Study enrollment began in 2001 and was completed in October 2002 (see ). Data collection was completed in January 2004. All analyses were conducted using SPSS 14.0 software. All analyses are intent-to-treat unless otherwise noted.
CONSORT figure of participant flow through the study.
Six hundred and eight smokers (57.9% women) participated in this study (see for demographic information). Treatment conditions did not differ significantly (p > .05) on any demographic variables or tobacco dependence indicators (e.g., Fagerström Test of Nicotine Dependence, cigarettes smoked per day) or depression symptoms (e.g., Center for Epidemiologic Studies Depression Scale score; data not shown). In addition, during treatment there was no differential attrition across treatment conditions (χ2 = 2.86, p = .24). At 73.1% of study visits it was judged that participants were taking their pills as directed by study staff (i.e., they returned the correct number of pills plus or minus two) and compliance did not differ amongst treatment conditions (χ2 = 3.69, p = .16). Participants across the three treatment conditions did not differ in gum use (M = 4.17 pieces per day [SD = 3.4], F (2, 412) = .26, p = .77). A total of 781 adverse events were reported in this study. The most common adverse events were insomnia (4.73% of all adverse events), headache (2.59% of all adverse events) and cold symptoms (2.57% of all adverse events).
The first hypothesis was that active pharmacotherapy (AA and AP groups) would improve cessation rates over placebo. A Cox Regression analysis of latency to relapse (defined as number of days to three consecutive days of smoking) conducted using 12-month follow-up data indicated that individuals who received active pharmacotherapy were statistically more likely to be abstinent than were individuals who received only placebo pharmacotherapy (AA: Wald = 11.64, OR = .64, p < .01; AP: Wald = 7.44, OR = .70, p < .01; see for survival curves). Logistic regression was used to predict CO-confirmed 7-day point-prevalence abstinence for the four different time-points with the double placebo condition (PP) coded as the baseline condition. See for abstinence rates. Analyses revealed that at 1 week post-quit, both the AA (Wald = 23.41, OR = 3.12, p < .01) and AP (Wald = 10.38, OR = 2.15, p < .01) groups were significantly more likely to be abstinent than the PP group. These results also were true at the end of treatment (AA: Wald = 18.44, OR = 2.95, p < .01; AP: Wald = 9.17, OR = 2.17, p < .01) and at 6 months post-quit (AA: Wald = 3.78, OR = 1.71, p = .05; AP: Wald = 5.92, OR = 1.88, p = .02). At 12 months post-quit the active treatment conditions no longer differed statistically from the double placebo condition (AA: Wald = 3.21, OR = 1.67, p = .07; AP: Wald = 1.85, OR = 1.48, p = .17).
Survival curves for the three treatment conditions
7-day point prevalence abstinence rates (%) at various follow-up time points for each experimental condition
To determine whether combining 4-mg nicotine gum with active bupropion (AA) improved cessation rates above those produced by active bupropion alone (AP) logistic regression was used to analyze biochemically-confirmed 7-day point prevalence abstinence. The AA condition was coded as the baseline condition. Results revealed that relative to the individuals in the AA condition, those in the AP condition were significantly less likely to be abstinent at 1 week post-quit (Wald = 3.74, OR = .69, p = .05). However, there were no statistical differences in abstinence rates between the AA and AP groups at the end of treatment, 6-month post-quit or 12-month post-quit time points. The lack of significant differences between the two groups beyond the first week was confirmed in survival analyses.
In addition to analyzing relapse data, we examined latency to relapse (smoking at least one cigarette on three consecutive days) following a lapse (smoking a cigarette, even one puff). On average, participants managed to delay relapse for 47.57 (SD = 105.37) days after having their first cigarette following their quit attempt. We found that 65.7% of individuals who relapsed by the end of the study returned to daily smoking the same day they lapsed. Results of a linear regression indicated that treatment condition did not predict latency to relapse after a lapse nor did gender, depressive symptoms or nicotine dependence.
Another outcome of interest is whether particular pharmacotherapies prevent or delay weight gain following a smoking cessation attempt. A univariate ANOVA revealed no significant effects of treatment condition on weight gain between baseline and the end of treatment, with participants gaining an average of 1.53 kilograms (SD = 2.41). This was also true when weight was analyzed only in participants who were abstinent at the end of treatment. There was a main effect of gender such that women gained fewer kilograms by the end of treatment (M = 1.23, SD = 2.38) than did men (M = 1.90, SD = 2.41; t (417) = 2.85, p < .05); however, the gender by treatment interaction was not statistically significant.
Moderation of Treatment Effects
Gender, age, race, number of cigarettes smoked per day, length of longest previous quit attempt, number of quit attempts, history of depression and dependence measures (Fagerström Test of Nicotine Dependence, Wisconsin Inventory of Smoking Dependence Motives-68 scales (WISDM-68), and Tobacco Dependence Screener) were tested as moderators using logistic regression. We modeled the main effects of treatment and the individual difference variables and the treatment by individual difference interactions on smoking status at all four follow-up time points (one week, end of treatment, six months and twelve months). There were significant main effects of gender at all three follow-up time points, after accounting for treatment, such that women were significantly more likely to be smoking than were men (one week: Wald = 7.97, OR = .61, p = .01; end of treatment: Wald = 5.75, OR = .65, p = .02; six months: Wald = 7.19, OR = .59, p = .01). There was a significant gender x treatment interaction, but only at one week post-quit (Wald = 4.46, OR = .61, p = .01). Women who received placebo pharmacotherapy were significantly more likely to be smoking (88.9%) than were men who received active (54.2%) or placebo (63.6%) pharmacotherapy or women who received active pharmacotherapy (60.7%). Length of previous quit attempts, race, the Fagerström Test of Nicotine Dependence, the WISDM-68 Automaticity subscale, the WISDM-68 Social/Environmental Goads subscale, and the WISDM-68 Tolerance subscale were all significantly related to outcome, but there were no significant interactions between these variables and treatment. There were no moderating effects for any individual difference variables.