研究者算错了。JAMA 的审稿人也是吃干饭的。美国人的数学：（， 已经更正了。 （见内）

https://www.bloomberg.com/news/articles/2020-04-22/almost-9-in-10-covid-19-patients-on-ventilators-died-in-study

......

Two days after releasing the study, the Journal (JAMA) issued a clarification, deleting the 88% figure from the abstract and saying that most ventilator patients in the study were still hospitalized and that the mortality rate is expected to come down as more results come in. The clarification also stated that 24.5% of the total number of patients who had received ventilator treatment had died, while 3.3% survived and were discharged. The rest, or 72.2%, remained hospitalized as of April 4.

所有跟帖： 

• 若按此比例，入院上呼吸机3.3%存活，24.5%死亡，存活/死亡=13.4%，就是说死亡比86.6%。 -pickshell- ♀ (602 bytes) () 05/09/2020 postreply 05:39:18

• Data massage -fuz- ♂ (0 bytes) () 05/09/2020 postreply 07:32:15

• 哈哈，massage下%大变：） -pickshell- ♀ (0 bytes) () 05/09/2020 postreply 08:18:11

• 我们上学的时候处理试验数据这招很常用啊，否则文章没法写了 -ephd- ♂ (378 bytes) () 05/09/2020 postreply 18:02:46

• 嗯，没造假，真实的也挑不出毛病，可稍揉捏下能呈现多种不同结果。 -pickshell- ♀ (0 bytes) () 05/10/2020 postreply 07:54:28