search

ccodwg / CovidTimelineCanada

A definitive dataset for COVID-19 in Canada
https://opencovid.ca/
Other
27 stars 11 forks source link

Verify QC hospitalization data #104

Closed jeanpaulrsoucy closed 11 months ago

jeanpaulrsoucy commented 1 year ago

Check for COVID/with COVID, as referenced in the figure here:

https://twitter.com/Aaron_Derfel/status/1628562019614949377

See how far back data making this distinction goes. For example, AB data making this distinction only goes back to early 2022 (#82).

jeanpaulrsoucy commented 12 months ago

The datasets:

have the same data and same start date (2020-04-10), but the former is updated more frequently.

The dataset: b78d46c8-9a56-4b75-94c5-4ace36e014f5 (INSPQ) is no longer updated but has data beginning earlier (2020-03-15) using an old definition.

These two datasets are combined to show the time series in figure 3.1 of the INSPQ website.

The dataset 3b93b663-4b3f-43b4-a23d-cbf6d149d2c5 (INSPQ) also has additional time series related to hospitalizations/ICU "with" and "for" COVID. These power figure 3.2 of the INSPQ website. This figure covers admissions rather than occupancy.

However, the tweet referenced above covers occupancy, not admissions.

image

The caption references MSSS. Are the for/with occupancy data part of one of the aforementioned datasets or was this figure derived from a privately requested dataset?

kbelisar commented 12 months ago

The datasets listed above do not contain any data regarding occupancy. I wonder if it could be mis-wording where they should have referenced admissions?

The MSSS do not list any variables regarding for/with distinction in their data dictionary here. Occupancy could be calculated from subtracting cumulative discharged from cumulative admissions, however I don't see any numbers regarding discharges. Additionally, there are numerous proportions, but none seem to correspond with occupancy. I've checked the dates and for/with numbers in the captured tweet, and do not see any matches with the available dataset.

jeanpaulrsoucy commented 11 months ago

The occupancy data reported in the tweet above appears to be a one-off: probably released as part of a media request on behalf of La Presse. While admissions data are broken down by with/for in the INSPQ dataset, this does not seem to be the case for occupancy data.