epi.herdtest {epiR}

Estimate the characteristics of diagnostic tests applied at the herd (group) level

Description

When tests are applied to individuals within a group we may wish to designate the group as being either diseased or non-diseased on the basis of the individual test results. This function estimates sensitivity and specificity of this testing regime at the group (or herd) level.

Usage

epi.herdtest(se, sp, P, N, n, k)

Arguments

se

a vector of length one defining the sensitivity of the individual test used.

sp

a vector of length one defining the specificity of the individual test used.

P

scalar, defining the estimated true prevalence.

N

scalar, defining the herd size.

n

scalar, defining the number of individuals to be tested per group (or herd).

k

scalar, defining the critical number of individuals testing positive that will denote the group as test positive.

Value

A list with one scalar and two data frames.

Scalar sfraction reports the sampling fraction (i.e., n / N). The binomial distribution is recommended if sfraction is less than 0.2.

Data frame dbinom lists APpos the probability of obtaining a positive test, APneg the probability of obtaining a negative test, HSe the estimated group (herd) sensitivity, and HSp the estimated group (herd) specificity calculated using the binomial distribution.

Data frame dhyper lists APpos the probability of obtaining a positive test, APneg the probability of obtaining a negative test, HSe the estimated group (herd) sensitivity, and HSp the estimated group (herd) specificity calculated using the hypergeometric.

Author(s)

Ron Thornton, MAF New Zealand, PO Box 2526 Wellington, New Zealand.

References

Dohoo I, Martin W, Stryhn H (2003). Veterinary Epidemiologic Research. AVC Inc, Charlottetown, Prince Edward Island, Canada, pp. 113 - 115.

Examples

## EXAMPLE 1:
## We want to estimate the herd-level sensitivity and specificity of 
## a testing regime using an individual animal test of sensitivity 0.391 
## and specificity 0.964. The estimated true prevalence of disease is 0.12.
## Assume that 60 individuals will be tested per herd and we have
## specified that two or more positive test results identify the herd
## as positive.

epi.herdtest(se = 0.391, sp = 0.964, P = 0.12, N = 1E06, n = 60, k = 2)

## This testing regime gives a herd sensitivity of 0.99 and a herd
## specificity of 0.36 (using the binomial distribution). With a herd 
## sensitivity of 0.95 we can be confident that we will declare a herd 
## as disease positive if it truly is disease positive. With a herd specficity 
## of only 0.36, we will declare 0.64 of disease negative herds as infected, 
## so false positives are a problem.