This paper has presented trends in population cigarette consumption in Great Britain using objective retail sales data. The analysis has shown that population cigarette consumption declined between 2008 and 2013 before stabilising. Cigarettes sold in 14- to 19-packs have substituted a sharp decline in 20-packs and now account for over half of all cigarettes sold in Great Britain. Cigarette consumption has been consistently higher in Scotland than England/Wales. This was due to higher sales of 20-packs in Scotland between 2008 and 2013, but has been substituted by higher sales of 14- to 19-packs in recent years. Convenience stores account for the slight majority of total cigarette sales in Great Britain with notably higher sales of 10-packs than large grocery retailers.
A key factor contributing the downward trend in population cigarette consumption in Great Britain as indicated by retail sales data is likely to be rising cigarette prices [15, 16]. Indeed, price is also likely to explain the rise of cigarettes sold in 19-packs. Manufacturers have made smaller pack sizes available to enable brands to maintain specific price points . In addition, lower priced or ‘value’ brands have become increasingly popular among consumers . This may, indirectly, have contributed to a greater reduction in cigarette sales than would have been expected based on the downward trend in sales of 20-packs before 19-packs emerged into the market. If it assumed that sales of 19-packs have directly replaced sales of 20-packs, 350 million fewer cigarettes were sold in Great Britain over the 5-year period March 2009–March 2014 due to a reduction in pack size. This is equivalent to 17% of the fall in total cigarette sales. More advanced statistical modelling of the data may help to strengthen this interpretation. Indeed, the availability of the cigarette retail sales data may be useful in evaluating the impact, if any, of the introduction of standardised packaging as well as the new EU regulations to introduce a minimum pack size of 20 cigarettes.
The rapid growth in the popularity of electronic cigarettes (‘e-cigarettes’) in Great Britain may also have contributed to the overall decline in population cigarette consumption since 2008. This assertion is supported by the results of Beard et al.  who showed that the prevalence of e-cigarette use by smokers in England increased from 2% in 2011 to 21% in mid-2013 with a stable trend thereafter. If it is assumed that smokers who start using e-cigarettes reduce the number of cigarettes they smoke [20, 21] this trend is consistent with those presented in Fig. 1: mean number of cigarettes sold per adult smoker declined between 2011 and 2013, with stability thereafter. Robust data on e-cigarette sales would help to substantiate this assertion but are currently lacking.
The specific reasons behind the consistently higher level of cigarette consumption in Scotland compared with England/Wales are unclear. The retail environment across Great Britain is similar with the same major retailers (supermarkets and chains of smaller stores) and tobacco multinationals operating in both jurisdictions. The policy context is also similar. For example, Scotland banned smoking in public places in 2006, with England/Wales introducing the same legislation in 2007. In addition, the increase in the minimum age of purchase to 18 years was brought into force within a day of each other (30th September 2007 in Scotland and 1st October in England/ Wales). However, besides the retail and policy environments, cigarette consumption patterns are influenced by a wide range of other factors, including historical smoking rates, demographics, cultural norms and socioeconomic deprivation. Given the strong social patterning of smoking , comparing Scotland with subnational areas of England/Wales with a more similar social and deprivation profile may have narrowed the differences observed [22, 23].
Strengths and limitations
This is the first time that retail sales data have been used to assess trends in population cigarette consumption and to explore differences between constituent countries in Great Britain. Using routinely available data on population size and smoking prevalence, we were able to express population consumption as cigarettes per adult smoker, thereby strengthening our interpretation. The cigarette sales data were available at frequent time periods (monthly and for more recent data weekly) with breakdowns by pack size. This enabled novel insights into market share changes. Data at Great Britain level were also available by retailer category and these revealed differences in the number of cigarettes sold in different pack sizes between large grocery retailers and smaller, convenience stores.
These features of cigarette retail sales data show the usefulness of the data for policy monitoring and evaluation. Indeed, data on cigarettes either cleared for sale or sold by retailers have been used to evaluate tobacco control efforts in other countries (for review see ). In Great Britain, recent tobacco control polices have included a ban on smoking in public places and a tobacco display ban for retailers selling tobacco products. Legislation to introduce standardised (or ‘plain’) packaging was introduced in May 2016 with tobacco manufacturers given one year to achieve full compliance . The legislation coincided with the introduction of the European Union (EU) Tobacco Products Directive which imposes new regulations on how tobacco products are manufactured, produced and presented in the EU, including minimum content for unit packs and increased size of health warnings . Cigarette sales data offer the potential to strengthen the evaluation of these recent legislative and regulatory controls in Great Britain.
The most important limitation of using retail sales data is their vulnerability to biases that can impact on their validity and reliability. This includes, for example, measurement error associated with Nielsen’s data collection methods, such as the non-inclusion of outlets that sell cigarettes and non-response bias from outlets invited to be part of their sample . The size of this bias is not quantifiable; however, levels and trends of annual population cigarette consumption derived from Nielsen data are broadly consistent with those based on tax clearance data, which provides reassurance that the data are representative and measure what they purport to measure . Furthermore, Nielsen data are now generally considered the gold standard source of data for monitoring and evaluating population alcohol consumption in Great Britain [7, 26]. This followed a detailed critique of the sampling methods used by Nielsen to estimate alcohol sales , which are very similar for cigarette sales . Although some methods may be considered commercially sensitive, developing good working relationships with data providers will help others to enhance the understanding and interpretation of cigarette sales data in their own context.
Other important biases affecting the validity and reliability of cigarette sales data were identified in a recent report . Using the best available data, it was estimated that cigarette retail sales data are likely to have underestimated actual population cigarette consumption in Scotland in 2013 by 15%, with a wide range of uncertainty around this estimate and notable changes over time. This was largely due to the consumption of illicit cigarettes, cross-border purchases and duty-free shopping. These biases present a challenge to the use of sales data to estimate population cigarette consumption. Nonetheless, we believe that the novel insights presented in this paper provide support for our earlier assertion that sales data offer the potential to strengthen the monitoring and evaluation of tobacco control policy when triangulated with other sources of data .
A final limitation is that the sales data presented here include cigarette sales only. As such, it is not possible to comment in trends in overall tobacco consumption. For example, there has been a recent growth in the consumption of loose, or ‘roll-your-own’, tobacco and this will have offset some of the overall decline in the consumption of manufactured cigarettes .