pertussis, validation, diagnostic accuracy, sampling strategy, partial verification bias, low prevalence