Negida et al reported a sample size calculation for a study of diagnostic test accuracy based on sensitivity and specificity.
Situation: determining accuracy of a diagnostic test
Parameters:
(1) sensitivity
(2) specificity
(3) prevalence in population (p)
Assumptions:
(1) 95% confidence interval for the normal distribution, with Z value of 1.96
(2) maximum acceptable width (W) set to 10%
(3) p value set to 0.05
TP + FN =
= Z^2 * (sensitivity) * (1 - (sensitivity)) / W^2
TN + FP =
= Z^2 * (specificity) * (1 - (specificity)) / W^2
sample size required for sensitivity =
= (TP + FN) / (p value)
sample size required for specificity =
= (TN + FP) / (1 - (p value))
total required sample size =
= (sample size for sensitivity) + (sample size for specificity)
