Pssst! Wanna boost kids’ smoking? Have a display ban!

Public health regulations frequently serve exclusively to prevent legal capitalistic acts between consenting adults. But the fact that prohibition’s history is a history of failure never stops the good people in the public health establishment from promoting the next ban, or the next one after that.

This brings us to the tobacco display ban, which would require shopkeepers to hide all tobacco products from sight.

Since the UK battle commenced over tobacco displays, those in favour have confidently argued that a ban here would have the same outcome as display bans in other countries, especially Canada. We continue to be promised that, critically, a display ban will reduce youth smoking.

What has the Canadian ban on tobacco retail displays done for public health in that country? Analysing the data produced by Health Canada leads me to conclude that display bans do not lead to lower rates of youth smoking. Nowhere – anywhere – in Canada is it possible to find evidence that display bans result in less youth smoking. Adult smoking is equally unaffected by display bans.

To be fair, however, I must acknowledge that display bans do have a positive impact upon youth cigarette consumption. By positive, though, I don’t mean good. By positive, I mean display bans boost the average number of cigarettes consumed daily. In the province of Nova Scotia, for example, youth consumption jumped 15% in the two years after introducing a display ban. Nationally, smoking among 11-15 years olds rose 46% and smoking among 15-18 year olds rose 16% between 2007 and 2009, according to Health Canada’s biannual survey released last month.

My statistical analysis comparing smoking prevalence before and after the implementation of display bans in the relevant provinces shows display bans are associated with increased prevalence for both young people and adults in Canada and no decline in consumption. 

Any careful reading and balanced assessment of the Canadian situation can only conclude that Canada’s display bans have been harmful to her public health.

Click here to download Patrick Basham’s report, “Canada’s ruinous tobacco display ban: economic and public health lessons”.

It is also mentioned in Dr Basham’s erudite paper but after a display ban was enacted in Iceland, youth smoking went up too.Regular smokers aged 15-21.2000 14.4%
2001 17.5% display ban
2002 17.9%
2003 16.9%
2004 17.4%
2005 15.7%
2006 12.2%
2007 15.2%Never smoked.1999 69.3%
2001 70.0% display ban
2002 70.6%
2003 65.8%
2004 72.6%
2005 69.7%
2006 67.7%
2007 69.3%http://www.statice.is/?PageID=1406&src=/temp_en/Dialog/varval.asp?ma=HEI07102%26ti=Smoking+habits+by+sex+and+age+1994-2007++++++%26path=../Database/heilbrigdismal/afengiogreyk/%26search=SMOKING%26lang=3%26units=percent

Today cigarettes to be hidden. Tomorrow alcohol, then cakes and sweet, and after that anything else that could be bad for our health.

Dr Basham’s paper undermines the hypothesis that smoking display bans are effective public health measures. Moreover, even if the bans led to big falls in smoking, there would still be strong grounds to object to them. Shops are private property and the decision as to what products are displayed within them should be made by the owners, not government. Display bans are yet another attack on private property rights.

It appears the IEA have protected private property rights and the free market. The paper is cited in this Guardian article.“Anti-smoking law may be overturned in government review. Government could scrap Labour legislation to ban cigarette machines from pubs and tobacco displays from retailers.”http://www.guardian.co.uk/politics/2010/jul/14/anti-smoking-law-government-review

And we have been described as “right wing”. I guess if you are pro euthanasia you are left wing and if you are pro allowing people to smoke you are right wing. I guess it would have confused the Guardian readers even more if we had been described as “liberal”

Dr. Basham’s analysis is patently misleading, and boarderline false. In “Canada’s ruinous tobacco display ban” he rails against the ESPAD data, while saying that the data from statistics Iceland is better. This is untrue. The reason for this is that the standard of error (”SoE”) for the Statistics Iceland data is around 3-4%, meaning that the percentage needs to change by 6-8% for the change to be statistically significant. ESPAD SoE is less than a percent. Basham conveniently ignores the standard of error for the CTUMS. Nova Scotia’s 15-19 data has a SoE of 12%. The 2% change he cites in his report is not statistically significant, and basing a conclusion on it is shoddy analysis.

My mistake. The Standard of error for the Nova Scotia data is 3.6%, the coefficient of variance is 12%. My source for this information is http://www.statcan.gc.ca/dli-ild/data-donnees/ftp/ctums-esutc/2008survey.... Use the “User guide” along with the “variance tables” to compute the standard of error for the data. My above analysis remains the same–Basham uses statistics with disregard to the standard of error and makes conclusions which are unsupported by the statistics he cites. Very poor work for one who makes himself out to be an expert.

First the ESPAD database reports only every 4 years and Stat Iceland annually. Also I can tell you how ASH arrived at the figures. Firstly I made a mistake, the age groups I quoted in the first comment were for 15-19 year olds. ASH concentrated on 15 year olds only who did smoke less, but the 16-19 ages showed an increase. The ESPAD database confirmed in 2007 “The proportion reporting cigarette smoking during the past 30 days[is](16%).” Stat Ice says 15.2% remarkably similar. Today the figure of those who smoke in 2009 is now 18.2%. We can conclude that the display ban is ineffective in reducing youth smoking.http://www.espad.org/sa/node.asp?nodehttp://www.statice.is/?PageID=1406

Dave,
The problem with your analysis is, like Basham, you either are not understanding or are ignoring some important statistical concepts. When people are trying to determine the prevalence of a social behavior, like smoking, it is not feasible to ask every single person in the population you are tying to study. Instead, they ask a representative portion of that population, and extrapolate from that proportion’s responses what the rest of the population’s response would be when confronted with the same question. This is a very effective tool for social research. However, it is not a perfect measure, because the sample never completely matches the population. The discordance between the

sample and the population is referred to as the “sampling error.” The smaller the sample, the larger the sampling error, and the larger the sample, the smaller the error will be. For example, if we are studying a population of 100 people, and we poll 5 people, the average of those responses will be less likely to represent the population than if we had sampled 30 or 40 people. Sampling error is measured a number of ways, but the easiest way to measure it is by a Standard of error. This is expressed by confidence level and a range. For example, the Canadian Tobacco Use Monitoring Survey sampled a large portion of the Canadian population and asked them questions regarding their smoking

John I do have a grasp of statistics and in epidemiology a sampling error is expressed as a P < x% number. The ESPAD database is based on: "..each country should draw a sample of about 2800 students of the target population. Since some students will be..sick or absent give about 2400 participating students."I have emailed Stat Iceland and have asked them for the size of sampling and expected variance. Methinks you are employing the tactics used in the faux critique of the BMJ published Enstrom/Kabat paper.http://www.espad.org/documents/Espad/Documents/The_ESPAD_History.pdf

John while I am here let us have a quick review of some epidemiology numbers. Perhaps you would like to hazard a guess which paper these numbers come from proving that second hand smoke (SHS) does not cause lung cancer and/or heart disease?ETS exposure during childhood was not associated with an increased risk of lung cancer (odds ratio [OR] for ever exposure = 0.78; 95% confidence interval [CI] = 0.64-0.96). The OR for ever exposure to workplace ETS was 1.17 (95% CI = 0.94-1.45)Yes the paper which proves how harmless SHS/ETS is the 1998 WHO/IARC paper by Professor Boffetta.An RR of 1.17 infers that there is an 85.47% that ETS/SHS was not the cause of an llness.

Dave,
First off, I have no comment on the SHS information, as I do not care one way or the other on the issue. My comments are directed at the failure of Basham to prove anything with his statistics, and a critique of his exceptionally poor research methods, not anything related to the dangers (or lack thereof) of SHS . I was unable to finish what I began posting yesterday because something came up at work. What I was saying: The upward change of smoking rates in Nova Scotia from 12.5% to 14.5% is well within the margin of error of +/-3.6% (at 95% CI), and so the change is not statistically significant. Certainly not anything which “proves” that a display ban boosts kids smoking.

However, here something that is consistent with the theory that a tobacco display ban REDUCES youth smoking. In 1999, Iceland’s youth s smoking prevalence was 28% among the 10th grade students who were surveyed with at SoE of less than +/-1%. In 2007, the smoking rate among the 10th grade students is 16%. That is a statistically significant change. While correlation does not prove causation, it is consistent with the hypothesis that a tobacco display ban reduces youth smoking, and completely inconsistent with the hypothesis that a tobacco display ban increases youth smoking. Source: European School Survey Project on Alcohol and Other Drugs, http://www.espad.org/keyresult-generator, (last visited July 12, 2010) (select key variable: “Cigarette use past 30 days”; select counties: “Iceland”; select years: “1995, 1999, 2003, and 2007”; select chart type: “Time line chart”; click “view”).

The link to the report is not working, so I can’t check the analysis, but John seems to have latched onto a claim by Patrick that the ban actually caused an increase, rather than it was just ineffective. In my opinion it would have been enough to show there was no evidence that the ban was linked to a decrease, which is clear from the figures, without doing any analysis.
Usual statistical methodology is to adopt two hypotheses: the null hypothesis, which is the status quo, i.e. the ban has not made a dfference; and the alternative, that it has. A useful analogy is the English justice system. The null hypothesis is innocence and it it is up to the prosecution to prove guilt beyond……..

John thank you for your response, I must say I find it more illuminating than you probably do. I said above: “ASH concentrated on 15 year olds..” which you confirm by referring to “10th graders.” However as I am at pains to point out if you include 16-19 as a group smoking went up.I am still waiting for Stat Iceland to get back to me but if you take the figure 69.3% as a (never smoker) number, because it is a prime number to one decimel place, by sheer mathematical logic dictates that Stat Iceland must use at least 1,000 people as part of the sample. So how near to 2,400 will it be?

Jonathan has said all that needs to be said in that even you John are confirming that there is no reduction in youth smoking after display bans. ASH have been making this incorrect point for a number of years.“ASH Scotland chief executive Sheila Duffy said the legislation could help reduce the 15,000 youngsters in Scotland who start smoking every year.”Perhaps John you could tell us where you are from?http://www.dailyrecord.co.uk/news/politics-news/2010/01/27/ban-display-of-tobacco-and-cigarette-vending-machines-msps-are-told-86908-21999252/

Since I am both a part of the Icelandic ESPAD research team and director of Research at the Icelandic Public Health Institute of Iceland, the body that collects the data reported by the Statice I must comment on the use of those figures.In the Icelandic ESPAD project almost the entire 10th grade cohort is sampled every 4th year. The method is valid to measure children smoking habits. The ESPAD survey shown reduction in youth smoking rate after the display banThe survey cited by STATICE is not designed to monitor changes in small age groups. The sample is annual N= 2400 age 15-89. ESPAD survey is far better for monitoring changes in teenage smoking behavior. Stefan Hrafn Jonsson

I have asked Statice to include comments about the small sample size when citing those numbers. They do that in some but not all places on their webpage. The smoking rates in Iceland for kids in 10th grade has according to number of cohort based studies (where sample is about 85-90% of each cohort) reduced from 15% in 2001 to about 10% in 2005 about 11% in Espad 2007, and down to less than 10% in unpublished research for 2010.When we want to monitor change in teenage smoking rate we tend to use the cohort based studies that are now available annually, ESPAD every 4th year and other studies are used for non-Espad years .
Few, if any, believe that this reduction is an effect on one factor

Based on the data we have, there is no support that the display ban has caused a increase. All cohort based surveys that we have show a decrease.The sample size for 15-19 years old for figures from Statice is about 200-400 each year. The sample (responses) size for ESPAD Iceland:in 1995 n = 3814
in 1999 n = 3555
in 2003 n = 3381
in 2007 n = 3678. The cohort size in 10th grade in Iceland is about 4200-4500 each year.

Here is a peer review article that address, among other, the trend in tennage smoking in Iceland. http://www.hr.is/lisalib/getfile.aspx?itemid=12303More when I return from vacation.

Stefan, I trust you are well and I hope you enjoy your holiday.You said “Few, if any, believe that this reduction is an effect on one factor.”That is precisely our point, any reduction in smoking cannot be attributed to the display ban.

I have read your scholarly and erudite paper in detail. I was shocked at the level of substance abuse in Iceland, with 42% of 14 and 15 year olds being drunk in the last 30 days, now down to 20%. It seems that with alcohol and cannabis, that are not subject to display bans, especially cannabis, there has been a steady reduction over 10 years in all abuse. Well done.My point is that some organisations in the UK have misrepresented your work with the ESPAD database. I.e. the major or even sole reason for the reduction in youth smoking in Iceland is down to the display ban. Your paper does not even mention it but your community and peer pressure initiatives being the critical interventions.

Dave AthertonFirst of all this is not my paper I cite earlier but address the issue.You state “That is precisely our point, any reduction in smoking cannot be attributed to the display ban.” As far as I have followed the discussion on the effect of display ban and teenage smoking rates in Iceland I recall arguments that states that smoking rates have gone up in Iceland after the display ban. Best available data shows that the true is that it has gone down. The title in this Blog “Pssst! Wanna boost kids’ smoking? Have a display ban!” is a clear sign of this misunderstanding of the Icelandic situation. Best

Post new comment

The content of this field is kept private and will not be shown publicly.
Type the characters you see in this picture. (verify using audio)
Type the characters you see in the picture above; if you can't read them, submit the form and a new image will be generated. Not case sensitive.