The clinical significance of ‘small’ effects of smoking cessation treatments

Questions are sometimes raised about whether the size of effects of some treatments to help people stop smoking are large enough to be clinically significant. This issue is of direct relevance when it comes to decisions about licensing medications to aid cessation, and decisions to cover the costs of smoking cessation treatments through health insurance or public funds. It may also affect the willingness of individual health-care practitioners to adopt these treatments as part of their practice and the standing of smoking cessation as a clinical speciality. In most countries, an increase in the percentage of smokers who abstain for at least 6 months of 7–9% number needed to treat (NNT): 11–141, 2 has been accepted as clinically significant by licensing authorities evaluating nicotine replacement therapies and bupropion SR (Zyban). National and international guidelines have accepted that effect sizes defined in the same way of 4% (NNT: 25) are clinically significant; this is the size of effect obtained by face-to-face counselling 3. Guidelines have also argued that an effect of as little as 2% (NNT: 50) is of clinical significance (e. g. 4; http: //www. treatobacco. net) ; this is the estimated effect of brief physician advice to stop smoking given opportunistically to unselected smokers as part of a routine consultation 5. In the United Kingdom, the technology review body National Institute for Health and Clinical Excellence (NICE) deemed that, given the cost of nicotine replacement therapy (NRT) and bupropion and the health benefits of stopping smoking, these medications represented good value for money as life-saving treatments 6. Even on worst-case assumptions, the cost per quality-adjusted life year gained was judged to be well below £5000, while the threshold normally set for judging a treatment to be worth funding was £20–30 000. Nevertheless, the issue of effect sizes continues to present an obstacle with regard to decisions concerning the licensing and funding of smoking cessation treatments. There appears to be no general analysis in the literature regarding what constitutes a minimum clinically significant effect size that could be applied across a range of health-care systems in different countries. The clinical significance of treatment for smoking cessation derives from three premises: (1) stopping smoking improves future health and prevents premature death; (2) stopping smoking is difficult for many smokers largely because of nicotine dependence, which is a neuropsychiatric disorder; and (3) treatments for this disorder increase the chances of long-term abstinence from cigarettes. These premises are considered in turn below. The effect of stopping smoking on premature mortality is now well established 7. For the average smoker in the United Kingdom, stopping before the age of 40 years almost normalizes life expectancy, adding approximately 9 years of life compared with continuing to smoke; stopping at the age of 50 years adds approximately 6 years while stopping at the age of 60 adds approximately 3 years. The years of life that are added are ‘healthy life-years’; in fact continuing smokers spend more of their lives with higher levels of disease and disability than ex-smokers 8. Nicotine dependence is a well-established disorder in which repeated ingestion of acute doses of nicotine from cigarettes leads to changes in neural pathways in the brain, leading to needs and urges that overwhelm and undermine attempts at restraint 9, 10. The precise mechanism by which this happens is unknown, but evidence points to an important role for the up-regulation of alpha4-beta2 nicotinic cholinergic receptors and functional changes in the dopaminergic pathways leading from the ventral tegmental area to the nucleus accumbens (see 11). Evidence shows that failure of attempts to stop smoking is not related to weaker beliefs about the adverse health effects of smoking or stronger beliefs about the benefits of smoking; it is tied much more closely to behavioural markers of neuropsychological processes that are unrelated to rational choice 12. It is important to recognize, however, that many factors contribute to whether or not someone smokes and that nicotine dependence is just one of these, albeit an important one. Treatments designed to aid smoking cessation work on the basis that nicotine dependence can, in some cases, be overcome by: (1) using psychological support to help sustain abstinence while the brain ‘normalizes’ andattempting to reshape the smoker's thought processes to reduce non-pharmacological smoking motives; and (2) using pharmaceutical aids to reduce the strength of needs and urges to smoke. A robust index of effectiveness of a course of treatment is the increase in the proportion of smokers who achieve at least 6 months of continuous abstinence 13. The reason for adopting this index is that our estimates of health gains from stopping smoking are based mainly on permanent cessation versus continued smoking, and 6 months of continuous abstinence provides sufficient information to make reliable estimates of permanent cessation. Some licensing authorities use a shorter period of abstinence, such as 4 weeks, but this faces a number of problems. The most serious of these is that estimating permanent cessation rates from 4-week abstinence rates is too imprecise and requires too many assumptions to be made about relapse rates after this point, especially when treatment continues past this point. Another approach is to examine the effect in the last 4 weeks of treatment. Again, this suffers from the problem of imprecise estimation of permanent cessation and the possibility that resumption of smoking may merely have been delayed by the treatment rather than prevented. It is also common to use ‘point prevalence’ abstinence at various follow-up points, which is simply whether the subject is smoking at the time. Again, this suffers from the drawback that it does not permit accurate prediction of permanent abstinence because the subject may have stopped smoking only very recently and may still be at high risk of relapse. Considering 6 months of continuous abstinence as the key outcome measure, we can calculate with some degree of confidence how different effect sizes will translate into reductions in premature deaths. On this basis, it is easy to demonstrate that a treatment effect of as little as 1% is clinically significant. One can also show that a treatment cost of £100 (€150 or US200) per treatment episode per 1% increase in 6-month continuous abstinence rates represents considerably better value for money than the majority of treatments currently adopted in western medicine. It is crucial to note, however, that this 1% effect size must be demonstrated robustly. A basis for this calculation is as follows: It has been found that there is approximately 20% relapse between 6 months and 12 months 14, approximately 60% of those abstinent for 12 months remain so for at least 8 years 15 and there is no difference in relapse rates after 12 months between treated and untreated abstainers 15. There is therefore what may be considered a permanent treatment effect on abstinence amounting to 50% of the effect at 6 months. This means that, from 1% of those who achieve 6 months of continuous abstinence, we would expect 0. 5% to maintain permanent abstinence. Untreated smokers stop smoking at an average rate of about 2% per year (see 16) and so the average age of stopping of 40-year-old untreated smokers is about 65 years. Smokers aged 40 years who stop smoking permanently gain an average of 9 years of life (see above), while stopping at 65 years gains an average of approximately 3 years. Therefore, the 0. 5% of treated smokers who stop at 40 years rather than an average of 65 years gain an average of about 6 years (9 years less 3 years) of life. This means that, for every 100 smokers treated, there will be roughly 3 years of life gained. This is clearly a significant health gain. If treating 100 of these smokers cost £10 000 (£100 per treatment episode), the cost per year of life gained would be approximately £3300 or £6600 ‘discounted’‘Discounting’ is an adjustment used in economics to represent a reduced value of gains made in future years versus in the present. If one discounts future life-year gains at 3. 5% per year over 20 years, the cost per life year gained estimates increase by about 100%. In a recent review, the median cost per quality-adjusted year of life gained for life-saving treatments was approximately 20 000 (£10 000), but the cost was higher for studies conducted in Europe and the United States, better-quality studies and studies not funded by industry 17. Clearly, in this context a 1% effect size at 6 months for a treatment costing £100 would represent extremely good value for money. The cost of an additional year of life for a 60-year-old smoker on this basis would be no more than £6600 (£13 200 discounted) (based on an average gain of 3 years of life), still around the average for life-saving treatments. Note also that this calculation is for cost per life-year gained, not quality-adjusted life-year gained. The figure for quality-adjusted life-year gained would be more favourable. It may be surprising to some observers that such small effect sizes can have such high levels of clinical significance and cost-effectiveness. The reason is, of course, that the health gains from stopping smoking are so large and £100 is very economical for a once-only course of treatment that prevents premature death. A course of NRT or bupropion increases 6-month continuous abstinence rates by between 7% and 9% (see above) and costs an average of less than £100 (given that it is usually purchased in weekly or fortnightly blocks and smokers stop treatment if they resume smoking). For every 100 40-year-old smokers treated, this yields between 21 and 27 years of life at a cost of less than £1000 per life-year gained (discounted) for most smokers. There can be no doubt that such effect sizes are clinically significant, as well as representing some of the best value for money available within any health-care system. In many countries, including the United Kingdom, the principle of prolonged treatment to aid smoking cessation has been accepted by regulatory bodies. This includes granting a licence for the ‘Cut Down Then Stop’ approach, in which smokers use nicotine gum or inhaler for up to 6 months while reducing their smoking, because it is estimated to increase subsequent 6-month abstinence rates by 4% (see 18). The clinical significance of this increase obviously remains the same. The cost-effectiveness will be reduced by an amount dependent on the mean duration of use. If it is the full 6 months, then the cost will still be only in the order of £4800 per life-year gained (discounted). More recently it has been found that extending usage of the medication, varenicline, for 3 months beyond an initial 3-month course of treatment increased abstinence rates over the 6 months following the extended treatment by 7% compared with placebo 19. The effect size using the usual criterion of continuous abstinence for 6 months from the quit date was even higher than this. This effect is over and above the 13% effect size demonstrated for an initial 12 weeks of treatment with this medication 19, and is the first demonstration of an effect of prolonging treatment on smoking cessation at a follow-up point that is well beyond the end of treatment. The uniqueness and significance of this finding should not be underestimated. If we take the most conservative position and use the 7% effect size figure, we can calculate an effect on permanent cessation of 3. 5% When interpreting relapse curves it is sometimes tempting to interpret treatment effects as merely delaying relapse. For example, in an untreated group 10% may be abstinent after 12 weeks compared with 20% in a treated group, but at 6 months the treated group may have a 10% abstinence rate, so the naive conclusion is that relapse was merely delayed by the treatment. However, this neglects the fact that relapse occurs in both treated and untreated groups. That is part of the basis for calculating that an effect size of x at 6 months translates into a permanent difference in cessation of approximately ½ x. This is over and above what would have happened in an untreated sample and would result in 21 life-years gained for every 100 40-year-old smokers treated. Small but robust effects of treatments that aid smoking cessation are clinically significant because of the very large health gains that accrue from stopping smoking. An effect of as little as 1% on 6-month continuous abstinence rates would result in at least 3 additional years of life for every 100 40-year-old smokers treated. A treatment that yields a 4% 6-month continuous abstinence rate would save 12 years of life per 100 smokers treated and a 7% rate would yield 21 years of life. There would, of course, be an effect in preventing disability over and above this. At a cost of £100 per 1% increase in 6-month continuous abstinence rates, the cost per life-year gained for 40-year-old smokers would be approximately £6600. This is far less than the estimated median cost of life-saving treatments generally. The clinical significance and cost-effectiveness of prolonged treatment to aid cessation is similarly beyond question. It should also be noted that smokers may use treatments to aid cessation repeatedly over the years so that, although the effect of a given treatment episode may be small, if smokers who resumed smoking after one course of treatment still receive an advantage from a subsequent course, over repeated courses the aggregate effect will be larger. This editorial has not dealt with the issue of whether it is cost-effective in societal terms to pay for the treatment through taxes or health insurance. The justification for providing such a subsidy is dependent upon acceptance that funding cost-effective life-saving treatments is the moral responsibility of health-care systems and that smoking cessation treatment should not be singled out for exclusion. Finally, this editorial has addressed the issue of treatment effectiveness for smoking cessation. The same principles obviously apply to population-level interventions, such as mass media campaigns, smoke-free legislation and measures to increase the cost of smoking, which are pivotal to any strategy to reduce smoking prevalence 20. Robert West undertakes research and consultancy for and has received hospitality and travel funds from companies developing and manufacturing smoking cessation treatments, including Pfizer and GSK. He is also part-owner of a patent for a novel nicotine delivery device.