A Novel Methodological Approach to Measure Linear Trends in Health Inequalities: Proof of Concept With Adolescent Smoking in Europe

Abstract

We describe a new method for presenting and interpreting linear trends in health inequalities, and present a proof-of-concept analysis of inequalities in smoking among adolescents in Europe. We estimated the regression line of the assumed linear relationship between smoking prevalence in low- and high-socioeconomic status (SES) youth over time. Using simulation, we constructed a 95% confidence interval (CI) for the smoking prevalence in low-SES youth for when this would be 0% in high-SES youth, and we calculated the likelihood of eradicating smoking inequality (<5% for both low and high SES). This method was applied to data on adolescents aged 15-16 years (n = 250,326) from 23 European countries, derived from the 2003-2015 European Survey Project on Alcohol and Other Drugs. Smoking prevalence decreased more slowly among low- than among high-SES adolescents. The estimated smoking prevalence was 9.4% (95% CI: 6.1, 12.7) for boys and 5.4% (95% CI: 1.4, 9.2) for girls with low SES when 0% with high SES. The likelihood of eradicating smoking inequality was <1% for boys and 37% for girls. We conclude that this novel methodological approach to trends in health inequalities is feasible in practice. Applying it to trends in smoking inequalities among adolescents in Europe, we found that Europe is currently not on track to eradicate youth smoking across SES groups.

Keywords: Europe; adolescents; eradication of inequalities; inequality; simulation methods; socioeconomic status.

Figure 1

Linear trend in socioeconomic inequalities in smoking in 23 European countries among boys (A) and girls (B), with fitted regression line and 95% confidence regions for each time point. Data from 23 European countries of the European Survey Project on Alcohol and Other Drugs in 4 survey waves (2003–2015). The proportion of weekly smokers in the low–socioeconomic status (SES) group is P(0) and in the high-SES group is P(1). The green dots indicate the point of reaching <5% prevalence, which is defined as the eradication scenario.