Who smokes in Australia? Cross‐sectional analysis of Australian Bureau of Statistics survey data, 2017–19

To assess the socio‐demographic and health‐related characteristics of people who smoke daily, people who formerly smoked, and people who have never smoked in Australia.


Methods
The team who undertook the cross-sectional analysis described in this article included investigators with lived Indigenous experience and experience in tobacco research, epidemiology, and public health.Our approach took Indigenous world views and relationality into account, and social justice principles, including the right of all Australians to good health, 15,16 have informed the project from its conceptualisation to dissemination of results, with the aim of better informing evidence-based targeted tobacco control for reducing tobacco-related disease and death.

Data sources and variables
We analysed response data for adult participants (18 years or older) in the Australian Bureau of Statistics (ABS) 2017-18 National Health Survey (NHS) 17 and adult participants in the ABS 2018-19 National Aboriginal and Torres Strait Islander Health Survey (NATSIHS) 18 (further details: Supporting Information, part 1).
The three smoking status categories for our main analyses were people who smoke daily, people who formerly smoked but no longer smoke, and people who have never smoked.We extracted survey data on self-reported socio-demographic and health-related characteristics for NHS respondents, as well as postcode-level socio-economic status (Index of Relative Socio-economic Advantage and Disadvantage, IRSAD 19 ) and remoteness (Australian Statistical Geographic Standard for Remoteness 20 ) information for NHS and NATSIHS respondents (further details: Supporting Information, part 2).To assess intersectionality, selected variables were combined (selected a priori: for the NHS, sex and age, and sex and country of birth; for the NATSIHS, postcode-level remoteness and socioeconomic status).

Statistical analysis
For characteristics according to smoking category, we estimated weighted population proportions with 95% confidence intervals (CIs) in binomial models using survey-specific microdata (ie, individual-level survey response data).Person-level population weights (derived by the ABS, and available in the ABS DataLab) were applied to adjust proportions for deviations from population representativeness resulting from the survey sampling strategy and missing responses.
The numbers and proportions of the Indigenous and non-Indigenous people in each smoking category were estimated on the basis of ABS estimates for the Australian population in the NHS 2017-18 and the Indigenous population in the NATSIHS 2018-19 (further details: Supporting Information, part 3).
Sensitivity analyses examined the influence of broadening the outcome of interest from the socio-demographic characteristics of people who smoke daily with those of all people who currently smoke (daily, weekly, or less often), separately for NHS 2017-18 and NATSIHS 2018-19 respondents.We also examined the influence of different estimates of change in population size and smoking prevalence between the two surveys on our estimates (Supporting Information, part 3).
All data were accessed and analysed (using Stata 16) in the ABS DataLab (https:// www.abs.gov.au/ stati stics/ micro data-table build er/ datalab), which provides a secure environment that complies with the Census and Statistics Act 1905 (Cth).

Ethics approval
Aboriginal and Torres Strait Islander people were involved in reviewing the concept, design, and conduct of this project through Thiitu Tharrmay, a national Aboriginal and Torres Strait Islander research reference group that provides input, advice and guidance on work undertaken by the Tobacco Free program at the Australian National University (ANU), consistent with the National Health and Medical Research Council guideline for conducting ethical research with Aboriginal and Torres Strait Islander peoples. 21The study was approved by the NSW Aboriginal Health and Medical Research Council (1730/20) and the ANU Human Research Ethics Committee (2021/424, 2014/208).

Education, employment
For an estimated 42.1% of people aged 25-64  For non-Indigenous respondents, intersectionality of postcodebased socio-economic status and remoteness was similar to that for the total Australian population.Specifically, the largest proportions of people who smoked daily lived in major cities and the two socio-economically most disadvantaged quintiles (Supporting Information, table 2).

Sensitivity analyses
Among the NHS 2017-18 respondents who currently smoked, the socio-demographic characteristics of people who smoked daily were similar to those who currently smoked (2667 people smoked daily, weekly, or less frequently) (Supporting Information, table 3).Among the NATSIHS 2018-19 respondents who currently smoked, the proportions by combined postcodebased remoteness-socio-economic status category were similar for the people who smoked daily and for those who currently smoked (3028 people) (Supporting Information, table 4).
The estimated proportions of people who currently smoked (any frequency) who were Indigenous or non-Indigenous people were similar to those for people who smoked daily (Supporting Information, table 5).In sensitivity analyses, different assumptions regarding change to population size and smoking prevalence between the two surveys did not markedly affect the proportions of Indigenous and non-Indigenous people for each smoking category (Supporting Information, table 6).

Discussion
Tobacco companies sell an addictive, deadly product to more than 2.5 million Australians.Based on the survey data we examined, most adults who smoked daily in 2017-18 had completed year 12 and were in good physical and mental health, despite popular perceptions; an estimated 92% were non-Indigenous people, 58.8% were men, 61.3% were 25-54 years old, 72.5% were born in Australia, and 54.3% lived in areas in the two socio-economically most disadvantaged quintiles; 68.5% of those of working age were in paid employment.
Lower level of educational attainment has been associated with increased likelihood of smoking, 9 but we found that 69.0% of working age people who smoked daily had completed year 12.
In Australia, smoking is more prevalent in certain occupational groups (logistics, hospitality, construction and mining), 22 consistent with our finding that the highest educational attainment for 42.1% of people who smoke daily was a trade qualification or diploma; occupation-based interventions could therefore be useful as adjuncts to measures for reducing general population smoking levels.
Most respondents who smoked daily reported good to excellent health (75.9%) and low to moderate psychological distress (73.0%).The proportions who reported chronic conditions were

Research
generally largest for people who had formerly smoked, probably for a combination of reasons, including their higher median age, the health effects of their earlier smoking, and the increased likelihood of smoking cessation after a medical diagnosis ("sick quitter effect").
Aboriginal and Torres Strait Islander people comprised an estimated 8% of Australians who smoked daily in 2018-19 (195 700 people).About two-thirds of Indigenous respondents who smoked daily in 2018-19 lived in areas in the most disadvantaged socio-economic quintile at all three remoteness levels.Exposure to factors associated with smoking, including social disadvantage, is particularly high among Indigenous peoples, reflecting the historical and ongoing consequences of colonisation, such as exclusion from the economy and from health and education systems. 22,23This underpins the higher prevalence of smoking among Aboriginal and Torres Strait Islander people than non-Indigenous Australians.Multi-sector interventions are needed to comprehensively overcome the structural drivers of disadvantage. 24,25 In our study, we found that 36.8% of people who smoked daily were men aged 25-54 years (Box 3).The proportion born overseas was smaller among people who smoked than for people who had never smoked, but the survey data were insufficient for estimating region-or country-specific proportions.The prevalence of smoking among migrants is influenced by smoking behaviours in their countries of origin, and also by their generally being younger, more financially secure, and healthier than people who do not emigrate. 27More than 60% of respondents born overseas who smoked daily were men, a predominance consistent with other reports. 28ompared with people who smoked daily, larger proportions of respondents who no longer or had never smoked were employed, had tertiary degrees, and reported good to excellent health and low to moderate psychological distress, suggesting complex interactions between social disadvantage and the health impact of smoking.
The higher prevalence of daily smoking in certain population groups indicates health and broader system failings and that comprehensive, multifaceted, systems-based approaches to tobacco control are needed.Understanding the interactions of factors that contribute to smoking is critical for protecting health, including that of people who smoke.Numerous policy

Research
and legislative options are outlined in the WHO Framework Convention on Tobacco Control, including structural changes and regulations, as well as intensifying measures such as tobacco taxation, media campaigns, tobacco advertising bans, restricting tobacco retail licensing and reducing the number of retail outlets, and expanding smoke-free areas. 5Acknowledging the multifaceted role of tobacco companies in the problem shifts the focus from individual blame, responsibility, and stigmatisation of people who smoke to the tobacco industry itself.Changing the socio-cultural, political, economic, health and regulatory systems in which tobacco companies promote and sell their products will encourage individual and community agency and support freedom from nicotine dependence.
Whole of population approaches 4,25 should be accompanied by targeted programs and policies.This includes specific interventions for Aboriginal and Torres Strait Islander people, and groups that include the majority of people who smoke daily, such as men, people aged 25-54 years, and those living in areas of greatest socio-economic disadvantage.Our findings also support applying intersectionality to investigating complex relationships to better understand disparities. 14Knowing that most people who smoke regularly are employed, educated, and mentally healthy can inform decisions about who should be targeted.Effective campaigns are those in which people recognise themselves and their social norms, so that they feel at risk and are motivated to action. 12Approaches based on stereotypes and stigmatisation are unethical and reduce selfefficacy and capacity for appropriate action in their intended audiences. 15Combining comparative and absolute perspectives is needed to inform a comprehensive approach to reducing tobacco use and supporting people to quit smoking or to avoid taking it up.

Limitations
We report the first detailed analysis of the characteristics of people who smoke daily, formerly smoked, or never smoked in Australia, based upon nationally representative data.The response rate for the two ABS surveys was high (76.0%), 29but their home interview sampling strategy may have biased their samples in favour of people more likely to be home because of their employment situation.Although the validity of selfreported smoking status has been reported to be excellent, 32 survey responses may not accurately reflect actual behaviour.
Estimates for Indigenous people were derived from two sets with data collected about twelve months apart; as using multiple data sources affects estimate accuracy, the figures in Box 7 are reported without decimal places or confidence intervals to avoid inappropriately implied precision.However, sensitivity analyses indicated that our major findings were robust to variations in assumptions regarding changes in population size and the prevalence of smoking between the two surveys (Supporting Information, table 6).Changes in smoking prevalence and increasing use of other nicotine products, such as e-cigarettes, means that population profiles should be regularly reviewed.

Conclusion
People who smoke regularly in Australia have much in common with the general population.However, there are also important priority groups for tobacco control.Understanding who smokes is needed to inform both whole of population and targeted programs and policies for reducing tobacco use and nicotine dependence.Further, comprehensive structural supply-and demand-based tobacco control is required to reduce smoking prevalence among adults to the current targets of 5% or less for all Australians and 27% or less for Aboriginal or Torres Strait Islander people.Reforming the health and broader systems in which tobacco companies promote and sell their products can improve individual and community agency to be free from nicotine dependence, and ultimately eliminate tobacco-related death and disease.

characteristics of people who smoke daily, formerly smoked, or have never smoked, Australia, based on 2017-18 survey data*
17Derived from Australian Bureau of Statistics 2017-18 National Health Survey data (respondents: 2453 people who smoke daily, 5384 who formerly smoked, and 8319 who have never smoked); proportions weighted at the person level.17Thedata for this figure are included in Box 1 and Box 4. † Response options: profound or severe core activity limitation; other (mild or moderate limitation, or limitation affecting education or employment); no disability or restrictive long term health condition.‡ Australian Bureau of Statistics 2016 Australian Statistical Geographic Standard for remoteness. 20 Australian Bureau of Statistics 2016 Index of Relative Socioeconomic Advantage and Disadvantage at the Statistical Area 1 level. 19◆

5 Postcode-level residential characteristics of Aboriginal and Torres Strait Islander people who smoke daily, formerly smoked, or have never smoked, Australia, based on 2018-19 survey data* Weighted proportion (95% confidence interval) Characteristic Smoke daily Formerly smoked Never smoked
* Derived from Australian Bureau of Statistics Australian Bureau of Statistics 2018-19 National Aboriginal and Torres Strait Islander Health Survey data (respondents: 2808 people who smoke daily, 1529 who formerly smoked, and 1866 who never smoked); proportions weighted at the person level. 18 Australian Bureau of Statistics 2016 Index of Relative Socioeconomic Advantage and Disadvantage at the Statistical Area 1 level. 19 Australian Bureau of Statistics 2016 Australian Statistical Geographic Standard for remoteness. 20◆ * Derived from Australian Bureau of Statistics 2017-18 National Health Survey data (respondents: 2453 people who smoke daily, 5384 who formerly smoked, and 8319 who have never smoked); proportions weighted at the person level. 17 Arthritis; asthma; dorsopathies; malignant neoplasms; chronic obstructive pulmonary disease; heart, stroke and vascular disease; kidney disease; mental and behavioural conditions; osteoporosis.‡ Mild or moderate limitation, or limitation affecting education or employment.◆

7 Estimated numbers and proportions of Aboriginal and Torres Strait Islander people aged 18 years or older who smoke daily, formerly smoked, or have never smoked, Australia, based on 2017-18 and 2018-19 survey data*
Derived from Australian Bureau of Statistics Australian Bureau of Statistics 2018-19 National Aboriginal and Torres Strait Islander Health Survey data (respondents: 2808 people who smoke daily, 1529 who formerly smoked, and 1866 who have never smoked); proportions weighted at the person level. 18 Australian Bureau of Statistics 2016 Australian Statistical Geographic Standard for remoteness. 20 Australian Bureau of Statistics 2016 Index of Relative Socioeconomic Advantage and Disadvantage (Statistical Area 1). 19◆ Derived from Australian Bureau of Statistics 2017-18 National Health Survey (NHS) 30 and 2018-19 National Aboriginal and Torres Strait Islander Health Survey (NATSIHS) data; 31 proportions based on NHS, 2017-18 population estimates.† Derived by deduction of Aboriginal or Torres Strait Islander population estimate from the total population estimate.◆ * * Sources:

1
Australian Institute of Health and Welfare.Australian Burden of Disease Study 2018: key findings.Updated 18 Aug 2021.https:// www.aihw.gov.au/ repor ts/ burde n-of-disea se/ burde nof-disea se-study -2018-key-findi ngs/ conte nts/ key-findings (viewed Apr 2023).Wilson H. How stigmatising language affects people in Australia who use tobacco, alcohol and other drugs.Aust J Gen Pract 2020; 49: 155-158.7 Iredale JM, Clare PJ, Courtney RJ, et al.Associations between behavioural risk factors and smoking, heavy smoking and future smoking among an Australian population-based sample.Prev Med 2016; 83: 70-76.French DJ, Jang SN, Tait RJ, Anstey KJ.Crossnational gender differences in the socioeconomic factors associated with smoking in Australia, the United States of America and South Korea.Int J Public Health 2013; 58: 345-353.10 Thurber K, Walker J, Maddox R, et al.A review of evidence on the prevalence of and trends in cigarette and e-cigarette use by Aboriginal and Torres Strait Islander youth and adults.The Australian National University, Sept 2020.