Canada just released a new report detailing the pandemic. Their report includes a section that measures the “success” of the vaccination campaign. Unfortunately, the data is heavily aggregated and hard to work compared to, say, Alberta, a province I have done some work on in the past. But I thought it would be an interesting exercise to look into the data and attempt to get a more accurate representation of the vaccine effectiveness. Furthermore, since the report is obviously meant to tell a one-sided story, it is only fair that we dig a little deeper.
The report includes this pretty little table, which completely lacks the needed context to evaluate individual risks. I will continue to harp on the point that these things need to be measured in life years lived. To do otherwise paints a slanted picture. While Canada may be 90% vaccinated (over the age of 12) right now, getting there did not take a day.
The Canadian vaccination campaign looks like this:
Obviously, it would be fantastic to break cases and deaths down by age to match this chart, but due to the level of aggregation this report used, and no underlying data provided, we need to aggregate to the same level.
Here are the approximate total life years lived (x1000) for each group since the vaccination campaign began until November 13th.
As you can see, Canadians actually lived more life year without the vaccine than with the vaccine up until November 13th (when our case data ends). Ceteris paribus, we would expect about 53% of cases to be in the unvaccinated; however, the report’s number is around 80%. But all things are not equal; there are extraneous factors here.
I would argue that the life years lived by those not vaccinated are confounded by the fact that their life years were harder than any other group. Most of their life years accumulated in the middle of the 2020-21 winter waves, and most of the vaccinated life years lived were accumulated during spring, summer, and fall of 2021. A vaccine enthusiast might argue, and not without room for truth, that the “hard years lived” factor is not completely exogenous to vaccination. We will see whether that is true first hand come winter 2021-22.
In a perfect world, we would evaluate the vaccine on a per day basis while controlling for location to account for seasonality, but again, we are constrained by the data government provides to us.
A less apparent confounding variable is the fact that we have little to no information on testing propensity. Who, indeed, is more likely to get tested for the virus? It is actually unclear, but in many places in Canada, the unvaccinated are being coerced into testing (for work), and the vaccinated are being disincentivized from testing (due to cost). This may be a problem. Ideally, we would have data on symptomatic cases to, somewhat, control for the social manipulation enacted by Canadian lawmakers; however, this data is hidden from us. But interestingly, the vaccinated are more likely to die per case regardless of vaccination status. Given one of the arguments for vaccination is that it is supposed to prevent severe symptoms and deaths, not make them worse then this is a strong signal that the cases are under-counted in the vaccinated cohort.
Furthermore, we must make another assumption — that the vaccination levels in the PHAC districts are similar to general vaccination levels in the population. As PHAC covers most of Canada, this is probably a pretty solid assumption.
Ok. So what do we have? A seemingly very effective vaccine.
According to the given numbers, the vaccine is more effective than being unvaccinated even in the first 14 days. But as someone that has closely watched Alberta data for a long time, these numbers do not make sense. For one, Alberta has consistently shown people are much more likely to test positive and die in the first 14 days. This is a big red flag that suggests the 2020-21 winter waves are heavily biasing the data. When I unbiased the Alberta numbers, I found that those 60+ were twice as likely to die from the vaccine in the first 14 days.
Let’s make two assumptions to remove some bias from the data.
My analysis of Alberta is robust and can be extrapolated to all of Canada.
Vaccinated individuals are equally likely to die of the virus when they get a case — not more likely.
These numbers are more in line with the data I have been tracking. It seems we have a reasonably effective vaccine that is well below the effectiveness needed to ever reach herd immunity even with low rate of reproduction (suggesting, once again, that the vaccination campaign is the lie of the century). The big X factor going forward is booster effectiveness. It will not make herd immunity more likely, but assuming the harder life years lived for the two dosers will not lower their effectiveness by that much (maybe a bad assumption), it should raise the overall vaccine effectiveness to somewhere between 47-67%
That 0.05% chance of dying for the unvaccinated is not daunting, but it is well overestimated for most age groups. Here is my personal cost benefit analysis. In my age group, the unvaccinated have lived (on the low end) 7,000,000 life years since April 1, 2020. According to the report, only 216 people have died in my age group. Even if we strongman the vaccinated case by making the supposition that all of those deaths were in the unvaccinated (they were not) and healthy with no comorbidites (greater than 95% of all deaths in Canada included comorbidites), the cumulative probability of me dying in a given year is 0.0031%. In other words, during a pandemic we were told would rival the Spanish flu, I still had a much greater chance of dying in a car crash. Eek.
If we assume the VE from the above chart, 47.54%, my chance of death from the vaccine would be about 0.0016% — or about 1.5 less deaths per 100,000 using our extremely pro-vaccine assumption.
I will probably do a full cost benefit analysis including the best information on adverse events at a later time. But this simple analysis begs a few questions, at least for me personally:
Can I be assured with 100% certainty that the chance of having a, as they claim, “rare” adverse event is less than 1 in 67000?
Can I be assured that there are no long term risks that we do not know about yet?
What if I place greater emphasis on unknown risks than known risks? The benefits, in that case, should greatly exceed the costs to discount for that fact.
What if some of the people that died over the last 2 years were actually vaccinated (they were) and I am overestimating my chance of death?
What if I have no comorbidities and I am overstating my risk of death by nearly 20-fold?
But most importantly, how is someone going to tell me I should not have autonomy over this decision?
What's your age group?