Statistical methods
All statistical analyses were performed in SPSS for Windows (SPSS Inc., Chicago, IL, USA). For the descriptive analyses of differences between intervention and control group simple Student's t-test and 95% confidence intervals (CIs) were used. Dichotomous variables were tested using χ2 test.
Mean differences in weight development between the two groups and the changes in blood parameters were analyzed using analysis of covariance with maternal age and parity included as fixed factors, and height and weight at inclusion included as covariates in the model. For analyses of the blood parameters, the concentration at inclusion was added as covariate as well. The model was reduced by excluding factors with P>0.1. Ninety-five percentage CIs were calculated for mean differences.
Group effects on dietary intake were tested using repeated measurements including intake, weight, and height at inclusion as covariates and maternal age and parity included as fixed factors in the model.