RESUMEN
BACKGROUND: Different land cover types in cities, including green areas, have impacts on mental health. Few studies, however, have been conducted in megacities in low-/middle-income countries, which have extremely complex urban arrangements. OBJECTIVE: We analyzed land cover using the population database of the São Paulo Megacity Mental Health Survey (n = 4287) and the associations with the diagnosis of depression/anxiety. METHODOLOGY: Automatic classification of the orthophotos of the metropolitan area was performed using the random forest algorithm to provide land cover variables. The association of mental health data with land cover was determined with logistic regression and multilevel regression models. RESULTS: The percentage of different green spaces, mainly grassed areas, within districts was negatively associated with the presence of anxiety (OR 0.994, 0.968, 0.994, respectively), while roofs, asphalt and shadow (OR 1.007, 1.021, 1.085, 1.021, 1.014, respectively) were positively associated with the presence of anxiety. These results were more significant in green areas within the fourth quartile [-0.352 (0.158) and -0.347 (0.155), respectively] and in the roofs in the fourth quartile [0.321 (0.159)]. No significant results were found for depression. SIGNIFICANCE: Our data indicate the need for intensive greenery in spaces with different vegetation compositions in urban environments, especially megacities, to improve the mental health of urban dwellers.