NOT PEER REVIEWED We welcome the timely review published by Hill et al. [1], and agree that more research is needed to assess the equity impacts of tobacco control interventions. The results of the review indicated that "increases in tobacco price have a pro-equity effect on socioeconomic disparities in smoking", but that "evidence on the equity impact of other interventions was inconclusive [...]". The inconclusiveness of findings with regard to smoking ban policies may be partly due to date limitations for the database searches, which included evidence from January 2006 through November 2010. It may also be partly due to the assessment of equity in outcomes that related only to active smoking. Although a benefit of smoking ban policy implementation may be a reduction in active smoking, this outcome is dependent upon an individual's personal response to the intervention. However, the primary purpose for implementing comprehensive smoking ban policies is to reduce secondhand smoke exposure among the population through environmental change. It is therefore important to assess whether the successful implementation of smoking ban policies has pro-equity health effects.

Our previous research indicated that the national smoking ban policy in the Republic of Ireland was associated with immediate reductions in all -cause and cause-specific cardiovascular, cerebrovascular, and respiratory mortality, and that these mortality reductions were primarily due to reductions in population exposure to secondhand smoke [2]. Our subsequent assessment of the socioeconomic differentials of these mortality reductions in the Republic of Ireland suggested that inequalities in smoking-related mortality were immediately reduced following smoking ban implementation [3]. Furthermore, given the higher rates of smoking-related mortality in the most deprived group, even equitable reductions across socioeconomic groups resulted in decreased inequalities in mortality [3].

Partial smoking ban policies do not fully protect health [4], and are likely to yield negative equity effects as a result of policy exclusions for workplaces and hospitality venues located in more deprived areas [5]. In contrast, comprehensive smoking ban policies provide equal protection for all against secondhand smoke exposure. Indeed, previous studies have indicated that comprehensive smoking ban policies are effective public health interventions for reducing both exposure to secondhand smoke and other indoor air pollutants [6-11], the benefits of which are experienced by all employees and patrons of restaurants, bars, and other public places, regardless of individual socioeconomic group. Therefore, we wish to highlight that the evidence for pro-equity effects of comprehensive smoking ban policies may be more conclusive when additional health-related outcomes are considered.

References

1. Hill S, Amos A, Clifford D, Platt S (2014) Impact of tobacco control interventions on socioeconomic inequalities in smoking: review of the evidence. Tob Control 23: e89-e97.

2. Stallings-Smith S, Zeka A, Goodman P, Kabir Z, Clancy L (2013) Reductions in cardiovascular, cerebrovascular, and respiratory mortality following the national irish smoking ban: interrupted time-series analysis. PLoS One 8: e62063.

3. Stallings-Smith S, Goodman P, Kabir Z, Clancy L, Zeka A (2014) Socioeconomic differentials in the immediate mortality effects of the national Irish smoking ban. PLoS One 9: e98617.

4. Tan CE, Glantz SA (2012) Association between smoke-free legislation and hospitalizations for cardiac, cerebrovascular, and respiratory diseases: a meta-analysis. Circulation 126: 2177-2183.

5. Lewis GH, Osborne DC, Crayford TJ, Brown AC (2006) Partial smoking ban would worsen health inequalities. Bmj 332: 362.

6. Fong GT, Hyland A, Borland R, Hammond D, Hastings G, et al. (2006) Reductions in tobacco smoke pollution and increases in support for smoke- free public places following the implementation of comprehensive smoke- free workplace legislation in the Republic of Ireland: findings from the ITC Ireland/UK Survey. Tob Control 15 Suppl 3: iii51-58.

7. Connolly GN, Carpenter CM, Travers MJ, Cummings KM, Hyland A, et al. (2009) How smoke-free laws improve air quality: a global study of Irish pubs. Nicotine Tob Res 11: 600-605.

8. Goodman P, Agnew M, McCaffrey M, Paul G, Clancy L (2007) Effects of the Irish smoking ban on respiratory health of bar workers and air quality in Dublin pubs. Am J Respir Crit Care Med 175: 840-845.

9. Mulcahy M, Evans DS, Hammond SK, Repace JL, Byrne M (2005) Secondhand smoke exposure and risk following the Irish smoking ban: an assessment of salivary cotinine concentrations in hotel workers and air nicotine levels in bars. Tob Control 14: 384-388.

10. Valente P, Forastiere F, Bacosi A, Cattani G, Di Carlo S, et al. (2007) Exposure to fine and ultrafine particles from secondhand smoke in public places before and after the smoking ban, Italy 2005. Tob Control 16: 312-317.

11. Eisner MD, Smith AK, Blanc PD (1998) Bartenders' respiratory health after establishment of smoke-free bars and taverns. JAMA 280: 1909- 1914.

Conflict of Interest:

None declared

INTRODUCTION Mejia et al1 argue that a harm reduction strategy based on promoting snus, the form of smokeless tobacco widely used in Sweden, is unlikely to result in any substantial health benefit to the US population. They divide the population into five tobacco groups (never tobacco users, former tobacco users, current cigarette smokers, current snus users, and current dual users), attaching to each group an estimate of the "tobacco-related health effect" (TRHE). By definition, TRHE is 0 in never smokers and 100 in current cigarette smokers, with other smoking groups having intermediate TRHE values, proportional to their relative excess disease risk. Mejia et al consider various scenarios (e.g. "aggressive smokeless promotion") which result in different predicted distributions by tobacco use, and hence different estimates of the overall average TRHE for the whole US population. For the "base case", with tobacco use distributions as they currently are, this is estimated as 24.2, and under the various scenarios considered the estimates lie between 19.2 and 30.5.

Their conclusion that snus promotion probably provides little health benefit seems surprising. Given the strong evidence that health risks from snus are much less than from smoking, one would intuitively expect a substantial benefit if increasing snus promotion led to many smokers switching to snus. It is useful therefore to look at the methodology used and assumptions made.

FAILURE PROPERLY TO ACCOUNT FOR PATTERNS OF TOBACCO USE

There are some deficiencies in the approach. First, there are clealy more than five relevant tobacco groups. Limiting attention to snus use and cigarette smoking, there are nine main groups, representing each combination of never, former and current use of each product. And within some combinations, there are subgroups by sequence of events. Why, for example, should TRHE be assumed similar in former tobacco users regardless of whether snus or cigarettes were previously used, or similar in current snus users who have or have not previously smoked cigarettes? Other deficiencies include failure to consider age, sex, amount used, and other tobacco products such as pipes or cigars. However, these are minor compared to the failure to account for time in its various guises - time since quit, time since switch, and time used snus or cigarettes. It is unsound to assume TRHE is the same for all former users of tobacco regardless of time quit, or the same for current snus users regardless of previous smoking history. Failure to consider time undermines the validity of the TRHE estimates for the different tobacco groups.

ESTIMATES OF TRHE BY SMOKING GROUP Quitters No justification is given for the TRHE estimate of 5 used by Mejia et al. It seems very low. Relative all-cause mortality rates for current, former and never smokers from the well- known CPS-II study2, indicates former smokers have about 40% of the excess all-cause mortality rate of current smokers, not 5%. The appropriate TRHE would be higher still for short- term quitters. Was the value of 5 intended to relate to long-term quitting?

Snus users The justification for the TRHE estimate of 11 is unclear. It is much too low, if applied to recent switchers from cigarettes, particularly following long-term smoking. However, if intended only to quantify effects of snus, it seems too high. Updates of published meta-analyses for snus use for heart disease3 and cancer4 (details available on request) suggest little or no increased risk, with combined relative risk (95% confidence interval) estimates of 1.01 (0.91-1.12) for ischaemic heart disease, 1.05 (0.95-1.15) for stroke, 0.97 (0.68-1.37) for oropharyngeal cancer, 1.10 (0.92-1.33) for oesophageal cancer, 0.98 (0.82-1.17) for stomach cancer, 1.20 (0.66-2.20) for pancreatic cancer, and 0.71 (0.66-1.76) for lung cancer. Given it is implausible that snus use might increase COPD risk, given the lack of confirmed reports of increased risks for other diseases, and given the much stronger relationships seen with smoking, the excess risk from snus use is probably no more than 2% of that from cigarette smoking and not as great as 11%.

Dual use The estimated TRHE of 90 derived from INTERHEART 5 is not relevant to snus, the smokeless tobacco use reported in that study being predominantly in Asian and African countries. Though data are lacking, one might imagine that if lifetime dual users get about half their required nicotine dose from each source, a TRHE of about 50 might be appropriate. Again, however, this would not apply to those changing from long-term smoking to dual use.

HIGH ESTIMATES OF DUAL USE The proportion of dual users predicted in some of the scenarios of up to about 20% seem implausibly high. Recent Swedish surveys (e.g.6,7 give estimates less than 3%. While adolescents in Sweden often try both products, adults usually only use one. Models based on studies in adolescents that do not take this into account may result in misleading predictions of the tobacco use distribution, especially when the data used8 relate to smokeless tobacco use, not snus.

SNUS AND INITIATION OF SMOKING Some Swedish retrospective studies9,10 claim snus users are less likely to initiate smoking than never tobacco users. While these claims are questionable (failing to adjust for time available to initiate), evidence that few Swedish smokers used snus before they started smoking9,10, and that most dual users started on cigarettes, suggest snus can be at most a minor determinant of smoking.

SNUS AND QUITTING SMOKING In theory snus use might discourage rather than encourage quitting. No published study in Sweden suggests discouragement, but many 9,11-15suggest encouragement. Although these studies have some limitations, concern regarding discouragement seems unjustified.

FURTHER THOUGHTS AND A SIMPLER APPROACH The approach of Mejia et al is complex and does not validly allow assessment of the effect on health of the various scenarios considered.

One problem is that promotion of snus cannot affect the risk resulting from past smoking (particularly so for those who quit before the promotion started), so that inclusion of this risk in the overall TRHE estimates obscures estimation of the effects of the various strategies discussed. It would seem better to compare the decline in risk for the given scenario of snus promotion with that in a comparable scenario where those assumed to switch to snus quit instead.

A second problem is that while their approach is complex, it ignores many factors, such as time quit or switched, age, sex, and quantity used. However, attempting to improve the model to account for these would likely be valueless, given the uncertainties involved.

Also Mejia et al do not define what they call a substantial health benefit. The strategy "aggressive promotion with most new users from smokers" reduces the overall TRHE from 24.2 to 19.2, i.e by about 20%. This seems quite substantial, especially so if it is a relatively short- term effect. Would strategies directly encouraging quitting do better?

To my mind, they have obscured a simple situation. Complete switching to snus seems likely to have a health effect virtually equivalent to quitting, with partial switching (dual use) having an intermediate effect. For smokers unwilling or unable to give up their nicotine, switching to snus is clearly a much better health alternative than continuing smoking. Promoting snus may produce some new tobacco users, but these will have little or no excess risk of disease, and be no more likely to take up smoking than are those who have never used tobacco.

(WORD COUNT: 1213)

REFERENCES

1. Mejia AB, Ling PM, Glantz SA. Quantifying the effects of promoting smokeless tobacco as a harm reduction strategy in the USA. Tob Control 2010;19:297-305.

2. US Surgeon General. Reducing the health consequences of smoking. 25 years of progress. A report of the Surgeon General. Rockville, Maryland: US Department of Health and Human Services; Public Health Services; 1989. DHHS Publication No. (CDC) 89-8411. http://www.surgeongeneral.gov/library/reports/index.html

3. Lee PN. Circulatory disease and smokeless tobacco in Western populations: a review of the evidence. Int J Epidemiol 2007;36:789-804.

4. Lee PN, Hamling JS. Systematic review of the relation between smokeless tobacco and cancer in Europe and North America. BMC Med 2009;7:36:

5. Teo KK, Ounpuu S, Hawken S, Pandey MR, Valentin V, Hunt D, et al. Tobacco use and risk of myocardial infarction in 52 countries in the INTERHEART study: a case-control study. Lancet 2006;368:647-58.

6. Persson J, Sj?berg I, Johansson S-E. Bruk och missbruk, vanor och ovanor. H?lsorelaterade levnadsvanor 1980-2002 (Health related habits of life 1980-2002). Statistiska centralbyr?n; 2004, (Accessed Oct 2010). (Levnadsf?rh?llanden (Living conditions).) 105. http://www.scb.se/statistik/le/le0101/1980i02/le0101_1980i02_br_le105sa0401.pdf With additional data supplied by E H?gstorp, Statistiska centralbyr?n, 2005.

7. Wadman C. Levnadsvanor - Tobaksvanor. Statens Folkh?lsoinstitut; 2009, (Accessed Oct 2010). http://www.fhi.se/sv/Statistik- uppfoljning/Nationella-folkhalsoenkaten/Levnadsvanor/Tobaksvanor/

8. Severson HH, Forrester KK, Biglan A. Use of smokeless tobacco is a risk factor for cigarette smoking. Nicotine Tob Res 2007;9:1331-7.

9. Furberg H, Bulik CM, Lerman C, Lichtenstein P, Pedersen NL, Sullivan PF. Is Swedish snus associated with smoking initiation or smoking cessation? Tob Control 2005;14:422-4.

10. Ramstr?m LM, Foulds J. Role of snus in initiation and cessation of tobacco smoking in Sweden. Tob Control 2006;15:210-4.

11. Lindstr?m M, Isacsson S-O. Smoking cessation among daily smokers, aged 45-69 years: a longitudinal study in Malm?, Sweden. Addiction 2002;97:205-15.

12. Lundqvist G, Sandstr?m H, ?hman A, Weinehall L. Patterns of tobacco use: a 10-year follow-up study of smoking and snus habits in a middle-aged Swedish population. Scand J Public Health 2009;37:161-7.

13. Rodu B, Stegmayr B, Nasic S, Cole P, Asplund K. Evolving patterns of tobacco use in northern Sweden. J Intern Med 2003;253:660-5.

14. Gilljam H, Galanti MR. Role of snus (oral moist snuff) in smoking cessation and smoking reduction in Sweden. Addiction 2003;98:1183 -9.

15. Ramstr?m L. Is snus a model for harm reduction: the scientific evidence from Sweden. In: The 13th World Conference on Tobacco OR Health: Building capacity for a tobacco-free world, The 13th World Conference on Tobacco OR Health: Building capacity for a tobacco-free world. Washington DC, July 12-15 2006. 2006;

Conflict of Interest:

I am a long-term consultant to the tobacco industry, and this work was supported by Philip Morris

Reduction as a permanent solution may give people false expectations Thanks to Dr. John R Hughes for his interesting remarks of 20 January 2007 to our article (TC 15:472-480). We have the following comments: 1. Dr. Hughes states that our main finding (no health benefit from reducing cigarettes) has not been found in the few prior prospective studies of this topic. This is not correct. Based on a large study population in Copenhagen, Dr. Nina S Godtfredsen and co-workers have reported the same results in a series of publications, references given in our article. Dr Hughes’ remarks imply that there may be other prospective studies that give other results. We have not been able to find other prospective studies that take up this problem. 2. Our article reports on results from three examinations; for the majority of participants the interval between the examinations was five years. A subgroup of the study population was nominated ‘sustained reducers’. They were heavy cigarette smokers at the first examination, had reduced their daily cigarette consumption by at least 50 % at the second examination, and had remained as ‘reducers’ at the third examination. Their mean consumption at the three examinations was 23.6 – 10.0 – 10.4 cigarettes per day (table 6 in our article). Dr Hughes states that “the question at each follow-up did not ask about smoking since the last follow-up”. This is correct, and we agree with Dr Hughes that it is unknown what the rate of smoking really was between follow-ups in sustained reducers. The sustained reducers had, however, a mean daily consumption that was almost the same at the second and third examination, and in our opinion, the most reasonable explanation is that their daily cigarette consumption had stabilised at a consumption level which actually was at least 50% lower than at the first examination. We also underline that at the second examination, reducers had a serum thiocyanate level that was lower than in heavy smokers, and close to the serum thiocyanate level in moderate smokers (table 3 in our article). 3. Dr Hughes states that reduction actually increases motivation to quit. In our paper, we state explicitly:”Undoubtedly, reduction in consumption may have a place as a temporary measure in systematic smoking cessation”. Our conclusion that advising reduction may give people false expectations, refer to reduction as a permanent solution. We think that the results of our study and of those of the Copenhagen Study, with study populations of more than 70 000 persons together, give a sound basis for this conclusion. Age Tverdal, Professor Norwegian Institute of Public Health, Oslo

Kjell Bjartveit Director Emeritus National Health Screening Service Oslo

We are mildly flattered that Philip Morris found it worthwhile to have Peter Lee criticize our framework [1] for assessing the likely population effects of aggressive promotion of smokeless tobacco as a harm reduction strategy in the USA. Peter Lee is a longtime tobacco industry consultant who has a history spanning decades criticizing important studies demonstrating the harms of tobacco and secondhand smoke [2], including the landmark Hirayama study [3-5] and publishing papers or letters to the editor contesting the health effects of secondhand smoke on cardiovascular disease [6], cancer [7], SIDS [8] and more recently the health effects of smokeless tobacco [9, 10] and menthol [11]. Lee's role in industry efforts to discredit the Hirayama study has been well documented in the literature [12, 13]. As Lee notes, "one would intuitively expect a substantial benefit if increasing snus promotion led to many smokers switching to snus." The whole point of our analysis was to move beyond "intuition" and make predictions based on data in a way that explicitly accounts for the uncertainty in the data on tobacco use behavior and the associated health costs. The fact that the likely health cost ranges overlap is what leads to the conclusion that, accounting for this uncertainty, the market changes likely to accompany aggressive smokeless promotion would not confer population-level health benefits.

Lee, a statistician, simply ignores the uncertainty associated with the estimates that form the core of the model.

He criticizes the fact that we do not account for the temporal dynamics of changes in tobacco use behavior and the associated risks over time. He is correct that our model is a steady-state, not a dynamic, model. We considered a dynamic model, but doing so conflicted with our fundamental goal of basing the results on data rather than the rhetoric and "intuition" that have characterized the harm reduction debate to date. We were unable to find the data necessary to model the dynamics Lee seeks. It is noteworthy that Lee did not provide citations to the data that one could use to develop the model he desires.

Lee found the justification for the health cost we used for snus as 11 to be "unclear." In our paper, we clearly stated that this estimate came from an expert consensus panel estimate of the health effects, reference 9 in our paper [14]. As we noted in our paper, the estimate that this panel produced is probably low because subsequent research has found higher risks for heart disease, that are larger than those considered in this reference, a case that has only grown stronger as evidence has continued to accumulate [15]. If anything, we are almost certainly underestimating these risks.

Lee questions our assumption of a risk of 90 for dual use. He is correct that there is little data available on dual use (a subject worthy of study). The reason we assumed a modest reduction in risk was that there might be less exposure to cigarette smoke, which could lower cancer risk but would have little effect on heart disease risk because of the highly nonlinear relationship between smoking and heart disease risk, with most effect occurring at low levels of smoking.

Lee criticizes us for including what he sees as unacceptably high levels of dual use (i.e., concurrent use of smokeless tobacco and cigarettes) in our scenarios. (It is important to define "dual use" as use of either product on some days rather than both products on all days. This latter definition does not reflect actual dual use, particularly as the snus products are being promoted for use when one cannot "light up", and substantially underestimates dual use.) The base levels of dual use we used in our model are from surveys of actual use patterns in the USA. The fact that we model large increases in dual use reflects the actual marketing of smokeless products by the tobacco companies, who are promoting snus products as cigarette line extensions, packaging them together, and explicitly promoting dual use in their marketing. Dr. Lee could make a real contribution to the debate if he were to present an analysis based on the market targets that his client, in this case Philip Morris, has established for both Marlboro snus and dual use of Marlboro snus and cigarettes together.

Finally, we were surprised that Dr. Lee did not simply put the health costs he asserts are accurate into the model and present the results to demonstrate that his assertions are correct and supported by actual data. (The full model is available on the Tobacco Control website at http://tobaccocontrol.bmj.com/content/19/4/297/suppl/DC1, something we pointed out to him when he contacted us asking for a copy of the model.) The whole object of this enterprise is to move beyond the "intuitive" arguments Lee presents to making decisions based on quantitative estimates of likely population effects: Lee failed to provide credible estimates demonstrating that smokeless promotion would actually be likely to reduce harm on a population level.

Adrienne Mejia

Pamela M. Ling

Stanton Glantz

University of California, San Francisco San Francisco, CA 94143

REFERENCES

1. Mejia AB, Ling PM, Glantz SA. Quantifying the Effects of Promoting Smokeless Tobacco as a Harm Reduction Strategy in the USA. Tob Control. 2010 Aug;19(4):297-305.

2. Lee PN. Many Claims About Passive Smoking Are Inadequately Justified. BMJ. 1997 Feb 1;314(7077):371.

3. Hirayama T. Non-Smoking Wives of Heavy Smokers Have a Higher Risk of Lung Cancer: A Study from Japan. Br Med J (Clin Res Ed). 1981 Jan 17;282(6259):183-5.

4. Lee PN. "Marriage to a Smoker" May Not Be a Valid Marker of Exposure in Studies Relating Environmental Tobacco Smoke to Risk of Lung Cancer in Japanese Non-Smoking Women. Int Arch Occup Environ Health. 1995;67(5):287-94.

5. Ong E, Glantz SA. Hirayama's Work Has Stood the Test of Time. Bull World Health Organ. 2000;78(7):938-9.

6. Lee PN, Forey BA. Environmental Tobacco Smoke Exposure and Risk of Stroke in Nonsmokers: A Review with Meta-Analysis. J Stroke Cerebrovasc Dis. 2006 Sep-Oct;15(5):190-201.

7. Lee PN, Hamling J. Environmental Tobacco Smoke Exposure and Risk of Breast Cancer in Nonsmoking Women: A Review with Meta-Analyses. Inhal Toxicol. 2006 Dec;18(14):1053-70.

8. Lee PN. Passive Tobacco Exposure and Sudden Infant Death Syndrome. Pediatrics. 1993 Sep;92(3):505-6.

9. Lee PN, Hamling J. Systematic Review of the Relation between Smokeless Tobacco and Cancer in Europe and North America. BMC Med. 2009;7:36.

10. Sponsiello-Wang Z, Weitkunat R, Lee PN. Systematic Review of the Relation between Smokeless Tobacco and Cancer of the Pancreas in Europe and North America. BMC Cancer. 2008;8:356.

11. Werley MS, Coggins CR, Lee PN. Possible Effects on Smokers of Cigarette Mentholation: A Review of the Evidence Relating to Key Research Questions. Regul Toxicol Pharmacol. 2007 Mar;47(2):189-203.

12. Hong MK, Bero LA. How the Tobacco Industry Responded to an Influential Study of the Health Effects of Secondhand Smoke. BMJ. 2002 Dec 14;325(7377):1413-6.

13. Yano E. Japanese Spousal Smoking Study Revisited: How a Tobacco Industry Funded Paper Reached Erroneous Conclusions. Tob Control. 2005 Aug;14(4):227-33; discussion 33-5.

14. Levy DT, Mumford EA, Cummings KM, Gilpin EA, Giovino G, Hyland A, et al. The Relative Risks of a Low-Nitrosamine Smokeless Tobacco Product Compared with Smoking Cigarettes: Estimates of a Panel of Experts. Cancer Epidemiol Biomarkers Prev. 2004 Dec;13(12):2035-42.

15. Piano MR, Benowitz NL, Fitzgerald GA, Corbridge S, Heath J, Hahn E, et al. Impact of Smokeless Tobacco Products on Cardiovascular Disease: Implications for Policy, Prevention, and Treatment: A Policy Statement from the American Heart Association. Circulation. 2010 Oct 12;122(15):1520 -44.

Conflict of Interest:

None declared