Comment on "Correlation inequalities and hidden variables"