Iron-deficiency anemia and thalassemia trait differentiated by simple hematological tests and serum iron concentrations

Abstract

The diagnostic value of common hematology tests in differentiating between thalassemia trait and iron-deficiency anemia (IDA) was assessed. Serum iron concentration was included in a second stage of the diagnostic model and its predictive contribution was determined. After applying discriminant analysis to the data, minimization of Wilks's lambda (Λ) criterion was used to select the best predictive variables. A training sample of 754 subjects previously classified as either IDA (428) or thalassemia trait (326) was used to determine the classification rule. When serum iron concentrations were included, the model showed a higher predictive capacity than that constructed from hematological variables only (D = 2.342 RBC - 0.079 Hb + 3.627 log[Fe] - 6.459, where D = discriminant score). The sensitivity, specificity, and diagnostic efficiency associated with this model, as assessed on a control sample of 256 patients (137 thalassemia and 119 IDA), were 94.2%, 91.6%, and 93.0%, respectively. This model, when compared with those of other authors, has the highest diagnostic efficiency.