Velindre NHS Trust, Cardiff, UK.
OBJECTIVES: To develop a hitherto unavailable risk factor model for accurately predicting anemia development in cancer patients before chemotherapy (CT) administration. METHODS: 2,070 nonanemic patients from the European Cancer Anaemia Survey (ECAS) with hemoglobin (Hb) > or =12 g/dl at enrollment who received their first CT during ECAS and underwent at least two CT cycles were divided randomly into split half (SH) 1 and SH2 (n = 1,035 each). The model was developed on SH1 using logistic regression to simultaneously evaluate predictive factors, and was validated using SH2 and an additional similar subpopulation of 5,901 ECAS patients. Anemia risk values were assigned to the predictive factors and the sum of the predictive factors gave the total anemia risk score; lower-, higher-, and highest-risk cutoff points of the total anemia risk score were determined. RESULTS: Variables ultimately identified as significant predictive factors for anemia were: lower initial Hb (< or =12.9 g/dl in females, and < or =13.4 g/dl in males); having lung or gynecologic cancer versus gastrointestinal (GI)/colorectal cancer; cancer at any other site versus GI/colorectal cancer; treatment with platinum CT, and female gender. CONCLUSION: Using this evidence-based risk model, nonanemic patients who are at the highest risk of developing anemia prior to receiving CT can be identified clinically, allowing appropriate anemia management to be planned