The impact of the introduction of tobacco product plain packaging on consumer responses in Ireland: a real-world policy evaluation stratified by socioeconomic groups

Abstract

Smoking prevalence remains high in Europe and widening socioeconomic group differences are driving health inequalities. While plain packaging policies disrupt tobacco industry tactics that sustain smoking, evidence of their equity impact is sparse. This study evaluated the implementation of plain packaging in Ireland in 2018 on consumer responses, overall and by the socioeconomic group. Consecutive nationally representative cross-sectional surveys (2018, n = 7701 and 2019, n = 7382) measured changes in 13 consumer responses among respondents who smoked across three domains: product appeal, health warnings effectiveness, and perceived harmfulness of smoking. Multiple logistic regression-derived adjusted odds ratios with 95% confidence intervals to compare responses post- versus pre-implementation adjusting for age, gender, educational level, and heaviness of smoking. A stratified analysis examined changes by socioeconomic group indexed using educational level. There were statistically significant changes in consumer responses to plain packaging policy implementation across 7/13 outcomes studied. Five changes were aligned with expected policy impacts (2/6 product appeal outcomes and 3/4 health warning effectiveness outcomes). Two responses were also observed which were not expected policy impacts (1 appeal-related and 1 perceived harm-related outcome). There was no change in five outcomes. Differences in consumer responses between educational groups were generally small, mixed in nature, and indistinguishable when interval estimates of effect were compared. Implementation of plain packaging in Ireland had intended impacts on consumer responses. Including plain packaging requirements in revising the European Union’s legislative frameworks for tobacco control will help build progress towards a Tobacco-Free Europe without exacerbating smoking inequalities.

Introduction

Smoking tobacco was responsible for almost eight million deaths and 200 million disability-adjusted life-years globally in 2019 [1]. While global progress is uneven, implementation of comprehensive tobacco control is reducing smoking prevalence [1, 2]. Large reductions in smoking prevalence have been observed in some European countries with stronger and more extensive tobacco control policies [3]. However, compared to other regions, smoking prevalence remains high in Europe. Furthermore, the slowdown in the decline of smoking in recent decades challenges the achievement of a Tobacco Free Europe by 2040 [4]. Widening smoking differences across socioeconomic groups are now a key challenge in Europe and a major driver of health inequalities [4, 5]. There is concern that population-level tobacco control measures may not reach people in lower socioeconomic groups [6, 7].

Tobacco industry marketing tactics have sustained the smoking-related harm epidemic [8, 9]. Packaging remained a key marketing opportunity for the tobacco industry and a compelling tobacco control policy target [10, 11]. The introduction of graphic health warnings (GHW) and, subsequently, standardized or plain packaging (PP), heralded the ‘death of a salesman’ [12]. By the end of 2022, 103 countries had adopted strong GHW and 22 countries had legislated for PP [13].

Implementation of GHW is a global tobacco control success [14, 15]. Commencing in Australia in 2012, PP implementation is at an earlier stage of implementation globally [16]. The Tobacco Products Directive enabled faster progress in Europe and PP policies are now in place in seven European Union member states, Norway, and the UK [17]. PP makes the initiation of tobacco product use less appealing for children and young people and impacts people who currently smoke [18]. Evidence from real-world implementation studies in countries including Australia, France, the UK, and Canada demonstrates that it reduces tobacco product appeal, increases GHW effectiveness, increases perceptions of the harmful effects of smoking, and ultimately reduces the likelihood of smoking by 10% [19–23]. In Europe, PP implementation has been linked with GHW strengthening under the Tobacco Products Directive and evidence suggests these policies act synergistically [24, 25].

Sharing real-world evidence of PP implementation helps overcome barriers to global progress and highlights learning [7, 26]. However, evidence regarding the equity impacts of GHW and PP was sparse [7].

Following a favourable evidence review for the government, in 2018, Ireland became the first European Union country to implement PP legislation [27]. This was preceded in 2016 by the transposition of the Tobacco Products Directive with regulations specifying minimum tobacco product pack size and requiring larger, combined text and graphic warning covering 65% of the surface [13, 17]. The stated objectives of the PP introduction were to decrease the appeal of tobacco products, increase the effectiveness of health warnings on tobacco packaging, and reduce the ability of the packaging of tobacco products to mislead consumers about smoking-related harm [28]. Smoking prevalence in Ireland among adults aged 15 years and older has reduced from 23% in 2015 to 18% in 2023 [29]. However, while smoking prevalence has declined among people in higher socioeconomic groups, it has stalled in lower socioeconomic groups [30]. The primary aim of this study was to measure the impact of PP on consumer responses among people who smoke in Ireland before (2018) and after (2019) implementation across the outcome domains of tobacco product appeal, health warnings effectiveness, and perceptions of the harmful effects of smoking. The secondary aim was to compare these impacts across socioeconomic groups indexed to the highest educational level.

Methods

The study was designed and is reported using the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement [31].

Study design

An uncontrolled before-and-after study was conducted through secondary analysis of consecutive representative cross-sectional population surveys to measure and compare changes in consumer responses across 13 PP outcomes pre- (2018) versus post-implementation (2019). Analysis was stratified to measure and compare consumer responses by socioeconomic group indexed to respondents’ highest educational level.

Setting

On 30th of September 2017, PP regulations were enforced for tobacco product manufacturers and retailers had one year (until 30th of September 2018) to sell-off branded stock, with oversight by regulatory inspections conducted by the Health Service Executive Environmental Health Service [28].

Data source

The Healthy Ireland Survey (HIS) is an annual population health survey commissioned by the Department of Health, which is administered face-to-face by IPSOS using consistent, quality-assured procedures [29]. A tobacco packaging module was added and repeated in HIS Wave 4 and 5. It was not included in other HIS waves. Data for the pre-implementation period were collected from September 2017 to June 2018 (Wave 4), which included the transition to PP. Data for the post-implementation period were collected from September 2018 to September 2019 (Wave 5 HIS), following full PP implementation.

Participants

Participants were those aged 15 years and older in Ireland [29]. A two-stage equal probability sample used the An Post/Ordnance Survey Ireland GeoDirectory of residential addresses as the primary data frame [29]. The module on tobacco packaging was administered only to respondents who reported smoking tobacco products. Overall, there were 7701 responses in HIS Wave 4 (September 2017 to June 2018, 62% response rate) and 7382 responses in HIS Wave 5 (September 2018 to September 2019, 61% response rate).

Variables

Participants who smoked only were administered items on self-reported exposure to plain-packaged products, policy approval, and PP outcomes. In total, 13 items across three PP outcome domains of tobacco product appeal (six items), GHW (four items), and perceived harm (three items) were adopted from the Australian PP evaluation of PP and responses were categorized using similar procedures (see Table 1) [19]. These outcomes aligned with evidence and objectives for PP policy in Ireland [27, 28].

Data for gender, age, cigarettes smoked per day, and highest level of educational attainment were also collected. Age was categorized into 15–29, 30–49, 50–69, and 70 years and older groups. The highest level of educational attainment was used to index socioeconomic position (categorized into primary level or less, secondary level, and tertiary levels for primary analysis, and then dichotomized as primary level or less versus secondary level or higher in the secondary analysis). Cigarettes smoked per day (categorized into ≤10, 11–20, 21–30, and 31+ groups) were used to measure the heaviness of smoking and nicotine dependence [32].

Statistical methods

The sociodemographic and smoking characteristics of respondents across HIS waves were compared with chi-square tests. Prevalence and unadjusted odds ratios (OR) with 95% confidence intervals (CI) were calculated to compare the likelihood of reporting various outcomes in consumer responses post-implementation versus pre-implementation, with significance assessed by chi-squared testing. After a review of assumptions [33], in the primary analysis logistic regression-derived adjusted ORs (aORs) for the likelihood of reporting 13 PP outcomes post-implementation versus pre-implementation with adjustment for covariates of gender, age, cigarettes smoked per day and highest level of educational attainment. In the secondary analysis, separate logistic regression analyses were conducted in higher (secondary or tertiary level) versus lower (primary level or less) educational attainment strata to derive aORs for the likelihood of 13 PP outcomes in consumer responses post-implementation versus pre-implementation stratified by educational attainment with adjustment for covariates of gender, age, and cigarettes smoked per day. Some younger respondents may not have reached an age where they could have completed secondary-level education so the sensitivity of the secondary analysis was tested by repeating it with the exclusion of respondents aged under 20 years (Supplementary Material).

The sample size was determined by usual HIS procedures [29]. HIS design and non-response weights were applied to data for all analyses [29]. Statistical analyses were conducted using JMP software. Alpha for statistical significance was set at 0.05 and a comparison of 95% CI was used to identify differences in outcomes across groups in the secondary analysis.

Data access and ethical considerations

HIS data were accessed through an application to the Department of Health in Ireland. Ethical approval was provided by the Royal College of Physicians of Ireland, Research Ethics Committee (RCPI RECSAF 170).

Results

Overview of respondents

At Wave 4 and Wave 5, current smoking was reported by 19.6% (n = 1513/7701) and 17.3% (n = 1279/7382) of respondents, respectively. Table 2 presents respondents’ sociodemographic and smoking characteristics, which were similar across both waves.

Changes in self-reported exposure to plain packaging and policy approval

Self-reported PP exposure in the last month increased from 24.1% pre-implementation to 65.9% post-implementation [OR 6.07 (95% CI 5.13–7.18), P < 0.01]. Policy approval also increased from 62.0% to 65.8% [OR 1.18 (95% CI 1.01–1.38), P < 0.04].

Changes in consumer responses to plain packaging

As illustrated in Table 3, statistically significant changes were identified in seven of 13 PP outcomes: three of six appeal domain outcomes, three of four GHW domain outcomes; and one of three perceived harm domain outcomes. Adjustment for covariates did not change these findings.

The greatest absolute and relative change was observed in reporting lower pack appeal than a year ago [27.89% pre-implementation versus 47.19% post-implementation; aOR 2.34 (95% CI 1.99, 2.76), P < 0.01]. Within the appeal domain, an increase in pack dislike was also observed. However, the belief that brands lack taste differentiation reduced [7.39% pre-implementation versus 3.95% post-implementation, aOR 0.51 (95% CI 0.36, 0.72), P < 0.01]. Consumer response changes were also observed in GHW domain outcomes. Change in one outcome was observed in the perceived harm domain. The likelihood of respondents reporting that their tobacco product was more harmful than a year ago reduced across the period [14.81% pre-implementation versus 11.48% post-implementation; aOR 0.74 (95% CI 0.59, 0.93), P < 0.02].

Changes in consumer responses to plain packaging by the highest educational level

As illustrated in Table 4, after adjustment for covariates, statistically significant changes in consumer responses were identified in 4 of 13 and 5 of 13 PP outcomes for respondents in lower versus higher educational groups, respectively.

In the appeal domain, increased likelihood of reporting lower pack appeal than 1 year ago was observed in both groups, and the changes observed were large and statistically significant. While the likelihood of reporting dislike of the pack also increased in both groups, the change was statistically significant only in the lower educational group [aOR 1.44 (95% CI 1.09–1.89), P = 0.01 versus aOR 1.13 (95% CI 0.93–1.37), P = 0.21 for lower and higher educational groups, respectively]. The likelihood of reporting a belief that brands do not differ in taste decreased in both groups, although the change was statistically significant for those in the higher education group only [aOR 0.62 (95% CI 0.36–1.05), P = 0.08 versus aOR 0.43 (95% CI 0.27, 0.70), P < 0.01 for lower and higher educational groups, respectively].

Regarding GHW responses, an increased likelihood of reporting noticing GHWs first when looking at the pack was observed in both groups, and the changes were statistically significant. Increased likelihood of attributing much more motivation to quit to GHWs and of reporting frequent pack concealment in the past month was observed in respondents in the higher educational group, and these changes were significant [aOR 1.66 (95% CI 1.19–2.31), P < 0.01 and aOR 1.33 (95% CI 1.00–1.77) P < 0.05]. Significant changes in these GHW responses were not observed in the lower educational group.

For the three perceptions of harm outcomes, in the higher educational group, no significant differences were observed between periods. Among those in the lower educational group, the likelihood of reporting higher harmfulness of tobacco products was reduced, and this difference was significant [aOR 0.53 (95% CI 0.35–0.81), P < 0.01].

The magnitude of point estimates of changes in PP outcome across the two educational groups were different and some statistically significant changes in one educational group were not observed in the other. However, interval estimates of the effect of PP implementation were indistinguishable between groups. These findings were not impacted by the sensitivity analysis which limited the educational groups to those aged 20 years and older (Supplementary Material).

Discussion

In this first study evaluating implementation of PP in Ireland, policy approval increased among people who smoke and there were significant changes in consumer responses across 7 of 13 outcomes. Five of these changes were expected outcomes of PP implementation in line with stated policy objectives (two of six appeal domain outcomes and three of four GHW domain outcomes) [27, 28]. In the appeal domain, a perception that brands lack taste differentiation was uncommon. However, this was reported less post-PP implementation, which was unexpected. In the perception of harm domain, the perception of higher product harmfulness than a year ago was also uncommon and it reduced after PP implementation. This was also unexpected given the stated policy objectives. While the observed pattern of short-term change across the 13 outcomes was substantially similar to that reported in other PP implementation evaluations [19], the null findings and unexpected changes may indicate that some proposed outcomes had greater proximity to the intervention than others under the PP implementation conditions that applied in Ireland. Importantly, socioeconomic group differences were also explored. There were significant changes in consumer responses in 4 of 13 and 5 of 13 PP outcomes for respondents in lower and higher educational groups, respectively (two of four and three of five, respectively, were aligned with stated policy objectives). While there were some differences between groups in point estimates of changes with the implementation of PP, these were generally small and mixed in nature, and interval estimates of changes indicate the effect of PP was indistinguishable between groups. This new finding concerning consumer responses to PP implementation across socioeconomic groups is relevant to policy planning.

The case for PP has been well-established, and this Irish study consolidates evidence of positive real-world impact [11, 12, 19–25]. Design of this study was adapted from Wakefield et al.’s [19] landmark evaluation of Australian adult smokers’ responses to the implementation of PP. That policy was implemented with a 2-month sell-off period for retailers following the imposition of manufacturing regulations and was linked with the intensification of GHW through new imagery and increasing pack coverage. In Ireland, there was a 1-year sell-off period, meaning a more gradual change in PP exposure to than in Australia. Furthermore, PP regulations were preceded by GHW strengthening through the transposition of the Tobacco Products Directive in May 2016, over 2 years before full PP implementation [29]. While the change in PP exposure was less abrupt in Ireland, findings were similar to those reported in Australia in the appeal-related domain, with both studies noting marked changes in reporting pack dislike and lower pack appeal than a year ago. Both studies found belief that brands lack taste differentiation was very uncommon before implementation (6.7% in Australia and 7.4% in Ireland). After PP implementation even fewer respondents reported this belief in Ireland, which was not an expected policy impact, while there was no change in Australia. Despite some differences in how GHW was implemented in conjunction with PP, both studies found impacts across GHW domain outcomes including warning salience, cognitive responses, and behavioural responses. Changes were more pronounced in Australia where GHWs were strengthened as part of the introduction of PP, rather than preceding PP, which was the case in Ireland. Changes in perceived harm outcomes were least pronounced in Australia, with a small change identified only in the perception that brands lack harmfulness differentiation. There was no change in that outcome in Ireland, albeit almost all respondents who smoked reported brands lack harmfulness differentiation. In Ireland, a lower likelihood of reporting that tobacco products were more harmful than a year ago was observed following PP implementation, especially among people in the lower educational group. This may suggest some wear-out of GHW effectiveness, despite strong GHW domain impacts. There was no change in this outcome in Australia, which may again be explained by differences in how PP introduction was linked with GHW strengthening. Although evidence suggests GHW salience and effectiveness increases with PP implementation [24, 25], which is consistent with the GHW domain findings in this study, this finding underscores the importance of warning rotation to maintain GHW effectiveness [34].

When changes in consumer responses to PP implementation in Ireland were analysed by respondents’ highest educational level, some differences in the estimates of effect size across the 13 outcomes analysed were observed. In the appeal domain, statistically significant changes were observed in reporting lower pack appeal than a year ago in both groups. However, a significant change in reporting pack dislike was only observed among respondents in the lower educational group. Respondents in the higher educational group reported significant changes in three of four outcomes in the GHW domain in line with intended policy impacts, including warning salience, cognitive response, and behavioural response, and no significant change in reporting a perception of higher product harmfulness than a year ago was observed. Respondents in the lower educational group reported a significant change in only one of four outcomes in the GHW domain (warning salience). However, this group also reported a reduced likelihood of perceiving higher product harmfulness than a year ago, and this difference in harm perceptions may relate to the difference in psychological reactance and may explain some observed differences in GHW domain outcomes across the groups [8, 27]. However, overall, differences between the educational groups were small, the pattern was mixed, and interval measures of effect were similar so no clear equity impact on the implementation of PP in Ireland can be concluded from this study.

To date, evidence on the equity impact of GHW and PP policy implementation has been limited [7]. Strengthening of GHW in Europe had a neutral equity impact, while in Australia and Canada GHW measures had a negative equity impact [35, 36]. Introduction of PP in Australia resulted in a stronger decline in tobacco product pack display in higher socioeconomic areas, while a greater decline in reported pack concealment was observed in lower socioeconomic areas [37]. Implementation of PP had a positive impact among Aboriginal and Torres Strait Islanders in Australia [38]. A more recent evaluation of PP implementation in England found smoking prevalence reduced, with no difference across socioeconomic groups [21]. Overall, in conjunction with this study, there is no strong evidence that PP has a negative equity impact, and its short-term impact is probably neutral. This is of importance to policy planning in countries, like those in Europe, where the reach of tobacco control interventions across population groups may be a concern [4, 6]. As real-world PP policy implementation extends in other countries, further evaluations should include equity impact assessment.

This study had some limitations. The time series comprised two consecutive years and it was not possible to evaluate longer-term trends in outcomes. Some pre-intervention fieldwork was conducted during transition to PP and in the post-intervention period some respondents reported a lack of exposure to PP. This may be explained by preceding pack changes and GHW strengthening interfering with the perception of PP exposure and by non-Irish duty-paid tobacco product use in Ireland [28]. These limitations increase the risk of type II error. Smoking prevalence reduced across the evaluation period raising the possibility that policy implementation may have motivated people who smoked with stronger consumer responses to PP to stop smoking. Since the module on PP consumer responses was only administered at each wave to those respondents who smoked, this may also have increased the risk of type II error. Data were limited to established proximal PP outcome measures and other important outcomes including increasing quitting attempts, preventing smoking relapses and reducing smoking initiation among children and young people were not examined [9, 18]. The use of educational attainment to index socioeconomic position has limitations, especially for younger respondents. However, this was explored through sensitivity analysis. There were relatively small numbers in the two educational groups examined in the stratified analysis leading to less precise measures of effect. Finally, while comparison of CI to identify differences between groups is a conservative method, crude differences between educational groups were generally small [39].

In conclusion, overall, this real-world evaluation of PP implementation in Ireland demonstrates that the policy led to strong and multiple intended changes in consumer responses [27, 28], especially in reducing pack appeal and increasing GHW effectiveness. It adds to the case for more widespread adoption of this game-changing tobacco control intervention. The findings are especially pertinent across the European Union where national-level PP policies are enabled, but not mandated, under the Tobacco Products Directive [17]. This step would help address the relative slowdown in the decline in smoking prevalence in the region [3]. Importantly, the study adds to the evidence that PP implementation is likely to have a neutral effect on socioeconomic inequalities in current tobacco use in the short term. This should help debunk rhetoric in countries with widening smoking inequalities where smoking inequalities may be misused to stall progress on PP policy. Future evaluations should extend this evidence by examining the equity impact of PP on the denormalization of tobacco use and initiation prevention among children and young people [9, 18]. As the European Beating Cancer Plan brings a welcome renewed focus to tobacco control in Europe by setting a goal of less than 5% smoking prevalence by 2040 [4], this study not only strengthens the argument for including PP as a requirement in future European Union legislative frameworks for tobacco control but also points to the importance now of shifting focus to new bold policies which can build a Tobacco-Free Europe that brings the harm caused by smoking to an end for everyone [40].

Acknowledgements

Permission to access the Healthy Ireland Survey Research Master Files was kindly provided by the Department of Health, Ireland.

Supplementary data

Supplementary data are available at EURPUB online.

Conflict of interest: None declared.

Funding

None declared.

Data availability

Healthy Ireland Survey Anonymised Master Files are available through the Irish Social Science Data Archive (ISSDA, https://www.ucd.ie/issda/data/healthyireland/). The data underlying this article used the Healthy Ireland Survey Research Master Files, which can be requested through application to the Department of Health, Ireland.

References