Background/Research Question: In the health literacy (HL) discourse there is debate about the ways by which HL is impacting health. Three different, logically non-exclusive hypotheses are proposed: (a) HL as a specific, direct, social determinant of health; (b) HL as a mediator between other determinants and health; and (c) HL as a moderator of the effect of other determinants on health. Only few examples of empirically testing the mediator or moderator hypothesis exist. The data of the European Health Literacy Survey allow testing of the three hypotheses comparatively in parallel for functional and comprehensive HL.
Methods: Data collection was based on multistage random samples of about 1000 European Union citizens aged 15 or older with Computer Assisted Personal Interviewing or Paper Assisted Personal Interviewing methodology in 2011 in eight Member States of the EU. Demographic and socioeconomic indicators, a comprehensive (European Health Literacy Survey Q47) and a functional (Newest Vital Sign Test) health literacy measure and one self-assessed health variable were surveyed. Correlation analysis, multiple regression analysis and path analysis were performed. Results: Comprehensive HL (and to a much lesser degree functional HL) is a relevant predictor for self-assessed health. Also, comprehensive HL is only to a limited degree mediating the effects of other determinants on self-assessed health and only for age does HL partly moderate the effect on health. Explained variance and strength of effects vary considerably by national context.

Conclusions/Implications: Comprehensive HL is a critical, direct determinant of health. Therefore comprehensive HL has a considerable potential for health promotion to improve population health and tackle the health gap. But comprehensive HL measurement should be standardised in every country to allow for designing adequate measures for the specific situation of the country and also for benchmarking. For better understanding of the causal structure of the impact of HL on health, longitudinal studies will be needed.