Press background

26th June 2000

ASH - Action on Smoking and Health

Efficacy of Zyban and NRT

Zyban launch - great for smokers but beware the hype

In its press release Glaxo makes the following claim: A large comparative study published in the New England Journal of Medicine found that almost one in three people (30.3%) treated with Zyban were not smoking at one year, compared to 16.4% of people using a nicotine patch. Zyban was thereby found to be almost twice as effective as a nicotine patch in helping people to quit smoking and remain abstinent at the one year follow up. ASH believes this is a carefully worded and true statement, but an incomplete and misleading summary of the evidence. This information sheet explains gives data on the efficacy of Zyban (bupropion) explains why many experts do not yet believe it is appropriate to describe Zyban as twice as effective as NRT.

There are two published studies on Zyban:

1. Hurt et al.A comparison of sustained release bupropion and placebo for smoking cessation. New England Journal of Medicine 1997; 337 (1195-1202.) This study of 615 (non-depressed) smokers showed 23.1% patients treated with Zyban were not smoking at one year, compared to 12.4% with placebo.
2. Jorenby et al.A controlled trial of sustained release bupropion, a nicotine patch, or both for smoking cessation. New England Journal of Medicine 1999; 340 (685-691). This is the only study in which Zyban has been compared directly to NRT. This study of 893 smokers is the basis for claims that Zyban is twice as effective as NRT. This is based on the following results from this study:
   • Not smoking at one year...
   • Placebo = 15.6%
   • Nicotine patch = 16.4%
   • Zyban = 30.3%

Given that 30.3% is nearly twice 16.4% - this has been used to support the claim that Zyban is twice as effective as NRT. The authors calculate 'odds ratios' of 1.1 for nicotine patch and 2.3 for Zyban compared to placebo at one year - [see note at end on 'odds ratios'].

Overall efficacy. The figures above are for 'point abstinence' - the number not smoking at a point in time12 months after the start of the trial. 'Continuous abstinence' - the number not smoking during the 12 months - is also used as an indicator and this is generally regarded as a more reliable indicator of the likely long term success of treatment. In the Jorenby et al study, the continuous abstinence figures are:

• Not smoking during one year
• Placebo = 5.6%
• Nicotine patch = 9.8%
• Zyban = 18.4%
• Zyban + NRT combination = 22.5%

The first thing to note is that the quit rate by this measure is closer to one in five rather than one in three and that this might be a more realistic basis for claims. However, there should always be caution with using a single study to make any generalised claim of efficacy - unpublished continuous abstinence data from the Hurt et al study suggests a lower efficacy than Jorenby et al.

It should also be noted that the Jorenby et al study was conducted in an intensive support setting - a smokers' clinic - and this may be different to the conditions in which Zyban is used in the UK, where it will be prescribed by general practitioners.

Comparison with nicotine patches. On making the Zyban to NRT comparison, the efficacy by point abstinence does show twice the efficacy. However, ASH believes it is premature and inappropriate to make this claim, given the current state of the evidence - for the following reasons.

1. This is a single study and should be replicated before drawing any conclusions.
2. The result for the nicotine patch is very poor = only 1.05 x placebo - odds ratio 1.1. Yet we know that NRT has a large evidence base (about 80 studies) suggesting that it is much more effective than this figure would suggest. So the 'twice as effective' claim is based on an anomalous result for the nicotine patch in a single study. Bupropion has only been compared with, and combined with, one fixed dosage form of NRT - a patch. The result should not be generalised to administration forms intended for ad lib use.
3. The Cochrane collaboration undertakes systematic reviews of evidence for the efficacy of smoking cessation treatments. According to the Cochrane review, the odds ratio for continuous abstinence with NRT compared to control was 1.72 (95% confidence interval 1.60 to 1.84), The odds ratios for the different forms of NRT were 1.63 for gum, 1.77 for patches, 2.27 for nasal spray, 2.08 for inhaled nicotine and 1.73 for nicotine sublingual tablet. For bupropion, the Cochrane review has pooled results of four studies (2 published) and given a combined odds ratio of 2.73. I shall resist translating these figures into a comparison of NRT and bupropion because.... www.update-software.com/cochrane/cochrane-frame.html
4. ... the Cochrane review (1999 issue 3, which is subscription only on CD) says of bupropion and nicotine replacement: "Nicotine replacement therapy (NRT) has proven efficacy in over 80 studies [...] and has a very benign side-effect profile. The early results for several anti-depressants, especially bupropion, are sufficient to endorse their use in medical practice. There is insufficient published evidence to recommend bupropion in preference to NRT or vice versa. Bupropion may also be helpful in those who fail nicotine replacement." Cochrane is probably the 'gold-standard' of evidence assessment.
5. This comparison question has also been evaluated by the US FDA (Food and Drug Administration) and is reflected in the US prescription information for bupropion: "The comparisons between Zyban, NTS [Nicotine Transdermal System] and combination treatment in this study [Jorenby et al] has not been replicated, and, therefore should not be interpreted as demonstrating the superiority of any of the active treatment arms over any other."

On the basis of the analysis above, Clive Bates, Director of ASH said:

"Zyban is an effective anti-smoking treatment, and that the first studies show very promising success rates. But until more studies have been done, it is too soon to be making comparisons between NRT and Zyban. Both products have an important role to play and will meet smokers' needs in different ways. The products may also be most effective when used in combination."

[Note on odds ratios; results and probabilities are sometimes expressed as 'odds', and comparisons made as 'odds ratios'. Roughly speaking, these work as follows... a 30% abstinence rate at 12 months means the odds of abstinence are 30:70 = 0.43. For 15% success rate, the odds are 15:85 = 0.176. Dividing these gives odds ratio - 2.4 in this example which is close to the bupropion result - this measure gives a more standardised comparison of treatment efficacy than comparing absolute success rates and enables results from treatment in different settings.

   Inhouse Pharmacy Europe