A comparison of sustained-release bupropion and placebo for smoking cessation

Authors Hurt R, Sachs D, Glover E, Offord K, Johnston J, Dale L, Sullivan P, et al. 
Source New England Journal of Medicine. 337:1195-202. October 23, 1997. 
Institutions Mayo Clinic; Palo Alto Center for Pulmonary Disease Prevention; West Virginia University; Glaxo Wellcome. 
Support Glaxo Wellcome 



The use of antidepressants as aids to smoking cessation seems reasonable. Nicotine may act as an antidepressant, and some smokers become depressed after quitting. Trials of antidepressants have had promising but also mixed results. This study was designed to evaluate a sustained release formulation of buproprion as an aid to smoking cessation (this drug was recently approved by the FDA for this purpose). 



Subjects were recruited (at three sites in the United States) through advertisements and were eligible if they were over 18 years of age, had smoked 15 cigarettes or more daily for at least one year and were generally in good health.

Exclusions: Predisposition to seizures; pregnancy or lactation; unstable medical or psychiatric conditions; current depression; anorexia or bulimia; alcohol or other non-nicotine dependence; current use of psychotropic medications, non-cigarette tobacco products or nicotine replacement therapy.


Subjects were randomized (stratified by study site) to placebo or to one of three dosages of sustained-release bupropion: 100 mg daily (50 bid), 150 mg daily (150 qd) or 300 mg daily (150 bid).

Treatment with the study drug was for 7 weeks. Smoking was to be stopped after the first week of treatment.


At baseline, historical data and smoking history were obtained as well as a physical examination, chest x-ray, EKG and laboratory work.

Follow-up visits were weekly during the 7-week treatment period, then at weeks 8, 12, 26 and 52.

Subjects filled out diaries recording smoking and withdrawal symptoms; questionnaires were administered, including evaluation of depression (Beck Depression Inventory).

Reported abstinence from smoking was validated by measuring the content of carbon monoxide in expired air. 


Smoking abstinence was assessed by two measures:

  • The point prevalence abstinence rate was defined as the rate of abstinence during the preceding week. Thus, if 30% of patients at week 6 were abstinent during the previous week (by history and confirmed by CO measurement), then the point prevalence abstinence rate at week 6 would have been 30%.
  • The rate of continuous abstinence required abstinence at all previous measurement points. Thus, if 20% of patients at week 6 had been continuously abstinent since quitting smoking at week 1, the rate of continuous abstinence at week 6 would have been 20%.
Smoking abstinence was compared between groups on an intention to treat basis. Subjects who missed a follow-up visit were assumed to have smoked during the preceding interval.

Withdrawal symptoms were recorded in a diary and a daily symptom score was calculated during the study drug period, using a 9-item instrument.

The Beck Depression Inventory was administered at baseline and at weeks 2 and 6 after the target quitting date.

Weight change was analyzed in subjects who were continuously abstinent during the treatment phase. 


Subjects and completion rates

612 subjects were enrolled, 153 in each of the four treatment groups.

Baseline characteristics were not significantly different between the treatment groups. These included (averaged across the four groups):

  • Age: 44 years; Female: 55%; White race: 96%
  • Cigarettes smoked / day: 27
  • Number of previous attempts at quitting: 4
219 subjects (36%) failed to complete all twelve months of the study; 148 dropped out during the 7 week treatment phase. The majority (89%) of dropouts withdrew consent for various reasons, including perceived lack of benefit; there were 15 dropouts due to adverse reactions and one death. Increasing doses of bupropion were associated with decreasing dropout rates (43% in the placebo group; 29% in the 300 mg group). 

Smoking cessation rates
Time Parameter Placebo 100 mg 150 mg 300 mg
1 week after quitting Not smoking 19% 31% 32% 49%
6 weeks after quitting
(end of treatment)
Not smoking (point prevalence) 19.0% 28.8% 38.6% 44.2%
Continuous abstinence 10.5% 13.7% 18.3% 24.4%
12 months
(end of study)
Not smoking (point prevalence) 12.4% 19.6% 22.9% 23.1%

Weight change

Among the 103 subjects who were continuously abstinent during the treatment period, weight gain was 2.9 kg in the placebo group, 2.3 kg in the 100 and 150 mg buproprion groups and 1.5 kg in the 300 mg group (p=0.003). At six months, however, there was no significant difference among the groups for the 59 subjects who remained continuously abstinent.


There was no significant effect of treatment on the Beck Depression Inventory scales.

Surprisingly, the only effect of treatment on the nicotine withdrawal score was a significant increase in the 100 mg group, compared with the placebo and the 150 mg and 300 mg groups.

Adverse effects

Adverse effects were similar in all four groups, although dry mouth occurred more frequently with the higher doses of buproprion (13%). Withdrawals due to adverse effects occurred in 8, 9, 7 and 13 patients in the four groups respectively. There was one death in a 63 year old woman with preexisting cardiomyopathy (cardio-pulmonary arrest 4 days after completing the 300 mg buproprion treatment). 

Author's discussion

The authors note that buproprion was effective in increasing the rate of smoking abstinence. Although there was no significant difference between the 150 mg and the 300 mg doses in terms of point-prevalence abstinence, the rate of continuous abstinence at 6 weeks was significantly increased only in the 300 mg group, and this is the dose that the authors feel is most appropriate.

The length of treatment (7 weeks) was chosen based on prior studies with nicotine replacement. Longer treatment with buproprion may be appropriate, however, particularly since the effect on limiting weight gain only persisted during active treatment.

The authors speculate that the efficacy of buproprion may be due to its effect on adrenergic and dopaminergic mechanisms in the brain, rather than on its anti-depressant properties. Although no increase in rates of seizures was noted in this study, the sample size was too small to reliably assess this potential issue (bupropion used as an antidepressant is associated with a small increase in seizure risk). [see note below] 


In this study, the antidepressant buproprion prescribed for 7 weeks was found to be effective in increasing the rates of abstinence from smoking, and was well tolerated. Although there was little difference between the doses of 150 mg and 300 mg daily, the authors recommend the higher dose based on its greater effect on continuous abstinence from smoking at the end of the treatment period.

Unfortunately, there was a fairly high dropout rate (between 29% and 43%). Importantly, the dropout rate decreased with increasing doses of the drug, and dropouts were automatically assumed to have resumed smoking (according to the methods section of the article). This fact could have biased the results in favor of buproprion. This issue might have been clarified had the authors reported the results in the subgroup that did not drop out, in addition to the "intention to treat" analysis of all subjects presented here.

This caveat aside, there is a great need for drugs that will help patients stop smoking, and buproprion may well turn out to be such an agent. Issues that will need to be clarified include the optimal duration of therapy, the effect of buproprion compared with and in conjunction with nicotine replacement therapy, and the role of other anti-depressant medications. 

November 7, 1997 

Important Note

In the original version of this summary, I stated that "antidepressents are associated with a small increase in seizure risk". This is inaccurate. In fact, "bupropion used as an antidepressant is associated with a small increase in seizure risk". This has now been rectified.

According to the PDR, the risk of seizure associated with sustained release bupropion at a daily dose of 300 mg is 0.1% (1/1000). This is small, but not negligeable. Multiple precautions are recommended to reduce the seizure risk (the drug should be avoided in patients with a history of seizure, with conditions that predispose to seizures or who are taking other medications that lower seizure threshhold).

January 7, 1998


References related to this article from the NLM's PubMed database. 

Reader Comments

January 7, 1998

In a letter to me, Dr. Richard Hurt (corresponding author of the study) points out that there was a dose-response effect throughout the treatment phase and up to one year. This was one of the major reasons the authors recommended the 300 mg dose as the best dose to be used. 

Date: Sun, 15 Mar 1998
From: melissa prew <mprew@oaktree.net>

I have a question regarding the way the results were reported in this article. I don't feel point prevalence data at 12 months necessarily reflects the effects of a medication administered 10 months ago. Isn't it possible patients resumed smoking at some point during the study, and then managed to quit again "cold turkey" independant of bupropion therapy? I don't understand why continuous quit data was reported only at 6 weeks and not reported at 12 months. 

In clinical practice I don't really care if my patients are able to remain abstinent from smoking for 1 week. I want to know if therapy with bupropion is more effective on a continuous rate for 1 year when compared to placebo. Why were the continuous quit rates at 12 months not included??

    The problem with continuous quit rates at 12 months is that any patient who smoked even once during the initial 6 weeks was classified as being not continuously abstinent. That is why the continuous quit rates at 6 weeks were so low (24.4% at the highest bupropion dose vs. 10.5% placebo). I'm not sure that a single, short-term relapse is significant. What I would have liked to see would be some measure like abstinence over the previous two months, at one year. I agree with you that point-prevalence at one year, based only on smoking during the previous 7 days, may not be the ideal measure of success. -- mj

November 19, 1998
Letters to the editor about this article were published in the February 26, 1998 NEJM. These are about the appropriate dose of buproprion and about whether the subjects were truly blinded.
Submit a comment

Journal Club homepage

Site Meter