Concurrent Alcohol and Tobacco Treatment: Effect on Daily Process Measures of Alcohol Relapse Risk

Abstract

Objective

The aim of this study was to compare the effects of alcohol treatment along with concurrent smoking treatment or delayed smoking treatment on process measures related to alcohol relapse risk.

Method

Alcohol dependent smokers (N = 151) who were enrolled in an intensive outpatient alcohol treatment program and were interested in smoking cessation were randomized to a concurrent smoking cessation (CSC) intervention or to a waiting list for delayed smoking cessation (DSC) intervention scheduled to begin three months later. Daily assessments of relapse process measures were obtained using an Interactive Voice Response (IVR) system for 12 weeks after the onset of smoking treatment in the CSC condition, and before beginning smoking treatment in the DSC condition. Smoking outcomes were assessed at 2 and 13 weeks after starting treatment.

Results

Seven-day CO-verified smoking abstinence in the CSC condition was 50.5% at 2 weeks and 19.0% at 13 weeks compared to 2.2% abstinence at two weeks and 0% abstinence at 13 weeks for those in the DSC condition. Drinking outcomes were not significantly different for CSC vs. DSC treatment conditions. On daily IVR assessments, CSC participants had significantly lower positive alcohol outcome expectancies relative to DSC participants. Multilevel modeling (MLM) analyses of within-person effects across the 12 weeks of daily monitoring showed that daily smoking abstinence was significantly associated with same day reports of lower alcohol consumption, lower urge to drink, lower negative affect, lower positive alcohol outcome expectancies, greater alcohol abstinence self-efficacy, greater alcohol abstinence readiness to change, and greater perceived self-control demands.

Conclusions;

Analyses of process measures provide support for recommending smoking intervention concurrent with intensive outpatient alcohol treatment.

Public Health Significance Statement

Study results support conveying a message to alcohol dependent smokers that smoking abstinence is accompanied by favorable changes in alcohol use, craving, mood, confidence, and motivation.

Although cigarette smoking prevalence among U.S. adults has declined to 19.5% (King, Dube, & Tynan, 2012), the majority of individuals with substance use disorders remain current smokers (Compton, Thomas, Stinson, & Grant, 2007). The negative health consequences of smoking among those with alcohol use disorders are substantial (Hurt et al., 1996). The Department of Health and Human Services Clinical Practice Guideline for Treating Tobacco Use and Dependence recommended that smokers receiving treatment for chemical dependency should be provided smoking cessation treatments including both counseling and pharmacotherapy (Fiore et al., 2000). However, the Guideline recommendation has not led to significant changes in clinical practice. Addiction treatment programs rarely provide tobacco treatment (Fuller et al, 2007; Gifford, Tavakoli, Wang, Hagedorn, & Hamlett-Berry, 2013; Hunt, Cupertino, Garretta, Friedman, & Richter, 2012; Knudsen, Studts, Boyd, & Roman, 2010). An analysis of the reasons for failure to treat tobacco use in addiction programs (Prochaska, 2010) focused on provider beliefs including assumptions that tobacco is less harmful than other drugs, and that continued tobacco use supports abstinence from other substances.

Studies examining the effects of concurrent smoking and substance treatment on alcohol and drug outcomes have not yielded consistent findings. Some studies reported that concurrent smoking cessation and substance use treatment did not significantly affect alcohol and drug outcomes (Burling, Burling, & Latini, 2001; Burling, Marshall, & Seidner, 1991; Carmody et al., 2012; Hurt et al., 1994; Kalman et al., 2001; Nieva, Ortega, Mondon, Ballbè, & Gual, 2011; Reid et al., 2008) while others report smoking treatment was associated with improved alcohol or drug outcomes (Bobo, Mcilvain, Lando, Walker, & Leed-Kelly, 1998; Winhusen et al., 2014). A meta-analysis of concurrent substance and tobacco treatment trials concluded that smoking interventions were associated with a 25% increased probability of long-term alcohol and drug abstinence (Prochaska, Delucchi, & Hall, 2004). However, a large trial comparing concurrent versus delayed smoking intervention found worse drinking outcomes in the concurrent smoking treatment group (Joseph, Willenbring, Nugent, & Nelson, 2004). In light of the conflicting outcome research literature, the 2008 Update of the DHHS Clinical Practice Guideline called for additional research on the impact of tobacco treatment concurrent with chemical dependency treatment (Fiore et al., 2008). Considering the number of outcome studies already completed, a study of treatment process rather than outcome could advance understanding and clear up misconceptions about the impact of concurrent smoking cessation and substance treatment. To our knowledge, no research to date has examined the impact of concurrent smoking and alcohol treatment on measures of processes thought to underlie alcohol relapse.

The present study was designed to examine the impact of concurrent smoking cessation treatment on process measures reflecting alcohol relapse risk during a three-month window after the start of the alcohol treatment. A control condition delayed smoking treatment for three months to allow a time window for comparison of participants who received alcohol treatment with and without concurrent smoking treatment. This study employed intensive longitudinal assessment methodology using daily interactive voice response (IVR) technology across the three-month follow-up period.

One limitation of previous studies of concurrent alcohol-tobacco treatment is that in many cases the smoking interventions used were no more effective than the control conditions (Bobo et al., 1998; Burling et al., 1991; Joseph et al., 1993; Kalman et al., 2001; Reid et al., 2008). Only very intensive long-term smoking cessation treatments have produced follow-up smoking abstinence rates near 20% (Burling et al., 2001; Carmody et al., 2012). A meta-analysis of tobacco and substance treatment (Prochaska et al., 2004) found significant treatment effects on smoking outcomes at posttreatment but not at follow-up. A low smoking quit rate in a concurrent treatment trial provides little opportunity to demonstrate the impact of smoking abstinence on alcohol and drug outcomes. The present study employed a smoking cessation intervention with contingency management (Higgins, Silverman, & Heil, 2008), evidence based behavioral counseling, and combination nicotine replacement therapy (NRT; Cooney et al., 2009) to maximize smoking quit rates.

A cognitive-behavioral model provided guidance for selecting process measures reflecting alcohol relapse risk. This model (Marlatt & Witkiewitz, 2005) identifies negative affect, low abstinence self-efficacy, positive drinking outcome expectancies, and craving as important proximal determinants of alcohol relapse. Low motivation for abstinence (DeClemente, 1999; Prochaska, DiClemente, & Norcross, 1992) and low self-control strength (Muraven & Baumeister, 2000) have also been suggested as determinants of relapse. These six key alcohol relapse risk factors were the focus of the present study.

Method

Overview

Individuals with both alcohol and tobacco use disorders were randomized to intensive alcohol treatment and concurrent smoking cessation intervention or to intensive alcohol treatment with the smoking intervention delayed until three months after alcohol treatment. Intensive longitudinal assessment examined process measures reflecting alcohol relapse risk during the three months after the onset of smoking treatment in the concurrent treatment group, and before onset of smoking treatment in the delayed smoking treatment group. We tested the hypothesis that concurrent and delayed smoking cessation conditions would have differential impact on alcohol relapse risk factors.

Participants

Participants were 151 individuals enrolled in a three-week substance use intensive outpatient treatment program at two Department of Veterans Affairs (VA) clinics. From May 2009 to October 2012, consecutive admissions to these clinics were screened and recruited. To increase female participation rates, women who were not military veterans were recruited from the community via newspaper, radio, and internet advertisements and enrolled in the trial. Inclusion criteria included DSM-IV diagnosis of current alcohol abuse or dependence with reported alcohol use in the past 30 days, current cigarette smoking rate of one or more cigarettes per day, minimum three-year smoking history, desire to quit smoking, and agreement to accept random assignment to either concurrent smoking cessation treatment or to delay smoking cessation treatment until three months after intensive substance use treatment. Exclusion criteria included allergy or hypersensitivity to nicotine or to adhesives used in transdermal delivery systems, body weight less than 100 pounds, severe generalized skin disorder, active peptic ulcer, uncontrolled hypertension, unstable insulin-dependent diabetes, history of clinically significant cardiovascular disease, homelessness or lack of a stable residence, need for detoxification treatment of alcohol withdrawal, presence of a medical or psychiatric condition that would make participation in this study hazardous, pending legal action that would inhibit participation in the study, pregnancy or lactation in females, females of childbearing age who are not practicing a medically accepted form of contraception (e.g., abstinence, oral contraceptives, diaphragm with spermicide, IUD, condoms with spermicide), or cannabis use with unwillingness to abstain from cannabis for study duration. Participants were not excluded if they had current stable use of prescribed opiate or benzodiazepine medications. All participants provided written informed consent, and the study was approved by institutional review boards at the VA and the affiliated universities. Treatment sites were equivalent with respect to baseline levels of cigarettes per day. Site differences were found on reported baseline heavy drinking days (proportion heavy drinking days = .63 versus .48) and abstinent days (proportion days abstinent = .32 versus .46).

Table 1 shows baseline demographic, smoking history, drinking history, and alcohol relapse process variables by treatment conditions. For 24 of the 26 variables examined, randomization was successful in balancing the treatment groups. The exception was that participants in the concurrent treatment condition were significantly more likely to meet criteria for other substance use disorders and to report a history of more previous alcohol treatments compared with those randomized to the deferred treatment condition.

Table 1.

Baseline Variables across Treatment Groups

Variable	Total N = 151 n (%)	Deferred n = 46 n (%)	Concurrent n = 105 n (%)	p
Women	21 (13.9)	7 (15.2)	14 (13.3)	.76
Married	16 (10.6)	5 (10.9)	11 (10.5)	.94
Employed or student	25 ((16.6)	10 (21.7)	15 (14.3)	.26
Military veteran	137 (90.7)	41 (89.1)	96 (91.4)	.65
Grade or high school only	91 (60.3)	28 (60.9)	63 (60.0)	.92
Race/Ethnicity
Hispanic	18 (11.9)	3 (6.5)	15 (14.3)	.18
White	82 (54.3)	25 (54.3)	57 (54.3)	.74
African American	63 (41.7)	20 (43.5)	43 (41.0)
Other race	6 (4.0)	1 (2.2)	5 (4.8)
Other substance use disorder
None (alcohol only)	79 (52.3)	30 (65.2)	49 (46.7)	.04
Marijuana	25 (16.6)	5 (10.9)	20 (19.0)	.21
Cocaine	50 (33.1)	14 (30.4)	36 (34.3)	.64
Opiate	11 (7.3)	2 (4.3)	9 (8.6)	.36

	Mean (SD)	Mean (SD)	Mean (SD)	p
Age	49.1 (8.95)	48.1 (8.68)	49.5 (9.07)	.37
FTND	5.2 (2.1)	5.3 (2.0)	5.1 (2.2)	.64
Previous smoking quit attempts	7.1 (15.4)	5.5 (13.6)	7.8 (16.2)	.41
Cigarettes smoked/day	16.2 (8.7)	16.8 (10.0)	16.0 (8.0)	.61
Baseline CO (ppm)	20.6 (13.4)	21.9 (13.6)	20.1 (13.4)	.74
EDSS	38.8 (11.1)	38.8 (12.5)	38.8 (10.5)	.99
Previous alcohol treatments	5.5 (10.4)	3.2 (3.2)	6.5 (12.1)	.01
Days since last drink	14.2 (13.1)	16.8 (15.3)	13.1 (12.0)	.11
Proportion days heavy drinking	.56 (.35)	.52 (.38)	.58 (.33)	.40
Penn Alcohol Craving Scale	8.7 (6.4)	8.7 (6.9)	8.8 (6.1)	.93
CES-D	16.9 (10.4)	15.4 (8.9)	17.6 (11.0)	.24
AASE confidence to abstain	70.1 (19.5)	68.8 (22.9)	70.7 (17.9)	.61
AEQ-68 total positive	44.8 (13.6)	43.7 (15.5)	45.3 (12.8)	.52
RTCQ-TV readiness score	19.3 (9.1)	18.3 (9.7)	19.7 (8.9)	.39
Self Control Scale	39.1 (8.0)	39.1 (8.2)	39.1 (8.0)	.98

Note. FTND = Fagerström Test for Nicotine Dependence. CO (ppm) = carbon monoxide (parts per million). EDSS = Ethanol Dependence Syndrome Scale. CES-D = Center for Epidemiologic Studies – Depression scale. AASE = Alcohol Abstinence Self-Efficacy scale. AEQ = Alcohol Expectancy Questionnaire. RTCQ-TV = Readiness to Change Questionnaire – Treatment Version.

Procedure

Screening interviews included the substance use sections of the Structured Clinical Interview for DSM-IV Axis I Disorders, Patient edition, version 2.0 (SCID-I/P; First, Spitzer, Gibbon, & Williams, 1996). Next, a physician or an advanced practice nurse obtained a targeted history and physical examination and laboratory tests. Women received a urine pregnancy test. Eligible participants were randomized to Concurrent Smoking Cessation (CSC) or Delayed Smoking Cessation (DSC) treatments using an urn randomization computer program (Stout, 1994) that balanced the two groups on baseline cigarette craving, alcohol self-efficacy, alcohol dependence, nicotine dependence and gender. Participants were randomized in an unbalanced ratio of 2:1 with more participants assigned to CSC. This was done to increase the number of participants who could be assessed at follow-up while abstinent from cigarettes.

Figure 1 shows a timeline for study procedures. Baseline assessments were completed in the first week of the three-week substance use program. At the end of this week, CSC participants began the smoking intervention. A two-week follow-up session was scheduled on the day before intensive outpatient program discharge, and another follow-up assessment session was scheduled 13 weeks after the start of intensive treatment. Note that the 13-week assessment was conducted at the end of CSC treatment and before the start of DSC treatment with the DSC condition acting as a waiting list control. Breath alcohol was assessed prior to these assessments with a requirement of a negative alcohol level for the assessment to proceed. Participants were paid $25 for the intake assessment, $25 for the 2-week assessment, and $50 for the 13-week assessment.

Figure 1.

Procedure timeline.

Interventions

Behavioral intervention

All participants were enrolled in a three-week intensive outpatient alcohol and drug treatment program meeting five days per week for five hours per day. Substance use treatment was primarily group-based and incorporated cognitive-behavioral coping skills training, motivation enhancement therapy and twelve step support. All participants also received an intervention consisting of evidence-based behavioral tobacco counseling (Fiore et al., 2008) combined with contingency management.

CSC treatment was provided in twelve 15-minute individual treatment sessions (180 minutes total time), delivered twice daily at the beginning and end of the substance use intensive outpatient treatment days. Smoking intervention fidelity was maintained with the use of a detailed treatment manual, centralized therapist supervision across sites, and a therapist checklist of handouts distributed to participants at sessions. Sessions 1–4 included motivational and behavioral strategies in preparation to quit, with the targeted quit date scheduled for session five. Sessions 5–12 were focused upon coping with nicotine withdrawal and relapse prevention as well as a contingency management (CM) procedure to reward smoking abstinence (Cavallo et al., 2007). CO levels were assessed at the beginning of the twice daily individual treatment sessions. Declining CO levels were rewarded on the target quit date until levels reached 5 parts per million (ppm), and then levels of 5 ppm or less were rewarded with a progressive monetary incentive starting at $5, increasing $0.50 for each consecutive negative CO, with a reset back to $5 for the first negative CO after any tobacco lapse. Using this protocol participants were eligible for up to a maximum of $140 voucher rewards for eight measured days of tobacco abstinence. Vouchers could be exchanged for gift cards at local stores.

Participants randomized to the DSC condition received the same intensive outpatient alcohol and drug treatment delivered immediately after enrollment and were assessed across three months, but the smoking cessation intervention was delayed until after the three-month study assessment period was complete. The DSC condition can be seen as a wait list control group with respect to evaluation of the smoking intervention. The DSC smoking intervention was modified to be administered in a non-intensive outpatient context. The twelve daily 15-minute smoking CSC intervention sessions were compressed to six 30-minute sessions scheduled twice weekly to make it feasible for participants to attend DSC smoking treatment sessions when they were no longer coming to the clinic daily.

Pharmacotherapy

Participants in both conditions received open-label nicotine patch for eight weeks in combination with up to twelve weeks of open-label 2 mg nicotine gum or 2 mg nicotine lozenge. Nicotine patch dose for participants smoking more than 10 cigarettes per day prior to quit date was 21 mgs for 28 days, 14 mgs for 14 days, then 7 mgs for 14 days. The recommended dosage of nicotine gum was at least 6 but not more than 20 pieces to be taken as needed in the first 6 weeks, and up to 20 pieces per day as needed in weeks 6 to 12. The maximum recommended dosage of nicotine lozenge was 15 pieces per day. Light smokers (those who smoked 10 or fewer cigarettes per day) were prescribed lower dose nicotine patches and were instructed to use the gum or lozenge on a purely as needed basis. Nicotine lozenges were prescribed for individuals with temporomandibular joint disease or dental problems and for individuals who expressed a preference for lozenge over gum. Combination NRT was chosen based on research showing combined patch and gum treatment led to better smoking cessation outcomes than patch used alone in the context of concurrent alcohol-tobacco treatment (Cooney et al., 2009).

Baseline Measures

Questionnaire measures of alcohol and tobacco relapse risk factors (craving, self-efficacy, outcome expectations, negative affect, motivation for abstinence, and self-control demands) were administered at baseline. These questionnaires were obtained to provide measures to control for baseline between-participants effects in the analysis of the daily process measures. The 20-item confidence scale from the Alcohol Abstinence Self-Efficacy scale (AASE; DiClemente, Carbonari, Montgomery, & Hughes, 1994) was used to obtain a measure of confidence to abstain from drinking (internal reliability α=.96). The revised 43-item version of the Relapse Situation Efficacy Questionnaire (RSEQ; Gwaltney et al., 2001) was used to assess confidence to abstain from smoking (internal reliability α=.92). A measure of alcohol outcome expectancies were obtained by summing items from the 68-item version of the Alcohol Expectancy Questionnaire (AEQ; Goldman, Greenbaum, & Darkes, 1997; internal reliability α=.94). Smoking outcome expectancies were assessed with the Brief Smoking Consequences Questionnaire – Adult (SCQ-A; Rash & Copeland, 2008), a 25-item questionnaire that yielded nine factors reflecting positive and negative effects of smoking. A positive expectancies score was calculated by subtracting the negative expectancies scale scores from the sum of the positive expectancies scale scores (scale scores correlated r=−.14; positive expectancies composite internal reliability α=.70). An alcohol craving score was derived from the five-item Penn Alcohol Craving Scale (PACS; Flannery, Volpicelli, & Pettinati, 1999; internal reliability α=.88). Craving for cigarettes was assessed with the 10-item Questionnaire of Smoking Urges – brief (QSU; Cox, Tiffany, & Christen, 2001; internal reliability α=.88). Motivation to stop drinking was assessed with the 15-item Readiness to Change Questionnaire Treatment Version (RTCQ-TV; Heather, Luce, Peck, Dunbar, & James, 1999). As described by Fosberg, Ekman, Halldin, and Ronnberg (2004), a continuous readiness to change score was calculated from this measure by adding Contemplation and Action scores and subtracting (the negatively correlated) Precontemplation scores (internal reliability for composite α=.72). Motivation to stop smoking was assessed with the Perceived Risks and Benefits of Cessation scale (PRBC; McKee et al., 2005). A motivation score was calculated from this scale by subtracting the risk scale scores from the sum of the cessation benefits scale scores. (Risk and Benefit scores were correlated r=−.03; internal reliability for Benefits composite score α=.89). Depressed mood or negative affect was assessed using the 20-item Center for Epidemiologic Studies – Depression Scale (CES-D; Radloff, 1977; internal reliability α=.90). Trait self-control was measured using the Brief Self-Control Scale (BSCS; Tangney, Baumeister, & Boone, 2004), a 13-item scale yielding a single score reflecting cognitive, affective, and behavioral aspects of trait self-control strength (internal reliability α=.90).

Daily Process Assessment

Daily process assessments were obtained using an IVR system, with data collection beginning 4 to 7 days after admission to the intensive outpatient substance use program and continuing after program discharge for a total of 12 weeks. Participants used the telephone numeric keypad to respond to questions asked by a recorded voice. System hardware and software were provided and supported by Telesage, Inc., of Chapel Hill, NC. Participants were provided a coded identification number and instructed to call the system at a dedicated toll-free line once per day between the hours of 5:00 and 8:00 PM. Those who did not call by 8:00 PM were called by the system and reminded to complete the interview. A research assistant monitored call compliance and addressed problems. Participants who did not have access to a telephone were provided a cellular telephone that was set up to call only the IVR system or the VA clinic. IVR call time averaged 5.0 (SD = 1.8) minutes per call and participants were compensated for participating in these IVR calls at a rate of $1 per call with a bonus of $7 extra compensation for participating in all seven calls in one week. The bonus increased by $1 per week for each week in which they participated in all seven calls. A perfect record of call participation across all twelve weeks resulted in $234 in compensation. Compensation for participation in IVR telephone calls was not dependent on whether or not participants reported abstinence or continued use of alcohol or tobacco.

Before beginning IVR assessments, participants were instructed in the definition of a standard drink unit for beer, wine and liquor. Daily drinking and smoking were measured by asking how many standard drinks and cigarettes the person had all day yesterday and how many today prior to the IVR call. The remaining items were rated to reflect “how you’ve been feeling in the past 24 hours using a scale from 1 to 5 where 1 = “not at all” and 5 = “very much.” Alcohol and tobacco craving were measured with single items “felt urge to drink” and “felt urge to smoke.” This single-item urge rating measure was reported to have high test-retest reliability (Cooney, Litt, Morse, Bauer, & Gaupp, 1997), and the smoking urge item was a significant momentary predictor of smoking relapse (Cooney et al., 2007; Holt, Litt, & Cooney, 2012). Affect was measured with 5-point ratings on eight items derived from the circumplex model of mood experience (Larsen & Diener, 1992) in which mood state was classed along two dimensions: pleasantness (negative vs. positive mood) and activation (low vs. high). Four quadrants of mood were created, all with internal reliabilities exceeding .80: positive high activation (active; peppy); positive low activation (quiet; relaxed); negative high activation (anxious or nervous; angry irritable or frustrated); and negative low activation (depressed or sad; bored). The negative high activation scale was a significant momentary predictor of smoking relapse (Cooney et al., 2007). For the purposes of the present study we constructed a single negative mood score computed from the sum of the four negative mood items (internal reliability α =.88).

Smoking outcome expectancies were measured using the method described by Gwaltney, Shiffman, Balabanis, and Paty (2005) using seven items derived from the Smoking Consequences Questionnaire (Copeland, Brandon, & Quinn, 1995; internal reliability α=.86). A four-item scale was created to measure drinking outcome expectancies summing three items from the Alcohol Expectancy Questionnaire selected based on their high factor loadings on Global Positive (“drinking makes the world seem brighter”), Physical and Social pleasure (“some alcohol has a pleasant taste”), and Social Assertiveness (“drinking makes it easier to talk to people”) subscales (Rubio, Bucholz, Neuman, & Rauch, 2003) plus one reverse-scored item from the Negative Alcohol Expectancy Questionnaire (Jones & McMahon, 1994; “if I drink I will feel depressed or anxious”). (Internal reliability of this scale was only α=.31, however). Also guided by Gwaltney et al. (2005), smoking and drinking abstinence self-efficacy were assessed with single items “felt confident that you can keep from smoking” and “felt confident that you can keep from drinking.” Gwaltney et al. found that these measures of smoking outcome expectancies and self-efficacy predicted smoking lapse. Other researchers also report that this measure of smoking self-efficacy predicted smoking relapse (Hot et al., 2012) and report that a similar measure of alcohol abstinence self-efficacy predicted alcohol relapse (Cooney et al., 2007).

Motivation to quit drinking and to quit smoking were assessed using single items (“felt motivated to quit drinking” and “felt motivated to quit smoking”). The smoking motivation item prospectively predicted the outcome of a smoking cessation attempt in a previous momentary assessment study (Turner, Mermelstein, & Flay, 2004). Perceived self-control demands were measured using four items developed by Muraven, Collins, Shiffman, and Paty (2005) for once-daily assessment. Participants rated whether they had to regulate their mood, control their thoughts, deal with stress, or felt overwhelmed that day. When combined, these items form a scale (α = .75) that was found to predict, on a daily basis, when participants tended to drink more alcohol, were more intoxicated, and were more likely to report violating a personal limit on alcohol intake (Muraven et al., 2005).

The 9-item version of the Minnesota Nicotine Withdrawal Scale-Revised (MNWS-R; Hughes, 2007; Hughes & Hatsukami, 1986) was included to measure the severity of nicotine withdrawal symptoms (anger, anxiety, craving, depression, difficulty concentrating, hunger, impatience, insomnia, and restlessness). Symptoms were rated on 0 to 4 severity scales, and the scale was scored to provide the mean score across all items (α=.85).

Outcome Measures

Alcohol, drug, and tobacco frequency-of-use data for the 90 days prior to intake and follow-up were assessed using the Form-90 (Miller & DelBoca, 1994), a calendar-based time-line follow-back survey. The primary drinking outcome was self-reported Proportion Days Heavy Drinking (PDH) assessed at the 13-week follow-up. The three month calendar data were divided into three 4-week measurement periods to allow assessment of change in drinking across time. Heavy drinking days were defined as ≥ 6 standard drinks per day for men and ≥ 4 standard drinks per day for women. Proportion Days Abstinent (PDA) for each time interval was also assessed. The primary tobacco outcome was self-reported 7-day point prevalence smoking abstinence verified by breath CO less than 10 ppm, measured at 2 and 13 weeks. Cases in which smoking outcome data were missing were treated as smoking. Nine participants (6%) were missing at 2 weeks and 29 participants (19%) were missing at 13 weeks. Among participants reporting smoking abstinence, 54 out of 54 were verified abstinent by breath CO at 2 weeks and 20 out of 27 were verified abstinent at 13 weeks.

Data Acquisition and Analysis

All analyses were conducted using IBM SPSS (v. 20). Smoking outcomes were analyzed using chi-square analyses of abstinence status at weeks 2 and 13. PDA and PDH drinking outcomes were evaluated with multilevel modeling analyses (MLM; using the SPSS Linear Mixed procedure) using maximum likelihood estimation. MLM procedures used in this way have advantages over repeated measures analyses, including their ability to use all the data available (e.g., Schwartz & Stone, 1998). In analyses of PDH and PDA all terms were treated as fixed. The level 1 (within-person) variable Time was measured in three monthly periods and treated as a repeated factor (due to the fixed time period between estimates). Treatment condition (DSC v. CSC) was treated as a level 2 (between-persons) variable. The general equation for these analyses may be represented as:

$${\text{Drinking}}_{ij}={\mathrm{\beta }}_{\text{Between}}{\text{TREATMENT}}_i+{\mathrm{\beta }}_{\text{Witihin}}{\text{TIME}}_{ij}+{\mathrm{\alpha }}_i+{\mathrm{\epsilon }}_{ij};$$

where β is an unstandardized coefficient. Drinking for each individual i at each time point j is a function of treatment and time measured j times for each person, plus the intercept α_i and the error for each person at each time point ε_ij.

Participants completed 68% of all scheduled Daily Process IVR calls, and 83% of the sample completed more than 20% of scheduled calls. Those completing less than 20% of scheduled IVR calls (n = 26) were dropped from the analyses, leaving n = 125 for analyses. Those who were excluded were not significantly different from those included on baseline measures of drinking, alcohol dependence, or smoking. The excluded participants scored higher on baseline nicotine dependence [mean FTND Scores: 6.4 (SD=1.6) v. 4.9 (SD=2.1); F (1, 148) =10.13; p < .01].

Daily process data were analyzed for each of the risk factors of interest (i.e., daily drinking or drug use, risk factors for relapse such as urge to smoke) using separate multilevel mixed model regression analyses for each outcome, with random effects and maximum likelihood estimation. Day of assessment (centered), treated as a continuous variable, was the level 1 variable and participants was the level 2 factor. Day of assessment was centered to allow interpretation of average effects at the midpoint of the 12 weeks, rather than at the beginning, when treatment effects are fresh, or at the end when recording becomes less reliable.

Daily risk factors were evaluated as a function of treatment condition, time in days, and the interaction of Treatment x Time. Day of assessment was treated as a continuous variable in these analyses. Terms representing Site and Site x Treatment were also included in models to reduce noise overall and to correct for baseline differences between sites (Stout et al., 2003). Number of records completed was also included to control for participants’ rates of responding. Subject was treated as a random effect, allowing the intercept to vary randomly across participants, and an unstructured covariance pattern was adopted based on AIC (Snijders & Bosker, 1999). For each daily outcome or risk factor examined, the baseline score from the questionnaire corresponding to that outcome or risk factor (PDH, PACS, AASE, AEQ, RTCQ-TV, CES-D, BSCS, CPD, QSU, RESQ, SCQ-A, PRBC, and MNWS-R) was entered as a level 2 covariate to further account for between-participants variance. (That is, if the outcome to be examined was daily Urge to Drink, the covariate used would be Penn Alcohol Craving Scale score). The general equation for these analyses was:

$$\begin{array}{l}\text{Risk}{\text{factor}}_{ij}=\mathrm{\beta }{\text{TREATMENT}}_i+\mathrm{\beta }{\text{DAY}}_{ij}+\mathrm{\beta }(\text{TREATMENT}i\times {\text{DAY}}_{ij})+\mathrm{\beta }{\text{SITE}}_i+\\ \mathrm{\beta }({\text{SITE}}_i\times {\text{TREATMENT}}_i)+\mathrm{\beta }{\text{COVARIATE}}_i+\mathrm{\beta }{\text{RECORDS}}_i+{\mathrm{\alpha }}_i+{\mathrm{\gamma }}_i({\text{SUBJECT}}_i)+{\mathrm{\epsilon }}_{ij},\end{array}$$

where γ is the coefficient for the random effect of subject (i.e., individual intercept).

The daily alcohol and smoking relapse risk variables were also separately analyzed as a function of smoking status on any given day. In these analyses each daily risk factor score was evaluated as a function of day of recording and smoking status on that day (i.e., any reported cigarettes that day, coded as 0 ‘no smoking’ or 1 ‘smoking’). As above, day of recording was grand-mean centered. Treatment site was also included as a covariate, as was the baseline value of the corresponding dispositional measure of outcome being evaluated in each analysis. Analyses were conducted only in those participants assigned to the CSC condition. This was done because only the CSC condition had participants with a sufficient number of non-smoking days to allow a comparison. The general equation for these analyses was:

$$\text{Risk}{\text{factor}}_{ij}=\mathrm{\beta }{\text{SMOKING}}_{ij}+\mathrm{\beta }{\text{DAY}}_{ij}+\mathrm{\beta }{\text{SITE}}_i+\mathrm{\beta }{\text{COVARIATE}}_i+\mathrm{\beta }{\text{RECORDS}}_i+{\mathrm{\alpha }}_i+{\mathrm{\gamma }}_i({\text{SUBJECT}}_i)+{\mathrm{\epsilon }}_{ij}.$$

We analyzed multiple daily measures of relapse risk because each is of interest, and each has a hypothesized relationship with treatment or with smoking status. Given these considerations, we did not adjust for multiple analyses of related risk measures. Instead, we have supplied confidence intervals for the parameter estimates in our models, in addition to p-levels, as recommended by Nakagawa (2004).

Results

Participant Flow

Figure 2 shows the participants’ flow and disposition. Many were excluded because they did not have current alcohol and tobacco use disorders. Among those who declined to participate, 73 did not want to quit smoking. Another 61 were not willing to accept random assignment to concurrent or delayed smoking treatment, with 71% of them expressing a preference for concurrent rather than delayed treatment. A sample of 151 participants (91% military veterans) was randomized to treatment, 142 (94%) completed the 2-week assessments at the end of the intensive treatment phase, and 122 (81%) completed the 13-week follow-up assessment. Participants lost at the 13-week follow-up were significantly younger than those retained (lost: M = 44.1, SD = 10.1 years; retained: M = 50.3, SD = 8.3 years), t(149) = 3.44, p < .01; no other significant difference was found on baseline demographic, smoking history, and drinking history variables. The follow-up attrition rate was comparable across treatment groups.

Figure 2.

Flow of participants through treatment and follow-up.

Treatment Effects

Concurrent smoking treatment participation was excellent, with 93% of participants attending all scheduled treatment sessions. Analysis of smoking abstinence at two weeks (end of intensive alcohol treatment) found 50.5% of CSC and 2.2% of DSC participants were classified CO-verified 7-day point prevalence abstinent [χ²(1) = 32.49; N = 151; p < .001]. At 13 weeks, 19.0% of CSC and 0.0% of DSC participants were classified abstinent [χ²(1) = 8.39; N = 151; p < .01].

Analyses of drinking over time from baseline to month 3 (week 13) indicated that participants in both treatment conditions dramatically decreased their rates of heavy drinking from 58% PDH to a mean of 3% PDH from month 1 to month 3 [F_time(3, 524) = 219.00; p < .001], with no significant differences by treatment condition [F_treatment(1, 524) = 2.04; p > .15], and no interaction of Treatment x Time [F_{treatment X time}(1, 524) = 0.16; p > .90]. Analysis of PDA also indicated that patients in both treatment conditions increased the frequency of alcohol abstinent days from a mean of 40% days abstinent at baseline to a mean of 95% days abstinent across the three-month follow-up [F_time(3, 526) = 207.66; p < .001]. Again there were no differences attributable to treatment [F_treatment(1, 526) = 1.85; p > .15] and no interaction of Treatment x Time [F_{treatment X time}(1, 526) = 0.28; p > .80].

Daily Process Variables

Relapse risk variables as a function of treatment

The assessment of alcohol relapse risk variables on a daily basis was intended to determine if frequent and proximal measurements might be sensitive indicators of treatment-related change in alcohol relapse risk. Table 2 summarizes the effects of treatment (CSC versus DSC), time (84 consecutive daily measures), and Treatment x Time interactions on each of the drinking and smoking-related risk measures. Site effects were all significant, and not shown in the tables that follow. As seen in the Table, participants provided concurrent smoking and alcohol treatment had significantly lower levels of positive alcohol outcome expectancies relative to participants provided alcohol treatment with smoking treatment deferred until three months later. A significant Treatment x Time interaction further indicated that positive expectancies in the CSC condition continued to decrease over time relative to those of the DSC participants. None of the other alcohol-related risk factors were sensitive to treatment effects, once baseline differences and site differences were taken into account. Examination of means on alcohol expectancies showed that treatment effects occurred within the first two weeks, during the intensive alcohol treatment phase, but that levels of positive expectancies continued to decrease in the CSC group, whereas the DSC group showed expectancies returning to baseline after 2 weeks.

Table 2.

Daily Process Measures Over Twelve Weeks as a Function of Smoking Treatment Condition. Each Risk Factor Analyzed Separately. Results Shown are for Treatment =1 (CSC)

	Effect Tested
Dependent variable	Treatment				IVR day				Treatment x IVR day
	B	t	Lower 95% CI	Upper 95% CI	B	t	Lower 95% CI	Upper 95% CI	B	t	Lower 95% CI	Upper 95% CI
Alcohol relapse risk factors
Number of drinks yesterday	−0.22	−1.74	−0.47	0.03	0.000	0.25	−0.002	0.003	−0.003	1.38	−0.008	0.001
Urge to drink	0.17	0.87	−0.22	0.56	0.001	0.75	−0.002	0.004	−0.001	0.53	−0.007	0.004
Alcohol abstinence self-efficacy	0.05	0.22	−0.40	0.51	0.001	0.67	−0.002	0.004	0.002	0.62	−0.003	0.007
Positive alcohol expectancies	−1.48	−2.12*	−2.86	−0.10	−0.017	−5.00***	−0.023	−0.010	−0.012	−2.01*	−0.024	−0.002
Alcohol abstinence motivation	−0.34	−1.45	−0.81	0.13	0.000	0.29	−0.003	0.003	−0.001	−0.39	−0.006	0.004
Negative affect	−0.25	−1.17	−0.52	0.18	0.002	1.70	−0.000	0.005	0.002	1.09	−0.002	0.007
Self-control demands	0.70	0.78	−1.08	2.47	0.012	2.80*	0.004	0.020	0.005	0.645	−0.010	0.020
Tobacco relapse risk factors
Number of cigarettes yesterday	−7.33	−5.33***	−10.04	−6.01	−0.015	−1.57	−0.034	0.004	0.017	1.00	−0.017	0.051
Urge to smoke	−1.38	−5.49***	−1.88	−0.89	−0.006	−3.99***	−0.010	−0.003	−0.004	−1.41	−0.010	0.002
Smoking abstinence self-efficacy	0.88	3.59***	0.40	1.37	0.000	0.12	−0.003	0.003	−0.001	−0.47	−0.007	0.004
Positive smoking expectancies	−4.89	−3.53***	−7.63	−2.15	−0.040	−4.28***	−0.058	−0.021	−0.013	−0.76	−0.046	0.020
Smoking abstinence motivation	0.87	3.49**	0.37	1.36	−0.001	−0.73	−0.004	0.002	−0.001	−0.41	−0.007	0.004
Nicotine withdrawal	−2.37	−1.28	−6.03	1.29	−0.002	−0.20	−0.020	0.017	0.003	0.18	−0.030	0.037

^*p < .05.

^**p < .01.

^***p < .001.

B = model parameter estimate; CI = Confidence interval. Df=8509.

Highly significant treatment effects were seen across all the smoking-related daily process measures. As expected, compared to DSC participants, the CSC participants smoked fewer cigarettes. In addition, however, CSC patients had lower urge to smoke, greater smoking abstinence self-efficacy, lower positive smoking outcome expectancies, greater motivation to stop smoking, and lower nicotine withdrawal scores. These effects were present from the beginning of the IVR monitoring period and remained stable throughout the 84 days.

Relapse risk variables as a function of daily smoking status

Table 3 shows a summary of the analyses of the individual alcohol and smoking risk variables as a function of time-varying smoking status in the CSC patients. As seen in Table 3, levels of all of the risk factors examined were significantly associated with smoking status. The overall means of the risk factor variables by smoking status indicate that, in every case except for self-control demands, more adaptive daily alcohol risk scores were seen for those persons who were not smoking on a given day. Non-smoking days were associated with lower numbers of drinks, lower urge to drink, more negative drinking outcome expectancies, lower negative affect, and with higher average levels of alcohol abstinence self-efficacy and motivation. Daily smoking status also had significant associations with smoking relapse risk variables. Non-smoking days were associated with lower urge to smoke, less positive smoking outcome expectancies, and lower nicotine withdrawal scores on those days, as well as with higher smoking abstinence motivation and self-efficacy.

Table 3.

Daily Process Measures Over Twelve Weeks as a Function of Smoking Status Each Day. Each Dependent Variable Analyzed Separately.

Dependent variable	Effect Tested								Mean of All Days by Smoking Status
	Smoking Status				IVR Day				Smoking	No Smoking
	B	t	Lower 95% CI	Upper 95% CI	B	t	Lower 95% CI	Upper 95% CI	Mean (SD)	Mean (SD)
Alcohol relapse risk factors
Number of drinks yesterday	0.20	3.90***	0.10	0.30	0.001	1.27	−0.001	0.004	0.44 (0.10)	0.09 (0.10)
Urge to drink	0.14	4.47***	0.08	0.20	0.002	1.32	−0.001	0.004	1.84 (0.04)	1.54 (0.04)
Alcohol abstinence self-efficacy	−0.19-	−5.87***	−0.26	−0.13	0.000	0.24	−0.002	0.003	4.04 (0.09)	4.34 (0.09)
Positive alcohol expectancies	0.59	10.19***	0.48	0.71	−0.012	−4.37***	−0.018	−0.007	9.31 (0.31)	8.92 (0.36)
Alcohol abstinence motivation	−0.15	−3.76***	−0.23	−0.07	0.001	0.48	−0.002	0.003	3.86 (0.05)	4.19 (0.05)
Negative affect	0.17	7.71***	0.13	0.22	0.001	1.41	−0.001	0.003	2.31 (0.09)	2.14 (0.09)
Self-control demands	0.05	1.24	−0.03	0.12	0.010	2.89**	0.003	0.017	11.75 (0.55)	12.16 (0.56)
Tobacco relapse risk factors
Urge to smoke	1.05	28.97***	0.98	1.12	−0.004	−3.70***	−0.007	−0.002	3.13 (0.17)	1.86 (0.15)
Smoking abstinence self-efficacy	−0.66	−20.41***	−0.72	−0.60	0.000	0.06	−0.002	0.002	3.26 (0.08)	4.24 (0.09)
Positive smoking expectancies	1.89	16.76***	1.67	2.11	−0.034	−4.52***	−0.049	−0.019	15.14 (0.99)	13.56 (0.91)
Smoking abstinence motivation	−0.50	−14.77***	−0.57	−0.44	−0.001	−0.98	−0.003	0.001	3.52 (0.04)	4.29 (0.04)
Nicotine withdrawal	2.03	12.45***	1.71	2.35	−0.001	−0.18	−0.017	0.014	13.16 (0.49)	11.04 (0.46)

Note. CI = Confidence interval. Df=5715.

Discussion

The concurrent smoking cessation intervention in the present study yielded 50% of intent-to-treat participants with seven days CO-verified point-prevalence smoking abstinence at the end of the two weeks of intensive treatment, and 19% smoking abstinence at the three-month follow-up. These rates are comparable to previous concurrent alcohol-tobacco treatment trials that employed very intensive and long term smoking cessation interventions (Burling et al., 2001; Carmody et al., 2012). Most previous trials of concurrent alcohol treatment and smoking cessation reported much lower follow-up smoking abstinence rates, with many finding no effect of concurrent smoking treatment on follow-up smoking abstinence (Bobo et al., 1998; Burling et al., 1991; Joseph et al., 1993; Kalman et al., 2001). The substantial smoking abstinence rates in the present study allowed for a meaningful test of the impact of concurrent smoking cessation and alcohol treatment on alcohol relapse risk.

Alcohol outcomes were not significantly different for those provided concurrent smoking treatment compared to those waiting for delayed smoking treatment. Ceiling effects may have prevented us from observing outcome differences. Across the three month follow-up, participants reported 95% days alcohol abstinent and only 3% days heavy drinking. Treatment conditions also did not differ on five out of six variables reflecting alcohol relapse risk assessed across 12 weeks of daily monitoring. There was one significant treatment difference on the daily measure of alcohol outcome expectancies. This measure was a composite of three positive expectancy items and a reverse-scored negative expectancy item. Those randomized to CSC had lower scores on this measure across the 12-week assessment compared to the DSC group, meaning they expected less positive and more negative effects from drinking. The direction of this finding suggests that concurrent smoking cessation might reduce alcohol relapse risk for alcohol-dependent smokers. Alcohol treatment studies have generally found that positive outcome expectancies are predictive of poor outcomes and negative outcome expectancies are predictive of good outcomes (Jones, Corbin, & Fromme, 2001). A proximal relationship between outcome expectancies and addiction relapse was seen in a daily monitoring study of smoking treatment (Gwaltney et al., 2005), with the finding that positive smoking outcome expectancies increased significantly on the day before smoking relapse.

The analysis of within-person covariation between daily smoking status and alcohol relapse risk factors (Table 3) showed that smoking abstinence on a given day was significantly associated with lower alcohol consumption, lower urge to drink, lower negative affect, lower positive alcohol outcome expectancies, greater alcohol abstinence self-efficacy, greater alcohol abstinence readiness to change, and greater perceived self-control demands across 12-weeks from the start of concurrent alcohol-tobacco treatment. A causal role cannot be established based on this analysis because smoking on any given day was a self-selected variable. Additionally, the predictors were measured on the same day as the smoking and drinking variables, and therefore temporal primacy cannot be established. The present results are consistent, however, with reports that smoking abstinence during or after alcohol and drug treatment is associated with better substance use outcomes (Friend & Pagano, 2005; Kohn, Tsoh, & Weisner, 2003; Lisha, Carmody, Humfleet, & Delucchi, 2014; Shoptaw et al., 2002; Tsoh, Felicia, Mertens, & Weisner, 2011; Winhusen, Kropp, Theobald, & Lewis, 2014).

The observation of an association between smoking abstinence and lower alcohol craving after concurrent smoking treatment differs from the findings of an earlier concurrent alcohol-tobacco treatment study (Cooney et al., 2007) which compared smokers who quit smoking with those who continued smoking and reported no difference in frequency of alcohol craving in the first two weeks after concurrent treatment. However, Cooney et al. also found alcohol urges were slightly more likely immediately after smoking episodes compared with presmoking assessments. The connection between smoking behavior and alcohol craving seen in our results is also consistent with a laboratory study showing that smoking cues elicited alcohol urges (Drobes, 2002). Cross substance cue reactivity could explain why smoking days in the current study were also days with higher alcohol craving. Lower alcohol craving on nonsmoking days suggests it is unlikely that smoking abstinence took away an effective means for coping with alcohol craving among alcohol dependent individuals in early recovery. Survey studies have found that only a small percentage of alcohol dependent smokers believe smoking decreases their urge to drink (Asher et al., 2003; Kalman et al., 2001; Monti, Rohsenow, Colby, & Abrams, 1995). The present results do not support the idea that smoking is an important strategy for coping with urges to drink.

Depressive symptoms were common at baseline in the present sample, with 51% scoring 16 or above on the CES-D. Our finding that negative affect was greater on smoking days is consistent with results from two prior smoking treatment studies. One study found a reduction in depressive symptoms among those who quit smoking with unchanged symptoms among those who relapsed to smoking (Kahler et al., 2002). Another study examined changes within individuals after smoking treatment and found that transitions from smoking to abstinence were associated with reductions in depressive symptoms (Kahler, Spillane, Busch, & Leventhal, 2011).

Taken together, daily process assessments comparing CSC and DSC conditions across 12 weeks provided no evidence for increased alcohol relapse risk among CSC treated individuals, with one of six process variables (drinking outcome expectancies) suggesting lower alcohol relapse risk with CSC treatment. The finding that cigarette abstinence days were associated with reduced alcohol consumption and with process changes suggesting reduced alcohol relapse risk on five out of six variables (alcohol craving, negative affect, alcohol abstinence self-efficacy, alcohol outcome expectancies, and alcohol abstinence motivation) further supports the conclusion that concurrent smoking treatment did not increase alcohol relapse risk. Given the well-known beneficial effect of smoking cessation treatment on health outcomes (U.S. Department of Health and Human Services, 1990), and the difficulty engaging alcohol treatment clients in delayed smoking treatment (Joseph et al., 2004), we recommend concurrent rather than delayed smoking treatment for those in alcohol treatment programs.

The present sample was 86% men and 91% military veterans, so we advise caution in applying these results to female nonveteran populations. Some individuals were screened out of the study because they were unwilling to accept randomization to concurrent or delayed smoking treatment. Most of these expressed a preference for concurrent rather than delayed tobacco treatment. Results may not generalize to individuals with a clear preference for the timing of tobacco treatment. Another limitation is that the study was designed to address treatment process questions in the first three months after the beginning of treatment. The relatively small sample size and short-term follow-up did not allow us to address long-term efficacy of concurrent alcohol-tobacco treatment. The present results are most applicable to individuals at the beginning of an episode of treatment for alcohol use. The mean time since last drink in our sample at baseline was two weeks. Smoking treatment was initiated within one week of the start of intensive alcohol treatment.

A limitation of this study is that it examined only one approach to concurrent alcohol and smoking intervention utilizing CM and combination NRT. This was done to achieve high smoking abstinence rates. One barrier to implementation of this intervention approach is the cost. Among CSC participants, mean cost of the medications was $117, and mean cost of CM vouchers was $90. Combination NRT is not unusual and has been recommended by the Department of Veterans Affairs (2008). On the other hand, CM protocols have not been widely implemented by community treatment providers (Roll, Madden, Rawson, & Petry, 2009). However, note that the CM protocol in this study was implemented for only a two-week treatment period. The relationships between smoking abstinence and reduced alcohol risk were observed across a 12-week longitudinal assessment period, so smoking abstinence was found to be associated with alcohol abstinence long after the CM protocol had ended. This suggests that this association is most likely not dependent on using a CM-based smoking intervention.

The current study extends evidence for recommending smoking intervention concurrent with intensive outpatient alcohol treatment. Alcohol and drug involved smokers tend to expect adverse outcomes from smoking abstinence, including expectancies of increased substance use, and these outcome expectancies are associated with decreased motivation to quit smoking (Hendricks, Peters, Thorne, Delucchi, & Hall, 2014). The current results support conveying an alternative message that smoking abstinence is accompanied by favorable changes in alcohol use, craving, mood, confidence, and motivation.