Effectiveness of nicotine gum in preventing lapses in the face of temptation to smoke among non-daily smokers: A secondary analysis

Abstract

Background and aims.

Non-daily smokers (NDS) comprise a large fraction of US smokers. Despite little or no dependence, as typically assessed, ITS have difficulty quitting smoking. A randomized clinical trial comparing the effect of nicotine gum with placebo on quitting smoking in non-daily smokers did not find an effect on overall abstinence. We undertook an analysis to assess whether using nicotine gum, vs placebo, when tempted to smoke, could reduce incidence of lapses in those situations.

Design.

Within a 6-week randomized, placebo-controlled clinical trial of nicotine gum, analyses contrasted the outcome of temptation episodes where gum was or was not used.

Setting.

Smoking cessation research clinic in Pittsburgh, PA, USA.

Participants.

255 adult ITS (131 nicotine gum, 124 placebo) seeking help for smoking cessation.

Intervention.

Nicotine gum (2 mg) vs placebo for up to 8 weeks, with as-needed dosing instructions.

Measurements.

Outcome was lapsing in temptation episodes, as reported by participants via ecological momentary assessment (EMA). Propensity scores predicting gum use from situational factors (e.g., mood, social setting, smoking cues) served as a control variable.

Findings.

Participants reported 2,713 temptation episodes, 46.0% (1,248) of which resulted in smoking (lapsing). There was a significant gum use x active treatment interaction (p=0.0009). Using nicotine gum decreased the odds of lapsing by 55% compared with using placebo (OR=0.45, 0.22–0.94); when gum was not used, the assigned gum condition made no significant difference (OR=1.53, 0.78–3.01; Bayes Factor=0.14). The nicotine effect was not reliably different when participants were trying to achieve abstinence versus when trying to maintain abstinence (OR=0.44 (0.10, 2.03), p=0.294; Bayes factor=0.11), for men and women (OR=1.68 (0.58, 4.87), p=0.343; Bayes factor = 0.10), or for participants with some or no dependence (OR=0.88 (0.30, 2.59), p=0.811; Bayes factor=0.06).

Conclusions.

When used in response to temptation to smoke, 2 mg nicotine gum can help prevent lapses among non-daily smokers.

While daily smoking has been the dominant form of tobacco use for decades, nondaily or intermittent smokers (ITS) now comprise a quarter to a third of US adult smokers (1). ITS show low or negligible levels of dependence (at least as traditionally construed (2)), voluntarily go multiple days without smoking (3), and experience no withdrawal or craving when abstaining (4). These data clearly show that ITS do not need or seek to maintain a particular minimal nicotine level. ITS do suffer health consequences from smoking (5, 6), and do make attempts to quit, in fact doing so more often than daily smokers (7). Despite their lack of nicotine dependence, ITS’ success rates in quitting are only slightly better than those of daily smokers (7).

Detailed analyses of ITS smoking patterns shows that their smoking is tightly linked to particular, sometimes idiosyncratic, situations (8); that is, it is under stimulus control. Also, a randomized trial with low nicotine cigarettes showed that ITS do engage in nicotine-seeking (9). This led to the hypothesis that ITS might seek the acute effects of nicotine in particular situations. That is, that they might be ‘peak-seekers,’ rather than ‘trough-maintainers’ (10). This suggested that ITS could be assisted in quitting if they used an acute nicotine replacement medication in such situations, to help avoid lapsing to smoking.

In a recent randomized clinical trial we tested this treatment approach, by randomizing ITS to active nicotine gum (2 mg) or placebo, with instructions to use the gum when they were tempted to smoke or anticipated being in a situation where they would be tempted to smoke (11). Studies with daily smokers have consistently showed that using nicotine gum in the face of cues that provoke craving is effective in reducing craving (12, 13), sometimes within minutes (14). However, our randomized trial (11) showed no effect of active nicotine gum on cessation outcomes (15) including initial quitting, avoiding lapses, or maintaining long-term abstinence.

As in many nicotine gum studies with daily smokers, it was evident that participants in this study had used very little gum, averaging one piece per day in the first week, and declining thereafter (11). Among DS, use of inadequate amounts of medication is associated with poorer cessation outcomes (16). Thus, it is possible that using the gum when faced with temptation to smoke was actually effective in preventing lapses to smoking but was not used often enough to improve overall quit rates. In this paper, we more directly examine a pre-specified hypothesis (recorded in the application for funding the study) that using nicotine gum, compared to placebo, can help ITS overcome temptation and avoid lapsing. Specifically, we compare the outcomes of temptations where gum (active vs placebo) was and was not used. An interaction effect would indicate a specific effect of nicotine gum on this ‘micro-outcome’ (17), more directly testing the effect of nicotine gum as a ‘rescue’ medication when ITS are facing temptation to smoke.

The data for these analyses were collected by Ecological Momentary Assessment (EMA [18, 19]), wherein participants used smart phones to record temptation episodes they experienced, whether they lapsed, and whether they had used their assigned gum. Analyses of temptation episodes modeled their outcome (lapsed or not) from the interaction between gum use (yes/no) and the randomly-assigned gum condition (nicotine vs placebo). A significant interaction would indicate that nicotine gum is more effective than placebo gum in preventing lapses during temptations. Individuals were randomized to nicotine or placebo gum, so between-subject differences were controlled. However, use of gum in particular situations was not randomized. To control for situational factors associated with using gum, we adjusted the outcome models using propensity scores that took account of situational predictors of gum use.

Within this model, we explored several potential moderators of the nicotine gum effect, examining the 3-way interaction (moderator x gum condition x gum use) (20). We assessed whether the effects of nicotine gum differed between the period when ITS were trying to achieve initial abstinence (defined as 7 days of non-smoking, achieved by 67.5% of participants) versus after they attained initial abstinence and were trying to avoid lapsing (achieved by 30.8% of those who achieved initial abstinence); see (15). We also explored two individual difference moderators: Gender, which has been hypothesized to moderate nicotine reinforcement and reward (21), and dependence, which was associated with poorer outcomes from nicotine gum in the parent trial (11).

Methods

Subjects

Participants were recruited from the Pittsburgh area using a variety of advertising methods. Eligibility criteria included smoking non-daily (4–27 days per month), for at least the past year (regardless of how many cigarettes they smoked), smoking at any rate for at least 3 years, and interest in making a quit attempt supported by behavioral counseling and either nicotine or placebo gum. (See Shiffman et al., 2019 [11] for full inclusion criteria.) To be included, participants had to have reported at least two temptations, so that gum use and outcome could potentially vary within-subjects. (Supplemental Table 1 shows that the 57 participants thus excluded were younger, with a less extensive smoking history.)

Procedures

The study was approved by the Institutional Review Board of the University of Pittsburgh and registered at ClinicalTrials.gov (NCT02168855). After a two-week baseline period of ad libitum smoking, participants reached the Target Quit Day, at which time they were randomized 1:1 to receive either active 2 mg nicotine gum or an inactive placebo gum for six weeks. All participants received behavioral counseling focused on identifying, avoiding, and coping with smoking cues, with two counseling sessions before the designated quit day, and sessions on the quit day and 1, 2, 4, and 6 weeks afterwards.

As described in detail in Shiffman et al. (11), participants were instructed to report in real time two kinds of temptation occasions: those in which they smoked (lapses) and those in which they were tempted but successfully avoided smoking (resolved temptations). The assessment that followed presented a series of questions regarding the participant’s mood, activities, location, and other contextual variables at the time of the event. Participants were also asked to make a record each time they used the assigned gum. Further, when participants reported a resolved temptation (i.e., no lapse), they were asked if they had used gum. When participants reported a lapse, they were also asked about gum use, unless they had already reported gum use in the preceding hour; a similar past-hour gum-use indicator was created for temptation episodes. Thus, participants are considered to have used gum if they reported it when recording the episode or in the preceding hour.

Analysis

To better assess the effect of gum use on lapse risk, we focused on temptation episodes where cigarettes were available, so that there was a material risk of lapsing, and the situational logistics did not dictate the outcome. (A previous analysis showed that the odds of lapsing were 98% lower when cigarettes were said to be unavailable than when they were easily available [22].) The resulting dataset included 255 subjects (131 active (51.4%), 124 placebo (48.6%)) with 2,713 temptation episodes, of which 1,248 (46.0%) resulted in lapses.

Generalized linear mixed models (i.e., logistic regression with random subject intercepts), specified with variance component covariance structures, modeled the likelihood of lapsing by treatment group and gum usage, with a focus on the interaction. That is, the analyses considered whether using nicotine gum was associated with different outcomes than using placebo gum. For non-significant results, Bayes factors (BF) were estimated using a function comparing the Bayesian information criterion of full and reduced models (23). All analyses were conducted using SAS PROC GLIMMIX.

Three potential moderators were considered: the phase of quitting (the time up to achievement of 7 days’ continuous abstinence, or later), gender, and Fagerstrom Test of Nicotine Dependence (FTND; [24]) scores (0 vs >0). Moderation effects would be indicated by significant 3-way interactions (gum use by gum type by moderator).

Models controlled for propensity scores to adjust for the probability of using gum in each episode. Propensity scores were generated from random intercept logistic regression models predicting the probability of gum use in the sample of 2,713 temptation episodes from EMA-based situational variables, using forward stepwise regression methods with a p<0.15 criterion for entry (to maximize prediction) and p<0.05 for removal.

Results

Participant characteristics

Participants averaged 45.1 years of age (SD= 14.3; 41.6% were male (n = 106) and 52.2% Caucasian (n = 133) (Supplemental Table 1). They smoked an average 3.6 days per week (SD=1.1), averaging of 3.6 (SD=2.4) cigarettes on those smoking days. Most (62.8%) scored 0 on the FTND.

Analysis of propensity to use gum in temptation situations

Overall, gum was used in 52.7% (n = 1430) of temptation episodes. The stepwise analysis predicting gum use from situational factors identified nine variables (largely overlapping with variables associated with lapse risk (22)(19) that were associated with gum use (see Supplemental Table 2), resulting in a c-statistic (area under the receiver operating characteristic curve) of 0.86 (0.85–0.88). The resulting estimated propensity scores covered the full range of probabilities, ranging from 0.01 to 0.98. As intended, propensity scores were higher for temptations where gum was used (Used Gum, Mean (95% CI): 0.50 (0.47–0.53), Didn’t Use Gum, Mean (95% CI): 0.45 (0.42–0.47), p<0.0001) but did not differ between the randomly-assigned active vs placebo groups (p=0.544), with no interaction between gum assignment group and reported gum use (p = 0.401, interaction coefficient = −0.007 (−0.023, 0.009)).

Relation of gum use and risk of lapsing

Overall, 46.0% (n = 1248) of temptations resulted in lapses. A model including gum-use propensity scores showed that using active nicotine gum was more effective at reducing lapse risk than using placebo gum (Table 1). Specifically, there was a significant gum-use by type-of-gum interaction (p<0.001) on the risk of lapsing (see Figure 1). In temptations when no gum of any kind was used, being in the nicotine versus placebo arm of the study did not matter (OR=1.53, 0.78–3.01) (p = 0.215; Bayes factor = 0.14). But when gum was used, using nicotine gum rather than placebo was associated with a 55% reduction in the odds of lapsing (OR=0.45, 0.22–0.94) (Table 2). This pattern indicates that the effects were attributable to use of either type of gum in the particular temptation situation, rather than due to a more general effect of assignment to nicotine gum.

Table 1.

Logistic regression model predicting lapse from gum use and type of gum

Main Model: Treatment + Gum Usage + Treatment*Gum Usage + Propensity Score (Dummy Coding)
Effect	Parameter Estimate	LCL	UCL	p-value
Intercept	2.89	2.23	3.55	<0.001
Propensity score for gum use	−3.87	−4.92	−2.83	<0.001
Nicotine gum (ref: Placebo)	0.12	−0.47	0.72	0.689
Used Gum (ref: Didn’t use gum)	−2.14	−2.52	−1.75	<0.01
Nicotine gum * Used Gum	−0.90	−1.43	−0.37	<0.001

^*The efficacy of nicotine gum for preventing lapses was evaluated by the nicotine gum Used gum interaction. The propensity score used variables associated with the temptation situation to model whether gum was used.

Figure 1.

Percentage of temptations resulting in lapse by gum use: nicotine vs. placebo

Note. Error bars are standard errors. Lapse proportions derived from the hierarchical random-effects model including propensity scores (see text). There was significant nicotine/placebo X gum use interaction, and the effect of random assignment to nicotine gum was significant when gum was used, but not when gum was not used (see Table 2).

Table 2.

Simple main effect contrasts within the nicotine gum ^* used gum interaction

Contrasts	OR	LCL	UCL
Within nicotine gum: Using gum vs not using gum	0.05	0.03	0.07
Within placebo gum: Using gum vs not using gum	0.12	0.08	0.17
When gum was not used: Active vs Placebo	1.53	0.78	3.01
When gum was used: Active vs Placebo	0.45	0.22	0.94

^*These odds ratios assess the simple main effects within conditions, as contrasts within the overall nicotine gum used gum interaction that is documented in Table 1.

Put another way: In the nicotine gum arm, using nicotine gum (vs no gum) was associated with a 95% reduction in the odds of lapsing (i.e., OR=0.05, 0.03–0.07). While using placebo gum was also associated with reduced risk of lapsing, the effect was smaller at 88% reduction (i.e., OR=0.12, 0.08–0.17). Thus, using gum, even placebo, was associated with a lower risk of lapsing, but the effect was significantly greater when active nicotine gum was used (Table 2).

Moderation effects

None of the tested moderators showed significant moderation effects (Supplemental Table 3). A model tested whether the effects of nicotine gum differed between the period when participants were trying to achieve initial abstinence (7 days) compared to after abstinence had been achieved. There was no interaction: the effects of gum were not observed to differ by milestone period (OR = 0.44 (0.10, 2.03), p = 0.294, BF = 0.11, preferring a model without interaction). Neither gender (OR = 1.68 (0.58, 4.87), p=0.343, BF = 0.10.) dependence (OR = 0.88 (0.30, 2.59), p = 0.811, BF = 0.06) moderated the nicotine gum effect (i.e., no interaction with gum use and gum assignment was observed).

Discussion

Findings from a detailed momentary analysis of temptation situations supported the hypothesis that use of nicotine gum can help prevent lapsing in the face of temptation to smoke. Among ITS trying to quit smoking, acute use of active nicotine gum, compared to placebo, reduced by over 50% the odds that a temptation would result in smoking/lapsing. The evidence that immediate use of nicotine gum can help avoid smoking in the face of temptation is consistent with evidence from several laboratory studies (12–14) showing that nicotine gum can reduce cue-provoked craving.

No difference in this effect was seen by cessation milestones (15), that is, between when ITS were first trying to establish abstinence, and when, having established abstinence, they were trying to avoid lapses. Nor was there any moderation by gender or degree of dependence.

The evidence for the situational efficacy of nicotine gum in this study stands in contrast to the findings that assignment to nicotine gum (vs. placebo) did not improve overall smoking cessation outcomes (11). The discrepancy is likely explained by the fact that study participants did not use gum frequently enough, or, more specifically use it often enough in temptations. Despite explicit directions to use gum to combat temptations, gum use was reported in just over half (53%) of reported temptation episodes, even where cigarettes were available¹. This suggests that if ITS could be induced to more often or always use gum when faced with temptation to smoke, nicotine gum could lead to improved smoking cessation outcomes.

Analysis of the overall outcomes in this trial showed poorer outcomes for those with FTND scores above zero and suggested, counter-intuitively, that this group might actually be disadvantaged by use of active nicotine gum (11). This led us to speculate that the small nicotine dose from the gum could potentially have primed smoking in this group. If that were so, we would have particularly expected to see this adverse effect of active gum in the acute setting of temptations, as examined in the present analyses. The data did not confirm such an effect: active nicotine gum was equally favorable for those with FTND>0 as those whose score was zero. The apparent adverse effect of gum in the original analyses remains unexplained.

Contrary to the hypothesis that nicotine would have different effects for women versus men (21), there were no gender differences in the efficacy of nicotine gum for preventing lapses in temptation episodes.

Besides demonstrating the efficacy of nicotine gum in preventing lapses, the data also suggested that even using a non-nicotine gum in the face of temptation could help prevent lapsing. Even in the placebo group, lapses were less likely when subjects chewed gum. This is consistent with prior research suggesting that chewing confectionary gum can help reduce craving (25, 26) (though see [27]), as well as with the very common advice to smokers that chewing gum or eating candy can help avoid lapsing (28). Because the decision to use gum in a given situation was made by the subjects themselves, inferences about causality are weak: for example, it is possible that the temptations when participants used gum were also ones where they were more motivated to avoid lapsing. Propensity scores were used to control for situational factors that distinguished the situations where participants used gum, and were very successful at modeling gum use, able to correctly predict gum use 86% of the time in pairwise comparisons (29). Nevertheless, such controls are necessarily incomplete, and residual confounding was possible. The fact that the situational factors associated with gum use were reported contemporaneously with reports of the gum use itself may also weaken the propensity score controls.

Use of acute nicotine replacement medication is only one of several interventions hypothesized to help smokers get through temptations without lapsing to smoking. For example, a variety of behavioral coping responses (“immediate coping” [30]) have been proposed and are often recommended for this purpose (31, 32). The specific findings aside, the analytic approach used here suggest the potential to use EMA data to assess such hypothesized effects. As in this example, they can help distinguish the efficacy of an acute immediate coping strategy, when enacted, from the effect of recommending that strategy on overall effect on outcomes, which depends heavily on the smokers’ consistent enactment of the intervention.

The fact that, in another study, switching ITS to very-low-nicotine cigarettes caused them to reduce their smoking was taken as evidence that ITS’ smoking is motivated by nicotine-seeking (9). The fact that acute use of nicotine gum helps prevent lapses confirms this, and further adds to the evidence that ITS smoke in order to experience the acute effects of nicotine, fitting Russell’s (10) definition of “peak-seekers.” The nature of the reinforcement that ITS may receive from smoking is unclear.

These analyses had some limitations. Although the assignment to active versus placebo gum was randomized, and this supports strong causal inferences, these analyses also depended on a non-randomized, subject-determined decision to use or not use gum in a given temptation situation. To account for this, the analyses controlled for factors associated with gum use, by including propensity scores that modeled situational factors associated with use and non-use of gum. In any case, the assignment to active vs placebo gum was randomized, so the crucial contrast between those two conditions supports robust causal inferences.

The study was conducted with ITS, who differ from daily smokers in multiple ways (1, 3, 33). Thus, generalization to daily smokers is uncertain. However, given the multiple studies showing that nicotine gum (and other short-acting nicotine medications) acutely reduces craving in daily smokers (13, 14, 34), and that this is likely a crucial part of its mechanism in promoting smoking cessation in daily smokers (12), one might speculate that the results do apply to daily smokers. This suggests that users should be encouraged to use the gum to acutely combat temptations to smoke.

In summary, these analyses showed that ITS who use nicotine gum in the face of temptation are more likely to avoid lapsing and suggests the potential utility of acute nicotine medications for this population.