Please note:  The current AAPHP tobacco policy is available at http://www.aaphp.org/tobacco

Update on the Scientific Status of Tobacco Harm Reduction, 2008-2010
Prepared for the American Association of Public Health Physicians

Brad Rodu, DDS and Joel L Nitzkin, MD
June 28, 2010

Abstract and Summary

In October 2008 the American Association of Public Health Physicians (AAPHP) became the first medical organization in the U.S. to officially endorse tobacco harm reduction as a viable strategy to reduce the death toll related to cigarette smoking.  Since the AAPHP endorsement there have been numerous contributions to the scientific literature that add to the scientific foundation for tobacco harm reduction.  This report describes these published studies, which include meta-analyses of the risk of cancer and cardiovascular diseases among smokeless tobacco (ST) users and evidence from clinical trials that ST is an effective substitute for cigarettes.  It also discusses American studies concerning whether ST use is a gateway to smoking and other topics relevant to tobacco harm reduction.  The new research significantly strengthens AAPHP’s position on harm reduction, which encourages inveterate smokers – who are unable or unwilling to abstain from all nicotine and tobacco – to switch to lower risk smokeless tobacco products.

 

I.  Introduction

In October 2008 the American Association of Public Health Physicians (AAPHP) became the first medical organization in the U.S. to officially endorse tobacco harm reduction as a viable strategy to reduce the death toll related to cigarette smoking.  The AAPHP based its decision on a review of the relevant scientific literature undertaken by Drs. Joel Nitzkin and Brad Rodu (1).  Since the AAPHP endorsement there have been numerous contributions to the scientific literature that add to the scientific foundation for tobacco harm reduction.  The objective of this report is to describe these published studies.

 

II. ST Use and Cancer

In 2008 and 2009 two separate meta-analyses examined the available epidemiologic evidence with respect to ST use and cancer.  The first, published in 2008 by Paolo Boffetta and colleagues, evaluated 17 studies from Scandinavia and the U.S. that contained information about cancer risk related to ST use (2). 

In 2009 Peter Lee and Jan Hamling, two epidemiologists based in the United Kingdom, co-authored a second meta-analysis regarding ST use and cancer (3).  Lee and Hamling compiled the statistics from 89 studies, and they used a straightforward technique to separate the risk related to ST use from the risk related to smoking and alcohol consumption.  In other words, the risks from ST use were adjusted for those from smoking and alcohol use, which is important because ST users often have a history of smoking and heavy drinking, both of which are risk factors for cancers of the oral cavity, throat and esophagus. 

Both meta-analyses reported summary RRs for cancer among ST users compared with non-users of tobacco.
 
  a. Oral Cancer

Boffetta et al. concluded that ST use in the U.S. was associated with an increased risk for oral cancer (Relative Risk, RR = 2.6, 95% Confidence Interval, CI = 1.3 – 5.2).  They found no evidence that ST use in Nordic countries was associated with oral cancer (RR = 1.0, CI = 0.7 – 1.3). 

Lee and Hamling found 41 studies that reported risks for oral cancer.  For all studies the RR was 1.79 (CI = 1.36-2.36), indicating a modest elevation in risk.  However, in the 19 studies that accounted for smoking the RR was 1.36 (CI = 1.04-1.77), and in the 10 studies that accounted for both smoking and alcohol the RR was 1.07 (CI = 0.84-1.37). 
Thus, the Lee-Hamling analysis found that the association between ST use and oral cancer is very weak, virtually nonexistent.  Lee and Hamling also found that, for studies published since 1990, the RR for ST use was 1.28 (CI = 0.94-1.76), a small, statistically significant increase that disappeared almost completely when only studies that accounted for smoking or smoking and alcohol were considered.  This means that any possible risks from using ST 40 or 50 years ago have not been seen in studies conducted since 1990.

  b. Other Cancers

Boffetta et al. reported that ST use in Nordic countries was associated with increased risks for esophageal cancer (RR = 1.6, CI = 1.1 – 2.4) and pancreatic cancer (RR = 1.8, CI = 1.3 – 2.5).  They did not find significantly elevated risks among American ST users for either disease (RR = 1.2, CI = 0.1 – 13 and RR = 1.4, CI = 0.7 – 2.7 respectively).

Lee and Hamling examined the epidemiologic evidence linking ST use with many other cancers.   It is noteworthy that the only statistically significant finding was a minimally elevated risk for prostate cancer (RR = 1.29, CI = 1.07-1.55).  Lee and Hamling commented that “Prostate cancer is not considered smoking related [original citations removed], and more information on its relationship with ST is needed before any clear conclusion can be drawn.”

  c.  Differences between the Boffetta and Lee-Hamling Meta-analyses

It is clear that the meta-analysis by Boffetta et al. reported significantly higher risk estimates for cancers of the oral cavity, esophagus and pancreas than that reported by Lee and Hamling.  The differences were the subject of a detailed commentary by Lee and Hamling, published in 2009 (4).  They offered the following explanation:

“One major reason for the difference is our more consistent approach in choosing between study-specific never smoker and combined smoker/non-smoker estimates.  Another is our use of derived as well as published estimates.  We included more studies, and avoided estimates for data subsets.  Boffetta et al. also included some clearly biased or not smoking-adjusted estimates.  For pancreatic cancer, their review included significantly increased never smoker estimates in one study and combined smoker/non-smoker estimates in another, omitting a combined estimate in the first study and a never smoker estimate in the second showing no increase. For oesophageal cancer, never smoker results from one study showing a marked increase for squamous cell carcinoma were included, but corresponding results for adenocarcinoma and combined smoker/non-smoker results for both cell types showing no increase were excluded.  For oropharyngeal cancer, Boffetta et al. included a markedly elevated estimate that was not smoking-adjusted, and overlooked the lack of association in recent studies.”

 

III. ST Use and Cardiovascular Diseases

Meta-analyses conducted by Peter Lee in 2007 (5) and Paolo Boffetta and Kurt Straif in 2009 (6) provided summary RR estimates for cardiovascular diseases (primarily heart attack and stroke) among ST users.

  a. Heart attack

Both studies found that ST use is not associated with statistically significant elevated RRs for heart attack (RRs = 1.12, CI = 0.99 – 1.27 (5) and 0.99, 95% CI = 0.89 – 1.10 (6)).  However, Boffetta and Straif (6) reported an elevated risk for fatal cases among ever users (RR = 1.13, CI = 1.06 – 1.21), almost entirely derived from one large Swedish study (7) and a very large study in the U.S. (8). 

Boffetta and Straif (6) did not find a dose-response effect for ST use and fatal heart attack, so the elevated risk from their study is somewhat tentative.  In addition, although no elevated risks were observed in the majority of studies, the large American study (8) that reported elevated risks comprised 85% (by weight) of the Boffetta-Straif analysis.  This is noteworthy because ST users in this study also had elevated risks for emphysema (RR = 1.28, CI = 1.03-1.59) and lung cancer (RR = 2.0, CI = 1.23-3.24), two diseases closely associated with smoking.  Thus, it is likely that the elevated risks for ST use in this study were due in part to residual confounding by smoking. 

  b. Stroke

The Lee (5) and Boffetta-Straif (6) meta-analyses also reported the risk of stroke among ST users.  Lee reported an increase in stroke risk among ST users (RR = 1.42, CI = 1.29 – 1.57)(5).  Boffetta and Straif (6) found no risk overall (RR = 1.19, CI = 0.97 – 1.47), but they found an elevated risk for fatal cases (RR = 1.40, CI = 1.28 – 1.54).  Boffetta and Straif (6) did not find a dose-response effect for ST use and fatal stroke, so this risk is also somewhat tentative.  The American study (8) comprised 89% of the Boffetta-Straif analysis, so the likelihood of smoking among ST users discussed in the previous paragraph is equally important for the elevated fatal-stroke risk.  

 

IV. Evidence That ST is an Effective Substitute for Cigarettes

During the past three years four clinical trials have provided evidence that ST is an effective substitute for cigarettes.  The first, published in 2007 by Mendoza-Baumgart et al., was a short-term (2-week) crossover trial comparing the effects of either a dissolvable tobacco pellet (Ariva, Star Tobacco) or a snus pouch (Exalt, Swedish Match) with a medicinal nicotine lozenge (MNL) (Commit, GlaxoSmithKline) among 65 adult smokers (9).  The authors concluded that “Physiological effects and subjective effects on withdrawal and craving were comparable among Exalt, Ariva, and the MNL. Ariva was preferred over the MNL, which was preferred over Exalt. With the exception of medicinal nicotine products, low-nitrosamine ST products have the greatest potential to result in reduced toxicant exposure compared with other combustible reduced exposure products and have promise for reducing individual risk for disease.”

In 2008 Tønnesen et al. reported the results from a randomized trial comparing the efficacy of group counseling with and without substitution of a ST pellet (Oliver Twist, House of Oliver Twist) among 263 adult smokers in Denmark (10).  The investigators reported that “Point-prevalence abstinence rates at 7 weeks were 36.4% versus 20.8% (odds ration, OR = 2.52, p = 0.001), respectively; and continuous abstinence rates from weeks 4 to 7 were 31.5% versus 19.2% (OR = 1.94, p = 0.023), respectively. The primary outcomes (i.e., 6-month point prevalence) were 23.1% versus 20.8%, respectively (OR = 1.31, ns). ST was relatively well tolerated, although 15 subjects (11.2%) stopped use due to adverse events. A total of 25 subjects (17.5 %) were still using ST after 6 months.  This trial demonstrated short-term efficacy of ST in combination with group support for smoking cessation but no long-term efficacy.” 

It is worth noting that the authors offered the following comment, which may partially explain the lack of long-term efficacy: “…we do not believe that the therapists induced a positive expectation in the ST group. Our impression is that the nurses were not convinced that ST would help or would be accepted by the smokers.”

In 2010 Matthew Carpenter and Kevin Gray published a small, but persuasive, study documenting that dissolvable tobacco products “led to a significant reduction (40%) in cigarettes per day, no significant increases in total tobacco use, and significant increases in two measures of readiness to quit, either in the next month or within the next 6 months.” (11)  They randomly assigned 31 smokers who were uninterested in quitting to receive dissolvable tobacco pellets Ariva or Stonewall (Star Tobacco) or to continue smoking cigarettes.  Smokers were given “minimal instructions on how to use” these products and were “told that there is no safe tobacco product and that the best thing they can do for their health is to quit entirely.”  Carpenter and Gray wrote that their findings suggest “that Ariva and Stonewall are effective products to curb withdrawal and craving.”  In short, these products satisfy smokers.
   
Perhaps the researchers’ most important conclusion was that there is “no evidence that ST (Ariva or Stonewall) undermines quitting. To the contrary, readiness to quit (in the next 1 month and within the next 6 months) significantly increased among smokers who used a ST product relative to those who continued to smoke conventional cigarettes.”

In 2010 Caldwell et al. published clinical trial results demonstrating that snus is a better cigarette substitute than nicotine gum (12).  After observing smoking patterns and consumption for one (lead-in) week, Caldwell et al. gave 63 smokers three different cigarette substitutes, each for two weeks.  The substitutes were Swedish snus (4-gram pouches in three flavors), Habitrol nicotine gum (containing 4 milligrams of nicotine) and Zonnic (a peppermint pouch containing 4 milligrams of nicotine embedded in microcrystalline beads, from Niconovum).  The researchers collected information from the participants about the “acceptability and the willingness of smokers to use” the substitutes.  They asked five questions gauging satisfaction, and they reported that “subjects scored Zonnic and snus more highly than gum for four out of the five…”  All three products significantly reduced craving for cigarettes, and all three “…enabled subjects to reduce their smoking significantly compared with the lead-in week.” 

Participants ranked Zonnic and snus higher than nicotine gum for both quitting and reducing smoking.  “At the conclusion of the study, subjects were asked to rank the three products in order of overall preference. For their first choice, an equal number (40%) chose snus or Zonnic, while 20% chose gum”

In 2006 Philip Morris launched a test market for Taboka snus in Indianapolis (in 2008 it was discontinued when Marlboro snus was launched), and in 2007 the city was one of several expansion markets for RJ Reynolds’ Camel Snus; that product went on to national distribution in 2009. 

Lois Biener and Karen Bogen, from the University of Massachusetts Center for Survey Research, analyzed data from the 2006-7 Indiana Adult Tobacco Survey, and their recently published study provides some valuable information about the consumer interest in snus (13).

Biener and Bogen reported that almost 20% of survey respondents throughout Indiana were aware of snus.  Awareness among smokers statewide was 44%, which was 4.5 times higher than awareness among non-smokers.  Awareness among respondents in central Indiana (i.e. around Indianapolis) was 29%.  More importantly, about 64% of male smokers in central Indiana had heard about snus, and 20% had tried it.  This is evidence that Philip Morris and Reynolds were targeting adult male smokers in their test-market campaigns, and that the manufacturers were fairly successful.

Biener and Bogen also reported that risk perception played an important role in getting people to try snus.  Respondents who correctly believed that ST is less harmful than cigarettes were almost 4 times as likely to try snus as those who had been misinformed about the differential risks.  Unfortunately, this study revealed that 88% of all respondents had been misinformed, so they incorrectly believed that ST was just as dangerous as cigarettes. 

Biener and Bogen offered some perceptive comments on the state of smoker misinformation:

“Both marketing and health education messages should include the information that all tobacco products are harmful and that abstinence from all tobacco products is the most healthful choice. At the same time, simply saying that ST is ‘not safe’ is not a sufficient stance for public health communications.  There is a recognized continuum of risk along which various tobacco products can be placed, with low-nitrosamine ST products much lower on the risk continuum than combustible tobacco, although it is not harmless. Devising an effective way to inform the public about the continuum should be an important research priority, as currently consumers are woefully incorrect in their assessments of relative risk of various tobacco products. This state of affairs could result in people deciding not to give up smoking in favor of a product lower on the risk continuum because they assume that all tobacco products are equally harmful.”

 

V.  Evidence That Misinformation About the Differential Risks of Smoking and ST Use is Widespread

In 2010 Peiper et al. published a study documenting that misinformation about the risks of ST use has led to widespread misperception among highly educated university faculty, even those in health-related schools (14).  They conducted a survey that quantified the risk perceptions of cigarette smoking and ST use with respect to general health, heart attack/stroke, all cancer, and oral cancer among full-time faculty on two University of Louisville campuses, the health science campus and another not focused on health. 

Peiper et al. found that misperception was common among all faculty; 51% incorrectly believed that ST use confers general health risks that are equal to or greater than smoking.  The misperception rate was lower for heart attack/stroke risk (33%) but higher for cancer (61%).  The misperception rate for oral cancer was stunning: 86% of all faculty incorrectly believe that ST use confers risks that are equal to or greater than smoking.  Although faculty on the health science campus had a somewhat lower rate than others (81% vs. 91%), the survey provided evidence that most health professionals have a poor understanding that ST use is vastly safer than smoking.

The authors offered some reasons for the misperceptions: “First, deficiencies in health education may exist with respect to tobacco use and health consequences.  Numerous U.S. studies have shown that medical, nursing and dental school graduates may have inadequate training to provide effective tobacco education or intervention.” (14)

They also commented that misperception results from misinformation from “…anti-tobacco advocates and organizations.  A 2005 review (15) found that websites providing health advice and information tend to conflate the risks of ST with the risks associated with cigarettes, using either direct or implied statements. This misinformation came from respected international and American federal health agencies like the World Health Organization, the U.S. Department of Health and Human Services and the National Cancer Institute, as well as nongovernmental organizations like the American Cancer Society and the Academy of General Dentistry. Another systematic review (16) of over 48 medical brochures from some of the same organizations (e.g., NCI and ACS) found that the risk of oral cavity cancer and of other conditions associated with ST use was frequently overemphasized, ‘reaching beyond the scientific data.’” (14)

 

VI. Evidence That ST Use Is Not a Gateway to Cigarette Smoking

The view that ST is a gateway to smoking is based mainly on two longitudinal studies comparing subsequent smoking among adolescent ST users and non-users (17,18).  The first study, which used the 1989 Teenage Attitudes and Practices Survey (TAPS) and its 1993 follow-up, found that young males who used ST were significantly more likely to have become smokers at follow-up than non-users of tobacco (OR = 3.5, CI = 1.8 – 6.5) (17).  However, a subsequent analysis revealed that the earlier study did not take into account well-known psychosocial predictors of smoking initiation that were in the TAPS, including experimenting with smoking, below average school performance, household member smoking, depressive symptoms, fighting and motorcycle riding (19).  Inclusion of these variables into a multivariate model reduced the odds ratio of smoking among regular ST users to 1.7, which was not statistically significant.  The investigators concluded that the earlier “analysis should not be used as reliable evidence that smokeless tobacco may be a starter product for cigarettes.”

The second study found that 7th and 9th grade students who had used ST (in the past 30 days) were more likely than nonusers to be smoking two years later (OR = 2.6, 95% CI = 1.5 – 4.5), after controlling for smoking by family and friends, low grades, alcohol use and deviant behavior (18).  However, Timberlake et al. (20) have commented that regression analysis may not adequately control for imbalances in covariate distributions between ST users and nonusers.  They analyzed data from the National Longitudinal Study of Adolescent Health after propensity score matching and found that adolescent ST use was not associated with an increased risk of smoking in later adolescence or young adulthood.

In 2010 Rodu and Cole examined the gateway issue with data from the National Survey on Drug Use and Health (NSDUH) during years 2003, 2005 and 2007 (21).  The NSDUH asks survey participants at what age they used cigarettes or smokeless for the first time.  Using this information, Rodu and Cole classified participants as cigarette initiators (meaning they smoked before they used smokeless), ST initiators, or both.  They determined the prevalence of current smoking among these groups using established criteria.  The analyses were restricted to white men age 18+ years, who are most likely to have used ST.  In addition, they looked at white boys aged 16-17 years, since the gateway claim often focuses on teenagers. 

Rodu and Cole showed that the prevalence of current smoking among white men who were cigarette initiators was 35% (21).  In comparison, the prevalence of smoking among ST initiators was only 28%, a significantly lower statistic (prevalence ratio, PR = 0.80, CI = 0.77 – 0.84).  If the gateway effect was real, smokeless initiators would have had smoking rates similar to – or higher than – cigarette initiators.

The results for boys were even more impressive.  Current smoking among cigarette initiators was 43%, but only 18% of ST initiators smoked (PR = 0.43, CI = 0.36 – 0.52).  This means that boys who had started with ST were less than half as likely to be smoking at the time of the survey.
   

VII. The Potential Impact of Swedish Smoking Rates on the European Union (EU)

In 2008 the EU Commission’s Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR) published a report on ST in 2008 entitled “Health Effects of Smokeless Tobacco Products.” (22)  Although the report acknowledged that “…particularly in Swedish men, there is a clear trend over recent decades for smoking prevalence to decrease and for use of oral tobacco (snus) to increase,” it concluded that “…these trends could also be due to successful smoking reduction programs or other socio-cultural factors, and it is therefore not clear whether or by how much the availability of snus has influenced smoking prevalence.”  The report also stated that “…it is not possible to extrapolate the trends in prevalence of smoking and use of oral tobacco if it were made available in an EU country where it is now unavailable.”

In 2009 Rodu and Cole addressed the SCENIHR indecision on extrapolation by estimating how smoking-attributable deaths would decline if EU countries had the smoking prevalence of Sweden (23).  They looked at lung cancer mortality trends in EU countries, starting about 1950 and ending in 2002.  Lung cancer is the sentinel disease of smoking, and a country’s lung cancer mortality rate (LCMR) provides a reasonable indication of the amount of smoking in that country.  The data came from the World Health Organization and the International Agency for Research on Cancer. 

In 2002, there were 172,000 lung cancer deaths among men in the EU.  If all EU countries had the LCMR of men in Sweden, there would have been 92,000 fewer lung cancer deaths. 

Using this data, the number of deaths from smoking in EU countries can be calculated and compared to the number in Sweden.  For men in the EU, 91% of all lung cancer deaths are attributed to smoking, and lung cancer accounts for 31% of all smoking-attributable deaths.  There were an estimated 509,000 smoking-attributable deaths among men in EU countries in 2002.  If all EU countries had the smoking rates of Swedish men, there would have been only 237,000 deaths.  In other words, 274,000 smoking-attributable EU deaths would have been avoided (23).  

References

1.  Nitzkin JL, Rodu B.  The case for harm reduction for control of tobacco-related illness and death.  Resolution and White Paper, American Association of Public Health Physicians.  Adopted October 26, 2008.  Available at: http://www.aaphp.org/special/joelstobac/20081026HarmReductionResolutionAsPassedl.pdf

2.  Boffetta P, Hecht S, Gray N, Gupta P, Straif K.  Smokeless tobacco and cancer.  Lancet Oncology 9: 667-675, 2008.

3.  Lee PN, Hamling J.  Systematic review of the relation between smokeless tobacco and cancer in Europe and North America.  BMC Medicine 7: 36, 2009.

4.  Lee PN, Hamling J.  The relation between smokeless tobacco and cancer in Northern Europe and North America.  A commentary on differences between the conclusion reached by two recent reviews.  BMC Cancer 9: 256, 2009.

5.  Lee PN.  Circulatory disease and smokeless tobacco in Western populations: a review of the evidence.  International Journal of Epidemiology 36: 789-804, 2007.

6.  Boffetta P, Straif K.  Use of smokeless tobacco and risk of myocardial infarction and stroke: systematic review with meta-analysis.  BMJ, Aug 18;339:b3060. doi: 10.1136/bmj.b3060, 2009.

7.  Hergens M, Alfredsson L, Bolinder G, Lambe M, Pershagen G, Ye W, 2007.  Long-term use of Swedish moist snuff and the risk of myocardial infarction amongst men.  Journal of Internal Medicine 262: 351-359, 2007.

8.  Henley SJ, Thun MJ, Connell C, Calle EE.  Two large prospective studies of mortality among men who use snuff or chewing tobacco (United States).  Cancer Causes and Control 16:347-358, 2005.

9.  Mendoza-Baumgart MI, Tulunay OE, Hecht SS, Zhang Y, Murphy S, Le C, Jensen J, Hatsukami DK.  Pilot study on lower nitrosamine smokeless tobacco products compared with medicinal nicotine.  Nicotine & Tobacco Research 9: 1309-1323, 2007.

10. Tønnesen P, Mikkelsen K, Breman L.  Smoking cessation with smokeless tobacco and group therapy: an open, randomized, controlled trial.  Nicotine & Tobacco Research 10: 1365-1372, 2008.

11. Carpenter MJ, Gray KM.  A pilot randomized study of smokeless tobacco use among smokers not interested in quitting: changes in smoking behavior and readiness to quit.  Nicotine & Tobacco Research 12: 136-143, 2010.

12. Caldwell B, Burgess C, Crane J.  Randomized crossover trial of the acceptability of snus, nicotine gum, and Zonnic therapy for smoking reduction in heavy smokers.  Nicotine & Tobacco Research 12: 179-183, 2010.

13. Biener L, Bogen K.  Receptivity to Taboka and Camel snus in a U.S. test market.  Nicotine & Tobacco Research 11: 1154-1159, 2009.

14. Peiper N, Stone R, van Zyl R, Rodu B.  University faculty perceptions of the health risks related to cigarettes and smokeless tobacco.  Drug and Alcohol Rev 29: 121-130, 2010.

15. Phillips CV, Wang C, Guenzel B.  You might as well smoke; the misleading and harmful public message about smokeless tobacco.  BMC Public Health 5: 31, 2005.

16. Waterbor JW, Adams RM, Robinson JM, Crabtree FG, Accortt NA, Gilliland J.  Disparities between public health educational materials and the scientific evidence that smokeless tobacco use causes cancer.  Journal of Cancer Education 19: 17-28, 2004.

17. Tomar SL.  Is use of smokeless tobacco a risk factor for cigarette smoking? The U.S. experience.  Nicotine & Tobacco Research 5: 561-569, 2003.

18. Severson HH, Forrester KK, Biglan A.  Use of smokeless tobacco is a risk factor for cigarette smoking.  Nicotine & Tobacco Research 9: 1331-1337, 2007.

19. O’Connor RJ, Flaherty BP, Edwards BQ, Kozlowski LT.  Regular smokeless tobacco use is not a reliable predictor of smoking onset when psychosocial predictors are included in the model.  Nicotine & Tobacco Research 5: 535-543, 2003.

20. Timberlake DS, Huh J, Lakon CM.  Use of propensity score matching in evaluating smokeless tobacco as a gateway to smoking.  Nicotine & Tobacco Research 11: 455-462, 2009.

21. Rodu B, Cole P.  Evidence against a gateway from smokeless tobacco use to smoking.  Nicotine & Tobacco Research 12: 530-534, 2010.

22. Scientific Committee on Emerging and Newly Identified Health Risks (SCENIHR). Health effects of smokeless tobacco products. Health & Consumer Protection, Directorate-General, European Commission, 6 February 2008. Available at: http://ec.europa.eu/health/ph_risk/committees/04_scenihr/docs/scenihr_o_013.pdf

23. Rodu B, Cole P.  Lung cancer mortality: comparing Sweden with other countries in the European Union.  Scandinavian Journal of Public Health 37: 481-486, 2009.