re model, four possible important prognostic variables were unbalanced and a center effect was not considered, which all can cause bias for study res