Socioeconomic gradients in smoking among young women: A British survey

Abstract

Marked and increasing socioeconomic gradients in cigarette smoking are well established. Tracking these differentials among women requires appropriate measures of their socioeconomic position (SEP) which are equivalent across older and younger age groups.

This study examines socioeconomic gradients in cigarette smoking by age among women aged 20–34, using a standard indicator of SEP (age left full-time education) and alternative indicators which take account of changes in women's educational levels across age cohorts. The study is based on a large cross-sectional British data set (n=12 398), the Southampton Women's Survey, conducted 1998–2002. Poisson log-linear regression with adjusted variance was used to predict smoking status (ever smoked ⩾1 cigarette a day for 1 year, currently smoking ⩾1 cigarette a day, and ex-smoker – ever-smoker not currently smoking ⩾1 cigarette a day) in models which controlled for SEP, age and year of interview.

Socioeconomic gradients in ever-smoking were marked but stable across age groups. With quitting more prevalent in the higher than lower socioeconomic groups, gradients in current smoking steepened across age groups, with significant age/SEP interactions. The socioeconomic patterning of ever, ex and current smoking was similar using both the standard and alternative measures of education, but interactions were less pronounced with the alternative measures.

Socioeconomic indicators which take account of the recent and rapid increase in women's educational participation rates may provide a more reliable indication of age-related differentials in smoking status than standard measures. Nonetheless, both conventional and alternative measures point to stable socioeconomic differentials in rates of ever-smoking and widening differentials in current smoking across age groups. Our study confirms that addressing the link between socioeconomic disadvantage and tobacco use remains a major challenge for the tobacco control community.

Introduction

Cigarette smoking is the major behavioural determinant of the diseases driving life expectancy in high-income countries (Peto, Lopez, Boreham, Thun, & Heath, 1994). Smoking rates have declined in these societies, but with men and women at different stages of the cigarette epidemic (Lopez, Collishaw, & Piha, 1994), the prevalence of cigarette smoking has typically fallen faster among men than women. The differential rates of decline have narrowed and, in an increasing number of countries, have eliminated the gender gap in prevalence among adults (Shafey, Dolwick, & Guindon, 2003). In adolescence, higher rates of smoking (⩾1 cigarette a week) are reported among girls than boys in an increasing number of European countries as well as in Australia and New Zealand (Currie et al., 2004; Hill, White & Segan, 1995; Statistics New Zealand, 1996). With adolescent smoking status tracking across early adulthood and into middle-age (Australian Bureau of Statistics, 2002; Jefferis, Power, Graham, & Manor, 2004; Office for National Statistics, ONS, 2004; Schooling & Kuh, 2002), the smoking population of high-income countries is set to become increasingly female.

In Europe, as in the US, Australia and New Zealand, women's smoking is increasingly characterised by the socioeconomic gradients first evident among men (Cavelaars et al., 2000; Fiore et al., 1989; Graham, 1996; Pierce, Fiore, Novotny, Hatziandreu, & Davis, 1989; Statistics New Zealand, 1996). Rates of regular smoking (⩾1 cigarette a day) have fallen more rapidly among women in higher than lower socioeconomic groups, widening differentials in smoking over time and across age cohorts (Evandrou & Falkingham, 2002; Fiore et al., 1989; Giskes et al., 2005; Pierce et al., 1989). In Britain where occupation provides the primary measure of socioeconomic position (SEP), the decline in smoking prevalence among women in the lowest socioeconomic group has stalled since the mid-1980s, with a prevalence of 33% in 1986 and 34% in 2000. Across the same period, smoking rates fell from 25% to 13% among women in the highest socioeconomic group (ONS, 2001).

The prevalence of current smoking is determined by rates of ever-smoking and quitting smoking. Quit rates are low in early adulthood and increase with age (ONS, 2004). In consequence in industrialised societies like the UK, smoking rates are at their highest in early adulthood, and decline with age and across age groups (Evandrou & Falkingham, 2002; Jefferis et al., 2004; Kemm, 2001). Linked prospective studies are the research design of choice for mapping changes in smoking status across age groups. However, the limited pool of prospective studies relate to older cohorts entering adulthood at a time when their smoking habits were less strongly related to social disadvantage (Pollitt, Rose, & Kaufman, 2005). Cross-sectional studies have therefore been employed (Evandrou & Falkingham, 2002; Kemm, 2001; Laaksonen et al., 1999), although the difficulties of separating out cohort, age and period influences are well-recognised (Evandrou & Falkingham, 2002; Hallqvist, Lynch, Bartley, Lang, & Blane, 2004; Laaksonen et al., 1999).

Whatever the research design, appropriate measures of SEP are required. For women, parental status mediates the relationship between educational level and occupational status, with mothers more likely than non-mothers with equivalent education to be economically inactive or in low-skilled and low-paid jobs (Joshi, 2002). As a result, occupation provides a non-equivalent measure for mothers and non-mothers (as well as for women and men). Asset-based measures like housing tenure have been used as an alternative but provide a less reliable guide to the SEP of younger women where transitional tenures (renting with friends, living with parents) are more common (Pugh, Power, Goldblatt, & Arber, 1991). Even education, the most widely used measure of SEP in high-income societies, does not avoid the problem of non-commensurability between different groups of women. In high-income societies, educational qualifications are increasing the gateway into employment, with the result that younger cohorts spend longer in full-time education and achieve higher levels of educational qualifications than older cohorts (Heinz, 1999). As a result, standard measures of educational level like years of full-time education and highest educational qualification are not equivalent across age groups. However, equivalent indicators are rarely employed in studies of changes in smoking status over time.

Developing and testing alternative measures of SEP is therefore an important methodological task. In a cohort analysis of smoking prevalence, Laaksonen et al. (1999) constructed a binary measure by dividing respondents on the basis of their position above or below the median length of education in their birth cohort. We extend their work by investigating socioeconomic gradients in smoking status among women in early adulthood using both conventional and alternative measures. Drawing on a contemporary British study, we aim to establish whether gradients in ever-smoking, ex-smoking and current smoking (i) are steeper in older than younger age groups, and (ii) are moderated when alternative measures are used.