Use of Nirmatrelvir-Ritonavir in Acute SARS-CoV-2 Infection and Implications for Long COVID Prevention

Nirmatrelvir-ritonavir (NMV/r, Paxlovid) is an oral antiviral demonstrated experimentally to reduce COVID-19-related hospitalization and mortality when administered early after SARS-CoV-2 infection. However, existing observational evidence on NMV/r has been constrained by coarse definitions, limited assessment of initiation timing and treatment completion, and insufficient causal evaluation of how acute-phase NMV/r use influences subsequent long COVID outcomes. To fill critical gaps, this dissertation integrates granular longitudinal measurement of NMV/r treatment in a prospective cohort study with causal target trial emulation to understand real-world patterns of NMV/r and the effect of acute-phase NMV/r treatment on long COVID prevention.In Aim 1, we characterized NMV/r initiation and completion among symptomatic adults and described heterogeneity across demographic and clinical subgroups. Fewer than half of participants at high risk for progression to severe COVID-19 and approximately one quarter of standard-risk participants initiated NMV/r within 5 days of symptom onset. Initiation and completion were consistently lower among younger adults, female, non-White/Caucasian participants, Hispanic/Latinx participants, individuals who were vaccinated more than 6 months earlier or unvaccinated, or those with prior known COVID-19 episode(s). In Aim 2, we emulated a sequence of nested target trials to estimate per-protocol causal effects of NMV/r use during acute infection on long COVID onset among high-risk participants. Using g-computation and inverse probability weighting, timely initiation within 5 days of symptom onset and completion of full-course treatment as prescribed were associated with modest reductions in long COVID risk through 72 weeks of follow-up, with slightly larger absolute risk reductions for initiation within 3 days. These findings support three overarching conclusions. First, NMV/r initiation and completion vary across host characteristics and access to care. Second, substantial gaps exist between prescribing recommendations and observed practice, with many high-risk individuals remaining untreated and considerable use occurring outside labeled indications. Third, NMV/r alone was unlikely to substantially reduce the risk of long COVID. By combining detailed treatment measurement with rigorous causal inference, this dissertation provides actionable evidence to inform COVID-19 clinical guidance, public health planning, and ongoing investment in longitudinal surveillance systems needed to guide responses to COVID-19 and future pandemics.