Comparing two case definitions for including severe acute respiratory infection patients in influenza and COVID-19 vaccine effectiveness studies among older adults using the test-negative design: VEBIS hospital network, seven European countries, 2023/24.
Abstract
INTRODUCTION Two severe acute respiratory infection (SARI) case definitions (CDs) are often used in test-negative design (TND) hospital-based vaccine effectiveness (VE) studies. SARI-CD1 requires ≥1 symptom (cough, fever, shortness of breath, anosmia, ageusia, or dysgeusia); SARI-CD2 requires both fever and cough. Using VEBIS hospital network end-of-season 2023/24 data, we compared the effect of these CDs on influenza and COVID-19 VE estimates in Europe.
Methods
Using the TND, we restricted analysis to seven countries recruiting patients using SARI-CD1; we created a subset of SARI patients also meeting SARI-CD2. Cases were RT-PCR-positive for influenza A(H1N1)pdm09, A(H3N2), or SARS-CoV-2; controls were influenza or SARS-CoV-2 RT-PCR-negative. We estimated VE among patients ≥65 years using logistic regression adjusted for country, onset date, age, sex, and chronic conditions. We estimated VE overall, by age group (65-79, ≥80 years) and time since vaccination (14-59, 60-119, ≥120 days). Absolute VE differences ≥10% were considered relevant.
Results
Patient characteristics and outcomes were similar across CDs. SARI-CD1 had slightly more vaccinated cases, with little difference in controls between the two CDs. For influenza, the absolute VE difference between SARI-CD1 and SARI-CD2 was 10% for influenza A(H1N1)pdm09 (34% vs 44%), 19% among those aged ≥80 years (21% vs 40%) and 18% at 60-119 days post-vaccination (4% vs 22%). For SARS-CoV-2, absolute VE differences were 11% overall (29% vs 40%), 11% among ≥80 years (33% vs 44%), and 12% at 60-119 days (34% vs 46%). All other absolute VE differences were < 10%.
Conclusions
VE was similar for influenza A(H3N2), adults 65-79 years and 14-60 days post-vaccination. We observed signals of higher VE with SARI-CD2 for influenza A(H1N1)pdm09 and SARS-CoV-2, particularly among adults ≥80 years and ≥ 60 days post-vaccination, although 95% CIs overlapped. These signals suggest CDs may contribute to VE variability and VE studies should clearly indicate CDs used.