The Effect of Varenicline on Smoking and Drinking Outcomes among Black and White Adults with Alcohol Use Disorder and Co-Occurring Cigarette Smoking: A Secondary Analysis of Two Clinical Trials

Abstract

Introduction.

Varenicline is an FDA–approved medication for smoking cessation and has demonstrated promise in reducing alcohol use. This study sought to compare the efficacy of varenicline in reducing smoking and drinking among Black and White people seeking alcohol treatment.

Methods.

Linear mixed modeling was conducted using data from two multi-site placebo-controlled randomized clinical trials examining the effects of varenicline for treatment of Alcohol Use Disorder (AUD; O’Malley et al., 2018 and Litten et al., 2013) among Black and White adults with AUD and co-occurring cigarette smoking. The primary analyses were conducted in a sample of 117 adults (O’Malley trial: 29.1% female, 55.2% Black), and replicated in an independent sample of 73 adults (Litten trial: 23.3% female, 45.2% Black).

Results.

Black participants smoked fewer cigarettes per day compared to White participants (O’Malley trial: F_1,116 = 8.95, p = .003; Litten trial: F_1,68.9 = 4.74 p = .03). Linear mixed models revealed a marginal effect of varenicline on reducing cigarettes smoked per day regardless of race in the O’Malley trial (F_1,109 = 3.34, p = .07), which was replicated in the Litten trial (F_1,67.1 = 20.77 p < .0001). Participants reduced the number of drinks consumed regardless of treatment condition or race in both trials (O’Malley trial: F_1,98 = 131.69, p <.0001; Litten trial:F_1,69 = 60.36, p <.0001).

Conclusions.

Our adjusted model findings suggest varenicline reduced smoking among Black and White people with AUD and co-occurring cigarette smoking. However, these findings should be replicated in a larger sample.

Introduction

Tobacco use is one of the leading causes of preventable death accounting for nearly 10 million deaths per year worldwide and over 1,000 deaths daily in the U.S.^1–3 In 2017, 14% of U.S. adults were current smokers, and the prevalence was similar between Black and White individuals at roughly 15%.² Notably, the rate of smoking doubles among those with alcohol use disorder (AUD).^4–6 Further, smoking predicts poorer alcohol treatment outcomes.^7–9 The remarkably high rate of smoking among those who also drink alcohol has led researchers to be interested in identifying medications that reduce both smoking and drinking outcomes, with varenicline tartrate being the leading candidate.

Varenicline, a partial agonist of α₄β₂ nicotinic acetylcholine receptors, is approved by the FDA for smoking cessation.¹⁰ Varenicline has also been investigated for reducing alcohol consumption^11,12 given that nicotinic acetylcholine receptors, the site of action for varenicline,^13,14 are also implicated in alcohol reward and alcohol drinking.^15,16 To date, there have been three multi-site clinical trials that have examined the effect of varenicline on drinking outcomes. The mixed findings from these trials indicated varenicline 1) reduced drinking,¹⁷ 2) had no effect on drinking,¹⁸ and 3) reduced drinking among men and reduced smoking in the overall sample.¹⁹ In addition, the efficacy of varenicline has been examined in smokers with an alcohol use disorder and was found to improve smoking and drinking outcomes.²⁰ However, none of these trials inform whether varenicline reduces smoking and drinking outcomes regardless of race.

Notable racial disparities in the consequences of smoking, drinking, and success in quitting each behavior suggest that race should be considered when investigating the effectiveness of varenicline for the treatment of individuals with AUD who also smoke cigarettes. Although Black people tend to smoke fewer cigarettes per day compared to White people,^21,22 Black people who do smoke cigarettes experience higher rates of morbidity and mortality related to smoking compared to other racial/ethnic groups.²³ Further, Black people who smoke are less likely to quit despite having more quit attempts.²⁴ Black people are also more likely to smoke menthol cigarettes, which is associated with greater nicotine dependence and difficulty quitting.²⁵ In addition, Black people are less likely to be prescribed medications for smoking cessation in U.S. safety net clinics than White people.²⁶ Although large clinical trials investigating varenicline for smoking cessation found no differences between Black and White people on varenicline versus placebo,^27,28 Black people had lower quit rates than White people regardless of treatment group.²⁷

Black people also tend to experience greater disparities with regard to alcohol-related problems compared to other racial/ethnic groups.²⁹ Despite more problems from drinking, some evidence suggests Black people are less likely to utilize treatment particularly those that provide pharmacotherapies.³⁰ Data from the National Survey on Drug Use and Health indicates that in 2016, 51% of Black individuals with AUD also reported smoking cigarettes.⁴ Further, all racial/ethnic groups studied showed a decline in smoking from 2002 to 2016 except for Black respondents with AUD.⁴

These noted racial disparities highlight the importance of considering race when investigating treatments for smoking and drinking. Therefore, the objective of the current study was to investigate the efficacy of varenicline on smoking and drinking outcomes across Black and White people with AUD and current cigarette smoking in secondary analyses of the only two U.S.-based clinical trials evaluating varenicline for the treatment of AUD.

Methods

Participants

We used data from two randomized, double-blind, parallel group, placebo-controlled trials testing the efficacy of varenicline for AUD. The trial by O’Malley and colleagues¹⁹ (clinicaltrials.gov Identifier: NCT01553136) comprised the primary analysis sample because it included comprehensive smoking in addition to drinking measures. This trial consisted of 131 individuals with AUD and co-morbid cigarette smoking (29.8% female, 52.7% Black) recruited from 2 academic sites in the U.S. from 2012–2015. All participants received medical management and either 2 mg of varenicline or matched placebo pills. For the current study, the sample was restricted to the Black and White participants (N = 117, 29.1% female, 55.2% Black). The second trial by Litten and colleagues¹⁷ (ClinicalTrials.gov Identifier: NCT01146613) for the replication sample consisted of 200 individuals with AUD (30.0% female, 28.5% Black) recruited without regard to smoking status from 5 academic sites within the U.S. from 2011–2012. In this trial, all participants watched Take Control videos,³¹ a computerized bibliotherapy, in addition to receiving either 2 mg of Varenicline or matched placebo pills. For the current study, this sample was restricted to Black and White participants with AUD who endorsed smoking at least one cigarette in the past week at baseline (N = 73, 23.3% female, 45.2% Black) to be comparable to the primary investigation which recruited individuals with AUD and concurrent smoking. Table 1 presents additional methodological details for each trial, and Table 2 presents participant characteristics for each sample used in the current study.

Table 1.

Methodological characteristics of each clinical trial

	Primary Analysis: O’Malley et al., 2018	Replication Analysis: Litten et al., 2013
N	131	200
Diagnostic Measure	DSM-IV AD	DSM-IV AD
Inclusion Criteria	≥18 years; heavy drinking ≥ twice weekly and ≤7 consecutive days of abstinence 90 days prior to intake; cigarette smoking ≥ twice weekly	≥18 years; consumed ≥28 or ≥35 drinks weekly for females and males, respectively during 28 days prior to consent and 7 days prior to randomization
Exclusion criteria	Clinically significant disease or abnormality; serious psychiatric illness; lifetime suicidal behavior; current suicidality; risk of aggression; current drug dependence (excluding nicotine and marijuana); need for detoxification; use of medication in past 3 months to treat SUD; pas month psychotropic medications (except SSRI); pregnancy, nursing or not using effective contraception	Past-year drug dependence (excluding nicotine); psychotic disorders; detox during screening; previous varenicline treatment; history of atherosclerotic cardiovascular disease; lifetime suicide attempts; current suicidality
Length of Trial	16 weeks	13 weeks
Follow-up	Week 52	Week 16
Drinking and smoking measures	TLFB	TLFB Form 90
Primary Endpoint	Percent heavy drinking days (≥4 for women, ≥5 for men) was summarized based on the last 8 weeks of treatment Cigarettes smoked per day (verified by biological measures) was summarized based on the last 4 weeks in treatment	Percent heavy drinking days (≥4 for women, ≥5 for men) assessed weekly during weeks 2–13 of the study Cigarettes smoked per day in the past week

Note. DSM = Diagnostic and Statistical Manuel of Mental Disorders. AD = alcohol dependence. TLFB = Timeline Follow Back.

Table 2.

Participant characteristics

	Black	White	Total
Primary Analysis: O’Malley et al., 2018
N	69	48	125
Female	15, 21.7%	19, 39.6%	34, 29.1%
Mean Age (SD)	44.9 (9.98)	41.1 (12.18)	43.3 (11.04)
Baseline Substance Use
% Heavy Drinking Days	63.2 (26.03)	64.2 (22.35)	63.6 (24.49)
Cigarettes Smoked per Smoking Day: mean (SE), range	9.8 (4.66), 1–20	14.0 (8.73), 2–25	11.5 (1.86), 1–25
% Daily Smokers	81.82%	72.50%	76.62%
Replication Analysis: Litten et al., 2013
N	33	40	73
Female	9, 27.3%	8, 20.0%	17, 23.3%
Mean Age (SD)	37.9 (10.49)	42.7 (9.97)	40.5 (10.41)
Baseline Substance Use
% Heavy Drinking Days	31.6 (6.62)	31.9 (6.45)	31.8 (6.48)
Cigarettes Smoked per Smoking Day	10.1 (6.05)	12.8 (7.44)	11.6 (6.92)

Note. Percent of heavy drinking days was defined as consuming 5 (for men) and 4 (for women) or more standard drinks. Participants were categorized as Black or White based on their response to which race they identified with most. Individuals who identified as Black regardless of whether they endorsed other races were categorized as Black, and individuals who identified as White only were categorized as White.

Measures

Participants in both studies were asked to select the race they identified with most. Individuals who identified as Black regardless of whether they endorsed another race/ethnicity¹ were categorized as Black, and individuals who identified as White only were categorized as White. Drinking was assessed using the Timeline Follow-Back Interview (TLFB)³² in the O’Malley et al.¹⁹ trial and the Form 90 interview³³ in addition to the TLFB for the Litten et al.¹⁷ trial. Percent heavy drinking days (PHDD) was the primary endpoint in both studies. A heavy drinking day was defined as consuming 5 (for men) and 4 (for women) or more standard drinks. The O’Malley et al.¹⁹ trial also assessed daily smoking using the TLFB and obtained carbon monoxide and plasma cotinine to confirm smoking status, whereas questions regarding smoking were assessed in the Litten et al.¹⁷ trial at weeks 1, 6, 10, and 14. Specifically, participants were asked, “Over the past week, on how many days did you smoke cigarettes?” and “If you smoked, what was the average number of cigarettes you smoked per day during the past week?”

Data Analysis

Linear mixed modeling examining moderation by race was conducted in SAS version 9.4³⁴ examining changes in smoking and drinking from baseline to the end of treatment. Specifically, we examined whether race moderated treatment group by time effects on smoking and drinking outcomes. The smoking endpoint was cigarettes smoked per day (given that both studies focused on reducing alcohol use and not smoking cessation and that reductions in smoking is associated with increased likelihood of a future quit attempt³⁵) and the drinking endpoint was percent heavy drinking days (PHDD). In the O’Malley et al.¹⁹ trial, change in smoking was based on the average number of cigarettes smoked per smoking day during the 28 days prior to intake and during last 28 days in treatment. PHDD was summarized based on the last 8 weeks of treatment. In the Litten et al.¹⁷ trial, change in smoking was based on cigarettes smoked per day in the past week assessed at weeks 1 and 14. PHDD was based on weekly assessments at weeks 2–13. The PHDD variable was log transformed for each trial to account for skewness of the variable. All models were adjusted for sex and site. Non-significant interaction terms were dropped from the final models. The effect of varenicline on smoking and drinking outcomes by race was examined first in the O’Malley et al.¹⁹ trial. We investigated whether the findings from the primary analysis replicated in a second independent sample.¹⁷

Results

Raw means, standard errors, and Cohen’s d effect sizes for change in smoking and drinking in the two samples are presented in Figures 1 and 2, respectively.

Figure 1.

Percent change in cigarette smoking (raw means and standard errors) by treatment group and race in the primary and replication analyses. Higher percent change indicates greater reductions and lower percent change indicates less of a reduction. Participants were categorized as Black or White based on their response to which race they identified with most. Individuals who identified as Black regardless of whether they endorsed other races were categorized as Black, and individuals who identified as White only were categorized as White. Cohen’s d effect sizes are reported. *<0.05, **<0.01, ***<0.001.

Figure 2.

Change in percent heavy drinking days (raw means and standard errors) by treatment group and race in the primary and replication analyses. Higher percent change indicates greater reductions and lower percent change indicates less of a reduction. Participants were categorized as Black or White based on their response to which race they identified with most. Individuals who identified as Black regardless of whether they endorsed other races were categorized as Black, and individuals who identified as White only were categorized as White. Cohen’s d effect sizes are reported. None of the within group differences were significant (p < .05).

Primary Analysis: O’Malley Trial¹⁹

Smoking.

Results from the linear mixed model indicated a main effect of race (F_1,116 = 8.95, p = .003) such that Black individuals (Least Squares Mean [LSM] = 6.83, SE = .79) smoked significantly fewer cigarettes per smoking day compared to White individuals (LSM = 10.49, SE = .83) regardless of time point or treatment group. A marginal 2-way interaction between treatment group and time was found (F_1,109 = 3.34, p = .07) suggesting a greater change in smoking among those who received varenicline (LSM Difference = 6.04, SE = .67, t₁₀₇ = 9.10, 95% CI: 4.73, 7.36) compared to those who received placebo (LSM Difference = 4.26, SE = .95, t₁₁₂ = 5.95, 95% CI: 2.84, 5.68). The 3-way interaction of treatment, time, and race was not significant (F_2,109 = .41, p = .52).

Drinking.

In terms of PHDD, interaction of race, treatment group, and time was not statistically significant (Figure 2). There were no significant main or moderation effects of race. A main effect of time was found such that drinking decreased from baseline regardless of treatment group and race (F_1,98 = 131.69, p <.0001).

Replication Analysis: Litten Trial¹⁷

Smoking.

A significant main effect of race was found (F_1,68.9 = 4.74 p = .03) indicating Black individuals (LSM = 8.06, SE = 1.20) reported smoking fewer cigarettes relative to White individuals (LSM = 11.40, SE = 1.18) regardless of time or treatment group. In addition, there was a significant 2-way interaction between treatment group and time (F_1,67.1 = 20.77 p < .0001) such that there was a greater change in smoking among those in the varenicline group (LSM Difference = 4.83, SE = .70, t₆₇ = 6.93, 95% CI: 3.44, 6.22) relative to placebo (LSM Difference = .53, SE = .64, t_67.2 = .83, 95% CI: −.74, 1.80). Like the findings from the primary analysis, the 3-way interaction between treatment group, time, and race was not significant indicating that the pattern of the treatment by time interaction was similar across Black and White individuals (F_2,109 = .22, p = .81). Notably, the 3-way interaction between race, treatment, and time was also tested in a subsample of daily smokers as sensitivity analyses and the findings replicated.

Drinking.

For PHDD, no significant interaction effects were found indicating there were no significant differences in drinking by race or treatment group. Like the primary analysis, there was a main effect of time such that everyone reduced their drinking across race regardless of treatment group (F_1,69 = 60.36, p <.0001). Notably, Figure 2 indicates that there was less of a reduction in drinking in the replication analysis relative to the primary analysis.

Discussion

This is the first examination of the efficacy of varenicline on smoking and drinking outcomes among Black and White people seeking treatment for AUD. The results suggest that Black and White people showed significant reductions in heavy drinking irrespective of treatment condition. However, individuals who received varenicline reduced cigarette smoking to a greater degree than those who received placebo among both Black and White people with AUD. Overall, these findings suggest that varenicline may help reduce smoking among Black and White adults with co-occurring AUD and cigarette smoking who are seeking treatment for AUD, though these findings should be replicated in a larger sample.

A major strength of this study is that we investigated our research questions using two independent samples. Further, the use of multisite trials increases the generalizability of the findings. The findings in both studies related to the efficacy of varenicline for reducing drinking indicate that participation in alcohol treatment, irrespective of race and whether individuals received varenicline, was associated with significant reductions in heavy drinking. The degree of drinking reduction, however, was greater in the primary investigation¹⁹; an observation that might relate to the more intensive behavioral platform as well as the longer duration of the trial compared to the Litten et al.¹⁷ trial. Although the primary investigation found sex differences such that varenicline resulted in reduced heavy drinking among men but not women,¹⁹ our sample size limited our ability to examine sex differences in smoking and drinking outcomes by race.

Regarding smoking, the results of our adjusted analyses of the main investigation indicated that varenicline reduced cigarette smoking regardless of race.¹⁹ Similarly, the adjusted model findings from the Litten trial indicated that varenicline, compared to placebo, reduced smoking regardless of race.¹⁷ Taken together, the findings from both trials suggest varenicline may reduce smoking for both Black and White adults seeking treatment for AUD, which is consistent with prior studies suggesting that varenicline is equally effective for Black and White people seeking smoking cessation treatment.²⁸

Unlike prior studies, we found that end of treatment reductions in percent days smoking were similar for both Black and White people whereas prior studies found that smoking abstinence rates, irrespective of treatment condition, were lower among Black compared to White people.²⁷ These differences could be due to many factors. For example, our analyses focused on smoking reduction (vs. cessation) as smoking cessation was not the goal of the current studies but rather the goal was treatment for AUD. In addition, biochemical verification of smoking status was not obtained in both trials and so was not incorporated into our smoking outcome, and our study enrolled non-daily smokers in addition to daily smokers (vs. only daily smokers). Regardless, the consistent finding of benefit from varenicline on smoking outcomes irrespective of race if replicated is important given that Black people with AUD have not experienced a reduction in smoking rates over time like people of other racial/ethnic groups⁴ and greater access to effective treatment could help reduce the considerable health disparities resulting from smoking.

Limitations

This study is not without limitations. Other contextual factors (e.g., childhood neighborhood, trauma history) that may have accounted for differences between Black and White participants were not assessed and could not be included in the analyses. We combined those who identified as Black only with those who identified as Black and another race/ethnicity, which increased the heterogeneity of the Black subgroup and within group differences in the Black subgroup were not examined due to sample size constraints. The original aim of the two clinical trials did not involve investigating race differences; randomization was not stratified by race. The sample for the Litten trial¹⁷ was reduced because we restricted the sample to those with AUD and who also reported smoking. Thus, the samples may have been underpowered to detect effects by race. Although we sought to replicate the findings from the primary study in an independent sample, there were methodological differences between the primary and the replication study that may have contributed to differences in the findings. These include differences in the duration of the studies (16 weeks versus 13 weeks), the behavioral intervention provided (medical management versus bibliotherapy videos), and the age of Black participants compared to White participants within each study (older versus younger). In addition, the smoking criteria were slightly different across trials such that participants in the primary analyses smoked at least two or more days in the past week, whereas participants in the replication analysis smoked at least one cigarette in the past week. While these inclusion criteria are lower than in studies of varenicline for smoking cessation and thereby limit comparisons to this literature, our criteria capture the smoking patterns seen in individuals with AUD who are current smokers, including daily and nondaily use.³⁶

Future Directions

This study also informs directions for future research. The findings from both the primary and replication investigations suggest that both Black and White people reduced their drinking regardless of the treatment received. Should these findings replicate in a larger sample, future research should identify the mechanism underlying changes in drinking (e.g., motivation, aspects of medical management/bibliotherapy). The O’Malley et al.¹⁹ trial found varenicline reduced heavy drinking among men but not women and this effect remained with race in the model; future research should use a larger sample to examine whether there is an interaction of race and sex in the effect of varenicline on smoking and drinking outcomes. Although race is not a biological construct but a social construct that contributes to racial disparities in health due to racism,^37,38 it is essential that future researchers include racially diverse participants and investigate the efficacy of clinical interventions across race. When differences in intervention effects are found across race, it is imperative that researchers identify the social constructs that contribute to these differences, so the interventions can be appropriately tailored. For example, extensive evidence links racial discrimination with increased alcohol use among Black people³⁹ suggesting that approaches for mitigating the effects of racial discrimination until racial discrimination is eliminated could be incorporated in treatment approaches.

Highlights.

We examined the effect of varenicline on smoking and drinking outcomes by race

Participants were Black and White adults with alcohol use disorder who also smoke

The effects were examined across two clinical trials of varenicline versus placebo

Varenicline reduced cigarette smoking in both Black and White adults

All participants reduced their drinking regardless of treatment group or race