Objective: To evaluate the role of a blood transfusion in a patient population with head and neck cancer that was treated with combined therapy (surgery and radiation).
Design: Retrospective, nonrandomized end point study. Univariate and multivariate analysis of 24 variables, including transfusion status.
Setting: The Veterans Affairs Cooperative Study (Cooperative Studies Program 268).
Patients: Patients in the surgical arm of the study (166 patients) underwent surgery and postoperative radiation therapy for advanced (stage III and IV) laryngeal cancer.
Main outcome measures: Identification of variables related to patient time to death.
Results: The univariate analysis identified age, clinical N (lymph node) class, primary tumor site, number of pathologically positive lymph nodes, extracapsular spread, pretreatment screening hematocrit, hematocrit and albumin level at 1 month after treatment, intraoperative fluids and units of blood, and total units of blood as significant (P < .05). The stepwise multivariate models identified independent significance in clinical and pathologic node status, hematocrits at both screening and 1 month, extracapsular spread, albumin level at 1 month, age, and primary site. The position of each variable within the individual regression models varied.
Conclusions: Clinical N class or pathologic node status occupied the first position of predictive significance in all models. Blood transfusion status never retained independent significance in any multivariate assessment.