COVID-in-wastewater for BC Lower Mainland, to 4-June-2022. πͺ‘
From bccdc.ca/Health-Info-Siβ¦
Red line at zero level added to emphasize that relatively high levels persist in wastewater, discordant with the low level of officially reported cases. (1/n) #covid19bc
The graph is dominated by the Jan-Dec peak, which effectively and conveniently scales down the current persistent high levels. The eye tends not to catch that the levels post-peak are much higher than the levels before. (2/n)
The description in the text seems determined to ignore these persistent high levels, and appears to focus on shorter-term bumps and noise in order to use the word βdecreasedββ¦ (3/n)
β¦which is unfortunate, because recent trends in hospitalizations and deaths make sense when one realizes that the wastewater data reveal a persistent high level of infection.
This high level of infection is not shown by the official case countβ¦ (4/n) bccdc.ca/Health-Info-Siβ¦
β¦the official case count that is based on limited and non-representative official PCR testing.
Official testing (MSP) is shown by the almost-illegible doted lines.
The solid lines are government + private PCR tests. (5/n) bccdc.ca/Health-Info-Siβ¦
This matters because we cannot manage personal/societal risk and resources without clear and timely data on infections.
It would not be difficult or expensive to do ongoing PCR testing on a representative sample, using statistics to extrapolate to the population as a whole. (6/n)
And yet, here we are.
(7/n)
A better COVID wastewater graph from the relentless robot @YVRCovidPlots π€should be available tomorrow.
@emilyakopp Offshoring of risky virological lab work to countries with lower or zero biosafety requirements is an old strategy.
Lassa/Ebola lab, Kenema Government Hospital, Sierra Leone.
Anyone who believes that China has or will provide reliable data to elucidate the origin of COVID should carefully study the statements made in this April 8 press conference.
Links to full conference appended.
@Ayjchan @JamieMetzl @mvankerkhove @mstandaert news.cgtn.com/news/2023-04-0β¦
The geospatial analysis in Worobey2022 relies on a centering model to determine the origin point of COVID in Wuhan Dec 2019.
This model is not valid. doi.org/10.1126/sciencβ¦
The centering model can be stated as follows: the spatial pattern of the home residence of severe cases is centered on the origin point, with spatial density decreasing away from the origin point.
Fig 2A and B of Worobey2022 provide insight into the authors' logic.
βWe hypothesized that if the Huanan market were the epicenter of the pandemic, then early cases should fall not just unexpectedly near to it but should also be unexpectedly centered on itβ
This thread examines two claims in Worobey et al:
Dec-2019 COVID case-residences in Wuhan were not concentrated in (1) areas of high population density or (2) areas with a high proportion of older persons. science.org/doi/10.1126/scβ¦
The specific claims in Worobey et al:
πDec-2019 cases did not reside in areas with high population density of (1) all age groups or (2) older persons.
πFig 1E, S9 and S10 are enlisted to support the claims.
Fig 1E purports to represent the spatial distribution of COVID cases in Wuhan in Jan-Feb-2020.
It includes no population density data, and therefore cannot be used to support the claims.
This thread addresses the claim in Worobey et al that KDE analysis shows centering of Dec 2019 COVID case-residences on the Huanan Market. science.org/doi/10.1126/scβ¦
This apparent centering is an artifact due to use of an overly-large bandwidth in the KDE calculation.
Adjusting the bandwidth parameter to more realistic values shifts the center of the KDE pattern away from the Huanan market, to an neighbourhood north of the market where there truly is a significant cluster of case-residences.
Worobey et al base their interpretation on simplified KDE maps in which the influence of each data point is smeared over a large area (oversmoothing).
The pattern is centered on the Huanan Market, but this is simply an artifact of oversmoothing.
Worobey et al. (2022) science.org/doi/10.1126/scβ¦
Consider the KDE probability contours for the residences of Dec 2019 cases.
Data from zenodo.org/record/6908012β¦
*Linked* cases in green.
The map of the linked-cases KDE was omitted from the article...
...although the KDEs for all-cases and unlinked-cases were prominently displayed on Fig. 1, and featured in various tweets emitted by the authors.
The map βοΈuses the following from the data files supplied with the article at zenodo.org/record/6908012
βΈ maps βΈ geojson
βΈ who_cases_dec-2019.linked.KDE.contours.geojson
βΈ who_cases_dec-2019.notLinked.KDE.contours.geojson