Technical variability and required sample size of helminth egg isolation procedures: revisited.

Mes [Vet Parasitol (2003) 115:311-320] recently reported a quantitative study of repeated measurements of nematode egg counts in faecal samples from dairy cattle, in order to compare the faecal egg counts resulting from two different laboratory techniques, the widely used McMaster method and a newer salt-sugar flotation (SSF) method. He concluded that the SSF technique requires much smaller sample sizes, and is also potentially simpler to carry out, making it the method of choice. Here I re-analyse these data to show that if the comparison is done on the most appropriate measurement scale (lognormal), and the large difference in multiplication factors is taken into account, then there is little to choose between the McMaster and SSF techniques as far as the required sample size is concerned. In particular, the treatment of the data as normally rather than lognormally distributed leads to incorrect statistical tests, power analyses and confidence intervals.