Attempt to understand differences πŸ‡¬πŸ‡§/πŸ‡©πŸ‡ͺ #Covid19 experience. Data comparability is challenging, and inevitably some of this may need further scrutiny. However, I do think there are a number of issues which warrant a discussion

1/16 (sorry)
Comparing some key indicators (πŸ‡¬πŸ‡§/πŸ‡©πŸ‡ͺ):

Cases +test
220/173k

Gender
M 56/48%
W 48/52%

# ICU beds used at peak
54%~3.3k (England)/<10% ~ 2.8k
Note: πŸ‡©πŸ‡ͺ 5x #ICU beds

Covid in πŸ‡¬πŸ‡§ hospital beds at peak 20k
πŸ‡©πŸ‡ͺ had 5k of Covid+ hospitalised at peak (not the same as beds)

2/16
Mortality in ICU
46/28% (!)

Ventilated in ICU
64/66%

% of all death
>70yrs 86/86%

Care home deaths
25/33%

3/16
Excess death (see @ft @jburnmurdoch and @ChrisGiles_ for great analysis generally)

61% πŸ‡¬πŸ‡§/6% πŸ‡©πŸ‡ͺ
50k/4.8k

i.e. without Covid πŸ‡©πŸ‡ͺ would have had negative excess deaths(?) For πŸ‡¬πŸ‡§ depends on how many of the excess deaths are actually Covid but untested
4/16
Conclusions:

πŸ‡¬πŸ‡§ has 46% higher ICU and 10x excess mortality and more patients in ICU. Age and gender distribution broadly similar. πŸ‡©πŸ‡ͺ has fewer cases even when compared to low testing levels in πŸ‡¬πŸ‡§ making it likely that true πŸ‡¬πŸ‡§ number much higher.
5/16
Particularly likely given that only 20% of πŸ‡©πŸ‡ͺ cases needed hospital where most πŸ‡¬πŸ‡§ testing took place.

πŸ‡¬πŸ‡§ therefore likely to have seen much worse outbreak.

Potential hypotheses:
6/16
a) πŸ‡©πŸ‡ͺ contained #Cvoid19 better/earlier and had fewer cases (πŸ‡¬πŸ‡§ has higher cases even with low testing and potentially more in hospital). One conclusion might therefore be πŸ‡¬πŸ‡§ simply has more cases, end of it.
7/16
b) πŸ‡¬πŸ‡§ had a more severe virus while πŸ‡©πŸ‡ͺ had milder and more asymptomatic cases – not sure there is much evidence for different severity of virus itself and we don’t yet know antibody levels at pop level though various studies underway
8/16
c) πŸ‡©πŸ‡ͺ has fewer high-risk groups - BMI (πŸ‡¬πŸ‡§ ranks 36, πŸ‡©πŸ‡ͺ 76 in world), BAME (πŸ‡©πŸ‡ͺ has hardly any) lower density (237/275 sqkm) London 4.5k/Berlin 3.9k sqkm; but πŸ‡©πŸ‡ͺ has higher per capita/km use of public transport (1.9/1.6k km/yrs)…
9/16
d) differences in treatment – need to know hospitalisation rates which we don’t have (people in πŸ‡©πŸ‡ͺ and beds occupied in πŸ‡¬πŸ‡§) to understand ICU ratios. E.g. πŸ‡©πŸ‡ͺ has 5x ICU beds, are more patients going to ICU earlier? 11/16
However, utilisation in πŸ‡¬πŸ‡§ at peak was only 50% of available i.e. no constraint. We also don’t have figures on nurse ratios though significant vacancy rate in πŸ‡©πŸ‡ͺ.
12/16
e) care home issue. πŸ‡©πŸ‡ͺ has a very low excess mortality rate (6%). RKI estimates at least 1/3 of deaths occur in care homes, in UK~25%. However,πŸ‡©πŸ‡ͺ&πŸ‡¬πŸ‡§ suspect real # is higher. This doesn’t explain diff in ICU mortality and potentially relatively high hospitalisation rates.
13/16
On balance, qs comes down to why are πŸ‡¬πŸ‡§ overall infection rates higher? Answer most likely a combination of late lockdown=more spread, ethnic mix, and susceptibility of underlying health issues (e.g BMI). Little evidence on differences in treatment. Needs deeper analysis.
14/16
All in all and given available data, UK seems to have a worse outbreak than Germany

15/16
Additional data needed mainly for πŸ‡©πŸ‡ͺ:

Covid hospitalisation
Nurse ratios in ICU
Ventilation levels
Mortality by occupation
Immunity level via antibody test

@kakape @DIVI_eV @rki_de anything you can help with?

16/16
Final obs on data availability. Huge advantages in having central bodies such as @ONS and @ICNARC making downloadable data available but still gaps and quality issues. @DIVI_eV and @rki_de doing a sterling job but much harder in fragmented πŸ‡©πŸ‡ͺ system and lots we don’t have.
END
You can follow @axelheitmueller.
Tip: mention @twtextapp on a Twitter thread with the keyword β€œunroll” to get a link to it.

Latest Threads Unrolled: