Bias versus causality : interpreting recent evidence of association between endometriosis and ovarian cancer