Article Text
Abstract
Introduction The aim of border screening for influenza is to prevent or delay the entry of a new pandemic strain, but the evidence base for this strategy is limited. This study aimed to assess the test characteristics of a health questionnaire and temperature measurement for predicting influenza infection in arriving international travellers.
Methods A questionnaire was distributed on board flights from Australia to Christchurch, New Zealand during 12 weeks in June–September 2008. It included questions on demographics, symptoms, contacts, and countries visited. All symptomatic travellers and a random sample of asymptomatic travellers were asked to provide a throat swab and have their temperature measured.
Results 175/307 (57%) flights were screened according to protocol, and 15 618/22 192 (70%) travellers on screened flights returned questionnaires. 48% of symptomatic travellers and 35% of sampled asymptomatic travellers agreed to provide a swab. The overall prevalence of influenza infection was 1%. The sensitivity for influenza infection of ‘any symptom’ was 85%, for cough 58%, sore throat 26%, self reported fever 15%, and for measured temperature >37.8°C it was 8%. The highest PPV was for self-reported fever, at 23%.
Conclusion The poor sensitivity of most screening questions or tests and low prevalence of influenza infection among travellers means that border screening would be resource-intensive yet would fail to identify all, or possibly even most, influenza infected travellers entering the country.