The application of faecal egg count results and statistical inference for clinical decision making in foals

This study investigated the impact of variability in Parascaris spp. and strongyle faecal egg counts (FEC)from foals on treatment decision-making and detection of a patent infection. A single faecal sample was collected once daily for three days from 53 foals and a FEC was performed on three separate portions of each sample (total of nine egg counts per foal). Differences in the decision to administer an anthelmintic using the results of a single count (C 1 ), the mean of three (X¯ 1–3 )or nine counts (X¯ 1–9 )and the upper 5% confidence limit of the gamma confidence interval (CI)of the estimate of the distribution mean (μ)from three (UCL 1–3 )and nine counts (UCL 1–9 )were determined for a range of egg count thresholds. The UCL 1–9 was used as the best estimate of μ, hypothesis testing for treatment and the comparison of treatment decision-making using C 1 , X¯ 1–3 , X¯ 1–9 and UCL 1–3 . The results of this study demonstrated that a point estimate (C 1 or X¯ 1–3 )was of limited value for estimating the distribution mean of egg counts in faeces and there was overall poor agreement in treatment decision-making for individual foals using C 1 compared with UCL 1–9 . Of the foals with C 1 of zero eggs per gram, 54% and 47% had Parascaris and strongyle eggs in subsequent counts, respectively. The egg density in faeces is inhomogeneous, resulting in considerable variability in egg count results for an individual foal: between faecal piles, different portions of a faecal pile and days. The use of the negative binomial distribution CI for μ takes this variability into account and is recommended for use when interpreting FEC data from horses.