ABSTRACT
The frequency distribution of Wuchereria bancrofti microfilariae counts in capillary blood, at a community level, is shown to differ significantly from the log-normal distribution, commonly used (and recommended by WHO). A much better fit is obtained with a truncated negative-binomial distribution, the k exponent of which is estimated to 0.3. This value does not seem to be affected by the endemic level, by microfilaricide mass-treatments or by the blood sample volume. This result has immediate practical interests: 1) it facilitates a posteriori comparisons between surveys in which different volumes of blood were examined; 2) at a given density it allows calculation the change in sensitivity (detection rate) as a function of a change in the volume of blood examined, thus allowing better a priori planning of surveys. The paper provides the formulae required to calculate k, or, given k, to calculate the expected numbers in the various density classes. It also provides nomograms (fig. 4a & b) giving estimates of the true mean M and of the detection rate TD, given any one of the three following observations, made on observed positives: the median Dmf. 50, the mean m, or the percentage Pl having only 1 mf. The effect of changing the volume of blood on the rate of detection can then be read easily from the nomogram. These results, likely to apply to other important parasitic diseases, should instigate more elaborated statistical research in this direction.
