RESUMEN
The effect of perioperative blood transfusion on the survival of patients with colorectal cancer was evaluated in 128 patients undergoing curative surgery between 1980 and 1988. The following clinical and histopathological variables were also studied: age, sex, duration of symptoms, presence of intestinal obstruction, tumour site, extent of spread through the bowel wall, lymph node involvement, Dukes' stage, grade of differentiation, venous invasion and type of surgical procedure performed. The need for perioperative blood transfusion was unrelated to the stage of disease. In the transfused patients (n = 73) the 5-year recurrence-free survival, calculated by the Kaplan-Meyer technique, was 37% and in the non-transfused (n = 55) was 60% (P = 0.0027, Mantel-Cox). Similar differences were found in the comparison of the groups with (n = 68) and without (n = 60) transfusions on the day of operation. The deleterious effect of transfusion was evident in patients who received only one unit of blood (n = 19)--these had a 5-year survival rate of 45% compared those who had more than one unit of blood (n = 54) (5-year survival rate 35%) (P = 0.0062). With a multivariate analysis, using a Cox proportional hazard model, taking into account all the variables studied, a significant and independent effect on survival was found for lymph node involvement (beta coefficient = 3.97), blood transfusion (beta coefficient = 2.16) and extent of bowel wall spread (beta coefficient = 1.75).(ABSTRACT TRUNCATED AT 250 WORDS)