Thiemo FetzerNov 29, 2022

I am going to share this as I think there is a much wider point to make. Last year, I studied the #Immensa #Dantelabs #COVID19 testing scandal.

A large #PCR lab basically produced tons of #FalseNegatives -- see my thread from last year:

https://twitter.com/fetzert/status/1460166688423489538

Today the #Inquiry results came out. Wrong equipment settings were to blame. This is what produced the #FalseNegatives.

More importantly, there is now an #Epidemiology paper with *13*
@UKHSA
authors. Let's compare...
@economics

Thiemo Fetzer 🇪🇺🇺🇦 on Twitter

“What is the epidemiological impact of a #falsenegative #COVID test? An important question in a high vaxx/low NPI context, but one that cannot be studied in a experiment for obvious reasons. Enter the UK, a reliable supplier of #naturalexperiments. 🧵⬇️ ➡️https://t.co/Zt0cHz9zoo”

Twitter
Show threadThiemo Fetzer

Findings: Broadly the same. My estimates are downward biased due to the limitatinos of public data. I find a "multiplier effect" of 0.6 to 1.6 for each misclassified case. They estimate this to be around 2 without providing a bound.

They also find effects on deaths & admissions that was not detectable statistically using the coarse public data back then. But I estimated back then this to have caused around 20 deaths simply looking at case mortality rates then.

Now Methodologically...

Nov 29, 2022 at 4:14pmWeb
Show threadThiemo FetzerNov 29, 2022

They basically do the same that I did. They estimate a #SyntheticControl based on binary treatment classification of areas from where tests were dispatched to #Immensa lab.

Its funny how their and my figures a year later are so similar.

It highlights: data access is a key barrier as they had slightly better data. But this was entirely avoidable.

In fact, I did launch a #FreedomOfInformation #FOI request last year to request the data that the #UKHSA team was using...

Now that FOI...

Show threadThiemo FetzerNov 29, 2022

Was point blank refused: https://www.whatdotheyknow.com/request/positive_lateral_flow_tests_not

The data that UKHSA used was lateral flow tests broken down to district level. I asked for exactly these breakdowns to be made available in the public interest and to help the research.

The reason for the refusal: it was deemed to cost more than GBP 450 to assemble the data -

I find this outrageous as ultimately it is our taxpayer money that paid for analysis that I would have been happy to do and even share all code etc....

Positive Lateral Flow tests not matched to positive PCR test regional/district-level breakdown - a Freedom of Information request to UK Health Security Agency

UKHSA is publishing "Weekly statistics for NHS Test and Trace (England)" (see https://www.gov.uk/government/collections/nhs-test-and-trace-statistics-england-weekly-reports) This includes tabular data on the "Tests conducted" (UKHSA publication gateway number GOV-10353) -- see e.g. https://www.gov.uk/government/publications/weekly-statistics-for-nhs-test-and-trace-england-28-october-to-3-november-2021. I am requesting a regional or district-level breakdown of the data that is included in this regular publication published in "Table 4 - Number of tests for COVID-19, Confirmatory PCR tests by route tested". At present the data presented in this table provides only aggregate information for the whole of England, I am requesting this data to be disaggregated by the region of residence of the individual tested. This data is available to UKHSA as testing with home kits or at asymptomatic testing sites requires individuals to supply their residential address. That is, for each of the variables listed below in weekly time series below, I am requesting this data to be disaggregated at the region, UTLA or LTLA level. That. is: Total number of LFD tests by week by region/UTLA/LTLA Total positive LFD tests by week by region/UTLA/LTLA Total number of positive LFD tests matched to a PCR test by week by region/UTLA/LTLA Total matched positive PCR tests by week by region/UTLA/LTLA Total matched negative PCR tests by week by region/UTLA/LTLA The format of the data should be in tabular form. I understand that this data is available to UKHSA due to the existing data being published in the weekly reporting. Nevertheless, if some of this data cannot be provided under the FOI provision... - section 12: please explain how it would take more than 18 hours to retrieve the form responses - section 21: please provide details (e.g. a URL) to where I can find this information - section 40(2): please redact the data that you cannot disclose, replacing it e.g. "Redacted under s40(2)". - section 43(2): please redact the data that you cannot disclose, replacing it e.g. "Redacted under s43(2)". Yours faithfully, Thiemo Fetzer

WhatDoTheyKnow
Show threadThiemo FetzerNov 29, 2022

The analysis is obviously well done and I congratulate the
@UKHSA researchers. I am questioning more the process behind. Why are we not empowering the research community to to what they can do best?

We see similar patterns with other work that researchers do where FOI that are intended to help improve data access are being refused or not supported. I am thinking of this one on #EatOutToHelpOut and many others: just look at my FOI track record:
https://t.co/HXWcq3nmkS

Thiemo Fetzer - Freedom of Information requests

WhatDoTheyKnow
Show threadThiemo FetzerNov 29, 2022

As society we need to learn to get better at making policy making evidence based. We need to increase transparency as there are tons of potential conflicts of interests when it comes to public procurement. As taxpayers we need to demand good stewardship of these resources.

#Immensa #COVID19 #Testing #Dantelabs #COVID19Inquiry

Show threadHector RufrancosNov 29, 2022
@fetzert Oh tell me about it! I tried to get access to Ambulance Scotland Statistics to look at spillover effects of students moving into student halls of residence, only to be stonewalled by the Ambulance service. The incentives to do research in the public benefit are not there for these organisations.
Show threadThiemo FetzerNov 29, 2022

@heelguru haha funny, I was trying to get access to ambulance service call of service records as well... great minds think alike!

There is a fear of too much transparency as it would expose the fragility of many of the public systems which naturally will raise questions: where does all the money go...

Show threadStijn MasscheleinNov 30, 2022
@heelguru @fetzert This is similar to what Zingales highlighted about access to the Uber data. Control over data determines which questions can be studied.
Show threadThiemo FetzerNov 30, 2022
@StijnMasschelein and that is why data ownership needs to be empowered. You own your own data and the regulator could mandate that firms share data for research use with consent to opt out for users but the default is to make data accessible. Because its not Uber that owns the data, you still have ownership rights as a user.