Fifty years since the United States Surgeon General’s first report on smoking and health, smoking prevalence has reduced globally,1 in part due to increased public awareness that smoking causes death and disease.2,3 However, it is possible that gaps in knowledge are contributing to health inequalities.4,5 In Australia, the prevalence of daily smoking has declined to just over 16% among adults but is higher in disadvantaged populations.6 Among the Aboriginal and Torres Strait Islander population, 42% of people aged 15 years or older smoked daily in 2012–2013.7 Understanding and tackling the causes of this disparity is a public health priority accepted by all Australian governments.8

Communicating information about the harmful effects of tobacco use is a major focus of programs to reduce smoking among Aboriginal and Torres Strait Islander peoples.9 Some evidence suggests that most Aboriginal and Torres Strait Islander people know that smoking causes lung cancer and heart disease,10–12 and that second-hand smoke (SHS) is dangerous.13–15 However, there is no current national research that describes knowledge of the harms of smoking and SHS exposure among Aboriginal and Torres Strait Islander smokers, or how it varies across this diverse population. Further, the extent to which lack of smoking-related knowledge contributes to the high smoking prevalence is unknown.

Greater knowledge and worry about future health effects of smoking have been shown to increase quit intentions and attempts in other settings.16–18 However, decisions to quit smoking are not one-dimensional, rational choices,19,20 and they may be obstructed by beliefs that diminish the likelihood or severity of smoking harms (risk minimisation).21,22 There has been some investigation into risk-minimising beliefs in Aboriginal and Torres Strait Islander tobacco control research. For example, perceived risk and worry may be low where there is discordance between information about the health consequences of smoking and the individual’s lived experience,14,23 or where there are fatalistic views of health effects that are perceived to be outside an individual’s control.12,24 This may explain why smoking persists in some contexts where knowledge of health effects is found to be high.

This is the first broadly representative description of smoking-related knowledge and health risk beliefs of Aboriginal peoples and Torres Strait Islanders. We also look at how this knowledge varies among smokers, and whether knowledge and health risk beliefs are related to quitting.

Methods

Survey design and participants

The Talking About The Smokes (TATS) project surveyed 2522 Aboriginal and Torres Strait Islander people (1643 current smokers, 311 ex-smokers and 568 never-smokers) from April 2012 to October 2013 (Wave 1, or baseline), and is described in detail elsewhere in this supplement.25,26 Briefly, we used a quota sampling design to recruit participants from communities served by 34 Aboriginal community-controlled health services (ACCHSs) and one community in the Torres Strait (project sites), which were selected based on the population distribution of Aboriginal and Torres Strait Islander people by state or territory and remoteness. In most sites (30/35), we aimed to interview samples of 50 smokers (or ex-smokers who had quit ≤ 12 months before) and 25 non-smokers (never-smokers and ex-smokers who had quit > 12 months previously), with equal numbers of men and women and those aged 18–34 years and 35 years or older. The sample sizes were doubled in four major urban sites and in the Torres Strait community. People were excluded if they were: not Indigenous, not aged 18 years or older, not usual residents of the area, staff members of the ACCHS, or unable to complete the survey in English (if there was no interpreter available), or if the quota for the relevant age–sex–smoking category had been filled.

In each site, different locally determined methods were used to collect a representative, albeit non-random, sample. The baseline sample closely matched the sample distribution of the 2008 National Aboriginal and Torres Strait Islander Social Survey (NATSISS) by age, sex, jurisdiction and remoteness, and also number of cigarettes smoked per day for current daily smokers. However, there were inconsistent differences in some socioeconomic indicators: our sample had higher proportions of unemployed people, but also higher proportions who had completed Year 12 and who lived in more advantaged areas.25

Interviews were conducted face to face by trained interviewers, almost all of whom were members of the local Aboriginal and Torres Strait Islander community. The survey, entered directly onto a computer tablet, generally took 30–60 minutes to complete. A single survey of health service activities, including whether there were dedicated tobacco control resources, was completed for each site. The project was approved by three Aboriginal human research ethics committees (HRECs) and two HRECs with Aboriginal subcommittees (Appendix 1).25

Survey questions

As the TATS project is part of the International Tobacco Control Policy Evaluation Project (ITC Project), survey questions were based on ITC Project surveys previously used in Australia and New Zealand (http://www.itcproject.org/surveys). The exact questions used for this article are listed in Appendix 2.

Knowledge and health risk beliefs

Four questions assessed knowledge of the direct health effects of smoking among smokers and non-smokers — whether it causes lung cancer, causes heart disease, makes diabetes worse and causes low birthweight (answer options: “yes”, “no” or “don’t know”). Three questions assessed knowledge of the effects of SHS exposure — whether it causes asthma in children (“yes”, “no” or “don’t know”) and whether it is dangerous to non-smokers and to children (both assessed on a five-point scale from “strongly agree” to “strongly disagree”). We also computed two summary items, for correct responses to all four direct effects questions (“yes” to all) and correct responses to all three SHS measures (“yes” or at least “agree”).

Two items assessed health risk beliefs among smokers. Smokers who responded “agree” or “strongly agree” to the statement that “Smoking is not very risky when you think about all the things that people do” (assessed on a five-point scale from “strongly agree” to “strongly disagree”) were assessed as holding risk-minimising beliefs. Those who responded “very worried” to the question “How worried are you that smoking will damage your health in the future?” (assessed on a four-point scale from “not at all worried” to “very worried”) were assessed as having health worry.

Wanting and attempting to quit

Two quit-related outcomes were used: wanting to quit (“yes” or “no”) and having attempted to quit in the past year (“yes” or “no”), which was derived from questions on ever having tried to quit and timing of the most recent quit attempt.

Statistical analyses

Percentages and frequencies were calculated for all knowledge and health risk belief questions. Logistic regression was used to assess: (i) variation in correct responses among smokers, by daily smoking status, key sociodemographic variables, and presence of tobacco control resources at the local health service; and (ii) the association of knowledge and health risk beliefs with quitting interest and activity among smokers. Stata 13 (StataCorp) survey [SVY] commands were used to adjust for the sampling design, identifying the 35 project sites as clusters, and the quotas based on age, sex and smoking status as strata.27 Both unadjusted and adjusted logistic regression analyses were performed, with daily smoking status and key sociodemographic variables included as covariates in the adjusted analyses. As unadjusted and adjusted calculations were very similar, only adjusted odds ratios (ORs) are reported here, with 95% confidence intervals.

Less than 1.5% of responses to each question were excluded (due to missing or refused responses), with the exception of quitting outcomes, which excluded a further 79 participants (4.8%) who did not know if they wanted to quit and 21 (1.3%) who did not know whether they had attempted to quit within the past year.

Results

Knowledge and health risk beliefs

Knowledge that smoking causes lung cancer and heart disease was high, and consistently over 90% of smokers and non-smokers knew about the harmful effects of SHS (Box 1). Knowledge that smoking makes diabetes worse was the lowest of all four direct effects, with 24% of daily smokers responding “don’t know” to this question (compared with 13% for low birthweight, 7% for heart disease and 3% for lung cancer). Among daily smokers, 44% held risk-minimising beliefs and 36% had health worry. Non-daily smokers had higher levels of risk-minimising beliefs and lower levels of health worry than did daily smokers.

Compared with daily smokers, non-daily smokers were more likely to respond correctly to all questions about the direct effects of smoking (OR, 1.79; 95% CI, 1.32–2.43; P < 0.001) and the harms of SHS (OR, 1.69; 95% CI, 1.08–2.62; P = 0.02) (Appendix 3).

There was some social patterning based on sociodemographic variables (Appendix 3). While knowledge of direct effects was significantly associated with employment and education, only area-level indicators were associated with both direct effects and SHS knowledge. Smokers were more likely to respond correctly to all questions if they were from a remote or very remote area (direct effects OR, 1.73; 95% CI, 1.16–2.57; SHS OR, 2.69; 95% CI, 1.61–4.52), compared with those from major cities, and smokers from an area of the highest level of disadvantage were more likely to respond correctly (direct effects OR, 1.83; 95% CI, 1.32–2.54; SHS OR, 1.33; 95% CI, 0.85–2.08) than were those from areas of least disadvantage.

Conversely, smokers from areas where the local health service had dedicated tobacco control staff or funding were less likely to respond correctly to all direct effects questions (OR, 0.64; 95% CI, 0.48–0.86) and all questions about the harms of SHS (OR, 0.58; 95% CI, 0.40–0.82), compared with those from areas where there were no dedicated resources (Appendix 3).

Relationship of knowledge and health risk beliefs with quitting

Smokers who responded correctly to all questions about harms of SHS were more likely to want to quit and to have attempted to quit in the past year, but those who responded correctly to questions about direct effects of smoking were not (Box 2). Similarly, smokers who responded correctly to all SHS knowledge questions were more likely to be very worried about their future health (OR, 4.74; 95% CI, 3.01–7.45; P < 0.001), but those with knowledge of all direct effects were not (Appendix 4). Those who were very worried about their health were more likely to want to quit and to have made a quit attempt in the past year (Box 2). Risk-minimising beliefs were not significantly associated with either wanting to quit or having attempted to quit in the past year.

Discussion

Our results show high levels of knowledge among Aboriginal and Torres Strait Islander people that smoking causes lung cancer and heart disease, along with strong awareness of the harms of SHS, consistent with previous tobacco control research in this population.10–14 Knowledge that smoking causes lung cancer and heart disease and is dangerous to others was assessed at very similar levels among Aboriginal and Torres Strait Islander daily smokers and those in the general population, based on comparable measures last assessed by Australian ITC Project surveys from 2002 to 2004.16,28

The main gap in knowledge, which has also been reported elsewhere,12 concerned the role of smoking in exacerbating diabetes. As Aboriginal and Torres Strait Islander people are more than three times as likely as non-Indigenous Australians to report a diagnosis of diabetes or high blood or urine sugar levels,7 with diabetes prevalence estimates ranging from 3.5% to 33.1%,29 this gap highlights the need for targeted education about the link between smoking and diabetes. This applies to clinicians as well as the broader Aboriginal and Torres Strait Islander population, particularly in light of updated evidence presented in the 2014 report of the US Surgeon General, which concludes that smoking increases the risk of developing type 2 diabetes in a clear dose–response manner.2

Our results also show a need to build knowledge that smoking causes low birthweight, which was either denied or not known by 18% of daily smokers, similar to previous findings.14,30 Messages that smoking causes lung cancer and heart disease and is dangerous to children have all featured on cigarette pack warning labels.31 Together with other sources of health information, such as mass media, news stories, local health promotion strategies and advice from health professionals, these are likely to have contributed to the high knowledge about these health effects among our Aboriginal and Torres Strait Islander participants.

Given health services are an important source of health information, it was surprising that knowledge was lower among smokers surveyed by sites with dedicated tobacco control resources. Though difficult to explain, this may be an indirect effect of the prioritisation of limited tobacco control resources to areas of greatest need, particularly as these resources included federally funded positions that had not long been established.9 Alternatively, it may suggest that information about the health effects of smoking is more effective when incorporated into established routine health service activities that include other areas of health and wellbeing.

Our findings suggest that gaps in knowledge are not responsible for the high prevalence of smoking or the social patterning of smoking among Aboriginal and Torres Strait Islander people. Contrary to the geographic and social patterning of smoking prevalence among Aboriginal and Torres Strait Islander people,7,32 we found that those from more remote and disadvantaged areas were more knowledgeable about the harmful effects of smoking and SHS. This is not to say that increasing knowledge is not important; prospective analyses from other ITC Project studies consistently show that knowledge, worry and risk beliefs contribute to motivation to quit.16,18,22,33 Though we have shown that knowledge is also related to interest in quitting among Aboriginal and Torres Strait Islander smokers, other factors are likely to be more important in influencing the success of quit attempts (and their translation to reduced prevalence), as found in other populations.17 For example, stress is commonly cited by Aboriginal and Torres Strait Islander smokers as a trigger for relapse,12,15,34,35 and it should be considered among other possible barriers including social normalisation of smoking, underlying social disadvantage, nicotine dependence and access to and uptake of services to support quitting.36

Among smokers, knowledge of SHS harms was associated with wanting to quit and attempts to quit, but knowing about direct, personal health consequences was not. Similarly, in an ITC Project survey in New Zealand, setting an example to children was more likely to be identified by Maori and Pacific peoples as a reason to quit, and was associated with SHS awareness and protective behaviour among smokers.37 Our findings are also consistent with qualitative research from the Northern Territory,15,24 in which Aboriginal participants expressed higher levels of concern for the health of others than for personal risk. Health is considered by many Aboriginal and Torres Strait Islander people to include the health of others.38 This may also explain why risk-minimising beliefs did not reduce interest in quitting, as predicted from research in the general population, despite being held at similar levels.21,22 It may be that these counterarguments are an ineffective shield to risks that include the health of others, and so have little or no effect on interest in quitting among Aboriginal and Torres Strait Islander people.

Our findings weaken the argument that risk-minimising beliefs explain why smoking persists in contexts where knowledge is high, and provide evidence that challenging these beliefs is unlikely to increase interest in quitting among Aboriginal and Torres Strait Islander people. Rather, health information may be interpreted with greater priority and relevance where negative health effects are framed in ways that include the health of others. This supports the approach used in the “Break the Chain” campaign, Australia’s first national Aboriginal and Torres Strait Islander antismoking campaign, launched in March 2011.39

Strengths and limitations

This is the first broadly representative survey of knowledge and health risk beliefs about smoking among Aboriginal and Torres Strait Islander peoples. The survey design made it feasible to interview a large number of people and to explore variation within our sample.

However, use of closed-ended questions may have led to overestimation of knowledge,40,41 which was assessed for a limited number of general health consequences of smoking. Knowledge may also have been overestimated if participants responded “yes” without fully scrutinising each question or because they did not want to appear uninformed. However, variation in the proportion of respondents who showed uncertainty in response to each item is evidence against this being systematic. Repeating the analyses with the “no” response as the dependent variable found the same general pattern of results (reversed). This increased our confidence in the validity of these outcomes, but did show that respondents from the most remote and disadvantaged areas were less likely to respond “don’t know”, consistent with biases to acquiesce or provide socially desirable responses in these areas. Some of the differences found, particularly area-level ones, may be due to social desirability biases, which are thought to be moderated by culture.42 Although face-to-face interviews can increase perceived pressure to provide socially acceptable responses, we attempted to reduce any such effects by engaging local interviewers, to minimise the social distance between the interviewer and participant.42

The questions used to assess health worry and risk minimisation showed good face validity, but have not been previously used to investigate these constructs with Aboriginal and Torres Strait Islander people. While these results paint a broad, representative picture of general health knowledge, concern and influence on quitting among Aboriginal and Torres Strait Islander people, more detailed assessments of knowledge may identify other gaps to target in future health information campaigns.

In conclusion, this national study found that lack of basic knowledge about the health consequences of smoking is not an important barrier to wanting and attempting to quit for Aboriginal and Torres Strait Islander smokers. Framing new messages in ways that encompass the health of others is likely to contribute to goal setting and prioritisation of quitting.

In 1988, the United States Surgeon General concluded that nicotine is the drug in tobacco that causes dependence on smoking.1 The nicotine that is delivered to the brain when smoking interacts with the habits and sensory stimuli associated with smoking to reinforce the behaviour.2 Genetic factors also influence the biological processes of nicotine delivery, metabolism and dependence.2

Clinicians and scientists have sought indicators to predict the success or failure of quit attempts, beyond indicators of motivation. The best such measure is the Heaviness of Smoking Index (HSI),3 or at least one of its two component items: cigarettes per day (CPD) and the time to first cigarette (TTFC) after waking.4,5 These two items are a subset of the six items in the Fagerström Test for Nicotine Dependence.6 There is also evidence that strong cravings (both before and after quitting) and shorter periods of abstinence on past attempts may independently predict failure of quit attempts.7–9 Identifying smokers who are most likely to have difficulty quitting is important in determining who might benefit from medications to assist cessation.

The age-standardised prevalence of smoking is 2.6 times higher among Aboriginal and Torres Strait Islander people as among other Australians.10 While both smoking prevalence and smoking intensity (based on self-reported CPD) are falling among the Aboriginal and Torres Strait Islander population, measures of dependence may differently predict which smokers will have the most difficulty quitting in this high-prevalence population where smoking is more normalised.10,11 Two small research reports have suggested that over-reliance on strategies that use stop-smoking medications may not be appropriate in this population, as nicotine dependence may be lower than in other populations.12,13 One of these studies found only low per capita consumption of cigarettes in remote Aboriginal communities,12 and the other found that only a small proportion of a sample of pregnant Aboriginal and Torres Strait Islander women who smoked were highly dependent.13

Here, we use a large national study of Aboriginal and Torres Strait Islander smokers to examine different indicators of dependence in this population and their association with sustained quit attempts, and to make comparisons with a national sample of Australian smokers.

Methods

The Talking About The Smokes (TATS) project surveyed 1392 Aboriginal and Torres Strait Islander daily smokers using a quota sampling design in the communities served by 34 Aboriginal community-controlled health services (ACCHSs) and one community in the Torres Strait, and has been described elsewhere.14,15 Briefly, the 35 sites were selected based on the distribution of the Aboriginal and Torres Strait Islander population by state or territory and remoteness. In 30 sites, we aimed to interview 50 smokers or ex-smokers who had quit ≤ 12 months before, and 25 non-smokers, with equal numbers of women and men and of those aged 18–34 and ≥ 35 years. In four major-city sites and the Torres Strait community, the sample sizes were doubled. People were excluded if they were aged less than 18 years, not usual residents of the area, staff of the ACCHS, or deemed unable to complete the survey. In each site, different locally determined methods were used to collect a representative, although not random, sample.

Baseline data were collected from April 2012 to October 2013. Interviews were conducted face to face by trained interviewers, almost all of whom were members of the local Aboriginal and Torres Strait Islander community. The survey was completed on a computer tablet and took 30–60 minutes. The baseline sample closely matched the national distribution of age, sex, jurisdiction, remoteness, quit attempts in the past year and number of daily cigarettes smoked reported in the 2008 National Aboriginal and Torres Strait Islander Social Survey (NATSISS). However, there were inconsistent differences in some socioeconomic indicators: our sample had higher proportions of unemployed people, but also higher proportions who had completed Year 12 and who lived in more advantaged areas.14

The TATS project is part of the International Tobacco Control Policy Evaluation Project (ITC Project) collaboration. Interview questions were closely based on those in ITC Project surveys, especially the Australian surveys.16 TATS project results were compared with those for 1010 daily smokers surveyed in Wave 8.5 of the Australian ITC Project between September 2011 and February 2012. That survey was completed by random digit telephone dialling or on the internet, and included smokers contacted for the first time and those who were recontacted after completing surveys in previous waves.

We asked questions about daily smokers’ usual smoking behaviour and variations in tobacco consumption, how easy it would be to not smoke, difficulties during their most recent quit attempt (eg, strong cravings, being around others who smoke), the duration of their longest quit attempt (to assess if any attempt had been sustained for at least 1 month) and sociodemographic factors. The questions are described in detail in Appendix 1.

The HSI was coded 0 to 6 based on the sum of the responses to the two questions about CPD and TTFC. These items were each coded as 0 (0–10 CPD; TTFC, ≥ 61 min), 1 (11–20 CPD; TTFC, 31–60 min), 2 (21–30 CPD; TTFC, 6–30 min) or 3 (≥ 31 CPD; TTFC, ≤ 5 min).3 We categorised HSI as low (0–1), moderate (2–3) or high (4–6).17,18 We also assessed the three criteria for dependence given in the Royal Australian College of General Practitioners (RACGP) cessation guidelines: TTFC ≤ 30 min, > 10 CPD, and withdrawal symptoms on previous quit attempts (defined in our sample as strong cravings during the most recent quit attempt).2

The project was approved by three Aboriginal human research ethics committees (HRECs) and two HRECs with Aboriginal subcommittees: Aboriginal Health & Medical Research Council Ethics Committee, Sydney; Aboriginal Health Research Ethics Committee, Adelaide; Central Australian HREC, Alice Springs; HREC for the Northern Territory Department of Health and Menzies School of Health Research, Darwin; and the Western Australian Aboriginal Health Ethics Committee, Perth.

Statistical analyses

We calculated the percentages and frequencies of responses to the TATS project questions, but did not include confidence intervals for these as it is not considered statistically acceptable to estimate sampling error in non-probabilistic samples. Therefore, we could not assess the statistical significance of differences with the Australian ITC Project results. The results for daily smokers in the Australian ITC Project were directly standardised to the distribution of age and sex of Aboriginal and Torres Strait Islander smokers reported in the 2008 NATSISS.

Within the TATS project sample, we assessed the association between sociodemographic variables and HSI using χ2 tests adjusted for the sampling design, using the 35 sites as clusters and the age–sex quotas as strata in Stata 13 (StataCorp) survey [SVY] commands.19 We assessed the association between indicators of dependence and sustained quit attempts using simple logistic regression, with confidence intervals adjusted for the sampling design and P values calculated for each variable using adjusted Wald tests.

Reported percentages and frequencies exclude those refusing to answer, answering “don’t know”, or for whom the question was not applicable (eg, questions about the most recent quit attempt excluded those who had not made an attempt in the past 5 years). Less than 2% of daily smokers answered “don’t know” or refused to answer each of the questions analysed here, except that 18 smokers (2.0%) answered “don’t know” to the question about difficulty in saying no when offered a cigarette during their most recent quit attempt, and 32 (2.3%) refused to answer the question about being unable to afford to buy cigarettes.

Results

There was little difference in the mean HSI scores for daily smokers in the TATS project compared with those in the Australian ITC Project (2.62 v 2.64; 95% CI, 2.45–2.83), but the TATS sample had fewer low and high scores and more moderate scores (Box 1). A higher proportion of smokers in the TATS project smoked 10 or fewer cigarettes per day (40% v 33.4%), but more also smoked their first cigarette within 30 minutes of waking (75% v 64.6%; 95% CI, 58.8%–70.0%). Lower proportions of Aboriginal and Torres Strait Islander smokers reported having strong urges to smoke at least several times a day (51% v 60.7%) or that it would be very hard to quit (39% v 47.9%).

Within the TATS sample, older smokers were more likely to have higher HSI scores, as were smokers who were not in the labour force, those with less education, those from both the most and least disadvantaged areas, and Aboriginal smokers compared with Torres Strait Islander smokers (Box 2).

Box 3 presents the results for questions that were only asked in the TATS project. Nearly half the smokers (47%) reported finding it very or extremely hard to go without smoking for a whole day, and most reported experiencing difficulties during their most recent quit attempt. A quarter (24%) of daily smokers had all three of the RACGP indicators of dependence.

Among the 61% of smokers in the TATS sample (833/1371) who had made a quit attempt in the past 5 years, all the indicators of dependence, except CPD and strong urges, were associated with being less likely to have made a sustained quit attempt of at least 1 month (Box 4). The indicators with the strongest negative associations with making a sustained quit attempt were the smokers’ assessments of how hard it would be to quit and their difficulties during the most recent quit attempt. Although the HSI and the RACGP criteria of dependence were negatively predictive of making a sustained quit attempt, CPD — one of their component measures — was not.

Nearly half the daily smokers in the TATS sample (45%, 606/1354) reported being unable to buy cigarettes for at least a few days in each fortnight before pay day, and 23% (314/1354) less often, while for 32% (435/1354) this was never a problem. When smokers were unable to buy them, 37% (342/916) reported they were often or very often given cigarettes, and 50% (460/916) were sometimes given them. As a result, 27% (245/911) said they smoked the same amount as usual when unable to buy cigarettes, while 50% (456/911) smoked a bit less and only 23% (210/911) smoked a lot less or not at all.

Compared with Australian smokers in the ITC Project, fewer Aboriginal and Torres Strait Islander smokers in the TATS project reported that the amount they smoked varied from day to day (42% [580/1392] v 58.5% [95% CI, 53.1%–63.7%]), but more reported that spending money on cigarettes left them with insufficient money for food or other essentials (23% [321/1378] v 12.9% [95% CI, 8.7%–18.6%]).

The Aboriginal and Torres Strait Islander smokers whose smoking led to insufficient money for essentials were less likely to have made sustained attempts to quit (odds ratio [OR], 0.70; 95% CI, 0.37–0.71; P < 0.001). Smokers who were never unable to afford cigarettes were less likely to have made a sustained quit attempt than those who were sometimes unable to buy them (OR, 0.51; 95% CI, 0.37–0.71; P < 0.001). Those who said they smoked about the same as usual when they were unable to buy cigarettes were also less likely to have made a sustained quit attempt, compared with those who at such times smoked a lot less or not at all (OR, 0.61; 95% CI, 0.41–0.91; P = 0.01).

Discussion

We found mixed relationships between indicators for dependence and sustained quit attempts in our sample of Aboriginal and Torres Strait Islander smokers. Based on CPD, frequency of strong urges to smoke and perceptions of how hard it would be to quit, dependence in this population appeared lower than among all Australian smokers. In contrast, our sample had a shorter TTFC. Nevertheless, the associations we found between dependence, as measured by the HSI, and being older and socially disadvantaged were similar to those in previous cross-sectional Australian ITC Project research.20

Previous research suggests TTFC is a more useful measure of dependence and a better predictor of successful quitting than CPD, although both are predictive and may contribute independently.4,5,21,22 Consistent with this, we found that longer TTFC was associated with having made a sustained quit attempt, while CPD was not. However, we also found no association for the frequency of strong urges while still smoking, which has been shown to be associated with successful quitting in longitudinal research, performing better than the Fagerström Test for Nicotine Dependence or its components, HSI, TTFC or CPD.7,8 These findings question the utility of existing indicators of dependence to predict successful quitting in Aboriginal and Torres Strait Islander smokers.

Aboriginal and Torres Strait Islander smokers’ perceptions of greater ease in quitting (quitting self-efficacy) may be falsely optimistic, perhaps reflecting less experience of unsuccessful quit attempts.23 In 2012–2013, only 37% of Aboriginal and Torres Strait Islander adults who had ever smoked had successfully quit, compared with 63% of other Australians.10 Some of the cross-sectional association we found between quitting self-efficacy and sustained quit attempts is likely to be in the reverse direction, with those who have not been able to sustain quit attempts understandably reporting that quitting will be harder. However, in other longitudinal research of the ITC Project, quitting self-efficacy has been associated with preventing relapses, both before and after a month.7 Nevertheless, we can take advantage of this optimism to encourage quit attempts.

Most Aboriginal and Torres Strait Islander smokers reported withdrawal symptoms (cravings) and situational difficulties during their most recent quit attempt, which have been described in more detail in previous qualitative research.24 It is notable that questions about the most recent quit attempt were consistently stronger predictors of being unable to sustain quit attempts than were traditional measures of dependence based on typical daily smoking patterns. Our results are consistent with more detailed recent research in other settings, which suggested that the components of the HSI are only predictive of early relapses in the first weeks of a quit attempt, whereas cravings and situational cues (such as the number of close friends who smoke) are important after 1 month.7,25

Current clinical guidelines recommend that clinicians ask smokers not only about CPD and TTFC, but also about their past unsuccessful quit attempts.26,27 Beyond emphasising the utility of the existing question about difficulties experienced during past attempts, we recommend waiting for further research on how the different measures prospectively predict quitting success before suggesting changes to the guidelines for Aboriginal and Torres Strait Islander smokers.

It is possible that estimates of CPD might be less accurate among Aboriginal and Torres Strait Islander smokers, where the relationship between purchase and consumption is more complicated because sharing and being unable to buy cigarettes are common. Two small studies of Aboriginal and Torres Strait Islander people showed that self-reported CPD is associated with urinary cotinine levels, but did not discuss whether the association was similar to that in other populations.28,29 However, we found that Aboriginal and Torres Strait Islander smokers were less likely than all Australian smokers to report variation in the number of cigarettes smoked each day, so it is difficult to suggest that such day-to-day variations are the reason for CPD being less useful in this setting. Those who managed to maintain usual consumption levels when they were unable to buy cigarettes were less likely to have sustained a quit attempt than those who smoked less at these times. Sharing of cigarettes therefore seems to increase in response to the inability to buy cigarettes among more dependent smokers, as has been reported elsewhere in response to pay cycles and the increased cost of cigarettes after tobacco excise rises.24,30

Strengths and limitations

The main strength of our study is its large national sample of Aboriginal and Torres Strait Islander smokers, providing detailed information about dependence directly from a population with a high prevalence of smoking. However, it is a non-random, albeit broadly representative, sample and caution is needed in making comparisons with the Australian ITC Project sample.

The cross-sectional associations we found warrant confirmation from future longitudinal analyses. There may have been some reverse causation, with past experiences of sustaining or not sustaining quit attempts influencing answers to the questions about dependence. Further, sustained attempts may have occurred years earlier, and the smokers’ dependence may have since changed. The use of past sustained quit attempts as an outcome necessarily meant excluding those who had not made any attempts. Predicting future quitting in this subgroup will be important but cannot include measures based on non-existent past attempts.

Our self-reported data are probably limited by incomplete recall of past quit attempts, and both forgetting and misremembering of symptoms. The effect of most of these biases will be to weaken reported associations, leading to greater confidence in the significant associations but requiring caution in the implications of findings of no association. For example, the lack of association of strong urges to smoke with sustained quitting found here, in contrast to other research, requires further exploration.8 More Aboriginal smokers than other Australian smokers use roll-your-own cigarettes, which may have caused greater misclassification bias of estimates of CPD.31 Future longitudinal analyses of the predictive association of these dependence measures with relapses and successful quitting should also control for the moderating effect of stop-smoking medication, which we were not able to do.25