Abstract
After the exclusion of iron deficiency and β-thalassemia, molecular research for α-thalassemia is recommended to investigate microcytic anemia. Aiming to suggest more efficiently the molecular analysis for individuals with a greater chance of having a symptomatic form of the disease, we have developed and validated a new decision tool to predict the presence of two or more deletions of α-thalassemia, increasing considerably the pre-test probability. The model was created using the variables: the percentage of HbA2, serum ferritin and mean corpuscular volume standardized by age. The model was trained in 134 patients and validated in 160 randomly selected patients from the total sample. We used Youden’s index applied to the ROC curve methodology to establish the optimal odds ratio (OR) cut-off for the presence of two or more α-globin gene deletions. Using the OR cut-off of 0.4, the model’s negative predictive value (NPV) was 96.8%; the cut-off point accuracy was 85.4%; and the molecular analysis pre-test probability increased from 25.9% to 65.4% after the use of the proposed model. This tool aims to assist the physician in deciding when to perform molecular studies for the diagnosis of α-thalassemia. The model is useful in places with few financial health resources.