Does a Change Over All Equal a Change in All? Testing for Polarized Alcohol Use Within and Across Socio-Economic Groups in Germany.

AIMS This study aimed at testing whether drinking volume and episodic heavy drinking (EHD) frequency in Germany are polarizing between consumption levels over time. Polarization is defined as a reduction in alcohol use among the majority of the population, while a subpopulation with a high intake level maintains or increases its drinking or its EHD frequency. The polarization hypothesis was tested across and within socio-economic subgroups. METHOD Analyses were based on seven cross-sectional waves of the Epidemiological Survey of Substance Abuse (ESA) conducted between 1995 and 2012 (n = 7833-9084). Overall polarization was estimated based on regression models with time by consumption level interactions; the three-way interaction with socio-economic status (SES) was consecutively introduced to test the stability of effects over socio-economic strata. Interactions were interpreted by graphical inspection. RESULTS For both alcohol use indicators, declines over time were largest in the highest consumption level. This was found within all SES groups, but was most pronounced at low and least pronounced at medium SES. CONCLUSION The results indicate no polarization but convergence between consumption levels. Socio-economic status groups differ in the magnitude of convergence which was lowest in medium SES. The overall decline was strongest for the highest consumption level of low SES.