3. Results and discussion
We found that even small misclassification probabilities may result in considerable increase of sample size necessary to reach the prescribed power. Table 2 illustrate dependence of sample size onp 0, Se , and Sp . Full details of the results are given in the supplementary material.