseline tobacco use and craving on alcohol use, during a 2×2 CM RCaT study targeting concurrent smoking and drinking. Consistent with our hypothesis, our findings point to the predictive effect of baseline tobacco use on future alcohol use and baseline alcohol use on future tobacco use among non-treatment seeking smokers with an alcohol use disorder. To our knowledge, this brief report is the first to examine such relationships using longitudinal biochemical outcomes. Our results fortify previous findings which were potentially limited due to the use of self-report measures alone (Schlauch, Crane, Connors, Dearing, & Maisto, 2019).
As expected, based on previous evidence about the associations between craving and relapse (Franken, 2003), baseline tobacco and alcohol craving were associated with future tobacco use. In contrast, different from previous evidence and our initial hypothesis, tobacco and alcohol craving were associated with less alcohol use over time. These findings suggest that alcohol and tobacco craving may play a more prominent role on tobacco use than alcohol use.
These results should be considered within the limitations of the study. First, we had a small sample size, which is characteristic of RCaTs. Second, the trial lasted for four weeks, substantially shorter than most substance use disorder trials.
Conclusion
Tobacco and alcohol use disorders frequently co-occur producing devastating effects worldwide (Hurt et al., 1996). Although the current study was associational and did not explicitly investigate mechanisms of change, it is possible that the same neural substrates underlying the cross-rewarding effects of alcohol and tobacco use may be involved in motivating an individual to use tobacco in an effort to alleviate cravings. Therefore, these crossover associations hold the promise that decreases in use of one substance may correspond with decreases in use of the other and vice-versa, if new interventions deliberately leverage findings such as these. This requires careful, specific targeting and timing when leveraging the behavioral and biochemical links, working on the “cascade-down” component of alleviating problematic co-use. Future trials should explore use of higher intensity or behavior shaping interventions for smokers and drinkers entering treatment with higher cravings for alcohol or tobacco or who are positive for use of either substance. In addition, trials designed to address mechanisms of behavior change in the context of novel treatments will promote a better understanding of the cross-rewarding effects related to the co-use of these substances and lead to the development of more integrated and appropriately intense treatments for individuals with concomitant tobacco and alcohol use disorders.
Disclosures and Acknowledgments
Drs. McPherson and Roll have received research funding from the Bristol-Myers Squibb Foundation. Dr. McPherson has also received research funding from Orthopedic Specialty Institute, and consulted for Consistent Care company. This funding is in no way related to the investigation reported here.
This original study was supported by the following grants: P20 MD006871 (McPherson and Buchwald); Alcohol and Drug Abuse Research Program (McPherson). All authors contributed in a significant way to the manuscript and all authors have read and approved the final manuscript.
None of the other authors have any financial, personal, or other type of relationship that would cause a conflict of interest that would inappropriately impact or influence the research and interpretation of the findings.
Financial Support: This project was supported by the following grants: P20 MD006871 (McPherson and Buchwald); Alcohol and Drug Abuse Research Program (McPherson).
References
- Cooke F, Bullen C, Whittaker R, McRobbie H, Chen MH, & Walker N (2008). Diagnostic accuracy of NicAlert cotinine test strips in saliva for verifying smoking status. Nicotine Tob Res, 10(4), 607–612. [DOI] [PubMed] [Google Scholar]
- Franken IH (2003). Drug craving and addiction: integrating psychological and neuropsychopharmacological approaches. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 27(4), 563–579. [DOI] [PubMed] [Google Scholar]
- Hurt RD, Offord KP, Croghan IT, Gomez-Dahl L, Kottke TE, Morse RM, & Melton LJ 3rd. (1996). Mortality following inpatient addictions treatment. Role of tobacco use in a community-based cohort. JAMA, 275(14), 1097–1103. [DOI] [PubMed] [Google Scholar]
- Littleton J, Barron S, Prendergast M, & Nixon SJ (2007). Smoking kills (alcoholics)! Shouldn’t we do something about it? Alcohol and alcoholism, 42(3), 167–173. [DOI] [PubMed] [Google Scholar]
- McDonell MG, Skalisky J, Leickly E, McPherson S, Battalio S, Nepom JR, … Ries RK (2015). Using ethyl glucuronide in urine to detect light and heavy drinking in alcohol dependent outpatients. Drug and alcohol dependence, 157, 184–187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- McDonell MG, Skalisky J, Leickly E, Orr MF, McPherson S, Roll J, … Javors M (2017). Pilot investigation of a phosphatidylethanol-based contingency management intervention targeting alcohol use. Psychology of Addictive Behaviors, 31(5), 608. [DOI] [PMC free article] [PubMed] [Google Scholar]
- McKee SA, & Weinberger AH (2013). How can we use our knowledge of alcohol-tobacco interactions to reduce alcohol use? Annual Review of Clinical Psychology, 9, 649–674. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Orr MF, Lederhos Smith C, Finlay M, Martin SC, Brooks O, Oluwoye OA, … Barbosa-Leiker C (2018). Pilot investigation: randomized-controlled analog trial for alcohol and tobacco smoking co-addiction using contingency management. Behavioural Pharmacology, 29(5), 462–468. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schlauch RC, Crane CA, Connors GJ, Dearing RL, & Maisto SA (2019). The role of craving in the treatment of alcohol use disorders: The importance of competing desires and pretreatment changes in drinking. Drug and alcohol dependence, 199, 144–150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Toll BA, Katulak NA, & McKee SA (2006). Investigating the factor structure of the Questionnaire on Smoking Urges-Brief (QSU-Brief). Addictive behaviors, 31(7), 1231–1239. [DOI] [PMC free article] [PubMed] [Google Scholar]