[M T]

... And the "more robust" analysis you cite says that "All CFR estimates should be viewed cautiously at the current time as the sensitivity of surveillance of both deaths and cases in mainland China is unclear." And their point estimates for CFR based on all infections are under 1%, albeit with wide confidence intervals.

That quote is from a British report published on or before Feb. 10 (based on data up to 8 Feb, reported in news on 10 Feb).   The WHO-China joint mission ended Feb. 24 and would seemingly supersede this, with on-site investigation.

As I read this report, ALL the numbers in this report were based on assumptions on top of a small data sample.  Wuhan (pop. 11M city; 18M metro) as of Feb 25, had 2000+ deaths, 7000+ in hospital currently, and 10,000+discharges.  This report's data for Hubei province (pop 59M, including Wuhan) is based on 29 deaths and 36 recoveries.  That's fine for a back-of-the-envelope, first rough estimate, but not for something one should quote when data is available on a much larger scale.  I wouldn't call it robust.
 
"assuming exponential growth at rate 0.14/day
 assume 5-day period from onset to report and 1-day period from death to report
 assuming infected individuals test positive for 14 days
 assuming infected individuals test positive for 7 days
 This estimate relies on information from just 2 deaths reported outside mainland China thus far and therefore has wide uncertainty.

"Given that the estimates of CFR across all infections rely on a single point estimate of infection prevalence, they should be treated cautiously. In particular,the sensitivity of the diagnostics used to test repatriated passengers is not known, and it is unclear when infected people might test positive, or how representative those passengers were of the general population of Wuhan (their infection risk might have been higher or lower than the general population)."

Because I consider this report to be untrustworthy, I won't quote their CFR for within Hubei, but it was much, much higher than has been mentioned here (for instance, from the WHO).  I consider those numbers to be evidence this report to be worthless.  The numbers they reported under 1% seemingly are even more dependent on assumptions than their numbers for Hubei.

[M T]

... And the "more robust" analysis you cite says that "All CFR estimates should be viewed cautiously at the current time as the sensitivity of surveillance of both deaths and cases in mainland China is unclear." And their point estimates for CFR based on all infections are under 1%, albeit with wide confidence intervals.

That quote is from a British report published on or before Feb. 10 (based on data up to 8 Feb, reported in news on 10 Feb).   The WHO-China joint mission ended Feb. 24 and would seemingly supersede this, with on-site investigation.

As I read this report, ALL the numbers in this report were based on assumptions on top of a small data sample.  Wuhan (pop. 11M city; 18M metro) as of Feb 25, had 2000+ deaths, 7000+ in hospital currently, and 10,000+discharges.  This report's data for Hubei province (pop 59M, including Wuhan) is based on 29 deaths and 36 recoveries.  That's fine for a back-of-the-envelope, first rough estimate, but not for something one should quote when data is available on a much larger scale.  I wouldn't call it robust.
 
"assuming exponential growth at rate 0.14/day
 assume 5-day period from onset to report and 1-day period from death to report
 assuming infected individuals test positive for 14 days
 assuming infected individuals test positive for 7 days
 This estimate relies on information from just 2 deaths reported outside mainland China thus far and therefore has wide uncertainty.

"Given that the estimates of CFR across all infections rely on a single point estimate of infection prevalence, they should be treated cautiously. In particular,the sensitivity of the diagnostics used to test repatriated passengers is not known, and it is unclear when infected people might test positive, or how representative those passengers were of the general population of Wuhan (their infection risk might have been higher or lower than the general population)."

Because I consider this report to be untrustworthy, I won't quote their CFR for within Hubei, but it was much, much higher than has been mentioned here (for instance, from the WHO).  I consider those numbers to be evidence this report to be worthless.  The numbers they reported under 1% seemingly are even more dependent on assumptions than their numbers for Hubei.