Correspondence analysis applied to grouped cohort data