Keeping Queenslanders Safe
Queensland Health released a study looking at the after-effects of infection from Influenza and COVID. The Queensland government took this study and made the giant leap to say that it shows vaccination reduces the effect of Long COVID. The study shows nothing of the sort. It could be showing that the boosted are more likely to be infected.
There are faults in the study which we go through and as usual we can still glean some useful information from the study.
Introduction
Queensland Health recently released a study and announced:
Vaccination reduces incidence of Long COVID.
The government protected Queenslanders by locking everyone out from other states in Australia and achieved a high rate of vaccination before the borders were open. Unfortunately, the study doesn’t show anything of the sort. In fact, if anything it may well show that the triple dosed are more likely to catch COVID than those with 0-2 doses!
Going through the study in detail, published on a preprint server, reveals what they actually found. Basically, a similar proportion of people feel sick 12 weeks after infection, irrespective of whether it is COVID or Influenza.
Should this surprise us?
From one’s own experience one often feels under the weather for months after a bout of the flu.
However, there may still be useful information to be mined from this study. As we have found on our data journey, even with useless studies, like investigation of car accidents of unvaccinated people, we can glean further information. We also learn from what obvious things have been left out. So, let’s go….
What the study doesn’t show
The study definitely does not show that vaccination reduces the effects of Long COVID.
The logic the Queensland government erroneously appears to have followed is that Queensland was highly vaccinated (90% they say but from looking at various sources I believe it was less than this). The study found the after-effects of COVID were no worse than for influenza based on the survey they performed. Long COVID is BAD. Therefore, vaccination must be protecting Queenslanders from Long COVID.
I note it’s the politicians making the ridiculous claims, not the study authors.
The Premier of Queensland makes a big deal about “keeping Queenslanders safe”. You can see her in action here and here (facebook and twitter).
Many Australians find the attitude really irksome. We are all Australians, except perhaps when it comes to football.
The study doesn’t try to look at other infections to see whether people feel similarly sick.
One of our best journalists, Rebekah Barnett from Western Australia, has done a fantastic job summarising the approach that has been taken by government here. She posted one of her classic articles yesterday “Public Health Propaganda in 7 Simple Steps”
Rebekah cleverly twists the mantra '”evidence-based policy making” saying the Queensland government is pioneering an innovative new concept:
“policy-based evidence making”
This is textbook stuff for the government playbook.
Incomplete data
Let’s look at the study in detail.
The study was over a period of 2 weeks, starting 12 June 2022. This is Winter in Australia and during a wave of infections. We presume the study took all the PCR tests for COVID and Influenza in Queensland, although that is not stated explicitly. They could be taking a sample. Out of interest the covidlive website also provides the number of COVID PCR tests. For the period 12-25 June 2022, in Queensland, it was 93,192 tests. So 13,694 COVID positive tests out of 93,192 tests. 15% positive rate. We don’t know though if they have taken a subset of all positive tests.
Figure 1 from the study shows how they came up with their cohorts.
I note the daily numbers reported on covidlive are erratic and there are sudden jumps and drops as adjustments are made. This has been a fundamental problem with Health Department pandemic data. There is limited, properly curated health data, in Australia except from the ABS.
The next thing that is strange is that one third of the records have invalid mobile phone numbers. In this day and age, when there was a time one couldn’t do anything without a phone, eg go into a shop, it is astounding that such a large number had no valid mobile number.
Perhaps they are giving a landline number? Perhaps people are intentionally giving an incorrect number?
After that only about a quarter of the people agree to do the Queensland Health survey. It ends up as a study of 3,146 participants. 951 had influenza and 2,195 had COVID. BTW remember the 2,195 number.
This is all suggestive of low-grade data, which is concerning for a study that is being used to guide government policy.
Mathematical Errors
There are numerous mathematical errors which will presumably be identified and corrected during peer review. Nonetheless, these are fairly basic errors, so it is again most surprising that this study was published without these errors being picked up.
Table 1 of the report is the important table that shows the characteristics of the study cohort.
Initially I was completely confused by this table. I asked around for advice from colleagues to check whether I had misunderstood the contents of the cells in the table. Thanks to those who helped.
First up it was a check of the percentages. This uncovered two things. Rounding of the decimal place seemed to be wrong and some numbers were completely out.
The percentages should be the value in the number column divided by the number in the cohort. Take for example COVID positive people over age 50. There were 1,241 and the cohort size was 2,195. Now 1241/2195 = 0.5653758542… Converting to a percentage with one decimal place is 56.5%. The value reported is 56.6%. OK it’s a very small difference. But computers don’t make this mistake. Did they somehow round up? I had to laugh when a teacher helping said the most common mistake students make is just truncating decimal places, they rarely make the mistake of rounding up.
Without going through all the other rounding errors it seems that in this case they could be explained if the number in the cohort is 2,194 rather than 2,195. It is probably some manual transcription somewhere and the data hasn’t been checked.
The second author on the paper is the Queensland Chief Health Officer. The paper was submitted to the preprint server on April 17, 2023.
Doing a search on terms “Queensland Health failures” comes up with plenty of material. From the country Queensland newspaper, Gympie Today
Long waits for elective surgery, unacceptable ambulance wait times. A plethora of problems. From Sky News
The Queensland Chief Health Officer certainly doesn’t have time to check the decimal places.
As I mentioned one of the people who checked for me is a mathematics teacher. I’ve observed that Math teachers are very good at spotting errors. I guess they are doing this day in and out checking assignments. In other lines of work once you see reports with many mistakes you usually just say this is rubbish and give up on it. Teachers have the obligation to keep going through the errors to find what is wrong. I remember in maths in a long working you only lost points for the initial error you made, even if it caused everything else to be wrong. Teachers have the obligation to follow the errors through and are good at this. I can suggest that Queensland Health enlist some of their Mathematics teachers to proof their work before it becomes government policy.
I think what may have happened is a mistake where the number 2,195 should be 2,194. We picked that up from another sum that didn’t add up. Small error but decimal places can be important (link provided for aviation people).
The next thing that is wrong is that the percentages for the Socio-economic status don’t make sense. Dividing the number for each category by the cohort total gives something completely different, always less. However, the percentages shown do add up to 100%.
What we think has happened is that not all people in the study gave a valid address. The socio-economic status is determined by postcode. If there was no valid postcode then status can’t be determined. Apparently, they did not exclude people with invalid address. They had to exclude people with invalid mobile number because then they couldn’t do the mobile survey.
Same thing for vaccination status. Look at COVID positive participants with 3 or more doses. n=1220. We assume there are 2194 participants in the cohort. 1220/2194 = 55.6% and 83.2% is shown. So what we think is that there are 1220/0.832 = 1466.3 people that had vaccination status known. That’s 1466/2195 = 67% of people.
So they didn’t know vaccination status for one third of the people?
For the row greater than or equal to 6 months since last dose we have 338/.231 = 1463 people. It’s a very similar number. There must have been 3 people they knew were vaccinated but didn’t know whether it was greater than 6 months ago.
This has been exhausting but, given the dearth of data the public has available, we have to make the most of what we’ve got.
In a similar way it seems there were about 83 people where indigenous status was not known.
They appeared to know the ages of all people in the study.
Strangely for people who tested positive for Influenza, the numbers add up meaning vaccination status was known. Perhaps they only included people in that cohort where COVID vaccination status was known? Very strange.
Is there unintended proof that the 3+ dosed are more likely to catch COVID?
83.2% of survey sample who tested positive for COVID had 3+ doses of vaccine.
According to covidlive website for Queensland, 2,389,675 3rd doses were delivered by 12 June 2022. Going to the ABS for population over 18 (ie those eligible for 3rd dose) in Queensland I estimated 4,161,317. Therefore 57 % of the population of Queensland were 3+ dosed at the time of the PCR tests.
Could this mean more doses makes one more likely to be infected?
Steve Kirsch published yesterday:
This is based on a Pfizer briefing document provided to the FDA back in 2021!
Steve also goes through other evidence on COVID infection vs vaccination status. For those who have been reading my substack, from when I started writing, you would know that this is what led me to write my first articles. The UK data showed this clearly in 2021
When it got so bad in the UK they stopped reporting it. NSW Health didn’t get the memo and started reporting it just before we opened the borders
They too had to stop reporting infection vs vaccination status data when it was clear that the vaccinated were more likely to be infected. Mandates were being enforced in Australia. People were losing their jobs. The world made no sense to me and I thought I might be able to contribute in some way by reporting on this data.
Back to the Queensland Health study
In a good study you would assume that the sample of the whole population they used was representative. But we have seen so many problems with this study we can’t assume that. A large proportion of people with COVID infection did not participate but we can include them in the statistics of the percentage of those with 3 or more doses. People with 3 or more doses were more likely to do the survey (83.2% vs 67.1%) than those who did not. Perhaps this is a surrogate of likelihood to follow government orders?
Taking into account the infections of those who did not participate in the study I estimate that 72% of those infected had 3 or more doses out of a population in Queensland where 57% had 3 or more doses. This is a 26% increase.
The other variable they decided to use is whether people had had a shot within the last 6 months. There is the inference that people with more than 3 doses and recent injection are somehow better off.
Use of the variable >= 6 months appears to be stupid in this context. Let’s look at vaccination in Queensland. Fourth doses had started being delivered.
The previous dose was the first booster.
It’s uncanny that both were being delivered right in the midst of waves of infection in Australia. But look 6 months previous to June 12, 2022. It is smack bang in the middle of delivering 3rd doses. People who had 3 doses only, will be randomly sitting on either side of 6 months previous to the study.
Perhaps Queensland Health thinks there is something different about a person who had their last dose 6 months and one day previous compared to 5 months and 30 days? My recommendation is that they possibly could have used 3 months as a useful variable.
Many of the people infected could have just had a 4th dose in the few weeks before the study period.
I think this is where I should put a face slap icon.
CONCLUSION
Queensland Health created a study that politicians used to make the unsupported claim that vaccination reduces long COVID prevalence. Instead, they appear to have showed that the boosted are catching COVID more than any other cohort.
We had a fantastic community event on the weekend. I’ll do a special write-up on it shortly. We recorded the speeches which I will post.