Sample size to estimate a true prevalence with an imperfect test
This utility calculates the sample size required to estimate true prevalence
with a specified level of confidence and precision, assuming a test with imperfect
sensitivity and/or specificity. The same method applies for estimating both animal and
herd-level prevalence, with herd-sensitivity and herd-specificity substituted for
animal-level values to estimate true herd-prevalence. The method is as described by:
Humphry RW, Cameron A, Gunn GJ, 2004. A practical approach to calculate sample size for
herd prevalence surveys. Prev. Vet. Med.
65: 173-188. Adjustment for finite
population size is described by Thrusfield M, 2005. Veterinary Epidemiology, 3rd Edition,
Blackwell Science, Oxford, UK (p 233-234).
Inputs are the assumed true prevalence, the desired level of confidence,
the desired precision of the estimate and the assumed values for sensitivity and
specificity of the testing regimen used. The desired precision of the estimate
(also sometimes called the allowable or acceptable error in the estimate) is half the width of the desired
confidence interval. For example if you would like the confidence interval width to be about
0.1 (10%) you would enter a precision of +/- 0.05 (5%).
To calculate sample size for herd-prevalence estimation, use herd-level values for
assumed prevalence, sensitivity and specificity instead of animal-level values.
Sample size is calculated for an assumed large (infinite) population. If the optional
population size is provided the sample size estimate is adjusted for the population specified.
Adjustment for finite population size may underestimate required sample size unless
this is also taken into account when estimating variance and resulting confidence interval.
The program outputs the sample size required to estimate the true prevalence with the
desired precision and confidence. Tables of sample sizes for a range of values for prevalence
and precision and for sensitivity and specificity are also produced.