The volume of attention to our study in the U.S. and international press is not surprising, considering the widespread promotion of nicotine replacement products to all smokers in the population, and their growing inclusion in government subsidized health plans. Prior to addressing criticisms made by Stapleton and others, we note that their comments reflect at least one important area of agreement. The fact that no advantage for long term abstinence was found for users of nicotine replacement therapies (NRT) comes as no surprise even to critics of the study findings underscores general consensus that NRT treatment has not solved the problem of relapse to smoking.

The argument is that our study ignores initial quit rates and that improved initial quit rates would be expected to have a long-lasting effect on population smoking rates. The logic of this argument is that use of NRT will increase the number of smokers in the population who quit initially, and even given the same relapse rate as of non-users, NRT would result in a larger number of long term quitters than would be the case had none of the smokers used it initially. That hope or expectation was prominent around the time that the U.S. Food and Drug Administration permitted NRT to be sold without prescription. Unfortunately, however, the predicted increases in both rates of quitting and long-term quitters failed to materialize, (Pierce et al., 2012) even though annual sales of NRT in the U.S. increased dramatically.

The meta-analyses by Etter et. al., which Stapleton suggests provides better information than our recent study, is also based on clinical trials, and subject to limitations for assessing population effectiveness as discussed in our paper. Further, none of the prospective studies reviewed in Hughes et. al. found a population effect of over-the-counter NRT.

The argument has been made that a form of selection bias may have occurred in which persons more addicted may have been more likely to seek treatment and also have been more prone to relapse. The likelihood of such a bias affecting the results was diminished by the study's control for level of dependence. On the contrary, the counterargument could be made that persons who made the effort to try and use NRT were actually more highly motivated to quit and consequently should have been less likely to relapse, in further support of our study's findings. Indeed, because of the primary role of motivation in cessation, clinical trials for NRT tend to be highly selective for subjects who are strongly motivated to try and quit.

The possibility of recall bias (that people who quit a longer time ago would find it more difficult to recall whether or not they used NRT) was addressed in the study by a sensitivity analysis, the results of which showed consistency of findings based on prior six months, one year, and two years reported abstinence. The sample size was reasonable for this study; and, a point missed by Stapleton, was sufficient to detect a statistically significantly <higher> rate of relapse among formerly heavy smokers who used NRT without counseling.

Although a major objective of the study was not to assess adherence to NRT use or reasons for its inappropriate use, the fact that many NRT users did not continue to use it for the recommended eight weeks raises more of a question regarding its effectiveness outside of the controlled trial setting than doubt regarding the findings.

Braillon and Dubois suggested the results might have differed had we analyzed covariates using alternative forms. We analyzed the data using categorical as well as ordinal variables for dependence, including a three -category scale from the Fagerstrom Test of Nicotine Dependence, and four- category variables representing numbers of packs smoked per day, age as well as education, respectively. Analyzing these variables as categorical is common practice. We do not think that the continuous form is necessarily a better representation than the ordinal form for each of these variables since the theorized relationships with relapse are not necessarily linear and monotonic. The results of these analyses with respect to NRT were qualitatively the same as the original analyses showing no differences except that use of NRT for at least six weeks without counseling was associated with higher rates of relapse.

Finally, Beard et. al. reports regarding smokers who "had not smoked for the last 4 weeks" and no misrepresentation was intended.

We are gratified by the attention that our study has received because we think it is important to examine the evidence both for and against this approach to one of the most vexing public health scourges of our time - tobacco addiction.

Pierce JP, Cummins SE, White MM, Humphrey A, Messer K. Quitlines and Nicotine Replacement for Smoking Cessation: Do We Need to Change Policy? Annu Rev Public Health. 2012 Apr 4. [Epub ahead of print]

Etter JF, Stapleton JA. Nicotine replacement therapy for long- term smoking cessation: a meta-analysis. Tob Control 2006;15:280-5.

Hughes JR, Peters EN, Naud S. Effectiveness of over-the-counter nicotine replacement therapy: a qualitative review of nonrandomized trials. Nicotine Tob Res. 2011;13:512-22.

Beard, E., McNeill, A., Aveyard, P., Fidler, J., & West, R. Association between use of nicotine replacement therapy for harm reduction and smoking cessation: a prospective study of English smokers. Tobacco Control, 10.1136/tobaccocontrol-2011-050007 Online 1 December 2011.

Conflict of Interest:

None declared

Throwing the baby out with the bath water

Alain Braillon(a) MD, PhD, Gerard Dubois(b) MD, MPH.

(a) 27 rue Voiture. 80000 Amiens. France (b) Public Health. Amiens University Hospital. France

In an observational study Alpert and colleagues concluded that persons who have quit smoking relapsed at equivalent rates, whether or not they used nicotine replacement therapy (NRT) to help them in their quit attempts and challenged the funding of cessation medication policy.(1)

They failed to discuss the severe methodological limitations of their study which conflicts the evidences from so many randomized controlled trials. These include, to cite a few : a) the recall biases which challenge covariates quality; b) a major bias of selection considering the high percentages of non screened and of lost of follow-up during the three successive rows (see methods); c) the use of cut-points to derive subgroups for covariates which is not appropriate as there is a continuous distribution of the values with no obvious modal values; d) the absence of the measure of the initial quit rates with nicotine replacement therapy (NRT) and others methods; e) the effect of the comprehensive tobacco control policy implemented since 2002, a major confounding variable. Massachusetts now ranks 9th among the 50 states: 16.1% of the adult population (aged 18+ years) are current cigarette smokers.(2)

Finally, as a population study, the Massachusetts program is more convincing. Since 2006, Massachusetts has offered free treatments to help poor residents (Medicaid) stop smoking. When the program started, about 38 percent of poor Massachusetts residents smoked. By 2008, the smoking rate for poor residents had dropped to about 28 percent. This is 30,000 people in two and a half years, or one in six smokers. No changes were observed in those not covered by the plan (3) Tobacco cessation benefit that includes coverage for medications and behavioral treatments can significantly reduce smoking prevalence.(4)

1 Alpert HR, Connolly GN, Biener L. A prospective cohort study challenging the effectiveness of population-based medical intervention for smoking cessation. Tob Control. 2012. Oneline 10 Jan 2012. doi:10.1136/tobaccocontrol-2011-050129.

2 Centers for disease control and prevention. Smoking & tobacco use. State highlights 2010. Available at http://www.cdc.gov/tobacco/data_statistics/state_data/state_highlights/2010/states/massachusetts/index.htm Accessed 19 Jan 2012

3 Goodnough A. Massachusetts Antismoking Plan Gets Attention. 2009 Dec 16. The New York Times Available at http://www.nytimes.com/2009/12/17/us/17smoke.html Accessed 19 Jan 2012

4 Land T, Warner D, Paskowsky M et al. Medicaid coverage for tobacco dependence treatments in Massachusetts and associated decreases in smoking prevalence. PLoS One 2010 18;5(3):e9770.

Conflict of Interest:

Dr Braillon, a senior tenured consultant, was sacked in 2010 from Prof Dubois' unit by the French Department of Health against the advice of the National Statutory Committee. Prof Dubois was sued for libel by the French Tobacconists Union (Abuse of libel laws and a sacking: The gagging of public health experts in France. Tobacco control blog 8 November 2010). Prof Dubois is honorary president of Alliance Contre le Tabac and chairs the Addiction Committee of the National Academy of Medicine. He has received consulting fees from Pfizer.

In their paper claiming to find that NRT is not effective long-term, Alpert et al [1] misrepresented findings from a paper for which I was primary author [2], citing it as evidence that other representative population studies have not found any beneficial effect of the use of NRT on annual smoking cessation rates. They state 'Beard et al found increased short-term abstinence only (sic) among persons who had reported using NRT six months earlier'. This is misleading given that we only looked at short -term cessation. The referencing is also erroneous, with our paper appearing as a sub-paper of Chapman and MacKenzie's [3], labelled 15a and 15 respectively. Our paper has no affiliation with these authors and we do not argue for the abandonment of clinical treatments for smokers.

1. Alpert, H. R., Connolly, G. N., & Biener, L. A. (2012). prospective study challenging the effectiveness of population-based medical intervention for smoking cessation. Tobacco Control, 10.1136/tobaccocontrol-2011-050129 Online 12 January 2012

2. Beard, E., McNeill, A., Aveyard, P., Fidler, J., & West, R. (in press). Association between use of nicotine replacement therapy for harm reduction and smoking cessation: a prospective study of English smokers. Tobacco Control, 10.1136/tobaccocontrol-2011-050007 Online 1 December 2011

3. Chapman, S, & MacKenzie, R. (2010). The global research neglect of unassisted smoking cessation: causes and consequences. PLoS Med,7(2), e1000216.

Conflict of Interest:

Emma Beard has received conference funding from Pfizer

The results of the recent study by Alpert et al. were interpreted incorrectly with respect to the efficacy of nicotine replacement therapy (NRT).(1) The study only considered relative relapse rates among people who had already stopped smoking according to whether they had used NRT or not. This is clearly an inadequate design to address the issue of efficacy because it ignores the initial quit rates in the two groups. Only if the results had indicated significantly higher relapse among those using NRT might they have offered evidence against long-term NRT efficacy, depending, of course, on the initial difference in quit rates (not measured) and the difference in relapse rates. However, this was not the case. There was no evidence of differential relapse. Therefore, the conclusion that these data provide evidence against the effectiveness of NRT is wrong.

Had the authors considered more thoroughly the literature they would surely have been enlightened by the meta-analysis review of relapse and long-term NRT effectiveness published in Tobacco Control.(2) It would have helped them understand the issues and to draw an appropriate conclusion, rather than a perverse one. That review included 4792 randomized subjects (not self-selected as in the new study) followed up for several years and found the same result as Alpert: the relapse rate did not differ between those using NRT and others. Consequently, because the initial NRT quit rate was higher, efficacy remained after a mean follow-up time of 4.3 years (Odds ratio =1.99, 95% C.I. = 1.50 to 2.64). In contrast to the new study, all the subjects in that review received some form of professional support, although often minimal. Therefore, the same finding with respect to relapse in the new population-based study tends, if anything, to broaden rather than diminish the evidence for long-term NRT effectiveness.

(1) Alpert HR, Connolly GN, Biener L. A prospective study challenging the effectiveness of population-based medical intervention for smoking cessation. Tob Control 2012 10.1136/tobaccocontrol-2011-050129 Online 12 January

(2) Etter JF, Stapleton JA. Nicotine replacement therapy for long- term smoking cessation: a meta-analysis. Tob Control 2006;15(4):280-5.

Conflict of Interest:

John Stapleton has conducted trials of nicotine replacement and other treatments for tobacco dependence supported by the Medical Research Council, the Department of Health and Cancer Research UK. He was formally an adviser on issues of study design and methodology to several manufacturers of smoking cessation medications, including NRT, bupropion and varenicline.

Many of Alpert, Connolly and Biener's population level NRT post- cessation findings are disturbing and worthy of further and deeper review. What's most baffling is that any government would invest so much confidence and so many lives in a product without demanding a shred of population level evidence as to its worth.

According to this paper, the odds of relapse for a heavily dependent NRT quitter who had quit less than six months were 3.53 times that of a heavily dependent quitter who quit without NRT or professional help. If true, that puts a rather hefty dent in NRT's most favored failure explanation, its selection bias theory.

This finding makes troubling the fact that varenicline (Chantix/Champix) failed to prevail in long-term point prevalence quitting over nicotine patch in the only head-to-head clinical trials to date (Aubin 2008 and Tsukahara 2010).

Alpert and colleagues do not attempt to explain the conflict between clinical trial and population level NRT findings. But I submit that this outcome was suggested by the first NRT clinical trial ever, the 1971 nicotine gum study by Ohlin and Westling.

Ohlin and Westling found that counseling and support ("ten visits and more persuasion") was superior to nicotine gum alone, but that nicotine gum could defeat placebo gum users. Even then, Ohlin and Westling documented obvious nicotine gum blinding concerns.

Try to name any other placebo-controlled study area where the condition sought to be treated (withdrawal) does not exist until researchers command its onset. Name any other study area where the placebo group is actually punished within 24 hours by a rising tide of anxieties.

Have three decades of referring to nicotine as "medicine" and its use "therapy" undermined natural learning and the quitter's ability to self- discover the most critical recovery lesson of all, that lapse almost always equals relapse, that one puff is too many and thousands never enough?

Nearly all population level quitting method surveys to date have found NRT less effective long-term than quitting without it. If true, are taxpayers today paying to reduce the quitter's odds of success? Are we responsible for undercutting their chances and costing many their lives?

John R. Polito Nicotine Cessation Educator

Conflict of Interest:

Pro bono director of a cold turkey stop smoking website.