Abstract:
Screening for β-thalassemia trait (βTT) is crucial for preventing β-thalassemia major
in offspring. Although hematological discrimination formulae (HDF), developed using complete
blood count parameters, are cost-effective tools for screening βTT, their performance varies
across different populations. This study evaluated the performance of 32 HDF for screening βTT
in the Sri Lankan population. Data were retrieved from laboratory databases and categorized into
confirmed βTT and non-βTT groups based on high-performance liquid chromatography results.
The βTT screening performance of the HDF was assessed using accuracy measurements, the
receiver operating characteristic (ROC) curve, and Youden’s index (YI). Furthermore, a
population-adjusted cut-off was determined using the Index of Union (IU) method to optimize
the predictive accuracy of HDF in screening βTT. Bordbar demonstrated excellent predictive
performance in males (AUC = 0.908; YI = 0.815), while Shine & Lal, Kerman-I, Nishad, Sehgal,
Bordbar, and Roth demonstrated high discriminative ability in females (AUC > 0.833; YI >
0.666). Applying a population-adjusted cut-off improved the βTT screening potential of Shine &
Lal, Kerman-I, Nishad, Bordbar, and Roth in males (AUC > 0.911; YI > 0.822) and enhanced
the performance of Kerman-II in females (AUC > 0.861; YI > 0.722). Notably, Shine & Lal
(AUC = 0.937; YI > 0.873) and Nishad (AUC = 0.897; YI > 0.794) demonstrated the best
performance for males and females, respectively, when a population-adjusted cut-off was applied
for screening βTT. In conclusion, determining a population-adjusted cut-off is a new initiative to
enhance the βTT screening performance of HDF across different populations.
