9.2 Socioeconomic disparities in tobacco exposure and use: are the gaps widening?

Show / hide chapter menu

While it is clear that smoking rates are higher among disadvantaged than among advantaged groups in Australia, much less clear is whether disparities have been widening over time. The answer one gets to this question depends most crucially on the period of time over which one analyses the data. In addition, the extent of differential changes in rates in different SES groups also seems to vary depending on the indicator of SES status, the data set and the jurisdiction being examined.

Insufficient data are available to examine trends over time in smoking during pregnancy. However, this section presents data on trends since 1980 in disparities in smoking by adults, together with limited data about emerging disparities among children.

9.2.1 Changes in the prevalence of smoking among adults in various socioeconomic groups

Trends over time in smoking prevalence among different social groups can be difficult to interpret because of changing social and economic conditions. With increasing school retention in Australia and introduction of financial assistance for tertiary students in the mid-1970s,[9] a much more diverse group of people in the 1990s are achieving higher levels of formal educational qualification compared with that group of people who undertook tertiary education in the late 1960s and early 1970s. In a period of low unemployment and a buoyant job market, the unemployed in the mid 2000s on the other hand may be less socially diverse than groups who were unemployed during times of low job vacancies in the 1980s and 1990s. Inconsistencies in methods of collecting data and in SES categories over time also make long-term analysis difficult. To get a reliable picture of trends in SES-related disparities in smoking, it is therefore useful to look at relative changes across several socioeconomic indicators and using several different data sets.

9.2.1.1 Changes in prevalence among those with varying levels of formal education

Table 9.5 sets out the prevalence of current smoking among people with various levels of education between 1980 and 2004.[10]

Adjusting for age and gender, analysis of these data shows that the prevalence of smoking found in every educational level in 2004 was significantly lower than that found in 1980.

As is evident from Table 9.5, the decline in prevalence of smoking for the total 24-year period covered by this survey was most substantial among those who had graduated from university, those who had finished secondary school and those who left school before year 10.

Figure 9.9 plots this data, omitting the figures for those who left school before year 10. (Given the historically lower minimum leaving age for school, and the historically lower rates of participation by women in tertiary education, this group would comprise mainly older people, with varying levels of income and living in varying circumstances. Older populations also exclude people who have died prematurely due to smoking.)

Figure 9.9 would seem to indicate that smoking fell roughly equally among people of various educational levels between 1980 and 1992. Declines have flattened since the mid-1980s in less qualified groups, but note the apparent fall in all groups between 1998 and 2001.[11] The pattern of these changes is discussed in more detail in Section 9.8 .

Rates of school retention have increased since 1998 among young people.66 Also, increasingly over time people born prior to World War II (who are much less likely to have completed school)67 are being lost from the total population. Thus, part of the explanation for the flattening of smoking rates in people who have not completed Year 12 is that this is becoming a group characterised by more social and economic disadvantage than was the case in previous cohorts.

9.2.1.2 Changes in prevalence in blue versus white collar groups

A quick reading of all the studies published on smoking prevalence in Australia over the past 30 years would seem to indicate that the gap in smoking prevalence between blue and white collar groups has widened considerably.68 However as discussed in Chapter 1, surveys have varied greatly in the age ranges reported and the ways that occupational status has been defined.

Table 9.6 sets out smoking prevalence between 1980 and 2004 for all occupational groups using data collected in surveys conducted by the Anti-Cancer Council of Victoria until 1998, and then the National Drug Household Survey, both re-analysed to include just people 18 years and over.[12]

Adjusting for age and gender, the decline in prevalence of smoking was significant for all groups but it was proportionately greater among white collar workers than blue collar workers. Overall, prevalence of smoking among upper white collar workers halved over this study period, while among lower blue collar workers, prevalence dropped by about 30%.28

But is that the full story? Combining upper and lower blue collar groups and upper and lower white collar groups obscures some interesting differences between the four groups at various points of time.

Figure 9.10 sets out the data from Table 9.6, omitting the figures for those not in paid work[13].

Examining Figure 9.10 it is clear that smoking rates over the 1980s and early 1990s declined roughly equally in absolute terms among various occupational groups.

It is also evident that one could get a very different picture of the relative declines among occupational groups in recent years depending on the period examined. Comparing 2004 to 1992 (showed by the dotted line in Figure 9.10) one would conclude that white collar groups did much better than blue. However comparing 2004 to 1995 (showed by the solid line), the decline would appear to be roughly equal. The implications of this pattern of decline is also discussed further in Section 9.8.

Over the entire period it is apparent that smoking has dropped profoundly in the upper white collar group. However, less apparent is the fact that gap between smoking in the lower blue collar group and the lower white collar group has barely changed, and the gap between smoking in the lower white and upper blue groups appears to have narrowed quite considerably.[14]

Figures 9.9 and 9.10 demonstrate the importance of avoiding simplistic point-in-time comparisons. Detailed statistical analysis is required to determine relative patterns of change over particular time periods.

9.2.1.3 Changes in smoking prevalence by area-level measures of SES

Similar to the picture with occupational status, it is easy to see a significantly greater decrease in smoking prevalence among groups living in more advantaged compared with those living in less advantaged areas.

However, once again such simplistic point-in-time comparisons obscure important differences between males and females in various categories of disadvantage. They also say little about the reliability of prevalence estimates given the quite small sample sizes that result from dividing the population into five groups.

In 2006 a useful statistical analysis of relative changes in smoking rates among Australians in different socioeconomic groups was published by the World Health Organization. The study analysed changes in the proportion of regular smokers in the 12 years between 1989–90 and 2001. The study tested the statistical significance of relative changes in the percentage decrease in each quintile of males and females according to the ABS's Index of Socioeconomic Disadvantage.32

Once again these data indicate a clear SES gradient within each survey year[15]. That is, the prevalence of smoking is lowest in the most advantaged group, slightly higher in Q4, slightly higher again in Q3 and highest in Q1. As in other data sets, researchers also found that smoking had declined between 1989 and 2001 in all socioeconomic groups among both men and women.

But has the differential between SES groups increased over time?

Interestingly, the degree of decline in smoking in women appeared to be unrelated to their socioeconomic status[16]. So, prevalence declined more in the most disadvantaged two quintiles than it did in the middle and most advantaged quintile. In men, the percentage of smokers did decline more substantially in Q5, the most advantaged group, than in Q1 the least advantaged group, so that the overall trend for an SES effect in men did just approach significance at the 0.05 level. However, with the size of changes in Q2 and Q3 exceeding those in Q4, there was also no clear evidence of a social gradient.

9.2.2 Differential uptake or differential cessation?

In the population overall, smoking has reduced due to a combination of fewer people taking up smoking, more people quitting, and more smokers than non-smokers dying prematurely.

Table 9.9 shows the percentages of people who have classified themselves as smokers and ex-smokers compared to those who have never smoked for the period between 1977 and 2000–01 using data from the ABS's National Health Surveys.

The analysis published by WHO discussed in Section 9.2.1.3 above also examined trends across socioeconomic groups in the proportions of males and females who reported never having been regular smokers.

The proportion of women classifying themselves as never smokers actually fell in all SES groups over this period, indicating an increase in the proportion of women who have ever smoked. This reflects the low rates of smoking among women born prior to World War II and the sharply increased rates of uptake among young women over the 1970s and '80s. But note that the fall in numbers of women who have never been smokers was actually highest among higher SES groups so that the social gradient actually reduced over the 12-year period (indicated by the X2 reducing between 1989–90 and 2001). Among males by contrast the proportion of never smokers increased in all groups. The level of increase in Q1 to Q4 did not show a clear SES gradient. However those in the most advantaged group were more likely and became increasingly more likely over time, to report never having been smokers.

A careful analysis of these data thus indicates no evidence of an increasing social gradient in women. Disparity did increase between lower and upper SES groups among males in the period between 1989 and 2001, and this was partly due to the fact that uptake did not decline much in the lowest socioeconomic quintile. However, overall, the increasing disparity can be explained not so much by lack of decline among lower SES groups, but rather by the extraordinarily large decline in smoking among males in the highest SES quintile. This decline would be partly due to the lower death rates and longer life expectancy of males in the highest SES quintile who never smoked (compared to smoking and non-smoking males in the lower quintiles who would have been more likely to have died prematurely due to a range of other causes). But it must substantially be due to lower rates of uptake in boys who by the 1990s were adults in the highest SES groups[17].

9.2.3 Changes in the prevalence of smoking among students in schools in areas of varying levels of disadvantage

Differences in uptake rather than cessation appears to have been the more significant driver of socioeconomic disparities in smoking in Australia[18]. To assess the likelihood of increasing disparity in the future, researchers could more carefully analyse rates of never smoking and cessation among different SES groups in cohorts of adults born since 1970.

Data on smoking rates among secondary school students of different socioeconomic backgrounds would also provide some indication of what future smoking disparities by SES may be.

Information on socioeconomic backgrounds of students is not collected in surveys of youth smoking in Australia. However, Figures 9.11 and 9.12 show smoking rates among 16- and 17-year-olds and 12-to-15-year-olds between 1987 and 2005 according to the level of disadvantage of the neighbourhood in which they lived.

Consistent with the conclusion in Section 9.2.2 above, the greatest overall declines over the 18-year period have been among young people living in the most advantaged areas. In 1987 students in these schools had the highest smoking rates. By 2005 their smoking rates were lower than students in schools in less advantaged areas.

Despite increasing rates of school retention between 1987 and 1996, smoking rates among 16- and 17-year-olds remained relatively flat over that period among students from the most disadvantaged areas. Prevalence increased in all groups between 1990 and 1996, and it appears to have declined roughly equally in all groups since 2002.

Following a sharp reversal of the socioeconomic gradient among 12-to-15-year-olds between 1990 and 1996, between 1996 and 2005 smoking declined equally among students from in areas at all levels of disadvantage.