[Kliment]

The way I read it, applying something with a nonnegligible false positive rate to a much larger population (that is, screening everyone without indications) will lead to many incorrect diagnoses because your error rate is multiplied by a large healthy population. Whereas applying diagnostic tests only to people with suspected a-fib gives you a smaller group with a higher base rate of what you're looking for, so false positives are not as much of a problem.