Gravar-mail: Overestimation of risk ratios by odds ratios in trials and cohort studies: alternatives to logistic regression