Studies that claim these prevalences either suffer from massive selection bias (e.g. only use COVID cases that were severe enough to end up in health records [0]) or consider even minor symptoms lasting more than a few weeks (like a persistent cough) as “long COVID.”
Upwards of 80% of the US had been infected with COVID as of June 2022 [1]. If serious long-COVID symptoms indeed occurred in 5-50% of COVID cases, that would mean 4-40% of Americans are currently suffering from debilitating long-term illness, which would be obvious on a societal level. We simply do not see this [2].
Do severe long term COVID cases exist? Absolutely, as is the case with other infections, like influenza [3]. But they are nowhere near as prevalent as otherwise reputable publications claim.
[0] https://arstechnica.com/science/2022/05/more-than-1-in-5-cov...
[1] https://www.healthdata.org/sites/default/files/files/Project...
[2] https://www.theatlantic.com/health/archive/2022/06/long-covi...