0
Clinical Investigations: SMOKING |

Bupropion for Smoking Cessation*: Predictors of Successful Outcome FREE TO VIEW

Lowell C. Dale, MD; Elbert D. Glover, PhD; David P. L. Sachs, MD; Darrell R. Schroeder, MS; Kenneth P. Offord, MS; Ivana T. Croghan, PhD; Richard D. Hurt, MD
Author and Funding Information

*From the Nicotine Research Center (Drs. Dale, Hurt, and Croghan) and the Section of Biostatistics (Messrs. Schroeder and Offord), Mayo Clinic and Mayo Foundation, Rochester, MN; the Robert C. Byrd Health Sciences Center (Dr. Glover), West Virginia University, Morgantown, WV; and the Palo Alto Center for Pulmonary Disease Prevention (Dr. Sachs), Palo Alto, CA.

Correspondence to: Lowell C. Dale, MD, Mayo Clinic, 200 First St SW, Rochester, MN 55905; e-mail: dale.lowell@mayo.edu



Chest. 2001;119(5):1357-1364. doi:10.1378/chest.119.5.1357
Text Size: A A A
Published online

Objectives: To identify predictors of smoking abstinence at the end of medication use that could assist in the optimal use of a sustained-release (SR) form of bupropion for treating cigarette smokers.

Design: A double-blind, placebo-controlled, dose-response trial.

Setting: Multicenter (three sites) study conducted in the United States.

Participants: Six hundred fifteen healthy men and women (≥ 18 years of age) who were smoking ≥ 15 cigarettes per day and who were motivated to stop smoking.

Intervention: Random assignment of patients to placebo or SR bupropion treatment, 100, 150, or 300 mg/d, for 7 weeks (total duration of study was 52 weeks: 7 weeks of treatment and 45 weeks of follow-up).

Measurements and results: Logistic regression was used to identify predictors of abstinence at the end of the medication phase. Univariate predictors included the following: bupropion dose (p < 0.001); older age (p = 0.024); lower number of cigarettes smoked per day (cpd) (p < 0.001); lower Fagerström Tolerance Questionnaire score (p = 0.011); longest time previously abstinent that was < 24 h or > 4 weeks (p < 0.001); absence of other smokers in the household (p = 0.021); greater number of previous stop attempts (p = 0.019); and study site (p = 0.004). Multivariate predictors of abstinence at the end of the medication phase were the following: higher bupropion dose (p < 0.001); lower number of cpd (p < 0.001); longest time previously abstinent from smoking (p = 0.002); male gender (p = 0.014); and study site (p = 0.021).

Conclusion: Bupropion SR therapy was effective in treating cigarette smokers independently of all other characteristics studied. Lower smoking rate, brief periods (ie, < 24 h) or long periods (ie, > 4 weeks) of abstinence with previous attempts to stop smoking, and male gender were predictive of better outcomes, independent of the dose of bupropion that was used.

Figures in this Article

The sustained-release (SR) form of bupropion (Zyban; Glaxo-SmithKline; Research Triangle Park, NC) is an effective treatment for smokers who are trying to stop smoking.1When compared to patients receiving placebo, patients receiving a 300-mg dose of bupropion SR had a significantly higher point prevalence abstinence rate at the end of the medication phase and at 1 year, and had a higher rate of continuous abstinence from the target quit date (TQD) through the end of the medication phase. From this same study, we also have reported that a history of neither major depression nor alcoholism were predictors of smoking abstinence, although the development of depressive symptoms during the medication phase was predictive of poorer outcome.2 If it were possible to identify smokers who would be more likely to succeed or fail with bupropion therapy, treatment decisions could be made that might increase the patient’s likelihood of achieving abstinence. Increasing the intensity of the behavioral treatment, adding nicotine replacement therapy to bupropion therapy, and considering residential treatment for smokers with severe nicotine dependence are all options that could be considered if accurate predictors of abstinence were available.

Despite extensive research, accurate and consistent predictors of successful treatment have not been identified. In the clinical program at the Mayo Clinic, it has been observed that a lower score on the Fagerström Tolerance Questionnaire (FTQ), the action stage of readiness of the patient, and current smoking-related symptoms were related to predicted outcomes for smokers trying to stop.3 In a long-term follow-up of patients in a community-based smoking intervention program, those who were white-collar workers, had previously stopped smoking for > 1 month or > 1 year, those desiring to stop smoking because of health concerns, and male patients had higher rates of abstinence.4Others have observed5that higher baseline cotinine levels were predictive of poorer outcome. Two clinical trials67 of patients using nicotine patch therapy clearly demonstrated that complete abstinence during the first 2 weeks after the TQD resulted in higher success rates compared to those patients who smoked even a puff during that time. Finally, lower FTQ scores have been predictive of long-term success in some trials,810 but not predictive in others.67,11

The goal of this analysis was to identify characteristics of cigarette smokers that would predict better outcomes and, thus, would assist in the management of patients using bupropion SR therapy for smoking intervention.

Subjects

The subjects for this analysis were part of a previously reported randomized, double-blind, placebo-controlled, dose-response study1 of bupropion SR therapy that was performed at three sites (Mayo Clinic, Rochester, MN; Palo Alto Center for Pulmonary Disease Prevention, Palo Alto, CA; and West Virginia University, Morgantown, WV). The methodology is detailed in that report. An institutional review board approved the study at each site.

Briefly, subjects were randomly assigned to receive bupropion SR treatment at a dosage of 100, 150, or 300 mg/d, or placebo for 7 weeks, after which they were observed for 45 weeks. Demographic information, data on smoking history, smoking rates, and symptoms of nicotine withdrawal during the medication phase were collected. Several questionnaires, including the eight-item FTQ and the Beck depression inventory, were administered.1213 Self-reported abstinence was considered to be validated by an expired air CO level of≤ 10 ppm.

Statistical Analysis

Logistic regression was employed to assess 19 factors as possible predictors of abstinence from smoking at the end of the medication phase (ie, 6 weeks following the TQD). The dependent variable was smoking status. Abstinence from smoking was defined as no smoking (not even a puff) in the past 7 days confirmed by an expired air CO level of ≤ 10 ppm. For each potential predictor, an initial univariate logistic regression analysis was performed in which the independent variables were the predictor being considered and the dose (treated categorically as placebo vs 100 mg/d vs 150 mg/d vs 300 mg/d). The dose-by-predictor interaction term was included in these models to assess whether the effect of the given predictor was dependent on dose (or whether the effect of dose was dependent on the given predictor). After verifying that the effect of the given predictor was not dependent on the dose, a logistic regression model that included only the main effects was used to assess the effect of the given predictor after adjusting for the dose. The p value associated with the main effect of the given predictor was used to test the univariate association with abstinence from smoking.

Multiple logistic regression with the backward elimination of nonsignificant variables was used to identify a set of multivariate predictors of abstinence from smoking at the end of the medication phase. Only variables with p < 0.10 from the univariate analysis were considered as candidates in the multivariate analysis. Main effect terms for all candidate predictors were included in the first step of the model selection process. For each subsequent step, the most nonsignificant (p > 0.05) predictor was removed. This process was repeated until all nonsignificant variables were eliminated. All two-way interactions were assessed for the final subset of predictors that were identified, and those found to be statistically significant were included in the final multivariate model.

Using logistic regression, an additional analysis was performed to assess the association of bupropion dose and week 2 smoking status with abstinence from smoking at the end of the medication phase. In all cases, two-sided tests were used with p values ≤ 0.05 to denote statistical significance.

This study included 615 cigarette smokers who received placebo (n = 153), or 100 mg (n = 153), 150 mg (n = 153), or 300 mg (n = 156) bupropion SR treatment as an intervention for smoking. The 1-week biochemically confirmed point prevalence abstinence rates at the end of the medication phase were 19%, 29%, 39%, and 44% for placebo, 100 mg, 150 mg, and 300 mg bupropion SR, respectively (p < 0.001).

Table 1Table 1 (cont.) shows the results of the initial univariate analysis of potential predictors associated with abstinence from smoking at the end of the medication phase. After adjusting for the bupropion dose, those predictors found to be univariately associated with abstinence included older age (p = 0.024), a lower average number of cigarettes smoked per day (cpd) (p < 0.001), lower FTQ score (p = 0.011), no other smokers in the household (p = 0.021), the longest time of previous abstinence from smoking of < 24 h or > 4 weeks (p < 0.001), a higher number of prior serious stop attempts (p = 0.019), and the study site (p = 0.004). A higher baseline cotinine level may be associated with a lower abstinence rate, although the comparison did not reach statistical significance (p = 0.057). In all cases, the dose-by-predictor interaction was found to be nonsignificant, indicating that the effect of the given predictor was independent of the bupropion dose.

Predictors from the univariate analysis with p < 0.10 were included in the multiple logistic regression analysis to produce a multivariate model for predicting abstinence from smoking at the end of the medication phase. After backward elimination of nonsignificant variables, the final subset of predictors included the dose of bupropion (p < 0.001), the average number of cpd (p < 0.001), the longest time previously abstinent from smoking (p = 0.002), gender (p = 0.014), and the study site (p = 0.021). Since study site is potentially confounded with many other predictors, the model selection process was repeated without including the study site as a potential predictor. The final set of predictors identified using this selection process remained the same. When two-way interactions were assessed for the final subset of predictors, including study site, the interaction of the longest time previously abstinent from smoking by the average number of cpd was the only one found to be statistically significant (p = 0.020).

Table 2 uses odds ratios (ORs) to show the interpretation of the final multivariate model. For each predictor presented, an OR of 1.0 is used to indicate the reference group. The likelihood of abstinence from smoking at the end of the medication phase was found to increase with higher doses of bupropion, and all active dose groups were significantly better than those receiving placebo (bupropion SR, 100 mg/d: OR, 1.8, 95% confidence interval [CI], 1.0 to 3.2; bupropion SR, 150 mg/d: OR, 3.2; 95% CI, 1.9 to 5.6; bupropion SR, 300 mg/d: OR, 4.0; 95% CI, 2.3 to 7.0). Women were less likely to be abstinent than men (OR, 0.6; 95% CI, 0.4 to 0.9). Since the model includes the two-way interaction of the predictors longest time previously abstinent from smoking by the average cpd, those variables must be interpreted simultaneously. For these variables, the reference group corresponds to those smoking 40 cigarettes per day whose longest period of previous abstinence from smoking was < 24 h. Note that in all cases the OR is smallest for those subjects whose longest period of abstinence from smoking was 1 day to 4 weeks. Interestingly, the OR increases with fewer cpd for all subjects except those whose longest period of previous abstinence from smoking was 1 day to 4 weeks.

Although the main effects for site and dose were included in the final model, these two factors were not found to be significant for any two-way interactions. This indicates that the effects on abstinence of site and dose are independent of the effects of the other predictors of abstinence that were included in the final model. For this reason, it is possible to isolate the multivariate effect of gender, length of previous abstinence from smoking, and average number of cpd. In order to simplify this multivariate effect, an abstinence index was constructed using the regression coefficients from the final multivariate logistic regression model. Table 3 gives the formulas for the calculation of this abstinence index. Using the 33rd and 66th percentiles of the distribution of the abstinence index scores as cut points, subjects were classified as having a low likelihood (abstinence index, ≤ −1.70), a moderate likelihood (abstinence index, ≥ −1.69 to ≤ −1.24), or a high likelihood (abstinence index, ≥ −1.23) of being abstinent from smoking at the end of the medication phase. Figure 1 displays the percentage of subjects who were abstinent from smoking at the end of the medication phase, according to the abstinence index and the dose.

Overall, subjects who smoked during week 2 (ie, days 8 through 14 following the TQD) were significantly less likely to be abstinent from smoking at the end of the medication phase (p < 0.001) than those who did not smoke during week 2. In an analysis restricted to subjects who were abstinent from smoking during week 2 according to dose groups (placebo, 33 subjects; bupropion SR, 100 mg/d, 43 subjects; bupropion SR, 150 mg/d, 55 subjects; and bupropion SR, 300 mg/d, 69 subjects), the percentage of subjects abstinent from smoking at the end of the medication phase was not significantly different across the dose groups (70%, 77%, 75%, and 75%, respectively). However, among subjects who smoked during week 2 (placebo, 120 subjects; bupropion SR, 100 mg/d, 110 subjects; bupropion SR, 150 mg/d, 98 subjects; and bupropion SR, 300 mg/d, 87 subjects), there was evidence of a significant treatment effect (p = 0.003), with higher doses of bupropion being associated with increased abstinence at the end of the medication phase (5%, 10%, 18%, and 20%, respectively). A further analysis of subjects who smoked during week 2 for whom smoking rate information was available (placebo, 94 subjects; bupropion SR, 100 mg/d, 85 subjects; bupropion SR, 150 mg/d, 77 subjects; and bupropion SR, 300 mg/d, 71 subjects) revealed significant differences in the smoking rate at week 2 across the four dose groups with median (25th, 75th percentile) rates of 5.4 cpd (1.7, 11.0 cpd), 2.1 cpd (0.3, 7.0 cpd), 2.6 cpd (0.7, 6.7 cpd), and 1.3 cpd (0.5, 3.4 cpd), respectively (p = 0.007 [one-way analysis of variance that adjusted for study site]). Across all dose groups, only 2 of 64 subjects who self-reported ≥ 10 cpd during week 2 were biochemically confirmed to be abstinent from smoking at the end of the medication phase. Among those subjects smoking during week 2 who self-reported < 10 cpd for week 2 (placebo, 68 subjects; bupropion SR, 100 mg/d, 71 subjects; bupropion SR, 150 mg/d, 61 subjects; and bupropion SR, 300 mg/d, 63 subjects), the biochemically confirmed end-of-medication-phase abstinence rates were 8.8%, 15.5%, 27.9%, and 25.4%, respectively (p = 0.016 [logistic regression analysis that adjusted for study site]).

We have identified the following strong predictors of successful abstinence from smoking in a population that used variable doses of bupropion: higher bupropion dose; lower number of cpd; length of abstinence with previous stop attempts; and male gender. Other variables found to be univariately predictive of greater abstinence at the end of medication phase included abstinence from smoking during the second week following the TQD, older age, lower FTQ, absence of other smokers in the household, and greater number of previous stop attempts. Identifying such predictors is a critical step in tailoring pharmacologic and behavioral interventions to the needs of the individual. Those cigarette smokers with many positive predictors may need only brief advice and appropriate therapy with pharmacologic adjuncts. Those smokers with few positive predictors likely will need more intensive interventions such as prolonged counseling, a combination of pharmacologic agents, and more frequent follow-ups. For the busy practitioner, using these predictors in conjunction with an assessment of the stage of readiness of the smoker14 should allow for rapid assessment and the determination of an appropriate treatment plan.

Although study site was found to be a significant predictor of outcome, we believe that the findings are generalizable. The multivariate backward elimination algorithm was performed twice, first with study site included as a potential predictor and then without it. With the exception of the study site, the same subset of predictors was identified in both of these analyses. All two-way interactions were assessed for the final subset of predictors, including study site, and no site-by-predictor interactions were found to be significant. Thus, we are confident that the effect of study site was independent of the other predictors included in the final model and cannot be explained by any of the other variables that we examined. The significance of study site may have been due to differences in the delivery of the intervention or differences in the overall conduct of the study. What is of most importance is that the treatment effect is consistent across study sites (ie, no site-by-treatment interaction) and that the predictors identified (dose, average number of cpd, longest time previously abstinent from smoking, and gender) were independent of the effect of study site, indicating that these predictors are generalizable.

As described in our previous report,1 higher doses of bupropion of up to 300 mg/d were associated with better abstinence rates. This efficacy of bupropion therapy was independent of any of the smoker characteristics tested. Based on this information, the practitioner can prescribe bupropion to diverse populations of patients and can expect beneficial results.

We also identified several characteristics of this study population that predicted abstinence from smoking that were found to be independent of the bupropion dose. These included male gender, length of previous abstinence, and number of cpd.

Interestingly, male gender predicted greater abstinence rates than did female gender. The 1996 Agency for Health Care Policy and Research Guideline review of the literature15revealed no consistent differences between men and women in abstinent rates. However, other studies1617 have noted that women have more trouble quitting than men. This has been attributed to women’s greater concerns about weight gain when they stop smoking18and to this fear of weight gain as a more frequently reported precipitant for relapse compared to men.19Women also have higher rates of depression than men and are more likely to use smoking as a means of managing negative affect.2021 It does not appear that bupropion therapy influences either of these issues in women significantly enough to result in abstinence rates that are comparable to those of men.

Longest time of previous abstinence exhibits a U-shaped association with abstinence outcomes; ie, better success was noted in those subjects reporting only brief periods (< 24 h) or long periods (> 4 weeks) of previous abstinence. This finding was also reported in a retrospective study22 of elderly smokers. Intuitively, those subjects who have experienced long periods of previous abstinence might be more successful because they could draw on their past success. However, higher success rates in the groups with no previous abstinence or with very brief periods are more difficult to explain. Perhaps the individuals in those groups had not tried to stop smoking previously or had not been very motivated until their participation in this study. Or, previous stop attempts may not have been of sufficient duration or of serious enough intent to foster fear of failure or to diminish self-efficacy, which may be encountered in those subjects with multiple failed stop attempts.

The average number of cpd over the previous year was also a significant predictor of abstinence. However, the importance of this variable was dependent on the subject’s longest time of previous abstinence. Those subjects whose longest time of previous abstinence was 1 day to 4 weeks had the poorest abstinence outcomes, and abstinence from smoking was not associated with baseline smoking level in this group. In all of the other duration groups, there was a negative association indicating that a lower smoking rate was associated with improved abstinence outcomes. We have been unable to explain this finding.

Over 70% of subjects who were not smoking at the end of week 2, regardless of the bupropion SR dose, were abstinent at the end of the medication phase. Among those subjects who did smoke during the second week following their TQD, there was a significant number of additional subjects receiving bupropion who became abstinent by the end of the medication phase. One of our nicotine patch trials found comparable results; 74% of subjects not smoking at the end of week 2 were abstinent at the end of the medication phase, but, of those smoking during week 2 after the TQD, only 6% of placebo patch subjects vs 23% of active patch subjects were abstinent at the end of the medication phase.7 Bupropion therapy also appears to improve abstinence at the end of the medication phase, especially for those subjects receiving higher doses of bupropion SR and fewer cpd during the second week after the TQD. The proposed mechanism of efficacy of bupropion for smoking cessation (ie, the inhibition of the reuptake of dopamine) could have a long lead-in time in some smokers, thus accounting for efficacy in those subjects who had decreased their smoking rate but were not abstinent during the first 2 weeks following the TQD.

How do we apply this finding to our patients? The first 2 weeks of a stop attempt are the most crucial. Complete abstinence at week 2 indicates a high likelihood of remaining abstinent, regardless of the method used or the dosage of medication received. The key is to aggressively treat the patient with pharmacologic agents during the first 2 weeks to achieve abstinence, in addition to other support that is currently not the norm. This support might include more frequent follow-up visits and/or telephone calls and more intensive behavioral therapy. Moreover, a significant additional benefit from bupropion therapy may be gained in those subjects who continue to smoke early in their stop attempt, and this may justify the continued use of bupropion therapy for several more weeks in those patients.

The findings of this study have been used to construct an abstinence index. Using the variables of longest time previously abstinent from smoking, male gender, and average smoking rate in cpd, this index predicts abstinence at the end of treatment. This score could be used as a stratification variable in future studies. From a clinical standpoint, those subjects with high abstinence index scores may have success with bupropion therapy and a brief behavioral intervention similar to the National Cancer Institute program used in this trial. However, those subjects with moderate and low index scores may need more intensive treatment such as combining bupropion therapy with nicotine replacement therapy (eg, patches, nicotine nasal spray, gum, or inhaler) and/or providing more intensive behavioral counseling. For those subjects in the lowest abstinence index score ranges, combination pharmacotherapy and even more intensive behavioral counseling with a consideration of residential treatment23 should be considered to maximize the chances for success. The validation of this index in subsequent studies would be worthwhile, although it is based on variables that have been repeatedly reported to be predictive of abstinence at the end of both the medication phase and long-term follow-up in other studies.

In summary, this analysis identified the following four important concepts that clinicians may find valuable in treating their patients who smoke:

  1. Bupropion therapy is effective for a wide range of smokers. We did not identify any characteristics of smokers that significantly influenced the efficacy of bupropion therapy.

  2. The following characteristics were identified that predicted greater rates of abstinence from smoking that were independent of bupropion use: male gender; length of abstinence from smoking with previous stop attempts (< 24 h or > 4 weeks); and lower number of cpd.

  3. Smoking status during the first 2 weeks after the TQD is an important predictor of long-term abstinence.

  4. An abstinence index constructed from this analysis is predictive of success in stopping smoking.

Abbreviations: CI = confidence interval; cpd = cigarettes smoked per day; FTQ = Fagerström Tolerance Questionnaire; OR = odds ratio; SR = sustained release; TQD = target quit date

Supported by a grant from Glaxo Wellcome, Inc, Research Triangle Park, NC.

Received December 30, 1999; revision accepted November 21, 2000.

Table Graphic Jump Location
Table 1. Summary of Univariate Analysis of Potential Predictors of Smoking Status at the End of the Medication Phase*
Table Graphic Jump Location
Table 2. ORs From Final Multivariate Model Predicting Abstinence From Smoking at the End of the Medication Phase*
* 

Predictors from the univariate analysis with p < 0.10 were included in the multiple logistic regression analysis to produce a multivariate model for predicting abstinence from smoking at the end of the medication phase. After backward elimination of nonsignificant variables, the final subset of predictors included dose (treated categorically, p < 0.001), average cpd (p < 0.001), longest time previously abstinent from smoking (p = 0.002), gender (p = 0.014), and study site (p = 0.021). When two-way interactions were assessed for the final subset of predictors, the interaction of longest period of prior abstinence-by-average cpd was the only interaction found to be statistically significant (p = 0.020). ORs were used to show the interpretation of the final multivariate model. For each predictor, an OR of 1.0 is used to indicate the reference group.

Table Graphic Jump Location
Table 3. Formulas for Calculating Abstinence Index*
* 

Although the main effects for study site and dose were included in the final multivariate model, these two variables were not found to be significant for any two-way interactions. This indicates that the effects of site and dose are independent of the effects of the other predictors included in the final model. For this reason, it is possible to isolate the multivariate effect of gender, longest period of prior abstinence, and average cpd. In order to simplify this multivariate effect, an abstinence index was constructed using the regression coefficients from the final multivariate logistic regression model.

Figure Jump LinkFigure 1. Percentage of subjects not smoking at the end of treatment. Smoking abstinence at the end of the medication phase according to the abstinence index (AI) and dose using the 33rd and 66th percentiles of the distribution of AI scores as cut points for the likelihood of being abstinent from smoking: low likelihood (AI,≤ −1.70); moderate likelihood (AI, ≥ −1.69 ≤ −1.24); or high likelihood (AI, ≥ −1.23).Grahic Jump Location
Table Graphic Jump Location
Table 1A. Continued
* 

Abst = subjects who had abstained from smoking at end of medication phase. Discrepancies in number totals are the results of missing data.

 

The p values presented are those from a logistic regression model which includes the main effects of the given predictor and dose. The p value associated with the main effect of the given predictor after adjusting for dose was used to test the“ univariate” association of the predictor with abstinence from smoking at the end of the medication phase.

 

For all analyses, dose was treated categorically (placebo vs 100 mg/d vs 150 mg/d vs 300 mg/d).

§ 

The given predictor was treated as a continuous variable in the logistic regression model.

We acknowledge Michelle Volpato for her assistance with the data analysis.

Hurt, RD, Sachs, DPL, Glover, ED, et al (1997) A comparison of sustained-release bupropion and placebo for smoking cessation.N Engl J Med337,1195-1202. [CrossRef] [PubMed]
 
Hayford, KE, Patten, CA, Rummans, TA, et al Effectiveness of bupropion for smoking cessation in smokers with a former history of major depression or alcoholism.Br J Psychol1999;174,173-178. [CrossRef]
 
Rohren, CL, Croghan, IT, Hurt, RD, et al Predicting smoking cessation outcome using stage of readiness: contemplation versus action.Prev Med1994;23,335-344. [CrossRef] [PubMed]
 
Hurt, RD, Offord, KP, Hepper, NGG, et al Long-term follow-up of persons attending a community-based smoking-cessation program.Mayo Clin Proc1988;63,681-690. [PubMed]
 
Hall, MH, Herning, RI, Reese, TJ, et al Blood cotinine levels as indicators of smoking treatment outcome.Clin Pharmacol Ther1984;35,810-814. [CrossRef] [PubMed]
 
Kenford, SL, Fiore, MC, Jorenby, DR, et al Predicting smoking cessation: who will quit with and without the nicotine patch.JAMA1994;271,589-594. [CrossRef] [PubMed]
 
Hurt, RD, Dale, LC, Fredrickson, PA, et al Nicotine patch therapy for smoking cessation combined with physician advice and nurse follow-up: one-year outcome and percentage of nicotine replacement.JAMA1994;271,595-600. [CrossRef] [PubMed]
 
Pinto, RP, Abrams, DB, Monti, PM, et al Nicotine dependence and likelihood of quitting smoking.Addict Behav1987;12,371-374. [CrossRef] [PubMed]
 
Killen, JD, Fortmann, SP, Kraemer, HC, et al Who will relapse?: symptoms of nicotine dependence predict long-term relapse after smoking cessationJ Consult Clin Psychol1992;60,797-801. [CrossRef] [PubMed]
 
Sachs, DP, Sawe, U, Leischow, SJ Effectiveness of a 16-hour transdermal nicotine patch in a medical practice setting without intensive group counseling.Arch Intern Med1993;153,1881-1890. [CrossRef] [PubMed]
 
Jorenby, DE, Smith, SS, Fiore, MC, et al Varying nicotine patch dose and type of smoking cessation counseling.JAMA1995;274,1347-1352. [CrossRef] [PubMed]
 
Fagerström, KO Measuring degree of physical dependence to tobacco smoking with reference to individualization of treatment.Addict Behav1978;3,235-241. [CrossRef] [PubMed]
 
Beck, AT, Speer, RA. Beck depression inventory. 1978; Center for Cognitive Therapy. Philadelphia, PA:.
 
Prochaska, JO, Goldstein, MG Process of smoking cessation: implications for cliniciansClin Chest Med1991;12,727-735. [PubMed]
 
The Smoking Cessation Clinical Practice Guideline Panel and Staff. The Agency for Health Care Policy and Research smoking cessation clinical practice guidelineJAMA1996;275,1270-1280. [CrossRef] [PubMed]
 
Bjornson, W, Rand, C, Connett, JE, et al Gender differences in smoking cessation after 3 years in the Lung Health Study.Am J Public Health1995;85,223-230. [CrossRef] [PubMed]
 
Swan, GE, Jack, LM, Ward, MM Subgroups of smokers with different success rates after use of transdermal nicotine.Addiction1997;92,207-217. [CrossRef] [PubMed]
 
Pirie, PL, Murray, DM, Luepker, RV Gender differences in cigarette smoking and quitting in a cohort of young adults.Am J Public Health1991;81,324-327. [CrossRef] [PubMed]
 
Swan, GE, Ward, MM, Carmelli, D, et al Differential rates of relapse in subgroups of male and female smokers.J Clin Epidemiol1993;46,1041-1053. [CrossRef] [PubMed]
 
Breslau, N Psychiatric comorbidity of smoking and nicotine dependence.Behav Genet1995;25,95-101. [CrossRef] [PubMed]
 
Livson, N, Leino, EV Cigarette smoking motives: factorial structure and gender differences in a longitudinal study.Int J Addict1988;23,535-544. [PubMed]
 
Dale, LC, Olsen, AD, Patten, CA, et al Predictors of smoking cessation among elderly smokers treated for nicotine dependence.Tob Control1997;6,181-187. [CrossRef] [PubMed]
 
Hurt, RD, Dale, LC, Offord, KP, et al Inpatient treatment of severe nicotine dependence.Mayo Clin Proc1992;67,823-828. [PubMed]
 

Figures

Figure Jump LinkFigure 1. Percentage of subjects not smoking at the end of treatment. Smoking abstinence at the end of the medication phase according to the abstinence index (AI) and dose using the 33rd and 66th percentiles of the distribution of AI scores as cut points for the likelihood of being abstinent from smoking: low likelihood (AI,≤ −1.70); moderate likelihood (AI, ≥ −1.69 ≤ −1.24); or high likelihood (AI, ≥ −1.23).Grahic Jump Location

Tables

Table Graphic Jump Location
Table 1. Summary of Univariate Analysis of Potential Predictors of Smoking Status at the End of the Medication Phase*
Table Graphic Jump Location
Table 2. ORs From Final Multivariate Model Predicting Abstinence From Smoking at the End of the Medication Phase*
* 

Predictors from the univariate analysis with p < 0.10 were included in the multiple logistic regression analysis to produce a multivariate model for predicting abstinence from smoking at the end of the medication phase. After backward elimination of nonsignificant variables, the final subset of predictors included dose (treated categorically, p < 0.001), average cpd (p < 0.001), longest time previously abstinent from smoking (p = 0.002), gender (p = 0.014), and study site (p = 0.021). When two-way interactions were assessed for the final subset of predictors, the interaction of longest period of prior abstinence-by-average cpd was the only interaction found to be statistically significant (p = 0.020). ORs were used to show the interpretation of the final multivariate model. For each predictor, an OR of 1.0 is used to indicate the reference group.

Table Graphic Jump Location
Table 3. Formulas for Calculating Abstinence Index*
* 

Although the main effects for study site and dose were included in the final multivariate model, these two variables were not found to be significant for any two-way interactions. This indicates that the effects of site and dose are independent of the effects of the other predictors included in the final model. For this reason, it is possible to isolate the multivariate effect of gender, longest period of prior abstinence, and average cpd. In order to simplify this multivariate effect, an abstinence index was constructed using the regression coefficients from the final multivariate logistic regression model.

Table Graphic Jump Location
Table 1A. Continued
* 

Abst = subjects who had abstained from smoking at end of medication phase. Discrepancies in number totals are the results of missing data.

 

The p values presented are those from a logistic regression model which includes the main effects of the given predictor and dose. The p value associated with the main effect of the given predictor after adjusting for dose was used to test the“ univariate” association of the predictor with abstinence from smoking at the end of the medication phase.

 

For all analyses, dose was treated categorically (placebo vs 100 mg/d vs 150 mg/d vs 300 mg/d).

§ 

The given predictor was treated as a continuous variable in the logistic regression model.

References

Hurt, RD, Sachs, DPL, Glover, ED, et al (1997) A comparison of sustained-release bupropion and placebo for smoking cessation.N Engl J Med337,1195-1202. [CrossRef] [PubMed]
 
Hayford, KE, Patten, CA, Rummans, TA, et al Effectiveness of bupropion for smoking cessation in smokers with a former history of major depression or alcoholism.Br J Psychol1999;174,173-178. [CrossRef]
 
Rohren, CL, Croghan, IT, Hurt, RD, et al Predicting smoking cessation outcome using stage of readiness: contemplation versus action.Prev Med1994;23,335-344. [CrossRef] [PubMed]
 
Hurt, RD, Offord, KP, Hepper, NGG, et al Long-term follow-up of persons attending a community-based smoking-cessation program.Mayo Clin Proc1988;63,681-690. [PubMed]
 
Hall, MH, Herning, RI, Reese, TJ, et al Blood cotinine levels as indicators of smoking treatment outcome.Clin Pharmacol Ther1984;35,810-814. [CrossRef] [PubMed]
 
Kenford, SL, Fiore, MC, Jorenby, DR, et al Predicting smoking cessation: who will quit with and without the nicotine patch.JAMA1994;271,589-594. [CrossRef] [PubMed]
 
Hurt, RD, Dale, LC, Fredrickson, PA, et al Nicotine patch therapy for smoking cessation combined with physician advice and nurse follow-up: one-year outcome and percentage of nicotine replacement.JAMA1994;271,595-600. [CrossRef] [PubMed]
 
Pinto, RP, Abrams, DB, Monti, PM, et al Nicotine dependence and likelihood of quitting smoking.Addict Behav1987;12,371-374. [CrossRef] [PubMed]
 
Killen, JD, Fortmann, SP, Kraemer, HC, et al Who will relapse?: symptoms of nicotine dependence predict long-term relapse after smoking cessationJ Consult Clin Psychol1992;60,797-801. [CrossRef] [PubMed]
 
Sachs, DP, Sawe, U, Leischow, SJ Effectiveness of a 16-hour transdermal nicotine patch in a medical practice setting without intensive group counseling.Arch Intern Med1993;153,1881-1890. [CrossRef] [PubMed]
 
Jorenby, DE, Smith, SS, Fiore, MC, et al Varying nicotine patch dose and type of smoking cessation counseling.JAMA1995;274,1347-1352. [CrossRef] [PubMed]
 
Fagerström, KO Measuring degree of physical dependence to tobacco smoking with reference to individualization of treatment.Addict Behav1978;3,235-241. [CrossRef] [PubMed]
 
Beck, AT, Speer, RA. Beck depression inventory. 1978; Center for Cognitive Therapy. Philadelphia, PA:.
 
Prochaska, JO, Goldstein, MG Process of smoking cessation: implications for cliniciansClin Chest Med1991;12,727-735. [PubMed]
 
The Smoking Cessation Clinical Practice Guideline Panel and Staff. The Agency for Health Care Policy and Research smoking cessation clinical practice guidelineJAMA1996;275,1270-1280. [CrossRef] [PubMed]
 
Bjornson, W, Rand, C, Connett, JE, et al Gender differences in smoking cessation after 3 years in the Lung Health Study.Am J Public Health1995;85,223-230. [CrossRef] [PubMed]
 
Swan, GE, Jack, LM, Ward, MM Subgroups of smokers with different success rates after use of transdermal nicotine.Addiction1997;92,207-217. [CrossRef] [PubMed]
 
Pirie, PL, Murray, DM, Luepker, RV Gender differences in cigarette smoking and quitting in a cohort of young adults.Am J Public Health1991;81,324-327. [CrossRef] [PubMed]
 
Swan, GE, Ward, MM, Carmelli, D, et al Differential rates of relapse in subgroups of male and female smokers.J Clin Epidemiol1993;46,1041-1053. [CrossRef] [PubMed]
 
Breslau, N Psychiatric comorbidity of smoking and nicotine dependence.Behav Genet1995;25,95-101. [CrossRef] [PubMed]
 
Livson, N, Leino, EV Cigarette smoking motives: factorial structure and gender differences in a longitudinal study.Int J Addict1988;23,535-544. [PubMed]
 
Dale, LC, Olsen, AD, Patten, CA, et al Predictors of smoking cessation among elderly smokers treated for nicotine dependence.Tob Control1997;6,181-187. [CrossRef] [PubMed]
 
Hurt, RD, Dale, LC, Offord, KP, et al Inpatient treatment of severe nicotine dependence.Mayo Clin Proc1992;67,823-828. [PubMed]
 
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

CHEST Journal Articles
CHEST Collections
PubMed Articles
Guidelines
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543