J. Kotwal et al., Erythrocyte indices for discriminating thalassaemic and non-thalassaemic microcytosis in Indians, NAT MED J I, 12(6), 1999, pp. 266-267
Background. Microcytosis is a common red cell change seen in anaemias of va
ried aetiology, These include iron deficiency, thalassaemia, chronic diseas
e and sideroblastic anaemias. The microcytosis of heterozygous beta-thalass
aemia needs to be distinguished From non-thalassaemic microcytosis for its
role in thalassaemia control, Red cell indices derived from automated red c
ell analysers have been used to discriminate between microcytic patients wi
th a high probability of thalassaemia minor from those with a low probabili
ty, There is a controversy on the choice of red cell indices to be used and
the cuts-off values for this distinction, because the prevalence of iron d
eficiency as a cause of nonthalassaemic microcytosis is variable. Since no
Indian study using receiver operator characteristic (ROC) curves was availa
ble to determine the above, we conducted this study.
Methods. Red cell indices (mean corpuscular volume, total red blood cell co
unt, red cell distribution width, linear discriminant function), serum iron
, total iron binding capacity and haemoglobin A(2) were estimated in 640 ad
ults with microcytosis (mean corpuscular volume 80 fl). The ROC curves were
plotted in all.
Results. Total red blood cell count was observed to be the most efficient s
ingle test followed by linear discriminant function and Bessman index. Mean
corpuscular Volume had the least efficacy. The cut-off values obtained for
the Indian population were mean corpuscular volume less than or equal to 7
6 fi, total red blood cell count greater than or equal to 4.9x10(12)/L and
red cell distribution width greater than or equal to 18% and a positive lin
ear discriminant function. These were different from those observed in the
West, possibly because of the high prevalence of iron deficiency in India,
Conclusion. In countries with a high prevalence of iron deficiency, cut-off
values for red cell indices should be recalculated using ROC curves.