INTRODUCTION AND OBJECTIVE: To evaluate the performance of red cell distribution width reported statistically as coefficient of variation (RDW-CV), standard deviation (RDW-SD), and mathematical deduction of 1 standard deviation (SD) around mean corpuscular volume (MATH-1SD) in identifying anisocytosis in automated blood counts when compared with the manual quantification of erythrocyte anisocytosis in peripheral blood smears. MATERIAL AND METHODS: 806 routine samples obtained from the hematology laboratory of Hospital de Clínicas da Universidade Federal do Paraná (HC-UFPR) were analyzed. Performance evaluations were carried out by dividing samples into microcytic, normocytic and macrocytic mean corpuscular volume (MCV). For each MCV range, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and efficiency were calculated. In addition, the Youden index (Y) was obtained and a comparative analysis with receiver operating characteristic (ROC) curves was done to evaluate the performance of RDW-SD, RDW-CV, and MATH-1SD on different MCV ranges. RESULTS AND DISCUSSION: RDW-CV had the best sensitivity (86.8%) and efficiency (86.8%) in detecting anisocytosis in microcytic MCV ranges. RDW-SD and MATH-1SD were more sensitive and efficient in normocytic (82.9% and 83.3%; 92.1% and 92.3%, respectively) and macrocytic (90.2% and 90.2%; 95.1% and 95.1%, respectively) MCV ranges. A ROC curve analysis indicated that RDW-CV was more efficient in detecting anisocytosis in microcytic MCV ranges (p < 0.05 vs. RDW-SD and MATH-1SD). In normocytic and macrocytic MCV ranges, RDW-SD and MATH-1SD showed similar efficiency in detecting anisocytosis (p < 0.05 vs. RDW-CV). CONCLUSION: RDW-SD, RDW-CV, and MATH-1SD deliver different performances in detecting blood smear anisocytosis according to MCV values. They are parameters that complement each other and should be used together to identify erythrocyte size heterogeneity.
laboratory automation; RDW-CV; RDW-SD; mean corpuscular volume; cellular analysis
