Figure 2.

Predicted probability of an alcohol-attributable event during follow-up, stratified by socioeconomic status

All models were adjusted for age, sex, study wave, smoking, body-mass index, and binge drinking in the past week. The lines represent the predicted probability of experiencing an alcohol-attributable event (estimated from logistic regression models) and the shading denotes the 95% CI. (A) Deprivation was categorised according to the most deprived two quintiles vs the least deprived three quintiles. (B) Social class categorisation was based on manual vs non-manual occupations. (C) Education was categorised according to attainment of none, other school, or Scottish standard grade qualifications vs Scottish higher grade, higher national certificate, higher national diploma, or degree or above. (D) Household income was categorised by the lowest two quintiles vs highest three quintiles. Follow-up for household income measures was shorter because participants were not asked about this information in the first two study waves of data collection.