
This image provided by Pfizer in October 2021 shows the company’s COVID-19 Paxlovid pills. Pfizer via AP
New Australian research suggests the routine use of expensive oral COVID-19 antivirals should be reassessed, concluding that clinical trials have failed to show a significant reduction in hospitalisation or death among vaccinated populations.
The Trial Versus Real-World Gap
However, the UNSW and Monash review notes that the medical landscape has changed significantly.
With high population-level vaccination rates, widespread prior infections creating hybrid immunity, and milder circulating virus strains mean the original trial results no longer reflect the current risk profile.
The UNSW–Monash review examined eight randomised controlled trials—the gold standard for scientific studies—alongside 35 observational studies. The review concluded that among COVID-19 vaccinated populations, clinical trials have “not demonstrated a reduction in hospitalisation or death.”
Observational (real-world) studies painted a somewhat more positive picture for Paxlovid, associating it with a 40 percent reduction in hospitalisation and a 67 percent reduction in death.
However, these studies are considered weaker evidence than randomised trials. Lagevrio showed no significant reduction in hospitalisation and mixed results for averting death even in the observational data.
Shifting Policy and Subsidy Pressures
Currently, antivirals remain heavily subsidised for anyone over 70, people over 50 with two or more risk factors, Indigenous Australians over 30 with one risk factor, and immunocompromised or previously hospitalised adults.
The study’s authors argue that as population risk and case severity continue to fall, Paxlovid’s real-world benefit is becoming very small. Given the lack of robust trial evidence for Lagevrio, they suggest its role in treating COVID-19 should be carefully reconsidered.
With the PBAC scheduled to scrutinise the subsidies later this year, the authors conclude that their findings support a policy reassessment.

