We examined changes in socioeconomic inequalities in smoking prevalence, smoking consumption, smoking initiation, and smoking cessation between 2001 and 2008 in the Netherlands. In line with previous studies [7–16], we found that respondents with lower education and income were more likely to be smokers than respondents with higher education and income in 2001 and 2008. Socioeconomic inequalities in smoking consumption, smoking initiation, and smoking cessation are less well documented than inequalities in smoking prevalence . We found that respondents with lower education smoked more cigarettes per day, had higher initiation ratios, and had lower quit ratios than respondents with higher education in 2001 and 2008. Therefore, it is important that tobacco control programs focus on decreasing smoking initiation and consumption and increasing smoking cessation among the lower educated. It seems that focusing only on either initiation or cessation is insufficient.
In this study, we found evidence that SES inequalities in smoking widened between 2001 and 2008 in the Netherlands, but this widening was not the same for women and men and it depended on which SES and smoking indicator was used. Among women, educational inequalities widened significantly for smoking prevalence, smoking initiation, and smoking cessation. Among men, educational inequalities widened significantly for smoking consumption. The widening pattern in smoking, initiation, and quitting among low educated women is especially worrying. In this group, smoking prevalence remained stable between 2001 and 2008 because both the initiation and quit ratio increased slightly. Among moderate and high educated women smoking prevalence decreased because initiation ratios remained constant, while quit ratios increased considerably. Other studies have also found that in countries in later stages of the smoking epidemic, SES inequalities in smoking prevalence are stable among men, while they are widening among women [10, 13]. Also, it was found earlier that SES inequalities in smoking consumption are stable among women and widening among men .
According to the literature, advertising bans, smoking bans in workplaces, reimbursement of smoking cessation therapies, and increased tobacco taxes have the potential to reduce SES differences in smoking . During our study period, smoke-free workplace legislation (2004), smoke-free hospitality industry legislation (2008), and tobacco tax increases (2001, 2004, and 2008) were implemented in the Netherlands. From our data, it looks like there was a change in SES inequalities in smoking prevalence and consumption between 2003 and 2004, when the smoke-free workplace legislation was implemented together with a tax increase. However, the SES inequalities seem to be widening instead of narrowing. An earlier study also showed that the workplace smoking ban in the Netherlands has increased SES differences in smoking cessation . An explanation could be that the smoke-free legislation was not comprehensive: the hospitality industry was exempted (until 2008) and designated smoking rooms were allowed.
Our study had some important strengths. We used very large, representative samples of the Dutch population over a period of eight years. Therefore, changes in SES differences over time could be detected and generalized to the Dutch population. Also, we examined both education level and income level differences in four smoking related outcomes, including smoking initiation and cessation for both men and women. Our study thus provides a more detailed picture of SES differences in smoking than most studies. This study also had some limitations. First of all, we relied on self reported smoking status. Since there can be differences in underreporting of smoking between SES groups, this could have influenced our results. Second, income levels were unknown for 21% of respondents. These respondents were excluded from the analyses and this may have introduced selection bias. However, since we also used another SES indicator (education level) with only a few missing values resulting in comparable findings, we expect that the missing income data did not bias our results. Furthermore, we used initiation and quit ratios to estimate population level smoking initiation and cessation. These ratios may not be sensitive enough to reflect short term changes in smoking initiation and cessation. However, initiation and quit ratios are suitable for understanding long term population trends .