Spatial confounding in univariate and multivariate disease mapping: differe