Identifying relevant intersections in relation to motivation and attempt to stop smoking by using a combination of...

To illustrate robust intersections of co-occurring factors for two predictors of smoking cessation, motivation to stop smoking (MTSS) and past year-quit attempts (QA), by using means to develop robust predictive models such as bootstrap resampling, scoring rules to evaluate the predictive accuracy and spline functions.

Cross-sectional data from the German Study on Tobacco Use (DEBRA). Past-years smokers (≥18 years, n = 13 245) from 22 survey waves (2016-2020) were included. The sample (mean age 46.8 years, 46.7% women) was randomly divided into learning (70%) and validation data (30%). Less than 20% in both data sets had tried to stop smoking within the preceding 12 months.

Multinomial regression (for MTSS) and logistic regression (for QA) were used to evaluate whether age, sex, education, monthly net household income per person and the region of residence form intersections with relevant differences in the two outcomes.

MTSS compared with the absence of MTSS was associated with middle [95% confidence interval (CI) = 1.02-1.39] and high education (95% CI = 1.37-1.98). Regarding MTSS, the highest probabilities were observed in participants aged 30 to 50 years from lower and middle (30-40 years) income groups. Regarding QA, the probability of at least one past-year QA was highest in females aged between 20 and 40 years and independent from educational level. Similar probabilities in males were seen only among those from the highest educated group. The predictive accuracy of the results was reduced by 3.1% for MTSS and 3.4% for QA when comparing learning with validation data.

This German study provides compelling evidence linking highest motivation to stop smoking to those aged 30 to 50 years with lower or middle household income. Regardless of educational level, females' probabilities of reporting at least one past-year quit attempt appears to be highest in those aged 20 to 40 years. These findings highlight the need for adopting an intersectional approach when studying predictors of smoking cessation.