Research Research Association between tobacco plain packaging and Quitline calls: a population- based, interrupted time-series analysis

Jane M Young nternationally, Australia is a leader MPH, PhD, FAFPHM, Abstract Professor in Cancer in tobacco control policies that Epidemiology1 Objectives: To investigate whether the introduction of tobacco plain packaging Ireduce community exposure to in Australia from 1 October 2012 was associated with a change in the number of Ingrid Stacey BSc(Hons), tobacco-related harm. In December calls to the smoking cessation helpline, Quitline, and to compare this with the Biostatistical Officer2 2012, Australia became the first coun- impact of the introduction of graphic health warnings from 1 March 2006. Timothy A Dobbins try in the world to enact legislation Design and setting: Whole-of-population interrupted time-series analysis in BMath, PhD, New South Wales and the Australian Capital Territory between 1 March 2005 Senior Lecturer1 mandating plain packaging for all tobacco products.1 The Tobacco Plain and October 2006 for the comparator, graphic health warnings, and October Sally Dunlop 2011 and April 2013 for the intervention of interest, tobacco plain packaging. PhD, Packaging Act 2011 (Cwlth) is designed Research Associate2 to prevent tobacco industry promotion Main outcome measure: Weekly number of calls to the Quitline, after adjusting Anita L Dessaix for seasonal trends, anti-tobacco advertising, cigarette costliness and the MPH, by simultaneously reducing pack number of smokers in the community. Manager, attractiveness and increasing the size Cancer Prevention2 Results: There was a 78% increase in the number of calls to the Quitline of graphic health warnings. The legis- David C Currow associated with the introduction of plain packaging (baseline, 363/week; peak, BMed, MPH, FRACP, lation required manufacturers to pro- 651/week [95% CI, 523–780/week; P < 0.001]). This peak occurred 4 weeks 2 Chief Executive duce plain packs with new warnings after the initial appearance of plain packaging and has been prolonged. The from 1 October 2012. From 1 Decem- 2006 introduction of graphic health warnings had the same relative increase in 1 Cancer Epidemiology and ber 2012, plain packaging became calls (84%; baseline, 910/week; peak, 1673/week [95% CI, 1383–1963/week; Services Research (CESR), P < 0.001]) but the impact of plain packaging has continued for longer. Sydney School of Public compulsory for all tobacco products. Health, University The new plain packs are olive green Conclusions: There has been a sustained increase in calls to the Quitline after of Sydney, the introduction of tobacco plain packaging. This increase is not attributable to Sydney, NSW. and devoid of brand design. Tele- anti-tobacco advertising activity, cigarette price increases nor other identifiable 2 Cancer Institute NSW, phone numbers for the national causes. This is an important incremental step in comprehensive tobacco Sydney, NSW. smoking cessation helpline, Quitline, control. jane.young@ feature prominently on the packs. sydney.edu.au Plain packaging legislation exists to introduction of the Tobacco Plain date the start of the intervention. Sim- encourage smokers to quit and dis- Packaging Act, complementing a ilarly, 1 March 2006 was the date of MJA 2014; 200: 29–32 courage the uptake of smoking.1 Quit- recent report of smokers’ feedback.7 the introduction of graphic health doi: 10.5694/mja13.11070 line is a free resource that can be used We did this by investigating the warnings on cigarette packaging. We by smokers who are motivated and impact of the introduction of tobacco looked at Quitline call numbers seeking support to quit. Therefore, the plain packaging on Quitline calls. To before and after these dates for the volume of calls to cessation helplines provide context, we compared the two interventions. Call data from 1 has frequently been used as one indi- impact on Quitline calls of the intro- April 2004 to 28 February 2006 were cator of changes in interest in quitting duction of tobacco plain packaging provided by Macquarie Telecom (Syd- in response to population-wide cessa- with the nationwide introduction of ney, Australia) and from 1 March 2006 tion policies and programs.2-6 graphic health warnings on cigarette to 31 March 2013 by the Telstra Ana- The best level of evidence for evalu- packaging in 2006.8 The null hypothe- lyser (Telstra, Melbourne, Australia). ating a whole-of-population initiative sis was that there would be no change This study did not require institu- such as tobacco plain packaging is an in call numbers, adjusting for known tional ethics approval as it did not interrupted time-series analysis. In confounders. involve data about individuals. The exactly the same way in which obser- study received no external funding. vational studies have been used to Reporting of the study complies with The Medical Journaldefine of the Australia association ISSN: 0025-between lung Methods the STROBE (STrengthening the 729X 20 Januarycancer 2014 200and 1 29-32tobacco (because ran- Reporting of OBservational studies in ©The Medical Journal of Australia 2014 domised studies are unethical in such We used an interrupted time-series Epidemiology) consensus guidelines www.mja.com.au analysis to investigate trends in the a context), before-and-after evalua- for reporting observational studies.9 Research tion, controlling for secular trends, is weekly volume of calls from New the optimal design for assessing the South Wales and the Australian Cap- effects of population-wide initiatives. ital Territory to the Quitline. Potential confounders In this study, we sought to examine As plain packs were phased in from An increase in anti-smoking advertis- Online first 13/01/14 behavioural change resulting from the 1 October 2012, we considered this ing in mass media such as television is

MJA 200 (1) · 20 January 2014 29 Research

1 Autoregressive integrated moving average (ARIMA) modelling results for the plain packaging and graphic health warning interventions

Peak call volume attributable to Number of calls in intervention (calls/week) Time Exponential decay week intervention to Duration rate of call volume Intervention was introduced Estimate (95% CI) Relative increase peak of effect following peak Plain packaging 363 651 (523–780) 78% 4 weeks 43 weeks 14% per week Graphic health warnings 910 1673 (1383–1963) 84% 12 weeks 20 weeks 60% per week

2 Weekly calls to Quitline, target audience rating points (TARPs) and cigarette Another potential confounder is price relative to income, before and after the introduction of plain tobacco cigarette costliness. We followed packaging and graphic health warnings Wakefield et al10 and calculated ciga- Plain tobacco packaging rette costliness as the ratio of the average quarterly recommended retail Calls to Quitline price for a pack of the two top-selling TARPs 2.0 800 Cigarette price Australian cigarette brands (obtained from the retail trade magazine Aus- tralian Retail Tobacconist, volumes 65 600 1.5 to 87) to the average weekly earnings

Plain packaging in the same quarter obtained from the Australian Bureau of Statistics.11 1.0 400 Finally, the number of smokers in the community can be a potential confounder. We obtained data on 200 0.5 smoking prevalence during the study periods from Health Statistics New Cigarette price relative to income (%) Total weekly calls to Quitline/TARPs 12 0 0 South Wales, and this was applied to Oct Apr Oct Apr quarterly figures for population size 2011 2012 2012 2013 from the Australian Bureau of Statistics13 to calculate the number of smokers in the community. Data for Graphic health warnings the first quarter of 2012 were not available, so we used results carried 2.0 forward from the final quarter of 2011. 3000 Statistical analyses 1.5 As the data for weekly number of calls 2000 to the Quitline were autocorrelated (each value was correlated with the 1.0 previous value) we used autoregres- Graphic health warnings sive integrated moving average 1000 (ARIMA) analysis in SAS version 9.3 0.5 (SAS Institute Inc). ARIMA models enable the investigation of changes Cigarette price relative to income (%) Total weekly calls to Quitline/TARPs 0 0 over time while accounting for sea- Apr Oct Apr Oct sonal variation and background 2005 2005 2006 2006 trends in such things as the effects of television anti-tobacco advertising, changes in cigarette pricing relative to a potential confounder as it increases tute NSW). TARPs are a product of weekly earnings and number of the number of calls to the Quitline.2,3,6 the percentage of the target audience smokers in the community. In ARIMA We ascertained weekly target audi- exposed to an advertisement (reach) modelling, comprising model investi- ence rating points (TARPs) for adver- and the average number of times a gation, estimation and diagnostic tisements broadcast in NSW and the target audience member would be checking, we followed the methods of ACT during the periods of interest, exposed (frequency), adjusted for the Box et al (Appendix; online at using OzTAM (Australian television length of the advertisement. For mja.com.au).14 audience measurements) for adults example, 200 TARPs might represent A single model fitted to the entire aged 18 years and older for free-to-air 100% of the target audience receiving 7-year period of Quitline call data did and cable television using established the message twice on average over a not meet technical criteria for model methods (unpublished report pre- specified period, or 50% reached four fit. Therefore, separate models that pared by OzTAM for the Cancer Insti- times. included data for 12 months before

30 MJA 200 (1) · 20 January 2014 Research

3 Change in numbers of call to Quitline after the introduction of plain packaging and graphic health warnings, adjusted for anti-tobacco advertising, seasonal trend, smoking prevalence and tobacco costliness*

Number of Quitline calls

Graphic health warnings Plain packaging

Parameter Estimate (SE) P Estimate (SE) P Intervention 12-week time to peak 4-week time to peak Peak number of additional calls 762.88 (147.88) < 0.001 288.18 (65.48) < 0.001 Weekly decay in number of additional calls 0.60 (0.24) 0.095 0.14 (0.10) < 0.001 Number of calls to Quitline attributable to confounders Target audience rating points (TARPs) 1.47 (0.20) < 0.001 0.08 (0.11) 0.47 New Year period 199.38 (169.49) 0.24 127.18 (43.78) 0.004

Smoking prevalence (per 100 000 smokers)  275.26 (230.15) 0.23 90.54 (595.37) 0.88 Cigarette costliness (per 1% increase relative to average earnings) 428.98 (546.25) 0.43

* Full model details can be supplied by the authors on request. ◆

and 6 months after each intervention Based on the estimated regression aging had a more immediate effect, a (1 March 2005 to 1 September 2006 parameters from the ARIMA models slightly smaller peak and a much and 1 October 2011 to 1 April 2013) (Box 3), the plain packaging interven- longer duration of impact. were fitted, as this was the longest tion resulted in an increase of 288 calls These changes need to be inter- duration of follow-up for tobacco (95% CI, 160–517 calls) to the NSW preted in the context of the earlier plain packaging available at the time Quitline, representing a 78% increase introduction of graphic health warn- of the study. from the number of calls in the week ings and a raft of other legislated The same modelling approach was the plain packaging legislation was tobacco control policies, and secular used for fitting models to both data introduced (Box 1). This effect was trends in smoking rates. It is possible subsets. Indicator terms were created sustained in subsequent weeks, with that the magnitude and duration of to represent the week of the intro- an estimated 86% of extra calls effect could be different in countries duction of the intervention (plain retained relative to the previous week that are implementing aspects of the packaging or graphic health warn- (Box 1). As a comparison, the intro- Framework Convention on Tobacco ings). A seasonal New Year’s Eve duction of graphic health warnings Control in different ways or different term was created to allow for an increased the number of calls by 763 orders.15 For example, in NSW, the increase in calls around the New Year (95% CI, 473–1053), an estimated most recently legislated changes period. The potential confounders of increase of 84% relative to the include the widespread introduction seasonal variation (New Years’ number of calls in the week graphic of smoke-free public spaces from peaks), anti-smoking advertising health warnings legislation was 2007,16,17 and banning the display of activity (TARPs) and number of implemented (Box 1). The effect of tobacco products at the point of sale smokers in the population were graphic health warnings diminished in 2010.18,19 Furthermore, smoking included in the models and were more quickly, with each post-peak prevalence has declined markedly retained regardless of statistical sig- week retaining only 40% of the previ- from 17.7% of NSW residents report- nificance on the basis of face validity. ous week’s extra calls (Box 1). ing daily or occasional smoking in Cigarette price was included in the 2006 to 14.7% in 2011.12 plain packaging analysis, but was A key strength of this study was the omitted from the graphic health Discussion ability to account for other known warnings analysis as there was mini- influences on quitting behaviour and We found a significant increase in the mal change in price during the use of the Quitline: number of calls to Quitline coincid- observation period. • anti-tobacco advertising activity; ing with the introduction of manda- • the costliness of tobacco; tory plain packaging of tobacco after • the number of smokers in the Results other known confounders had been community; and taken into account. Australia has • seasonal peaks that are observed Overall, there was a general decrease taken a lead on mandating plain in the New Year period. in the number of calls each week to packaging, now supported by evi- Although the volume of calls to the the Quitline over the period of obser- dence of an immediate impact of this Quitline is an indirect measure of vation. There were 910 calls in the legislation. This should encourage quitting intentions and behaviour, it week before the introduction of other countries that are preparing does provide an objective outcome graphic health warnings in 2005 similar legislation. measure that is not subject to selec- compared with 363 calls in the week We also found that, compared with tion or social desirability biases that before the introduction of plain the introduction of an earlier smoking may occur in community surveys. tobacco packaging in 2012 (Box 1, cessation initiative — graphic health Quitline data are available almost in Box 2). warnings — the effect of plain pack- real time so that the immediate

MJA 200 (1) · 20 January 2014 31 Research

impact of community-wide programs acceptability of smoking as well as 9 von Elm E, Altman DG, Egger M, et al; STROBE 21 Initiative. Strengthening the Reporting of or policies can be evaluated soon after smoking prevalence. Observational Studies in Epidemiology implementation. Interrupted time- In conclusion, we showed a signifi- (STROBE) statement: guidelines for reporting series analysis provides one of the cant increase in the number of calls to observational studies. BMJ 2007; 335: 806-808. 10 Wakefield MA, Durkin S, Spittal MJ, et al. Impact most robust methods of evaluating the Quitline that coincided with the of tobacco control policies and mass media the impact of programs and policies introduction of plain packaging campaigns on monthly adult smoking that affect the whole population.20 We nationally in Australia. Future evalua- prevalence. Am J Public Health 2008; 98: 1443-1450. examined legislation enacted nation- tions of other smoking-related out- 11 Australian Bureau of Statistics. Average weekly ally simultaneously, precluding the comes need to include the impact on earnings, Australia, Nov 2012. (ABS Cat. No. use of a control group. smoking prevalence. 6302.0.) http://abs.gov.au/ausstats/[email protected]/ mf/6302.0 (accessed Oct 2013). A weakness of our study is its Acknowledgements: We thank Donna Perez and James 12 Centre for Epidemiology and Evidence. Health inability to differentiate the impact of Kite for help with obtaining the data. This study was statistics New South Wales. Sydney: NSW internally funded by the Cancer Institute NSW. Thanks the increased size of graphic health Ministry of Health. http://www.healthstats.nsw. also to Bruce Armstrong, Sanchia Aranda, and Rebecca gov.au (accessed Oct 2013). Kenyon for their helpful comments on the manuscript. warnings that happened simultane- 13 Australian Bureau of Statistics. Australian ously with tobacco plain packaging Competing interests: No relevant disclosures. Demographic Statistics, Dec 2012. (ABS Cat. No. from the impact of plain packaging Received 16 Aug 2013, accepted 22 Oct 2013. 3101.0.) http://www.abs.gov.au/ausstats/ [email protected]/mf/3101.0 (accessed Oct 2013). itself. Quitline calls are one measure 1 Australian Commonwealth Government. Tobacco Plain Packaging Act 2011 (Cwlth) (No. 148, 2011). 14 Box GEP, Jenkins GM, Reinsel GC. Time series of changes in people’s behaviour in http://www.comlaw.gov.au/Details/ analysis: forecasting and control. 4th ed. response to legislative change. The C2011A00148 (accessed Oct 2013). Hoboken, NJ: John Wiley & Sons, 2008. 15 World Health Organization. WHO Framework 2 Pierce JP, Anderson DM, Romano RM, et al. rate of calls to Quitline may have been Convention on Tobacco Control. Geneva: WHO, Promoting smoking cessation in the United 2003. http://whqlibdoc.who.int/publications/ confounded by the Quitline tele- States: effect of public service announcements 2003/9241591013.pdf (accessed Oct 2013). phone number appearing to be more on the Cancer Information Service telephone line. J Natl Cancer Inst 1992; 84: 677-683. 16 New South Wales Government. Tobacco prominent in the absence of proprie- Legislation Amendment Act 2012 (NSW) 3 Siahpush M, Wakefield M, Spittal M, Durkin S. tary branding. Also, our study has No 56. http://www.legislation.nsw.gov.au/ Antismoking television advertising and shown an association but cannot sessionalview/sessional/act/2012-56.pdf socioeconomic variations in calls to Quitline. (accessed Oct 2013). J Epidemiol Community Health 2007; 61: 298-301. prove causation. 17 Australian Capital Territory Government. Smoke- Longer-term investigation of the 4 Chan SS, Wong DC, Fong DY, et al. Short-term Free Public Places Act 2003 (ACT). http://www.l impact of new smoke-free legislation on the impact of the legislation should egislation.act.gov.au/a/2003-51/current/pdf/ utilization of a quitline in Hong Kong. Nicotine Tob 2003-51.pdf (accessed Oct 2013). Res 2009; 11: 356-361. include assessment of more direct 18 New South Wales Government. Tobacco retailing measures: 5 Miller CL, Wakefield M, Roberts L. Uptake and laws. Public Health (Tobacco) Act 2008 (NSW). effectiveness of the Australian telephone • smoking uptake in non-smokers; http://www0.health.nsw.gov.au/publichealth/ Quitline service in the context of a mass media healthpromotion/tobacco/information_for_ • prevention of relapse among ex- campaign. Tob Con tro l 2003; 12 Suppl 2: ii53-ii58. tobacco_retailers_and_consumers.asp (accessed smokers; and 6 Farrelly MC, Davis KC, Nonnemaker JM, et al. Oct 2013). • cessation of smoking among cur- Promoting calls to a quitline: quantifying the 19 Australian Capital Territory Government. Tobacco rent smokers. influence of message theme, strong negative Amendment Act 2008. (repealed) http://www. emotions and graphic images in television legislation.act.gov.au/a/2008-50/default.asp This would include the need to advertisements. Tob Con trol 2011; 20: 279-284. (accessed Oct 2013). investigate variations in impact on 7 Wakefield MA, Hayes L, Durkin S, Borland R. 20 Lagarde M. How to do (or not to do) . . . assessing subgroups of the population. A com- Introduction effects of the Australian plain the impact of a policy change with routine packaging policy on adult smokers: a cross- longitudinal data. Health Policy Plan 2012; 27: prehensive evaluation of the policy sectional study. BMJ Open 2013; 3: e003175. 76-83. will require ongoing monitoring to 8 Australian Commonwealth Government. Trade 21 Hammond D, Wakefield M, Durkin S, Brennan E. assess factors such as changes in Practices (Consumer Product Information Tobacco packaging and mass media campaigns: Standards) (Tobacco) Regulations 2004. http:// research needs for Articles 11 and 12 of the WHO brand recognition, awareness of the www.comlaw.gov.au/Details/F2007C00131 Framework Convention on Tobacco Control. health risks of smoking, the social (accessed Oct 2013). Nicotine Tob Res 2013; 15: 817-831. ❏

32 MJA 200 (1) · 20 January 2014