The role of discriminant functions in screening for beta-thalassaemia traits during pregnancy.

INTRODUCTION

The mean red cell volume (MCV) has been shown to be useful in a 2-stage screening process for beta-thalassaemia traits among pregnant women though associated with a large number of false positive results. We tested prospectively the ability of 5 discriminant functions (DF), (England & Fraser, Shine & Lal, Mentzer, Srivistava and Klee et al) to reduce the number of false positives when used as additional screening determinants for beta-thalassaemia in antenatal patients with red cell microcytosis.

METHODS

The diagnostic performance of each DF was compared in 493 patients with microcytosis and known beta-thalassaemia status. Truth table analysis and Receiver Operation Characteristic curves for each function were determined.

RESULTS

11.4% of the patients with microcytosis were diagnosed to have beta-thalassaemia traits. DFs incorporating the red cell indices: haemoglobin or total red cell count are unsuitable during pregnancy. Shine & Lal's index [(0.01 x MCH x (< MCV)2] reduced the number of people recalled for confirmatory testing by 31.1% and increased the diagnostic yield to 38.7% while maintaining a negative predictive value for the test of 0.993.

CONCLUSION

We conclude that a 3-stage screening process for beta-thalassaemia among pregnant women in Singapore involving the MCV, Shine & Lal's index and a confirmatory test to be both valid and cost-effective.