Study of over 2M participants in Uk and USA to investigate racial and ethnic determinants of Covid 19 risk.
www.medrxiv.org/content/10.1101/2020.06.18.20134742v1
Found that even after adjusting for other risk factors including co-morbities and sociodemographic factors still significant disparities.
Hoping that this might count as sufficiently‘robust’ research for the Katie Hopkins type statistical manipulators on this thread.
It would be good if you would engage in good faith rather than with ad hominems.
As I said in my last post we do know that ethnic minorities in the UK are much more likely to have been infected and that this will straightforwardly lead to higher death rates.
And as I said, no study is going to counter the overwhelming fact that sex and age are by the far most important factors in death risk.
What I have said is that nobody has provided robust evidence showing an inherent ethnic link to covid-19 mortality. That remains true after your post, of which the 2M participants is not particularly important in that vast samples aren't necessarily better than smaller ones. Sampling is more important.
You haven't provided a specific finding from the link you provided, but thanks for providing the link
Anyway, I read the first paragraph and it was exactly what I said in my next-to-last 2 posts
"We used a smartphone application (beginning March 24, 2020 in the United Kingdom [U.K.] and March 29, 2020 in the United States [U.S.]) to recruit 2,414,601 participants who reported their race/ethnicity through May 25, 2020 and employed logistic regression to determine the adjusted odds ratios (aORs) and 95% confidence intervals (CIs) for a positive Covid-19 test among racial and ethnic groups"
so this is not random sampling. Rather, it is the Zoe app, which people download and it records data and suggests you might have covid-19. This non-representative sample is not necessarily a problem, but we should understand what has been tested.
en.wikipedia.org/wiki/COVID_Symptom_Study
"Compared with non-Hispanic white participants, the risk for a positive Covid-19 test was increased across racial minorities (aORs ranging from 1.24 to 3.51). After adjustment for socioeconomic indices and Covid-19 exposure risk factors, the associations (aOR [95% CI]) were attenuated but remained significant for Hispanic Hispanic (1.58 [1.24-2.02]) and Black participants (2.56 [1.93-3.39]) in the U.S. and South Asian (1.52 [1.38-1.67]) and Middle Eastern participants (1.56 [1.25-1.95]) in the U.K"
So, the app found that South Asian and Middle Eastern participants in the UK were about 50% more likely to have caught covid-19.
And as per the link in my previous post there is no correlation between catching covid-19 and Vitamin D levels.
And as I said in my previous post we know ethnic minorities are more likely to have caught covid-19.
And it's reasonable to assume that different ethnic groups spread covid-19 in different ways - there was much speculation and stereotyping on here about Italians being tactile and familial and Germans being more reserved and this being a factor in spread.
We also know that British Jews have the highest death rate of all races or religions. We know Christian and Islamic gatherings have caused thousands of infections for singl emeetings.
So when we have a statistical fact that we have no reason to doubt, namely that UK South Asians are more likely to be infected with covid-19 than is predicted by their circumstances, what do we conclude? That there's something inherent to South Asian biology that makes them more likely to catch covid-19? Or the blindingly obvious fact that South Asian people will have different social, religious, etc. practices from other races.
So please actually READ even the precis of the study before asserting that it proves someone else is a Nazi.
If we continue to the actual full text of the study, it says
"However, after adjustment, the risk of a positive Covid-19 test remained significant for several racial and ethnic minorities, which is likely due to additional contributing factors for which we were unable to account, including insurance coverage, access to healthcare, use of public transit, and other essential occupations not specifically queried. sian and Hispanic populations are also more likely than non-Hispanic whites to live in multigenerational households, and, like Black populations, are more likely to live in densely populated urban areas."
In other words they aren't asserting for one moment anything different from what I have been saying, namely that ethnic minorities are more likely to have been infected and that they haven't even addressed many likely reasons for this.
In fact
"Additional covariates were selected a priori based on putative risk factors, including sex, body mass index, history of diabetes, heart, lung, or kidney disease, current smoking status, isolation, community interaction with individuals with Covid-19, frontline healthcare worker status, population density, income, and education"
so they had a priori assumptions about risk factors, some of which are not likely to be relevant, and some serious risk factors of which they missed.
I downloaded the app to check, and they ask
"Are you a healthcare worker (including hospital, elderly care, or in the community"
No/Yes I interact with patients/Yes I do not interact with patients
"Do you care for multiple people in the community, with direct contact with your patients"
No/Yes
And that's it.
They do not ask for income NOR occupation, but instead it is apparently implied from LSOA
A LSOA is 1500 people approximately and will be somewhat homogenous generally, but clearly in urban areas social housing and private housing are going to be in the same LSOA, and they don't have anyway of determining that.
I.e. if Joe White lives in a rented private flat and works in IT and is furloughed/working from home, while Michael Black lives in social housing in the same development and works in a meat plant, going to work every day, then this won't be identified at all by the app, since they don't ask about your occupation, and while they do ask the postcode (which might be different), this is only to identify a LSOA, which will be the same.
It's unfortunate that with such a large data pool they didn't ask people if they were going to work, and at what!
It turns out that the study doesn't even show that, say, ethnically South Asian factory workers are more likely to have caught covid-19 than white factory workers.
And asking about 'frontline healthcare' isn't that helpful in that we know in general that male 'care home workers' are more likely to have died from covid-19 than the general population, but that male 'doctors' have no such elevated risk. However the death risk doesn't distinguish between 'frontline' and 'non-frontline'.It'
If we scroll down to the data on page 25
www.medrxiv.org/content/10.1101/2020.06.18.20134742v1.full.pdf
it turns out that the sample is massively biased, being 94.2% white, as against the latest estimates of 84.9% for the UK.
The sample also is massively biased against young people with disproportionately many people aged 35-64, and is mostly women.
Also people are fucking liars about their weight, with 53.3% of South Asians claiming to be underweight or normal weight, and 47.1% of white people.
Anyway, from their very large but really exceptionally shitty and unrepresentative sample, they found that black and South Asian people were more than twice as likely to have a positive test compared to white people. Here we should note that black people in their sample were more than 2.3x more likely to state 'frontline healthcare worker' status, and 1.8x for South Asian, than white people.
Here we should note that positive test results have long been misleading in that if you test more people you get more positive, and there's clearly bias in that people who are older/sicker/healthcare workers are far more likely to be tested.
They performed a
an "inverse probability weighting (IPW) as a function of race/ethnicity and other factors, such as age, symptom burden, COVID-19 exposure risk factors, and socioeconomic status, followed by inverse probability weighted logistic regression" [note that they don't know socioeconomic status, and can only extrapolate from LSOA and have nothing else to go on]
however this IPW is really quite shit - they ignore the fact that healthcare workers are much more likely to be tested because of their jobs, for example, and a positive test result tells you that someone was positive, but the lack of one doesn't say they are negative.
A more useful study would be to track people as part of a regular testing program, so that every user of the app would be tested weekly or whatever. As it is they know:
they had around 2.25 million people in the app
an undisclosed number had tests which were negative
8,335 white people and 1000 people not identifying as white had positive tests
Anyway, aside from their IPW, once they asked
In the last week how many times have you been outside your house with limited interaction with other people
In the last week how many times have you visited somewhere with lots of people (e.g work/for groceries/public transport/school)
In the last week how many times have you visited a healthcare setting including for work (hospital/clinic/dentist/pharmacy)
and took into account community exposure, frontline healthcare worker status, the excess risk fell from 117% to zero for black people, but remained at 52% for South Asians.
Overall the app is an exercise in massive data, where massive is as in 'massively unrepresentative'.
There will be all sorts of sources of bias beyond the obvious that there are too many women, too many white people, not enough young people. We can also consider that the very ill are unlikely to use the app, that those who get a negative test are less likely to report it, that there is likely to be a sex bias in terms of continuing use of the app (diligence).
I'm sure the app has some useful metrics, but these racial data are not among them.....