Stupid Study - Winner for 2022
But we can still get useful insights
My wife and I are back from a short break to Queensland. During this time, we visited family and friends, some newly met through the adversity of the last few years in Australia.
We drove from Sydney and while on the way legacy media articles reported on a paper, just published in the American Journal of Medicine, on the dangers of traffic accidents, caused by unvaccinated people!
I’d already seen the reports in my phone’s news app, but thanks to friends reading this who kindly thought to warn me 😊.
Igor Chudov was first off the mark, noting it as the “dumbest study of the month”.
But I think it could be in the running for “stupidest of the year” and the authors must have rushed it in to be eligible for the 2022 awards.
My first thought, after seeing the headline, was that I bet they didn’t take into account the fact that people who have more accidents are probably those who spend more time on the road. And so it was. Igor describes it well. There are so many things wrong with the paper it would take a full report to explain the details.
The inference is that drivers who are unvaccinated take risks. They also don’t trust the government. They therefore do not take notice of traffic rules like give way signs and cause accidents.
It was not only drivers (43%) included in the study but also passengers (18%) and pedestrians (39%). How would the vaccination status of a passenger affect the likelihood of accident? Perhaps they are sitting in the back spouting conspiracy theories (like the CIA was involved in assassination of JFK) and distracting the driver.
People were not considered vaccinated until 14 days after injection. This is the trick that keeps giving. It is as if antibodies are suddenly developed and people wake up to themselves.
People who died at the scene (albeit a small number) were not included. There was no recognition of accident being caused by medical emergency of the driver, like heart inflammation leading to heart attack. That would certainly distract the driver.
They also would have underestimated the number of unvaccinated in the population (the denominator).
One can only assume that the authors wanted to find a certain outcome and worked hard to achieve the result they wanted.
Vinay Prasad’s conclusion was that:
We wanted unvaccinated older people to get vaccinated in first half of 2021. Everything else was trivial. Boosting teenage boys and vaccinating babies has no RCT data, and observational studies will be confounded, and only a misguided person would pursue those policies.
Unfortunately, many of those people work for this White House.
Dr John Campbell quickly weighed in. His short analysis is here and his full analysis here. He concludes that the study shows us:
“Staying at home reduces the chances of being in an accident”.
Professor Fenton’s Substack
Professor Norman Fenton has just started a substack with his colleague Martin Neil. Please sign up to his substack. I am grateful to Professor Fenton, as he has provided advice to me and when he tweeted one of my early articles this helped people become aware of my work and led to my first subscriptions other than people I already knew. By the way I am grateful to all readers subscribing, it gives me confidence in continuing the work.
His article on the accident paper is here:
and includes his usual clear video explanation.
Who Writes this Stuff?
One of the first things I do now with stupid papers is check who the authors are and their financial disclosures. University of Toronto has significant Gates funding. The first author has done other such research. In 2017 he reported on increased risk of motorcycle accidents when there is a full moon. The implication is that riders are distracted by the moon. Apparently, it was worse when there was a supermoon. Imagine if it was a lunar eclipse.
I’m surprised they didn’t incorporate the phase of the moon as a variable in this traffic study. Other researchers have looked into road fatalities including the coincidence of a full moon and Christmas.
The author can be seen in a video posted on Twitter from Canadian TV.
I was a bit alarmed watching this.
Looking at the secondary analysis there are various other breakdowns.
The table above also shows the risk is the same for a passenger as a driver, if unvaccinated. I note that the risk for unvaccinated is much greater at night! Imagine if it was a full moon.
Looking at the acknowledgements I was discouraged to see a well-known Stanford statistician, who is an author on one of the best books I’ve found on Data Analysis, “Introduction to Statistical Learning”, had apparently made helpful comments on the paper. I can only hope his comment was: don’t publish this rubbish!
Clare Craig posted her take on the paper on Twitter which you can find here. She points out the obvious bias in the results. Disparate subgroups seem to show the same result. A line through most of the values likely takes into account the bias due to miles driven and represents the neutral position where we expect no difference between unvaccinated and vaccinated. I’ve drawn it in the table copied below.
Igor Chudov picks up that the subgroup, those over 65 (3rd row), are basically a control group. The amount of driving is probably similar for vaccinated and unvaccinated in this subgroup. Most over 65’s typically don’t have to drive to work. So this group is not subject to the offset. People greater than 65 years old therefore have a slightly higher risk of accident if vaccinated, according to the results. But it is very close to even risk. People >65 years that are still working probably had to get the vaccine. They probably drive more. That may have swung the risk slightly.
Several of the subgroups basically contain most of the people and are practically the same as total cohort. They include:
Cancer - No
COVID infection – No
Dementia – No.
They all sit at about the same value. This is probably the correction factor for the amount of miles driven and also correcting for the time vaccinated people were miscategorised as unvaccinated.
What can we tell after considering the offset?
Some subgroup results sit distinctly away from our new effective neutral line (orange dashed line above). I circled a few of these in the plot above.
We see people with COVID infection are more likely to be vaccinated. We have known this since late 2021, from the negative effectiveness of the vaccines. A paper from the Cleveland Clinic has just come out where they studied the effectiveness of the bivalent vaccine on staff in their system. However, one of the outcomes they found is that the incidence of getting infected with COVID-19 increased with the number of doses! This is something that we already know, but this paper makes it glaringly obvious.
It is progressively worse the more shots taken. Some people actually had 6 shots in the study.
I have only briefly scanned the paper at this stage and a few substack authors have already reported on it. A sad thing I saw in the paper is that they had to handle the fact that people were being terminated during the study, presumably because of staff not taking the injections. It’s really quite tragic what has happened. They are doing a study showing how useless the injections are to stop infection and at the same time terminating those who refused.
Unvaccinated people with cancer appear to have lower risk of accident. Those people with cancer who did not have vaccine may be people not advised to take it. They could be close to end of life and therefore not expected to travel. It’s certainly unlikely they are driving. Those well enough people with cancer and managing it tend to have a lot of travel to appointments etc.
People with dementia involved in accidents have higher risk of accident with no vaccine. I think what this tells us is that the group with dementia and not vaccinated are a special group. We have seen this in our NSW data. Any unvaccinated COVID deaths are of elderly people and many with dementia.
Almost all elderly people in Australia are vaccinated. Of those that aren’t there is a group that is super healthy. We know of an elderly couple who survived Mugabe’s regime in Zimbabwe who refuse vaccination (one can’t be surprised at their mistrust of government). Another group is those where an informed decision has been made not to not risk adverse effects of vaccination.
People with diabetes who are vaccinated seem to be at higher risk. This could be something worth looking into. Hypoglycaemic attacks definitely cause diabetic people to unfortunately have accidents. Are diabetics having more hypoglycaemic attacks with vaccination?
I did a quick google and it is a known thing, Vaccine-Induced Hyperglycemia (ViHG). A reference was found to a clinical trial underway titled : “Suspicion of Hypoglycemic Effect of mRNA Based Covid-19 Vaccines (Polyethylene Glycol) Will Happen, Observational Study on Diabetic Patients at Saudia Arabia”.
Putting people with Sleep Apnoea, yes or no, as subgroups is stupid because we also need to know if they are treated. Do they use CPAP machine? If treated then there should not be a problem. For example professional drivers are required to use CPAP by insurance companies. Certainly untreated sleep apnoea will lead to more accidents.
In one of the comments to Clare’s twitter post there is a useful (sarcastic) observation of where they saw the paper first promoted.
A useful thing about this type of study is to see who promotes this nonsense. We can calibrate the bias of these people. For those who follow COVID news closely it will be no surprise that one of the two champion Australian “misinformation correctors” was quick off the mark.
The replies to Dr Correcting Misinformation’s post are priceless. It’s worth scrolling through them. This is one had me laughing out loud:
This reply is insightful:
The comment is exactly on the mark. If this research group had access to the detailed personal information they appear to have, they could have solved the excess mortality question. We ultimately need to know vaccination status for all deaths.
Miss-information Corrector makes a single attempt, in characteristic sarcastic way, to correct someone:
But someone else points out:
Miss-information Corrector gives in to the comments onslaught.
Summary
We can still learn things from Stupid Studies. I hope the study is not retracted as it teaches us so much. It will be used in the future as an example of how experts can get befuddled.
I see a signal that hypoglycaemic attacks may be increased in vaccinated persons with diabetes.
And of course we learn that if you stay at home you won’t be involved in traffic accidents.
Oh dear - can't believe that the media gave this airtime. What is with Dr Redelmeier's eyes? Maybe he has been driving too much in the full moon?
Oh no, not the bonkers Dr Oliver again! She takes great delight in rubbishing anything even slightly against the Covid vaccination narrative, including accusing Malone, McCullough et al of being misinformation spreaders, whilst carefully omitting to mention she's not medically qualified herself, (she works in Polymer Research). Next time I need advice on Polymers I'll be sure to give her a call (well, maybe)