Abstract Aim This study assesses the influence of various specific health conditions as activity limitations and groups of chronic diseases, in estimating poor self-rated health for the Spanish population aged 30–60, stratified by the intersections of gender and social class. Moving beyond unidimensional social gradients, the research utilises decision tree analysis to identify specific configurations of SRH responses that align with the epidemiological profiles of distinct intersectional strata. Methods Data were drawn from the 2023 Spanish Health Survey (ESdE), focusing on a sample of 10,210 adults aged 30–60. This age bracket was selected to capture health inequality dynamics during the working life-course. Specific machine learning decision trees (J48 algorithm) were employed to assess the pathways for classifying individuals with poor SRH across six distinct intersectional groups. Results Resulting decision trees differ significantly in composition, shape, and length across strata. A clear trend emerged: less skilled groups exhibited higher tree complexity, reflecting a more rapid accumulation of chronic illnesses at earlier ages. Conclusion This research contributes to the emerging field of study regarding what truly underlies individuals’ general assessment of their health. The resulting decision trees illustrate the social gradient in health in a novel way that enriches the understanding of health inequalities. Instead of simply showing a linear trend across a one-dimensional variable, these trees reveal that as disadvantages accumulate through the intersection between gender and social class, the combination of relevant diseases do not only become more complex, but also distinct, in specific and unique ways for each group, as the intersectional framework predicts.
Gumà et al. (Fri,) studied this question.