Sensitivity and specificity of surveillance case definitions in detection of influenza and respiratory syncytial virus among hospitalized patients, New Zealand, 2012–2016

Abstract

BackgroundThe WHO is exploring the value of adding RSV testing to existing influenza surveillance systems to inform RSV control programs. We evaluate the usefulness of four commonly used influenza surveillance case-definitions for influenza and RSV surveillance. MethodsSHIVERS, a multi-institutional collaboration, conducted surveillance for influenza and RSV in four New Zealand hospitals. Nurses reviewed admission logs, enrolled patients with suspected acute respiratory infections (ARI), and obtained nasopharyngeal swabs for RT-PCR. We compared the performance characteristics for identifying laboratory-confirmed influenza and RSV severe acute respiratory infection (SARI), defined as persons admitted with measured or reported fever and cough within 10 days of illness, to three other case definitions: 1. reported fever and cough or shortness of breath, 2. cough and shortness of breath, or 3. cough. ResultsDuring April-September 2012–2016, SHIVERS identified 16,055 admissions with ARI; of 6374 cases consented and tested for influenza or RSV, 5437 (85%) had SARI and 937 (15%) did not. SARI had the highest specificity in detecting influenza (40.6%) and RSV (40.8%) but the lowest sensitivity (influenza 78.8%, RSV 60.3%) among patients of all ages. Cough or shortness of breath had the highest sensitivity (influenza 99.3%, RSV 99.9%) but the lowest specificity (influenza 1.6%, RSV 1.9%). SARI sensitivity among children aged <3 months was 60.8% for influenza and 43.6% for RSV–both lower than in other age groups. ConclusionsWhile SARI had the highest specificity, its sensitivity was limited, especially among children aged <3 months. Cough or shortness of breath was the most sensitive.