The effect of counting duration on quantitative fecal egg count test performance

Fecal egg counts are the primary diagnostic tools of equine parasitology and use of the McMaster test and its variants in clinical practice is widely recommended. Manual counting is, however, prone to various sources of human error. For example, in real-world situations analysts can be under significant pressure to process high numbers of samples in a limited time. This practice could affect test result quality, but yet no studies have determined whether this is the case. This study’s purpose was to assess the effect of shortened counting duration (from either restricting counting time or counting only one grid of a slide) on McMaster test performance, and to compare the results to those of an automated test whose output is not subject to such limitations. Fifteen fecal samples from horses infected with strongylid parasites were divided equally into three groups based on high, medium and low levels of egg content (201–500, 501–1000 and 1001+ eggs/g). Slurries were produced from each sample and 10 subsamples of each were counted by both the McMaster and automated methods. McMaster slides were first counted at leisure, and then twice again with counting time being restricted to either one or two min. The effect of reducing sample processing time by counting only one grid of the McMaster slide was also assessed. Counting for one min significantly decreased manual egg counts by 50–60% relative to counts conducted at leisure (p < 0.001). While these decreases were somewhat ameliorated by counting for two min, the results were still approximately 10% lower than the at-leisure counts, a difference that was also statistically significant (p < 0.001). Furthermore, restricted counting duration also resulted in a significant decrease of approximately one-third in McMaster test precision, as assessed by the coefficients of variation (CoVs) of the 10 replicates of each sample, as did counting just a single grid of the McMaster slide. These differences effectively further improved the observed superior precision of the automated method compared to at-leisure manual McMaster counting, and the automated counts and their precision remained relatively unaffected following multiple analyses of the same processed samples. Taken together, these results indicate that analysists should carefully assess the possible effects on test performance of modifications to standard egg-counting procedures that are designed to account for real-world pressures, in order to achieve an optimal compromise between test accuracy and precision on one hand and practicality on the other.