Which statement best defines paired percent agreement?

• A. The agreement between two tests divided by the total number of pairs where at least one test indicates a positive result.
• B. The agreement between two tests/observers divided by the total number of pairs where at least one test indicates a positive result.
• C. The proportion of observed agreement that occurs by chance.
• D. The total number of positive results agreed upon by both observers.

Answer: (B)

Paired percent agreement measures how often two tests or two observers agree on results when you look at paired observations. For each subject, you have a pair of results from the two sources. To calculate it, take all pairs where at least one result is positive. Among those pairs, count how many times both results are positive (they agree on a positive finding). Then divide that count by the total number of pairs in which at least one result is positive. This gives the proportion of concordant positives among the relevant pairs.

This approach focuses on agreement where a signal exists (positive results) and avoids inflating agreement by counting the many pairs that are negative on both sides. It differs from overall percent agreement, which would include all pairs (including those where both are negative), and from measures that assess agreement by chance or simply tally positive agreements without forming a proportion. For example, if out of 30 pairs with at least one positive, 20 are both positive, the paired percent agreement is 20/30 = 66.7%.