Just a note on this. It is true that case figures on CV are very flakey in China and the US. And elsewhere.
It is also true that China is constitutionally unable to deliver good figures for something like this because of a longstanding policy of ignoring and/or punishing those who deliver bad news. That means local and national officials at all levels will be inclined to cover things up.
In the US things are a bit different - there is strong political pressure both locally and nationally to deny the existence, or severity, of the virus, from the Republicans, the Trumpite version of which seems also to distrust science and hence to be happy denying it.
There equally strong pressure to overdo the efefct of CV from the democrat side. the difference is the Democrats generally trust science and scientists, and do not have a strong tendency to lie about it. (Of course, if you truly believe all scientists are, like politicians, just giving opinion, then departing from what science knows may not seem like lying, just like choosing a better opinion).
I'll make an apolitical political point. A site like this that is interested in science should be as appalled as I am by the Trump tendency to deny clearly known science fact. Politicians always twist and color things. Democracies are protected from this by having a bedrock of reasonably accurate truth delivered via journalism and science. The public debate in the US is fast losing that. it is a tragedy, whether you are left-wing or right-wing. And (again apolitical) the political circumstances that understandably lead 85% of the Republican party to go along with this wholesale truth denial are very very bad for the future of US democracy.
Anyway I want to point out that in spite of this most of the large inaccuracy in the case figures everywhere si not political, but due to the inherent inaccuracy of case counts - and that has been true for every new pandemic.
At the start testing is limited, and without widespread testing you miss 90% or more of cases.
Even after the start the people who are actually tested will be a frcation of those who catch the disease for one like this where symptoms of mild disease are indictinguisahable formsymptoms you have with no disease.
Take the UK. Our initial peak wildly underestimated the real case count - we did not have random sample testing then so cannot know by how much except looking at deaths or hospital admissions.
Currently we have good testing, and critically we have high quality random sample population testing throughout the whole country. The random sample testing shows the true rate, the case rate is then about 20% of that (quite difficult to calculate because it depends on how many days an infected person stays test-positive):
Current:
UK case rate (around 25 Nov) 15000/day
random sample positive rate:
22-28 Nov 1 / 105
UK people: 70,000,000
So from case count we get approx .22 / thousand / day
Suppose test positivity for 10 days (?) that translates to 2.2 /thousand from case rate versus 9 / thousand from random sample testing. These figures I'm afraid still not quite right because the random sample testing was community-based, whereas many of the cases occur in hospitals etc.
The only change in death rate stats I remember was to count as CV only deaths within 28 days of a positive test. That was badly needed at the time in order for the death rate to be meaningful.
We were coming out of a high peak and getting about 50% or more of our death rate from those who had caught the virus in the peak. It made looking at the death rate a very poor estimate of the infection rate (3 weeks previously) because of the very large number of positive cases 7 or 8 weeks previously and people dying randomly after a test.
Overall that did not much affect the stats (maybe + 20%) but the temporal affect was larger and the change much needed for death rate to be meaningful.
THH