Unravelling the alcohol harm paradox: a population-based study of social gradients across very heavy drinking thresholds

BACKGROUND There is consistent evidence that individuals in higher socioeconomic status groups are more likely to report exceeding recommended drinking limits, but those in lower socioeconomic status groups experience more alcohol-related harm. This has been called the 'alcohol harm paradox'. Such studies typically use standard cut-offs to define heavy drinking, which are exceeded by a large proportion of adults. Our study pools data from six years (2008-2013) of the population-based Health Survey for England to test whether the socioeconomic distribution of more extreme levels of drinking could help explain the paradox. METHODS The study included 51,498 adults from a representative sample of the adult population of England for a cross-sectional analysis of associations between socioeconomic status and self-reported drinking. Heavy weekly drinking was measured at four thresholds, ranging from 112 g+/168 g + (alcohol for women/men, or 14/21 UK standard units) to 680 g+/880 g + (or 85/110 UK standard units) per week. Heavy episodic drinking was also measured at four thresholds, from 48 g+/64 g + (or 6/8 UK standard units) to 192 g+/256 g + (or 24/32 UK standard units) in one day. Socioeconomic status indicators were equivalised household income, education, occupation and neighbourhood deprivation. RESULTS Lower socioeconomic status was associated with lower likelihoods of exceeding recommended limits for weekly and episodic drinking, and higher likelihoods of exceeding more extreme thresholds. For example, participants in routine or manual occupations had 0.65 (95 % CI 0.57-0.74) times the odds of exceeding the recommended weekly limit compared to those in 'higher managerial' occupations, and 2.15 (95 % CI 1.06-4.36) times the odds of exceeding the highest threshold. Similarly, participants in the lowest income quintile had 0.60 (95 % CI 0.52-0.69) times the odds of exceeding the recommended weekly limit when compared to the highest quintile, and 2.30 (95 % CI 1.28-4.13) times the odds of exceeding the highest threshold. CONCLUSIONS Low socioeconomic status groups are more likely to drink at extreme levels, which may partially explain the alcohol harm paradox. Policies that address alcohol-related health inequalities need to consider extreme drinking levels in some sub-groups that may be associated with multiple markers of deprivation. This will require a more disaggregated understanding of drinking practices.