This study demonstrates that new graphic cigarette packet warnings coincided with increased awareness among smokers of the health consequences of smoking observed in cross-sectional surveys of South Australian smokers across four years. While it is possible that these increases in awareness of smoking related illnesses may have happened due to other influences or by chance, new graphic cigarette packet warnings are the most likely cause of the increases in awareness of smoking related disease.
Over the time that new graphic cigarette pack warnings were introduced, we observed substantial increases in top-of-mind awareness of diseases that were the subject of new warnings, and no increases in awareness of other health effects. Further supporting evidence was provided by the increased proportion of smokers who reported noticing warnings on cigarette packets after the new warnings were introduced. We also observed significant increases in smokers' unprompted recall of pack warnings as a source of anti-tobacco information. Again, this effect was isolated to pack warnings and not generalised to other sources such as television. After the new warnings were introduced, cigarette packets became second only to television as a recalled source of anti-tobacco messages for smokers. Arguably, noticing anti-tobacco messages on television could be at saturation point, after 20 years of regular anti-smoking campaigns. However, there was no increase in noticing messages on the less used media of radio or on the internet.
As observed in this study, it has been demonstrated previously that new messages delivered via television campaigns can markedly increase awareness of smoking related diseases in a 6-month period. As was the case in this study, the effects on awareness were specific to the diseases highlighted in the advertisements and not generalised to all smoking related illnesses .
Similar to the Canadian experience , cigarette packets became a prominent important source of anti-tobacco information, after graphic cigarette warnings were introduced, and they remained so in the 2 years after they were introduced. In this study there was some evidence of a spike of recall of new warnings with some short term attrition, followed by more steady results. Importantly, most of the data in this study are on unprompted recall, so it is to be expected that top-of-mind recall of warnings and associated health beliefs would be highest in the year that new warnings are introduced, and that it might subsequently be displaced from top-of-mind by more recent warnings. This would be consistent with Fishbein & Ajzen's  contention that people's salient beliefs about the consequences of any contemplated action do not exceed 5-9 in number. Warning and health effect recall does appear to stabilise but more longitudinal data are required to ascertain longer term effects.
The impact varied greatly between warnings. "Smoking causes heart disease" and "lung cancer" are warnings that have been on Australian cigarette packets for a long time (as text-based warnings). They are also diseases which a high proportion of smokers were already aware were caused by smoking, at baseline. Awareness of these diseases and recall of these pack warnings remained high but demonstrated no improvement once the new graphic warnings were introduced. "Smoking is addictive" was also a graphic adaptation of an old text-based warning. Like "heart disease" and "lung cancer" messages, no significant increase was observed in awareness of the relationship with smoking, or in recall of the warnings. However, unlike "heart disease" and "lung cancer", "addictive" stayed at a low level on both measures. At baseline, smokers already had a high awareness of the relationship between smoking and emphysema. The introduction of the completely new "emphysema" warning (with a familiar graphic - see Table 1), did increase recall of the warnings but did not shift the already high awareness of the disease among smokers. These four cases suggest that adding a graphic image (or at least these graphic images) to an old warning or an "old news" disease did nothing to improve awareness or recall.
By contrast, when baseline awareness of a disease/damage caused by smoking was low, and the disease/damage had not previously been used as a pack warning, awareness grew very significantly. Greatest growth in awareness was observed in relation to gangrene (4% to 27%) and mouth cancer (10% to 24%), both of which were "new news" and contained new images. Even topics that involved new warnings but images and messages which had been the subject of previous tobacco control campaigns (see Table 1) induced significant growth in awareness: "blocked arteries" increased 8%; "blindness" increased 9%; and "stroke" increased 8%. Hence, based on these examples, adding a new graphic image (or at least these graphic images) to a new warning would seem to improve awareness considerably, as does adding a familiar graphic image to a new warning.
Moreover, those warnings most recalled across the board and in different subgroups were those which were "new news", and used new images and particularly images of body parts likely to elicit a visceral "yuck" response. "Gangrene" (40% unprompted recall) and "mouth cancer" (32%) were dominant in this sense. Although "heart disease", "lung cancer" and "harms unborn babies" were also recalled well (49%, 31% and 31% respectively) and contained visceral or emotive images, the new packs failed to provoke an improvement in recall over baseline, suggesting their high recall cannot be attributed to the graphic imagery.
The warnings with weakest recall were "Tobacco smoke is toxic" (4%), "Smoking is addictive" (8%), "Quitting will improve your health" (6%), "Smoking - a leading cause of death" (11%). With the exception of "Smoking is addictive" all of these warnings are general rather than specific about the consequences of smoking and none of these warnings contain images of body parts. Anti-tobacco television campaigns have consistently demonstrated that images and messages eliciting a visceral response and messages that are novel or "new news" are more likely to be attended to and have impact on quitting behaviour [20, 22, 30]. This study demonstrates that these findings are generalisable to cigarette pack warnings. This study also demonstrates that these findings apply to smokers in general, as well as to different subgroups of smokers.
The addition of the Quitline number to the cigarette packet appears to have increased general top-of-mind awareness of the availability of the Quitline service. This is noteworthy because the Australian Quitline has been operating for over two decades and already enjoyed high levels of awareness. Although not significant, a coincident trend was observed in increased use of the Quitline as a source of help to quit. The proportion of smokers who knew the Quitline number doubled; and in 2007, one in eight smokers could recite the number accurately. An independent study demonstrated that calls to the Australian Quitline doubled in the year after the new warnings were introduced .
Health promotion often aims to segment different messages for different markets in the expectation of having greater impact. The case has been made, using mass-media quit campaigns as the example, that this is unnecessary and even counter-productive because it comes at a cost, namely the dilution of resources required for population-wide campaigns . Comparisons between warnings as well as comparisons between population sub-groups show that what "works", works well across the board and what "doesn't work" across the board, also doesn't work well with any subgroup. The only exception in this study was the greater propensity shown by women and younger smokers to respond to warnings about unborn babies and children, presumably because their closer specific personal relevance. Overall, this study provides another example of a population-based intervention working well with both smokers generally and within subgroups, building the case for non-segmented interventions.
This study provides clear evidence that Australia's new graphic cigarette packets succeeded in attracting the attention of Australian smokers. A limitation of this study is that it did not explicitly ask smokers what, about the different warnings, attracted their attention, nor did it ask smokers directly about their perceptions of the credibility of different warnings. Some warnings may have been better recalled than others because smokers thought they made outrageous and incredible claims. However, this study provides evidence that smokers did find the highly recalled warnings credible. The fact that smokers' unprompted recall of illnesses caused by smoking increased in line with the increased recall of warnings is evidence of this. Changes in awareness about the harms of smoking are an important antecedent to behaviour change for many smokers. Whether behaviour change did follow was not measured in the current study.
This study provides support for the Framework Convention on Tobacco Control Article 11, mandating large cigarette packet warnings and recommending graphic imagery. Tobacco control policies, such as the FCTC and Australia's National Tobacco Control Strategies, recognise the complexity of smoking behaviour and the multiple behavioural and structural interventions required to reduce tobacco's toll. Graphic cigarette packet warnings play a role as one component of a comprehensive suite of tobacco control interventions.