Data sources

We searched six databases of peer-reviewed literature (EMBASE, Web of Knowledge, Scopus, PsycInfo, Medline and the British Library catalogue) using a search strategy (see supplementary file) based on the terms ‘tobacco’, ‘packaging’ and ‘illicit’ and then adapted it to suit individual databases using Boolean operators. Reference lists of included studies were hand searched for additional papers. Data sources were searched for English language references from the period 1st January 1987 (when it is generally accepted that the first study into standardized packaging was conducted²⁷) to 28th November 2016. After eliminating duplicates, this search produced 902 papers.

Study selection

Two authors (CAH and CMT) shared screening of titles and abstracts (where available) for empirical studies that addressed the topic of standardized packaging and illicit tobacco. A 10% random sample of titles and abstracts was screened by both authors at the outset to check for consistency, with any disagreements resolved by consensus or third party adjudication. Empirical studies were defined as those based on observed and measured phenomena and which derived knowledge from actual experience rather than from theory or belief. No restriction was placed on empirical study design. Papers were excluded if they focussed only on illegal sales in relation to underage sales (as opposed to sales of illicit tobacco) or packaging in relation to health warnings and product labelling. See Table 1 for detailed inclusion/exclusion criteria. This resulted in 153 full text papers being retrieved for screening against the inclusion criteria. Following exclusions, ten papers were included in the review (See Figure 1).

Figure 1

Process of identifying references for inclusion in systematic review²⁸

Data extraction

Data extraction was shared between two of the authors (CAH and CMT) using piloted standardized data extraction forms for consistency, control of bias, validity and reliability²⁹. A 10% random sample of included papers (corresponding to one complete study³⁰) was extracted by both authors at the outset to check for consistency with any disagreements resolved by consensus or third party adjudication. A standardized data extraction form was developed by CMT and then piloted on an included study. As a result some redundant fields were removed and others included and expanded upon. The data extracted included study aims and design, the sample studied, sampling strategy, recruitment and consent process, data collection and analysis and findings. Each study was also assessed for quality and relevance using the most appropriate critical appraisal tool from the Critical Appraisal Skills Programme (CASP)³¹, depending on the study design. Quality was included as part of the data extraction process but was not used as a basis on which to exclude studies.

Synthesis

We were prevented from conducting a meta-analysis as there was little homogeneity between the included studies. The results of the papers are therefore presented through a narrative synthesis with studies grouped according to purchasing, retailing or industry behaviour.

Purchasing behaviour (intended)

Three studies examined the potential impact of standardized packaging on intended illicit tobacco purchasing behaviour^{30, 32, 33}. Two qualitative studies carried out in Glasgow in 2010³⁰ and 2012³², respectively, used focus groups to explore current smokers’ illicit tobacco purchasing intention if standardized packaging was subsequently introduced. Moodie, Hastings and Joossens³⁰ used purposive sampling to recruit 54 adult smokers aged 18-35 years, as at the time, this age group contained more than half of all smokers in the UK and previous research had suggested that younger smokers were buying more illicit cigarettes than other age groups. The consensus among male participants, who were more likely to buy illicit tobacco, was that if all cigarettes came in the same colour packs, their illicit tobacco purchasing behaviour would not change. The main reason given for this was that illicit tobacco use was typically the result of the availability and the price of the product. Females similarly reported that standardized tobacco packaging would not change their use of illicit tobacco, or make them start buying it because of the packaging. Participants also reported that it was unlikely that they would purchase counterfeit cigarettes by mistake as they were generally easy to identify³⁰.

Later Moodie, Purves, McKell and de Andrade³² used purposive sampling to recruit 49 young women smokers aged 16-24 years old. This age group was selected because between 2004 and 2009 the rate of smoking among women this age increased while the popularity of smoking in women aged 20-24 years old remained similar to what it had been 25 years previously. Research had also revealed that the attractiveness of cigarette packaging was more influential to women this age than men of a similar age. Therefore removing the visual allure of packaging, through standardization, may have a greater impact on young women’s opinions of illicit tobacco. Participants reported that, apart from reduced cost, there were few benefits to counterfeit tobacco. Again the decision to purchase counterfeit tobacco was usually based on price. Participants, who had bought counterfeit cigarettes before, believed them to be of a lower standard and less enjoyable than their usual legitimate brand, however they would consider purchasing counterfeit cigarettes if their finances dictated it. While cost was definitely a consideration, participants who had never purchased counterfeit cigarettes before reported that they would not switch to counterfeit products if this meant forgoing the positive attributes of their legitimate brand. In terms of the cigarette brand that participants would choose if standardized packaging was introduced, most participants reported that if all cigarette packaging (legitimate and counterfeit) were standardized they would continue to purchase their current brand. Participants who actively opposed the introduction of standardized packaging also reported that they would continue to purchase their usual brand because they would know what was in the box, although some respondents reported that they would consider purchasing cheaper legitimate brands. Several young women reported that they would consider purchasing counterfeit cigarettes; however on being told that these brands could not be purchased from shops, they reported that they would continue to purchase their current legitimate brand. Current users of counterfeit cigarettes confirmed that they would continue to purchase counterfeit cigarettes from time to time as a result of cost and convenience. However, users of counterfeit cigarettes said that, in general, they would continue to purchase their usual legitimate brand and if finances demanded they would purchase cheaper cigarettes (legal or counterfeit) regardless of packaging.

In contrast to these two qualitative studies, Freeman³³ conducted a content analysis of commentary posted on Australian online news articles that covered the topic of tobacco standardized packaging legislation in Australia after the measure was announced in 2010. The aim of the analysis was to examine how online readers reacted to standardized packaging, in order to understand how those opposed to policy reform framed their arguments thereby allowing; (a) anticipation of such arguments in countries where standardized packaging was being discussed and (b) the subsequent planning of communication strategies to counteract oppositional arguments that have the potential to gain political traction. Of 117 standardized packaging news items, 41 included 1818 reader comments. Of these 1187 (65.3%) comments contained no reference to standardized packaging and mainly addressed a tobacco tax rise announced at the same time. The comments about standardized packaging were more than 2.5 times more likely to oppose than support the policy. The dominant argument, comprising 27% (206) of oppositional comments was that standardized packaging would be ineffective in reducing smoking. However, the study also found 45 comments (5.9%) reporting that standardized packaging as a policy would fail as smokers would be more inclined to purchase counterfeit or smuggled branded packs. These arguments framed standardized packaging as pointless legislation as it would present loopholes to exploit by the tobacco industry and smokers.

In terms of quality, both qualitative studies presented for the first time, data which provided insight into young adult³⁰ and young women³² smokers’ perceptions of the potential impact of standardized packaging on intended illicit tobacco purchase behaviour. Both studies used relatively large samples for qualitative research; however it is unlikely that findings could be generalised to the wider population of older smokers. More research is needed to consider a broader age range of smokers as well as level of addiction. The reality was that standardized packs were not available in the UK and that socially desirable responding may have impacted on the responses from participants. A further limitation was the use of a visual prompt that did not accurately represent the appearance of standardized packaging, either as proposed in the UK or as used in Australia. In addition neither qualitative study provided detail with regard to how the data were analysed, whether data saturation was achieved or reported any method used to validate findings. However, Moodie, Purves, McKell and de Andrade³² reported methods to minimise potential moderator and question order bias. While Freeman³³ was first to systematically document and analyse the common opposition arguments to standardized tobacco packaging legislation in Australia in readers of online news, the time period searched was small (11 days) and the method by which data were collected (via online news articles) was weak. These limitations were exacerbated by a lack of knowledge about the people who regularly post comments on online news items. The views gathered were from self-selecting individuals who were unlikely to be representative of the Australian population. Further, it has been reported that negative emotions motivate people to express their opinions and the most active people who post online comments are those with negative views on events⁴⁰. It was not clear within the published paper whether commentators were posting more than once³³. However, Freeman³³ ensured that two researchers each coded the sample independently and a total of six researchers coded a random sample. Inter-coder reliability tests revealed a high level of reliability.

Purchasing behaviour (actual)

Three studies examined the actual impact of standardized packaging on illicit tobacco purchasing behaviour following implementation of legislation in Australia^{21, 36, 37}. Two studies used repeated³⁶ and continuous³⁷ cross-sectional population telephone surveys, respectively, to examine early evidence of whether Australian smokers were more likely to use illicit tobacco following the standardization of tobacco packaging. Scollo and colleagues³⁶ carried out a survey one year prior to implementation of the policy (November 2011), during roll-out (2012) and a year after implementation (2013). Participants were smokers aged 18 years and over, identified in an annual population survey in the Australian State of Victoria (2011: n=754; 2012: n=590; 2013: n=601). Results revealed the prevalence of low cost Asian brands was low and did not increase between 2011 (1.1%) and 2013 (0.9%) (adj OR=1.02, 95% CI 0.28 to 3.75, p=0.98). The proportion who reported use of unbranded, i.e. illicit tobacco, was 2.3% in 2011 and 1.9% in 2013 (adj OR=0.69, 95% CI 0.25 to 1.88, p=0.46). In 2013, 2.6% of cigarette smokers reported having purchased one or more packs of cigarettes in non-compliant packaging in the past 3 months.

Scollo and colleagues³⁷ also carried out a larger, continuous national cross-sectional telephone survey from 6 months prior to implementation of standardized packaging (April 2012) until 15 months after implementation (March 2014). Respondents were current smokers (n=8678). Results showed there were no significant increases in purchase of ‘illicit whites’ (<0.1%; OR=0.24, 95% CI 0.04 to 1.56, p=0.134), international brands purchased for 20% or more below the recommended retail price (0.2%; OR=3.49, 95% CI 0.66 to 18.35, p=0.140) or packs purchased from informal sellers (<0.1%; OR=0.24, 95% CI 0.04 to 1.47, p=0.124). The prevalence of any use of illicit tobacco remained at about 3% (adjusted OR=0.79, 95% CI 0.58 to 1.08, p=0.141). The results of this large National study³⁷ confirmed those of the smaller State-based study³⁶. Both found no increase in purchase of unbranded illicit tobacco and low levels of purchase of cigarettes likely to be contraband.

Finally, using the preferred methodology of the tobacco industry (empty-pack surveys), a third study²¹ also reported no increases in illicit tobacco use in Sydney following the introduction of standardized packaging. Peters and Saffron²¹ reported research based on two empty-pack surveys carried out 12 months apart in Sydney. In phase 1 of the study 1802 cigarette packs were collected, 12.7% (range 0-33% by local government area) of which did not comply with Australian legislation. In an extension phase of the study in which only non-conforming packs were collected, 727 were found. Of these, 500 were manufactured in or for South Korea and 66 in or for China. Only 1.4% of the cigarettes found were illicit or contraband. It was clear from this survey too that there was no increase in illicit tobacco use following the introduction of standardized packaging.

The State-based telephone survey³⁶ used both mobile and landline phones to survey smokers, which is likely to result in a more representative sample than using landlines alone. However, recent purchase of potentially illicit branded (contraband) tobacco was only assessed in the final year of the survey. As with other surveys, socially desirable responding and recall bias may have affected prevalence of illicit tobacco (as it is illegal), however this is likely to be the same over time. The survey was only conducted in one Australian State and with English speaking residents. The authors suggested that a larger National sample would be required to provide more robust data. However, the large National survey³⁷ still suffered from some of the same limitations as the earlier survey such as socially desirable responding, recall bias and exclusion of non-English speaking individuals. Also, because it is known that the prevalence of illicit tobacco purchasing behaviour increases with regularity of smoking, a sample which is restricted to individuals who smoke frequently will produce higher reported levels of this behaviour. Also when participants are surveyed regarding illegal behaviour, respondents may be unlikely to report this illegality or may be unsure as to whether their behaviour was illegal or not. This is despite ensuring anonymity and when consequences for the illegal behaviour are mild. Finally, the surveys were not able to reach heavy smoking subpopulations, such as the homeless or those in prison.

Limited information was provided on which to judge the quality of the empty-pack survey²¹. Regardless, empty-pack surveys do allow researchers to use pack markings to identify purchase location and to determine the location of use based on where the pack was discarded. Unfortunately, empty-pack surveys suffer from lack of generalisation, as collection areas are often not in regions representative of the general population but limited to specific parts of a city. This may be further exacerbated by patterns of commuting and tourism. Other limitations of this method include problems identifying times of purchase and of consumption of the discarded packs as well as type of packs identified and packs purchased legally but discarded in the collection area.

Retailing behaviour (intended)

Only one study examined retailers’ perception of how the proposed introduction of standardized packaging would influence illicit trade³⁹. Guthrie, Hoek, Darroch and Wood³⁹ reported a qualitative study employing in-depth interviews with 23 retailers of small convenience stores, small supermarkets, and service stations in two New Zealand cities (Wellington, a large city and Dunedin, a provincial city). Retailers were purposively sampled, from a previous study database, and stratified by store type and area deprivation level. Retailers reported concern about the implications of standardized packaging, financially and on their ability to provide a fast and efficient customer service. However, few retailers believed standardized packaging would promote illicit trade of tobacco products and even fewer had heard of a black market for tobacco or thought one likely. If a black market was to occur, retailers thought continued price increases would be more likely than standardized packaging to lead to illicit trade.

This was the first study³⁹ to examine New Zealand retailers’ views of standardized packaging on the illicit tobacco trade using well-designed qualitative methods and collecting data until saturation. However, this was a comparatively small sample with many participants only speaking English as a second language and some who had sought support and advice from tobacco companies. Once again this study explored participants’ views of standardized packaging rather than their actual experiences of the policy and socially desirable responding may have impacted on the responses from participants. While two members of the research team reviewed interview transcripts in order to provide some validation of findings, no methods were reported aimed at minimising bias.

Retailing behaviour (actual)

Again, only one study explored how the introduction of standardized packaging legislation actually affected the retail of illicit tobacco. Scollo, Bayly and Wakefield³⁸ assessed the availability of illicit tobacco in small mixed business retail outlets before and after the introduction of standardized packaging in Australia. Retail outlets (n=303) were visited at baseline (June and September 2012) and following implementation of the legislation (December 2012, February 2013, April 2013 and July 2013). The store sample was derived from sixteen postcode-based areas in each of Melbourne and Sydney, and eight in each of Adelaide and Perth. Areas were stratified by socio-economic status (SES) and randomly selected from each SES quartile. Secret shoppers requested a particular low cost brand of cigarette and then pushed for an even cheaper brand. The cheapest pack of cigarettes offered was purchased. In 179 (59%) of the shops the secret shopper then asked the retailer about the availability of unbranded, i.e. illicit tobacco. At baseline 13 (2.2%) of 598 packs were either non-compliant with Australian legislation and/or purchased for a suspiciously low price. Four packs (1.3%) of 297 met either or both criteria in the implementation month (December 2012) and five (0.6%) of 878 did so in the three assessment months following implementation. Illicit tobacco was offered on two (0.6%) of 338 occasions before implementation, one (0.6%) of 170 occasions in the implementation month and on three (0.6%) of 514 occasions after implementation.

This study³⁸ was likely to reflect the experience of ordinary customers attempting to locate illicit tobacco from everyday small mixed business tobacco retailers. However, the study did not include sales of illicit tobacco from informal sources such as friends/family and informal sellers/venues. Detailed methods reported that fieldworkers were well trained, rotated across stores and demographically similar to known users of illicit tobacco. Nevertheless, it is possible that the retail stores approached by the secret shoppers may have been more likely to sell illicit tobacco to known regular customers rather than to strangers and the study might have failed to approach retail outlets known by users to sell large amounts of illicit tobacco. This study also benefitted from a large sample, random selection of areas and a low rate of store loss from the sample.

Tobacco industry behaviour

Finally, two studies^{34, 35} examined the nature and quality of industry data on illicit tobacco and the potential effect of standardized packaging. Rowell and colleagues³⁴ quantified the volume, nature and quality of UK newspaper articles citing tobacco industry data on illicit tobacco between March 2008 and March 2013. They found that stories citing industry data on illicit tobacco started in June 2011, two months after the publication of the Tobacco Control Plan for England (including the proposal for standardized packaging). Most of the data reported were based on industry empty-pack surveys for which no methodology was available. Almost all industry data suggested an increase in illicit cigarettes that was not supported by the independent data to which they were compared (HM Revenue & Customs figures and survey data from Pricing Policies and Control of Tobacco in Europe)³⁴.

In addition, Evans-Reeves and colleagues³⁵ carried out a content analysis of the four TTCs submission to the consultation on standardized packaging for relevance and quality of evidence that TTCs cited to support their arguments that “it will harm business and increase illicit trade”. They found 74 pieces of TTC cited evidence; 45% opinion (from TTC connected third parties), 25% research, 15% strategy documents and 15% facts and figures, all of which were of poor quality. Sixty three pieces (85%) of the evidence related to the illicit trade but the authors did not report results separately for this and the economic evidence. However, there was no independent, peer-reviewed evidence that supported either of their arguments and 47% of the evidence was industry connected. Much (66%) of the industry connected evidence was opinion only and all industry connected evidence was of significantly poorer quality than independent evidence (p<0.001).

Rowell et al.³⁴ were the first to analyse tobacco industry manipulation of data on the illicit cigarette trade, in the UK press, in response to the proposal for standardized packaging. Analysis of newspaper articles covered a 5-year period, although no other forms of media were considered. In addition the intent of the TTCs could not be measured and the independent data that were used for comparison of industry data were provided by one organisation only. Evans-Reeves and colleagues³⁵, analysis of the four TTCs submission to the consultation on standardized packaging failed to report results separately for evidence relating to economic and illicit trade consequences and they were not able to identify industry connections to three of the organisations who co-authored with other TTC linked organisations. Evans-Reeves and colleagues³⁵ also did not assess the accuracy of the TTCs’ interpretation of the evidence reported and only counted unique pieces of evidence (although several were reported by more than one company and often by all four).

Strengths and Limitations

One of this review’s limitations is that it considers a public health intervention, standardized tobacco packaging, that is in its early stages in only three countries and about which the first evaluations are only beginning to emerge. The majority of data on the illicit tobacco trade appears to have been collected outside of the peer review domain and as such are not available for review. Therefore, only ten peer-reviewed published studies were identified and these were of relatively weak study design and most were not of high quality. The lack of homogeneity between studies also resulted in a narrative synthesis of results rather than being able to combine data in a meta-analysis. The optimum time to assess the veracity of the tobacco industry’s main argument that standardized packaging will lead to an increase in illicit tobacco would be after a considerable period of time following implementation, and ideally would include studies from more than one country in order to reveal associations. However, this review was carried out at an important time of policy development in the UK and is important to provide timely refutes to arguments that are not built upon evidence and are designed to oppose effective public health interventions.