Oh, that sounds really interesting. Thanks. That was not what I understood from the aforementioned study and obviously that changes the figures a lot (by a factor 4). Too bad it's behind a paywall :/
Taking that in account, out of the 45k who died in hospital, 12k died after being admitted in ICU. In the meantime, ICU saved 32k people (73%)[1], which is indeed far lower than my previous estimation, I stand corrected.
[1]: https://drees.solidarites-sante.gouv.fr/IMG/pdf/dd67.pdf (page 17) which is the document quoted in your article.