Clinical Epidemiology of Emerging Viral Variants

Abstract

The emergence of antigenically distinct viral variants has repeatedly outpaced the capacity of public health systems to characterise their clinical impact in real time. Genomic surveillance platforms now generate sequence data faster than clinical outcome registries can annotate it with meaningful patient-level information. We evaluated five epidemiological surveillance frameworks applied across 74,382 patients with confirmed viral respiratory illness at 22 European sentinel sites between 2020 and 2023, covering six viral variants of concern. Sites were located in Vienna, Rome, Amsterdam, Prague, and Lisbon. Frameworks ranged from passive hospital-based reporting through genomic-linked clinical registries to ML-integrated syndromic surveillance with real-time severity prediction. Performance was assessed by our Epidemiological Surveillance Composite Score (ESCS), integrating variant detection speed, clinical severity characterisation accuracy, population representativeness, policy actionability, and forecasting reliability. The ML-integrated genomic framework achieved the highest ESCS (0.924), detecting severity shifts 8.4 days earlier than passive reporting and predicting ICU admission with AUC 0.89. Hospital-based genomic registries scored 0.901, offering the most complete clinical characterisation. Passive surveillance (ESCS 0.762) systematically underestimated severe outcomes by 34% due to care-seeking bias. Variant transmissibility -- measured by site-adjusted effective reproduction number -- was the strongest independent predictor of healthcare system strain across all frameworks.