Do Not Seek and Ye Shall Not Find

Did flu surveillance and reporting systems (intentionally?) fail to detect flu?

Summary

• There appears to have been a collective and systemic failure in flu surveillance and flu death reporting systems in the UK during 2020 and into 2021.

• The fact that this failure cuts across all parts that comprise the surveillance and reporting system suggests that this failure is not coincidental, as do the observed inconsistencies in changing patterns of flu surveillance and reporting across different branches of UK public health.

• The UK FluDetector system, which uses machine learning to track flu using Google Trends data, reported that flu disappeared in 2020/21. This is contradicted by Google’s own data and UKHSA reports, both of which report a clear signal for flu in the UK in 2020/21.

• The UK FluSurvey system tracked flu incidence only until week 20 of 2020 and never updated this data. They then abandoned flu tracking until January 2021 when they announced that the flu season had now restarted, after supposedly witnessing the near eradication of flu in the preceding year.

• Routine mortality reporting by the ONS switched almost entirely to Covid-19 deaths in 2020 and flu is only included in surveillance reports from 2021, thus giving the impression flu had disappeared in the preceding year.

• Anecdotal evidence from Australia and Germany suggests routine flu testing, done but GPs was dramatically reduced or stopped in favour of Covid-19 testing.

• In response to an FOI about flu in 2020 the ONS obfuscated its answer by using a different death code ‘respiratory disease’ rather than flu for the period up to May 2020 and may have done so to hide the true deaths that should have been attributed to flu.

• This FOI request shows that there were 2,287 flu deaths in March 2020; while this is less than the 3,324 deaths that were classified as Covid-19 in the same month it is strange that SARS-CoV-2 was considered to be such a significantly greater threat to public health.

• In January 2021 the ONS reported that there were almost as many deaths involving flu (7,610) as there were involving covid-19 (5,719), yet for only 5.2% of these flu deaths was flu recorded as the underlying cause of death.

• The NOMIS system shows that for the whole of 2020 there was no significant change in the number of deaths in the UK caused by flu (20,130), compared to previous years, directly contradicting the ONS, which deliberately published no comprehensive data on flu. Similarly in the USA the CDC WONDER system reports flu deaths for 2020 at similar levels seen from 2015-2019.

• Despite these facts the WHO’s international flu surveillance system, FluNet, shows no significant flu for 2020/21 in either the UK or USA.

Loading video

(Short snippet about flu from Martin Neil’s recent interview with Freedom Research)

Introduction

This is the fifth article in our flu series. The fourth article presented the discovery that Google searches for flu, by the general public, did not cease even when flu was said to have vanished in 2020/21 and 2021/22. In our third article we investigated how quarantines, and other policy decisions, may have caused the flu to vanish. In our second article we raised questions about flu testing and viral interference between SARS-CoV-2 and flu, and in the first we noted that the flu had not completely disappeared in 2020/21 as had been claimed

¹

.

In this article we look into flu surveillance systems and flu data reporting, mainly in the UK.

A hodge podge of flu surveillance systems have consistently reported on flu prior to 2020 and continued to have some role to play post Spring 2020 until flu returned at the end of 2021. In the UK flu surveillance is done via clinical surveillance by primary care (based on networks of GPs), and the FluDetector and the FluSurvey systems.

Our investigations into flu surveillance use in the UK suggest they were either not picking up signals, that were in fact present, or were (intentionally?) deactivated for the duration of what would have been a normal flu season in 2020-2021.

We also looked at UK ONS data for influenza and pneumonia in 2020-2022. We discovered a reporting gap for flu in 2020 when all respiratory disease reporting switched to focus myopically on Covid-19. And then, in 2021 and 2022 the ONS re-started flu reporting and influenza and pneumonia deaths returned.

However, we discovered a hitherto much ignored source of government statistics: The UK NOMIS system. This showed that for the whole of 2020 there is no significant change in the number of deaths caused by flu, compared with previous years, directly contradicting the ONS data. Likewise, the 2020 deaths recorded as caused by flu in the USA was also similar to that seen in previous years.

FluDetector

Credence has been given to the idea of tracking pandemics, such as the UK’s flu-detector system

²

, which tracks internet searches using Google trends. The chart below is the result produced by UK the flu-detector system from April 2017 to the present.

Clearly, there doesn’t appear to be any flu signal in the winter of 2020-21.

It is important to realise that the output of the system isn’t raw data but is rather modelled output that processes the data through a sophisticated machine learning system (a non-linear Gaussian process model).

We can easily verify this ‘modelled’ output against the raw data by turning to Google Trends itself for the ground truth on searches for “Flu”. What we find is very surprising indeed:

Google Trends shows there is a clear signal in winter 2020-21 showing that people were in fact searching for flu. Yet the UK surveillance system, that claims to use Google’s own data as the basis for its calculations, has either not detected this signal from Google, or somehow suppressed it. How can this difference be explained? Have the Gaussian process models simply failed and performed so catastrophically, or is there some fault in Google’s search data archives?

What the actual Google data does show is that people were concerned enough about flu that they searched for flu on the internet in significant numbers even though they were in the middle of a deadly pandemic caused by SARS-CoV-2, a virus they have been told is more novel and more deadly than the flu. It is also notable that the number of searches for flu in 2020 were not significantly different from those conducted in previous years in the UK.

Data from the UKHSA national weekly flu reports confirm that the NHS were recording people with cold/flu symptoms at the time they were searching for flu on Google:

Note that the peak for cold/flu of just over 20,000 online assessments in September 2020 (when flu is said to have been absent) is actually slightly higher than the peak for covid during the same period. Yet, despite this, the ‘modelled’ data from the flu surveillance system fails to pick up this signal.

So here we have independently checked the FluDetector system against that provided by UKHSA and Google itself, both of which report a signal for flu in 20/21, and yet FluDetector shows no signal for that season.

It is hard to explain away this failure and the implications of this difference are potentially quite sinister. Were machine learning systems introduced in order to suppress the flu signal and help convince the world that flu had disappeared from the UK? Or is there some innocent explanation? Or is this story repeated in different countries which also used automated flu surveillance systems?

FluSurvey

The UK FluSurvey was originally conceived to survey a panel of self-selecting participants for signs and symptoms of influenza-like illnesses (ILIs). In March 2020 it was repurposed to cover both Covid-19 and flu symptoms, and the routine questionnaire was adapted to capture covid-19 specific information.

In June 2020 the FluSurvey reported

³

the 2019/20 provisional results for flu, which only ran up to week 12 - the end of March 2020. Notice that the last 3 months’ worth of data up until June 2020 is entirely missing.

By October 2020 this had not been updated despite the survey continuing. However, ‘health care seeking behaviour’ reports were produced instead, and the symptoms fever, cough and loss of smell were tracked but not explicitly attributed to either flu or covid-19.

This kind of reporting continued until January 2021 when the FluSurvey website announced that they had decided back in November 2020, at the start of the normal UK flu season, to discontinue using FluSurvey to track both flu and covid-19:

In January 2021 they then helpfully updated the ILI incidence chart to include up to week 20 of 2020 (only half a year’s data):

Fast forward to the next update of the website on November 30 2021, almost a full year later, and we find this statement on the website:

So, they make a spontaneous decision to do an about turn and track both covid-19 and flu together using FluSurvey at the end of 2021. And they state that they know the 2021/22 flu season has now begun! How do they know this when the chart previously reported in January 2021 had apparently not been updated at that time, in November 2021? And how do they know it would come back when flu had disappeared?

FluSurvey reports fed into the UKHSA national flu report for the first time from 8 October 2020. They did not explicitly appear in any national flu report before that date, including before the covid pandemic.

The UKHSA week 28 national flu report (15 July 2021) contains the FluSurvey ‘healthcare seeking behaviour’ chart can be found on page 32.

Notice that the rate of fever and coughs between weeks 37 of 2020 and week 3 of 2021 is consistently above 25 per 1,000 participants and peaks at 50 per 1,000 participants (right hand scale on chart). Also note that the ILI line in the chart is blank up to week 43 confirming the decision to not formally track ILIs as noted on the FluSurvey website and instead formally attribute cough and fever symptoms to covid-19.

The UKHSA week 4 national flu report (27 January 2022) contains the FluSurvey ‘healthcare seeking behaviour’ chart, which can be found on page 35. Here the rate of fever and coughs peaks at approximately 70 per 1,000 participants during the 2021/22 flu season, weeks 37 2021 to week 3 2022. Notice that throughout this period the ILI (black line) is charted and acknowledges the rate of ILIs.

The peaks marked in each of the above charts are remarkably similar. Given the purpose of the FluSurvey is to track flu symptoms, and fever and cough are a primary method for doing so, this surely suggests that there is evidence of a flu peak in 2020/21 when it was claimed to be absent. And at a time when FluSurvey decided to change its reporting standards (November 2020).

Let’s review the sequence of events:

• FluSurvey tracked ILI incidence only until week 20 of 2020 and never updated this data.

• They decided to no longer track both flu and covid from November 2020 but only report this change in policy in January 2021.

• FluSurvey stated they (presciently) knew in January 2021 that the flu season had now begun and did so after witnessing the near eradication of flu from March 2020 to May 2020.

• Fever and cough symptoms were tracked and peak in a similar way and this continued, but tracking ILI incidence was abandoned, thus reducing the strength of the flu signal.

This surveillance data and timeline of reporting policy decisions is suspicious:

• Why decide to track both viruses March 2020 to November 2020 (outside of the normal flu season), then stop doing so without explanation?

• Why was ILI tracking abandoned when fever and cough symptoms were peaking in 2020/21?

• Why claim the 2020/21 flu season had begun and stop reporting flu and covid-19 using the system?

• Why then switch back to reporting both viruses again in November 2021 and declare the flu season had begun with ZERO surveillance evidence flu had returned?

Anecdotal reports about flu testing in Germany & Australia

A subscriber posted this self-explanatory comment on an earlier post. It concerns flu testing in Germany:

Australia did not experience flu, nor did it experience SARS-Cov-2 in 2020. This report attributes the disappearance of flu in Australia to physical distancing and good hand hygiene.

There are obvious questions about whether NPIs can really explain the absence of flu in Australia given they weren’t mandated in summer 2020 and were not stringently applied. Furthermore, they subsequently failed to prevent Covid-19.

But, in the comments section comments by Dr Emilie Larkin draw our attention to changes in testing policy:

And this from Dr Irandani Anandi Ranasinghe-Markus:

Here is another report highlighting the same issues from the WHO:

This evidence suggest testing resources were indeed diverted to SARS-CoV-2 testing from routine flu testing, and hence routine surveillance systems were dismantled.

ONS and FOI response

The UK ONS report mortality statistics for each year. The statistics for England and Wales are available for download here. The 2020 dataset contains information that excludes any report on influenza and pneumonia deaths, and instead is wholly focused on deaths ‘involving’ Covid-19:

In contrast the 2021 dataset does contain data on influenza and pneumonia deaths, starting on January 8th 2021.

Note that, unlike the 2021 dataset, the section “Weekly figures by cause” is not present in the 2020 dataset. The ONS do not provide weekly figures by cause data for anything other than Covid-19 before 2021.

In 2020 all Covid-19 deaths are categorised as ‘involving’ Covid-19 (codes U07.1, U07.2, U09.9, U10.9) rather than due to Covid-19 (codes U07.1, U07.2, U10.9) and in 2021 Covid-19 and pneumonia & flu deaths are categorised as ‘involving’ (J09-J18) and ‘due to’ (J09-J18):

2020:

2021:

So in 2020 official ONS statistics any reporting of flu deaths was completely abandoned in favour of covid-19 reports.

An official response to an FOI request for deaths from influenza January - April 2020 stated there were 2,287 recorded deaths, in England, caused by influenza and pneumonia in March alone, and up until May 2020 there were 31,315 deaths from ‘respiratory disease’ as the underlying cause of death:

Hence flu deaths have been included in a catch-all category with ICD code J00-J99 for respiratory deaths up until May 2020, but we have specific influenza death data only for the single month of March 2020, when the pandemic started.

The ONS 2020 mortality report makes no mention of reported flu deaths in 2020 giving the impression flu was absent, but when we examine the ONS reports for deaths caused by pneumonia & flu in 2021 and 2022 there were 16,079 and 20,008 reported respectively, so for all years apart from 2020 they are happy to provide accurate flu data.

AlterativeNarrative noticed that the ONS report on deaths registered weekly in England and Wales, from January 2021 reported these flu deaths:

Summary:

• In response to an FOI about flu in 2020 the ONS obfuscated its answer by using a different death code ‘respiratory disease’ rather than flu for the period up to May 2020 and may have done so to hide the true deaths that should have been attributed to flu.

• This FOI request shows that there were 2,287 flu deaths, in March 2020 which is not greatly different from the 3,324 Covid-19 deaths that same month; yet SARS-CoV-2 was considered to be a significantly greater threat to public health.

• In January 2021 the ONS reported that there were almost as many deaths involving flu (5,719) as there were involving covid-19 (7,610), yet for only 5.2% of these flu deaths was flu recorded as the underlying cause of death.

NOMIS and WONDER mortality statistics

Of course, absence of evidence is not evidence of absence - can we find evidence for flu from other data sources? Luckily, we can.

Data for pneumonia and flu deaths can be extracted from the UK NOMIS (official census and labour market statistics) system. However, this only provides yearly totals rather than weekly figures. The coding system used shows deaths ‘due to influenzas or pneumonia’ rather than ‘involving’ (codes J09-J18).

When we queried this system, we were quite shocked to find that it returns influenza and pneumonia deaths for 2020, and these are reported at a rate consistent with previous years (the data for 2021 is flagged as provisional in the system).

So why does the NOMIS system list flu deaths in 2020 but the ONS reports do not in 2020. Perhaps they never got the memo? Why did the ONS only release flu data in response to an FOI request? The FOI request stated there were 2,287 deaths in England in March alone, at a time there was supposed to be little to no flu. At the same time the 3,234 Covid-19 deaths was taken to be a sign of a pandemic caused by a novel and deadly virus.

The presence of flu deaths in 2020 in the UK is repeated elsewhere. In the USA influenza and pneumonia numbers are similar in 2020 to previous years, as reported in the CDC WONDER system, with 53,544 deaths in 2020 compared to between approximately 50-60k deaths in each year from 2015 to 2019:

Summary:

• Routine mortality reporting by the ONS switched to covid-19 in 2020 and flu is only included in surveillance reports from 2021, thus giving the impression flu had disappeared.

• The NOMIS system shows for whole of 2020 that there was no significant change in the number of deaths caused by flu and pneumonia (20,130), compared to previous years, directly contradicting the ONS, which deliberately published no comprehensive data on flu.

• In the USA the CDC also report flu deaths in 2020 at levels consistent with historical patterns.

The Flu Vanishing Act Exposed?

Despite these facts the WHO’s international flu surveillance system, FluNet, shows no significant flu for 2020/21 in either the UK or USA.

UK:

USA:

1. This vanishing phenomenon was reported back in 2020 by anti-media.

2. Note that this guidance document was withdrawn in October 2022, without explanation and replaced with new systems.

3. Note that all reports FluSurvey use links to archives of their website accessible using the web archive Wayback machine.