To identify spatial variation in the risk for metabolic complications (RMC) by means of a semi-parametric approach for multinomial data.

Cross-sectional study.

We visited 730 households selected in the first of a two-stage sample in South health district in Campinas, São Paulo, Brazil, 2006–2007.

We interviewed 651 individuals and obtained their respective anthropometric measures and geographical coordinates of their house location. They were classified according to a combination of BMI and abdominal circumference as having no risk, increased, high or very high RMC.

Gender, age and schooling were associated with RMC. Crude spatial risk for the three levels of RMC in relation to the absence of risk suggested different patterns in each level. Adjusted spatial risk for the RMC showed smaller significant areas, but the pattern remained similar to crude risk.

Spatial point analysis with a multinomial approach improves the understanding of differences in RMC found, as we could identify specific areas in which to intervene. The public health significance of these findings may lie in the additional evidence provided that spatial location and its features can influence patterns of RMC.