search

owid / covid-19-data

Data on COVID-19 (coronavirus) cases, deaths, hospitalizations, tests • All countries • Updated daily by Our World in Data
https://ourworldindata.org/coronavirus
5.66k stars 3.64k forks source link

data: overshoots in covid-attributed death count for Chile, Ecuador, Colombia #2795

Open czka opened 1 year ago

czka commented 1 year ago

coronavirus-data-explorer

Country

Chile, Ecuador, Colombia

Domain

Deaths

Which data is inaccurate or missing?

Please refer to https://ourworldindata.org/explorers/coronavirus-data-explorer?facet=none&Metric=Confirmed+deaths&Interval=7-day+rolling+average&Relative+to+Population=false&Color+by+test+positivity=false&country=ECU~CHL~COL (screenshot attached).

On that chart, the covid death count for Chile around 22.03.2022 greatly exceeds the all-cause mortality in that period. Same thing for Ecuador around 21.07.2021 and 07.09.2020.

Aditionaly, I noticed 3 less severe, but still artifficially-looking spikes: Chile ~18.07.2020, Colombia ~09.05.2020, Ecuador ~13.01.2022.

Why do you think the data is inaccurate or missing?

As explained above.

Are these issues a matter of processing on the OWID's side, or do they originate from the WHO dataset you have switched to recently? As I compare with my charts rendered from the OWID's data before 08.03.2023, these spikes are something new.

lucasrodes commented 1 year ago

Hi @czka, Thanks for reporting!

I just checked the source, and this seems to come from them. You can see this in their chars in the following links (scroll down once on their site):

I just contacted them to understand these values better. You can also contact them (more info at https://covid19.who.int/info, under the "Contact us" section).

czka commented 1 year ago

Hey @lucasrodes

Thanks for clarifying and forwarding this to the source. I suppose an OWID team member will get a better reply from WHO than an average Internet being like me ;), so let's wait to see what they tell you.

I'd love to know what others think, but for me it's surprising to see such huge, singular differences between Johns Hopkins University and WHO covid death counts, which are otherwise quite similar. Has OWID considered assessing those 2 data sources from the perspective of such annomalies? Because maybe there are more such cases besides those that I spotted (while updating my charts).

lucasrodes commented 1 year ago

Hi @czka, Thanks for your thoughts and insights.

Before the source switch, we ran several tests comparing both sources to ensure no significant differences. Back then, we did not observe any of these singularities in these countries, which is aligned with what you say:

As I compare with my charts rendered from the OWID's data before 08.03.2023, these spikes are something new.

We are currently not making such comparisons, but I think that we should. I'll try to come up with some weekly JHU-WHO sanity checking so we can report any anomalies to the WHO.

czka commented 1 year ago

@lucasrodes Hello, it's been a while 👋🏿. Are there any chances this issue could be fixed?

lucasrodes commented 1 year ago

Hi @czka, Thanks for your message.

We reported the issues to the WHO, but we did not get any response.

We are reporting values for confirmed deaths and cases as reported by the WHO, without any alterations. We do not plan on adding any processing on our side.

We have reported this incident again to the WHO, but please feel free to report this yourself as you encounter new issues.

lucasrodes commented 1 year ago

Hi @czka, This is what the WHO replied:

The spikes observed are due to changes in either case definitions of COVID-19 deaths or the exclusion of probable deaths by the respective Member States.

See below for links to the information.

Chile

Ecuador: