Wednesday March 26 2014
Study was not peer reviewed
“Unbranded cigarettes could mean people smoke more, experts warn,” The Daily Telegraph reports.
However, the evidence provided by experts and reported on by the media has not been peer reviewed, and their work has been funded by a major cigarette manufacturer.
The news story is based partly on an analysis of smoking trends in Australia and partly on new industry data on tobacco shipped to retailers in the same country. We have only appraised the former of the two.
In 2012, Australia introduced a plain tobacco packaging law. The country's law makers believed that stripping branding from cigarettes packs and including graphic images of the health risks smokers face may stop youngsters from taking up the habit.
The study indicated that smoking rates in Australians aged 14-17 have been steadily declining over the past decade. However, there is tentative evidence to suggest this long-term trend didn’t dramatically alter in the year after the plain packaging law was introduced (from December 2012 to December 2013).
Importantly, there was only a year’s worth of data after the law came into effect on which to assess whether it had had the desired reduction in smoking uptake among the young. This is a relatively short period to assess such an impact.
Because of this, it is difficult to make any firm conclusions on whether plain packaging affected smoking prevalence.
It is also worth noting that the tobacco industry is estimated to kill five million people a year worldwide, according to the US Centers for Disease Control and Prevention.
Where did the story come from?
The study was carried out by researchers from The Department of Economics at the University of Zurich (Switzerland) and Saarland University (Germany), and was funded by Philip Morris International (PMI). PMI describes itself as “the leading international tobacco company”. The authors stated that “at no time did we provide Philip Morris International with access to the underlying data”. However, the researchers do not say whether PMI had any control over the study design and other factors that could influence the results.
The study was published as part of a “working paper series” for the Department of Economics at the University of Zurich. The research does not appear to have been peer reviewed, meaning it has not been scrutinised by independent experts in the field for methodological rigor, or to check if the conclusions are reliable. This increases the risk of misleading findings, which can reach the public and mainstream media before they have been properly scrutinised.
There is a clear potential conflict of interest in receiving funding from a leading tobacco company when attempting to carry out impartial research assessing smoking data. The risk of misleading information being presented is further increased when the research is not peer reviewed. Given that both these factors are present in this particular study, the results should be interpreted with caution.
Generally, the media reported the story accurately. However, few mentioned the potential conflict of interest surrounding the funding or how the study was conducted, and none mentioned the lack of peer reviewing.
What kind of research was this?
This was a repeated cross-sectional study (time-trend analysis) using data to estimate the impact of plain packaging on smoking prevalence in 14 to 17 year olds in Australia.
In December 2012, the Australian Plain Packaging Act 2011 came into force, with the aim of reducing smoking prevalence. However, the law was particularly intended to prevent young people taking up smoking. Australia was the first country to enforce a law of this nature, and other governments are eager to see if it works before deciding whether to do something similar.
What did the research involve?
The study used market research data from the Roy Morgan Single Source (Australia).
Roy Morgan is a major Australian market research firm, and the Single Source data set has been drawn from surveys. The data is reported to be weekly surveys completed through computer-assisted personal interviews (CAPI), which were administered door-to-door and sampled about 50,000 Australians a year.
The participants were Australians aged 14-17, and were assessed between January 2001 and December 2013. The surveys aimed to see whether the plain packaging law introduced in December 2012 was achieving a reduction in smoking prevalence.
The researchers used yearly prevalence data to plot a long-term trend of smoking from 2001 to 2013. Their main analysis then looked at month-on-month prevalence variations, to see if there was any obvious acceleration of the downward trend after the packaging ban was introduced.
Smoking prevalence was based on a binary variable on whether the person smoked. There was no further description of how this was arrived at, or what participants were asked as part of the CAPI assessing their smoking.
What were the basic results?
There was a decline in the annual long-term trend for smoking prevalence in Australians aged 14-17 years. Smoking prevalence reduced from nearly 12% in 2001 to almost 6% in 2013 – an annual average decline of 0.44%.
The month-on-month estimates were based on samples of between 350 and 200 people – some months saw slightly less people taking part. Because of this, there was a large variation surrounding the longer-term downward trend.
The month-on-month estimates since the introduction of the packaging ban saw a similarly large variation in smoking prevalence. There was no obvious acceleration of the long-term downward trend based on the data.
In a related press statement, new figures on tobacco sales were released from Philip Morris International. This statement says the information “shows that legal tobacco sales actually rose incrementally, by 59 million cigarettes in the first year plain packaging was introduced. This increase reversed the long-term decline of legal sales volumes in the country since before 2009”.
How did the researchers interpret the results?
Based on the trend analyses, the researchers said they failed to find any evidence for an actual plain packaging effect on reducing smoking in young people.
Conclusions in the press release based on the trend analysis and tobacco sales figures indicated that “plain packaging in Australia has not reduced smoking rates and has had no impact on youth smoking prevalence. Consumers aren’t smoking less, they are just buying cheaper alternatives like roll-your-own cigarettes, or turning to branded packs available on the black market”.
This study indicates that smoking rates in Australians aged 14-17 have been experiencing a steady decline over the last decade. The evidence tentatively suggests this long-term trend didn’t dramatically alter in the year after the plain packaging law was introduced (between December 2012 and December 2013).
Importantly, there was only a year’s worth of data after the law was introduced on which to assess its impact. This is a relatively short period to assess such an impact, and it may be too narrow a timescale to identify a movement in the long-term trend. Furthermore, the month-on-month prevalence estimates are highly volatile, so cannot be relied upon to give an accurate picture either.
It was also not clear how smoking prevalence was assessed, so the exact estimates of smoking rates may be erroneous. It was also not clear if the same method was used over the entire decade, allowing prevalence estimates to be accurately compared year-on-year. Changes in how smoking was assessed during the questionnaire and categorised into smokers and non-smokers could distort the results.
The separate data released containing tobacco sales data was not critically appraised as part of this article, so we cannot comment on whether this data is reliable or informative.
The smoking data for this trend study was obtained from a market research company. Census data containing information smoking prevalence may also be available, and it would be useful to see if this alternative data source matches the information presented in this study. If this were possible, we would at least be able to cross-validate the longer-term trend findings from this study.
The research does not appear to have been peer reviewed, meaning it has not been scrutinised by experts in the field for methodological rigor, or to check if the conclusions are reliable. This significantly increases the risk that misleading findings can reach the mainstream media and public before it has been properly scrutinised.
Based on this data alone, it is difficult to make any firm conclusions on whether the plain packaging affected prevalence rates.
While the researchers stated that there was no access to the analysis of the data by the tobacco company, it raises questions – and eyebrows – that this research has been released to the press without being peer reviewed by independent experts.
Analysis by Bazian. Edited by NHS Choices. Follow Behind the Headlines on Twitter. Join the Healthy Evidence forum.