Improved detection of β-thalassaemia carriers by a two-test method

Summary

Seven red cell parameters, taken one at a time and in their 21 possible pairs, were investigated for their power to discriminate between adult carriers of the β-thalassaemia allele and adult normal subjects. The red blood cell count (RBC), haemoglobin concentration (Hb), haematocrit (Hct), mean cell volume (MCV), mean cell haemoglobin (MCH), mean cell haemoglobin concentration (MCHC), and haemoglobin A₂(HbA₂) fraction were measured in 24 obligate heterozygotes and in 99 adult controls with comparable age and sex distributions. Quadratic discriminant functions were computed using Bayesian analysis of univariate and bivariate Gaussian density functions. Classification errors were then calculated by integrating the density function for one genotype over the region assigned to the other.

In the univariate case, MCH led to the lowest cost of misclassification while MCV was the second best discriminant for all posterior probabilities considered. In the bivariate case, MCV combined with percentage Hb A₂ yielded the best discrimination and generated misclassification costs roughly 1/30 of those generated by the most efficient single parameter. When use of MCV alone cannot classify an individual reliably either as a heterozygote or as homozygous normal, combined use of MCV and percentage Hb A₂ is recommended for maximum accuracy.

Application of this screening method to 260 adult subjects at risk for thalassaemia heterozygosity yielded an unbiased frequency of 0.067 for the adult carrier in the Montreal Greek community, a value similar to that reported in the source population in Greece. The improved discriminations thus achieved is particularly useful for sibs of affected subjects whose high prior probability of heterozygosity (0.67) impairs classification.