The sample size matters: evaluating minimum and reasonable values in prevalence studies

Estimating sample size is important for prevalence studies, as it directly influences the validity of the research outcomes. Our objective was to highlight constraints in the prevalence assessment and to provide guidance on the delineation of minimum and reasonable sample size. We also assess the prevalence properties as a function of sample size visualizing the median prevalence, confidence intervals, precision, and changes in precision. Constraint analysis indicates that a sample size of less than 15 host individuals will likely result in unacceptable precision in the most cases. Because the prevalence estimate accuracy depends on both sample size and the estimate itself, the minimum sample size may vary widely, from 16 to over 450 individuals, when the prevalence is between 1% and 99%. A sample size of 16-45 elements can be used as a minimum for estimating true prevalence between 10% and 90% with an acceptable precision. However, caution should be exercised with a such small sample size as the prevalence will have a high degree of uncertainty. A simple, practical suggestion for selecting a minimum sample size is to sample until at least 5 infected (cases) and 5 uninfected (non-cases) hosts are detected. This approach is effective in most situations, except in cases of extreme prevalence (1% or 99%). The design of a reasonable sample size should be based on a flexible strategy that takes into account the study objectives, available resources and desired precision. This strategy may based on finding the plateau phase within the precision or confidence intervals curves. As the uncertainty in prevalence decreases rapidly with increasing sample up to 110-135 individuals, but not much more with further increasing sample efforts, opting for a sample size exceeding this threshold, could be considered an optional choice within the prevalence range of 5-95%. We advise authors, editors and reviewers to track sample size in conjunction with the actual prevalence of the parasites and other pathogens. If the minimum sample size is unattainable, authors should acknowledge this limitation, as all data contribute to understanding parasite distribution.