Drinking as a risk factor for sustained smoking

Abstract

Data from a nationally representative sample of US adults (n=42 862) were used to explore the associations between drinking and smoking, on a lifetime and past-year basis, and between drinking and smoking cessation for a subsample of past-year smokers (n=12 586). Smoking cessation was defined as current nonsmoking (as of the time of interview) that had lasted at least 3 months. The proportions of both lifetime and past-year smokers increased with volume of alcohol intake and frequency of heavy drinking and were greater for individuals with DSM-IV alcohol abuse or dependence than for individuals without these disorders. For example, the proportion of past-year smokers rose from 22.5% of lifetime abstainers to 53.0% of heavy drinkers, from 23.8% of those who never drank 5+ drinks on any drinking day to 61.8% of those who drank 5+ drinks weekly or more often and from 27.6% of individuals without past-year DSM-IV alcohol abuse or dependence to 55.5% of those with either of these disorders. The proportion of past-year smokers who had stopped smoking decreased from 7.8% of former drinkers to 4.6% of heavy drinkers, from 7.3% of those who never drank 5+ drinks to 3.4% of those who did so weekly or more often and from 6.7% among individuals without past-year abuse or dependence compared to 4.4% among those with either disorder. In a multivariate analysis that adjusted for background variables and smoking history, average daily ethanol intake was not significantly associated with the odds of smoking cessation, but drinking 5+ drinks at least once a month reduced the odds of cessation by 42%. Having an alcohol use disorder increased the odds of smoking cessation below the age of 30 (an effect that disappeared after adjusting for the interaction between age group and having children in the home), but had an increasingly negative effect on smoking cessation at older ages.

Introduction

More than 20 years of studies have shown a strong positive association between smoking and drinking in the United States. Much of this evidence has been drawn from studies of clinical samples of alcoholics, which have shown that persons with alcohol abuse and dependence are more likely to smoke and are heavier smokers than are members of the general population (see reviews in Bien & Burge, 1990, Bobo, 1992, Shiffman & Balabanis, 1995, Toneatto et al., 1995). Studies of adolescent substance use have confirmed that drinkers are more likely than nondrinkers to be smokers and vice versa and that the use of either one of these substances increases the risk of starting to use the other (Kandel & Yamaguchi, 1993, Torabi et al., 1993, Schorling et al., 1994, Chen & Kandel, 1995). In a longitudinal study of a small sample of college students, tobacco dependence and alcohol use disorders (i.e. abuse and dependence) each predicted the other over time (Sher et al., 1996). Finally, studies of the general adult population have confirmed an association between drinking and smoking. Craig and Van Natta (1977), who studied 1089 residents of a single Maryland county in the early 1970’s, found a modest association between alcohol and nicotine consumption that became stronger with increasing levels of drinking, and Friedman et al. (1991), who studied data on 13 673 patients receiving check-ups between 1979 and 1985, found that alcohol use was strongly related to the number of cigarettes smoked per day. In some but not all population subgroups, alcohol use also was associated with duration of smoking, time from arising to first cigarette, frequency and depth of inhalation and proportion of cigarette smoked. Similarly, laboratory studies with human subjects have described ethanol-associated increases in puff volume (Nil et al., 1984, Mintz et al., 1985) and decreases in intercigarette interval (Mello et al., 1987). Some of these studies have noted ethanol-induced increases in smoking only in response to high doses of ethanol or only among individuals whose previous alcohol intake levels were in the moderate to high range (see review in Zachny, 1990).

Several mechanisms have been proposed as underlying the association between smoking and drinking. One of these is cross-tolerance, i.e. the chronic use of one substance resulting in tolerance to the effects of another. Partial cross-tolerance to the effects of ethanol have been found in mice chronically infused with nicotine, and partial cross-tolerance to the effects of nicotine have been found in mice chronically treated with ethanol (Burch et al., 1988, Collins et al., 1988). These studies have suggested that cross-tolerance may be associated with ethanol-induced disruption of the membrane lipids that regulate nicotine receptor function (Collins, 1990). Other proposed mechanisms include the putative effects of ethanol on nicotine metabolism (Adir et al., 1980), the ability of each substance to counteract the aversive effects of the other, e.g. the stimulant effects of nicotine offsetting the performance deficits caused by excess ethanol intake (Zachny, 1990, Kerf et al., 1991), the disinhibiting effects of ethanol overcoming restraints against smoking (Shiffman et al., 1994), shared genetic predisposition and/or environmental influences (Swan et al., 1990), and cue reactivity and expectancies that foster the concurrent use of the two substances (Gulliver et al., 1995).

In addition to being correlated with current smoking, drinking bears a relationship with smoking cessation. A number of studies have found heavy levels of baseline drinking to be negatively associated with smoking cessation. A study of World War II veteran twins interviewed at two times 16 years apart found higher levels of baseline alcohol consumption in those who continued smoking than in those who quit (Carnelli et al., 1993). In a community sample of 2115 adults, Zimmerman et al. (1990) found that heavy drinking was negatively associated with the likelihood of having attempted to quit smoking and with the likelihood of success among those who had tried; however, their measure of drinking level was at time of interview rather than at baseline. Based on data collected in 1975 and 1981 from members of the Finnish Twin Cohort, frequency of consuming spirits and frequency of consuming beer at baseline reduced the likelihood of smoking cessation (Kaprio and Koskenvuo, 1988).

Alcohol problems, recovery from alcohol problems and having stopped drinking also are associated with smoking cessation. DiFranza and Guerrera (1990) found that only 7% of alcoholic smokers successfully stopped smoking, compared to 49% of a control group of nonalcoholics, and a study of 235 smokers recovering from noncardiac surgery found that a history of alcohol abuse was positively associated with smoking relapse (Simon et al., 1992). In a study of problem drinkers receiving behavioral self control training, Miller et al. (1983) found that having stopped drinking increased the probability of smoking cessation. Similarly, in a study of Canadian adults, Sobell et al. (1995) found that the probability of having stopped smoking was lower for individuals with unresolved alcohol problems than for those whose alcohol problems were resolved. Interestingly, rates of smoking cessation were even higher among recovered problem drinkers who were drinkers in the past year than among those who were abstinent from alcohol.

Studies of samples from smoking cessation programs or of self quitters have provided results consistent with those described above. In a study of a nicotine gum trial, Hughes (1993) reported lower 1-year quit rates among individuals with past alcohol or drug problems than among those without a history of alcohol or other drug use disorders (those with present alcohol or drug problems were excluded from the trial), and a study of 235 self-quitters recruited from newspaper advertisements found that heavier drinking at baseline was positively associated with relapse over a 1-year period (Garvey et al., 1992). Complementing these are studies of smoking relapse episodes that have implicated alcohol consumption as a risk factor for relapse (Shiffman, 1982, Baer & Lichtenstein, 1988).

Despite the large number of studies that have linked alcohol consumption, alcohol problems and smoking cessation, other studies have reported contradictory or mixed findings. Data collected over 6 years as part of the Multiple Risk Factor Intervention Trial indicated that baseline number of drinks per week was not associated with the likelihood of attempted smoking cessation but that it was positively associated with smoking relapse among those who attempted to quit (Hymowitz et al., 1991). In a study of 3923 participants in the Lung Health Study’s smoking cessation program, baseline number of drinks was negatively associated with the initial odds of quitting for women but not for men, and drinking was not associated with relapse for either men or women (Nides et al., 1995). In a study of 220 smokers enrolled in a smoking cessation program, recovering alcoholics had success rates similar to those for nonalcoholics, except for those with comorbid alcoholism and major depression, who had lower rates of success (Covey et al., 1993). Hughes and Oliveto (1993), reporting on one sample of 105 smokers receiving physician advice and placebo gum and another sample of 630 self quitters, stated that amount of baseline alcohol intake was not predictive of relapse in either study.

Many of the existing studies of the role of alcohol in smoking cessation have been based on samples of individuals who were either in smoking cessation programs or attempting to stop smoking on their own. These provide a limited view of the effect of alcohol on smoking cessation, because they account only for the effects of alcohol on the success of a given attempt and not on its overall effect, which incorporates the probability of making an attempt in the first place. Other studies have been limited by inadequate measures of drinking patterns and alcohol-related problems. None have distinguished the effects of a heavy volume of ethanol intake from those of alcohol abuse or dependence or of episodic heavy drinking, despite the importance of these distinctions in attributing support to theories such as those of common genetic susceptibility, cross tolerance, cue reactivity and disinhibition. Moreover, few studies have considered the confounding effect of major depression, a condition that is highly comorbid with alcohol dependence (Grant and Harford, 1995) and for which nicotine may be a moderately effective form of self-medication.

This study represents an attempt to remedy those limitations. It is based on a nationally representative population sample of 42 862 adults, 12 586 of whom were past-year smokers. In addition to identifying past-year drinkers, former drinkers and lifetime abstainers from alcohol, it contains measures of ethanol intake volume, frequency of heavy drinking and alcohol abuse and dependence. Bivariate analyses demonstrate the associations of these various aspects of alcohol consumption with smoking and smoking cessation, on both a lifetime and past-year basis, and their effects are further distinguished in multivariate analyses that predict the odds of smoking cessation among past-year smokers, before and after adjusting for major depression, use of other drugs, aspects of smoking history, familial alcoholism and a host of other background variables.