Fig. 2

Tested conceptual models for the interplay between stunting, socioeconomic factors, the fecal microbiota and child neurodevelopment. (A) The simplest SEM model includes only direct effects of the socioeconomic status, the fecal microbiome and stunting on neurodevelopment; (B) the simple SEM model includes direct effects of the socioeconomic status, the fecal microbiome and stunting on neurodevelopment and an indirect effect of the socioeconomic status on neurodevelopment through stunting and through the fecal microbiome; (C) The complex SEM model includes the socioeconomic status, the fecal microbiome, stunting, anemia, BCAA, α1-antitrypsin level (AAT), age and reported birth size, allowing for direct and indirect effects on neurodevelopment; (D) The complex path model includes only observed variables (number of rooms, maternal age at first pregnancy, the fecal microbiome, stunting, anemia, leucine, α1-antitrypsin level, age and reported birth size), without latent constructs, allowing for direct and indirect effects on the overall developmental score. Observed variables are represented as squares and latent variables are represented as circles. Indicator variables used to build latent constructs are squared with dashed lines. HAZ: Height-for-Age Z-score; BCAA: branched-chain amino acids.