Association between graph literacy, health literacy and numeracy and understanding of diverse formats of numerical health information in a socioeconomically disadvantaged population: A cross-sectional study

Background Risk communication remains a major challenge for public health. The use of graphical displays such as bar charts or icon arrays could help, particularly in socioeconomically disadvantaged populations.Aim The aims of this study were to 1) measure graph literacy (GL) and investigate factors associated with high GL (including health literacy and numeracy) and 2) assess different formats for presenting numeric health information to a socioeconomically disadvantaged population according to their level of GL.Methods We conducted an online cross-sectional study in a socioeconomically disadvantaged population (household income <€12,000) in France. In addition to sociodemographic variables and existence of a chronic condition, we measured GL, functional health literacy, numeracy, comprehension of data comparing treatments and the preferred format for presenting such data. GL was measured using an adapted version of the short, validated 4-item version of the Graphic Literacy Scale. Factors associated with higher GL were investigated using ordinal regression models. We assessed the impact of GL on numerical comprehension (3 questions after presentation in various formats: tables, bar charts, and icons) and used the McNemar test to compare the correct responses across formats.Results We recruited 401 people, aged 40 on average and predominantly female (64%); 27% had high GL, 70% had adequate functional health literacy, and 48% had high numeracy. In the ordinal regression, both inadequate health literacy (OR = 0.51; p = 0.002) and low numeracy (OR = 0.46, p < 0.001) were associated with lower GL. Although people preferred bar charts (39%), comprehension scores for tables were significantly higher compared to bars and icons on average (p < 0.05). Participants with high GL understood treatment comparison better (regardless of format) than those with low GL (p < 0.001).Discussion/Conclusion GL was low among this socioeconomically disadvantaged sample and lower GL was associated with both lower numeracy and health literacy. Graphical displays of risks might not be better understood than numbers for conveying health risks, but tables or a combination of tables and graphical displays could be a suitable alternative.