Generation-time distortion is an emergent property of epidemic control

We find that a vast space of public health interventions necessarily reshape the intervals between successive infections, known as generation times, for directly-transmitted infectious diseases (e.g., COVID-19, Ebola virus, influenza). While some studies have connected specific interventions (e.g., contact-tracing, isolation) to shifted generation times, most epidemiological analyses assume these intervals are stationary. We reveal that this assumption is substantially more fragile than previously recognised. We develop a universal microscopic representation for all intervention mechanisms that reduce individual-level infectiousness and establish generation-time reshaping as a fundamental macroscopic feature of those interventions, emerging at the population level. By constraining the topology of our representation to reflect specific mechanisms, we explain and unify previous studies as well as predict generation-time phenomena yet to be observed. We further show that neglecting generation-time reshaping can under- or over-estimate intervention effectiveness and disease control- lability by 100%, under realistic parameter settings. A stringent intervention can even produce no estimated transmissibility difference. Generation-time distortion causes unavoidable and underrated biases that can severely limit model-driven public health policy and pandemic preparedness.