Why are these voluntary samples of the asymptomatic testing programs biased? Voluntary samples are always biased. They tend to be healthier, wealthier, with more time to spare, have better self-care habits, among many, many other traits.
bmcpublichealth.biomedcentral.com/articles/10.11…
Even if you go to a hot spot, the population that shows up for voluntary testing is non-representative of the hot spot population. Sampling is the bread and butter of field epidemiology, but even the recent Santa Clara sampling fiasco was already forgotten the-scientist.com/news-opinion/h…
And it was really a fiasco, loudly discussed, everywhere:
statmodeling.stat.columbia.edu/2020/04/19/fat…
Even across the pond:
theguardian.com/world/2020/apr…
Online surveys are notoriously problematic, and online surveys of covid follow the same rule. They always err, and we seldom know in which direction. Similarly, recruiting individuals by e-mail, letter, or school website invariably introduces bias pesquisa.bvsalud.org/global-literat…
Voluntary participation has been known, for decades, to bias results. It is invariably and strongly non-random, with the volunteers in these samples being substantially different from the target population.

nature.com/articles/s4146…
In this study, participation in the UK biobank is shown to be higher among those with more years of education, older age at menarche and taller stature, and lower among those with higher adiposity, Alzheimer’s and schizophrenia.
nature.com/articles/s4146…
Even genetic scores confirm the self-selection bias in non-random samples. academic.oup.com/ije/article/47…
So, if you remember when a study concluded smoking was associated with lower risk of infections, you likely also know that selection bias explained the odd finding qeios.com/read/37F3UD
Even early estimates of pandemic parameters used in the initial modelling studies suffered from selection bias.The table below lists just a few of the distortions that can happen. arxiv.org/pdf/2004.07743…
Biases are not only an issue for the external validity of the study, they violate mathematical assumptions done during modelling and invalidate the findings. The paper above goes in a lot of detail about how that happens, if you are interested.
But we can remove the math and easily realize that saying the population is not adhering to public health measures based on a study that interviewed people that were arriving at a shopping mall during a pandemic is ridiculous.
Finding no cases in your sample when incidence is as high as it has ever been is not reassuring. It does not fit any theoretical framework for infectious diseases, nor conforms to any biological theory. It is just deficient sampling, low response rates and self-selection bias.
Ps.: If you knew there was no paid sick leave nor any school arrangements for quick transition to online learning during your 14 days of isolation following a positive test, would you volunteer to go find out know if you are infected?
Along this line of thought - here is a really great paper discussing some theories that may explain how people decide what they want to know, including individual biases that can lead to both insufficient and excessive information-seeking. nature.com/articles/s4156…

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Diego Bassani, PhD 🏠😷

Diego Bassani, PhD 🏠😷 Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @DGBassani

9 Apr
Covid is airborne, but the continuing insistence in pushing fomites as the major source of infections makes ‘the idea of fomites as the main infection route’ a large contributor to infections. In a certain twisted way, fomites are the problem.
This insistence in fomites as a mode of transmission is especially problematic with the more infectious variants. For example: Indoor contact with cases for any amount of time as long as wearing a mask is generally classified as low risk in this province.
This means that potentially infected individuals are not considered high risk, and in many situations, small clusters of cases (the majority of transmission events) may not be classified as outbreaks, for not meeting epidemiological link criteria.
Read 12 tweets
7 Apr
As if we did not have big problems with child mental health issues... back in 2009 we estimated that ~ 570,000 children under 12 lived in households where the an adult met criteria for one or more mood, anxiety or substance use disorders in the previous 12 months...
That corresponded to 12% of Canadian children under the age of 12. Almost 3/4 of those children had parents that reported receiving no mental health care in the 12 months preceding the survey.
And for 17% of all Canadian children under age 12, the individual experiencing a psychiatric disorder was the only parent in the household. capmh.biomedcentral.com/articles/10.11…
Read 5 tweets
6 Apr
Now here are the real devastating mental health outcomes of this pandemic, and they’re consequences of the infection, not public health measures. Who would have thought? thelancet.com/journals/lanps…
Seems so much worse than the flu.
And very worrying long-term consequences even for those under 65 years of age. Brace up Ontario, this will cost a fortune. Much more expensive than testing, tracing and isolation, sick days, safe schools...
Read 6 tweets
31 Mar
New preprint from OAHPP, Sunnybrook and SickKids about Pediatric household transmission of SARS-CoV-2 infection in Ontario with data from individuals living in private households (N=132,232 cases in 89,191 households) between June 1 and Dec 31, 2020. medrxiv.org/content/10.110…
This covers 84% of all cases (157,087) recorded during this period. Age-to-age transmission suggests that frequency of transmission increases with age, with kids infecting other kids, as well as adult household members. No data on adult index cases shown.
Younger children (0-3 years) were more likely to transmit SARS-CoV-2 compared to older children, but after adjustments ORs of household transmission were similar for ages 4-8 and 0-3. Testing delays also increased the odds of household transmission considerably.
Read 8 tweets
18 Mar
In the news: In 3 of the 7 previous years the average monthly number of youth suicide attempts presenting to Emergency Dept in the City of Hamilton PHU has been higher than the variability in the 4 month observation from McMaster Children's Hospital.
In all three previous years in which the presentations to ED were higher than the Oct-Jan reported increase of 4.8 additional cases/month they were above 6.0 additional cases/month. And in some, they have been much lower than the previous year.
Using longer time-series before going to the media with a before and after comparison that ignores the much longer time-series available for anyone to see is not reasonable. The hospital certainly could have calculated the monthly rates for a much longer time-series.
Read 7 tweets
5 Mar
Few people know houses in Brazil have no basements, and even fewer know it’s because they were filled to stop mouse infestations in the 1900s. Epidemiology’s history is full of examples of physical actions to solve problems, such as removing pump handles and filling basements.
Dr. Oswaldo Cruz was the Brazilian sanitarian who filled basements and tore down ‘unsanitary’ housing to control major infectious diseases that were killing the population at the time, including bubonic plague & yellow fever. scielo.br/pdf/rsbmt/v53s…
Working with the mayor of Rio de Janeiro to control infections during 1902-06, the poor were evicted and displaced to makeshift housing in the hills, giving origin to Rio’s now famous ‘favelas’. images.app.goo.gl/qNPE8sf722h9SW…
Read 9 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!