Follow-up to Vaccine Effectiveness around the World
My aim is to look at data from as many angles as possible to try to understand what is going on, with a bit of a focus on my own country, Australia. In my previous newsletter we looked at COVID-19 death rates around the world. There are some good comments to the Substack article. Professor Fenton tweeted the article and there are some “interesting” comments there. So, I thought it was worth following up with some thoughts based on the various comments, including what would chocolate and Nobel prizes have to do with anything? And why I would be characterised as being “pusillanimous”, a term which I not heard of before. Read on to find out.
BTW I’m grateful for the likes and comments and to those subscribing to get the emails automatically. Also, to friends and colleagues referring articles and making suggestions. It all helps us get to the bottom of what has happened. In a recent Australian Parliament Senate hearing the newsletter got a mention. Dr Wendy Hoy, Professor of Medicine, University of Queensland, testified at the hearing and quoted from the article on vaccine deaths vs COVID deaths. You can find the video of her talk here from the website of the Senator that organised the inquiry. If you have time, there are some other very interesting testimonies during the session worth listening to. Peter McCullough also testified. One talk from a mother, Raelene Gotze, whose daughter died following vaccination, tells of the gaslighting by the medical community and the Australian regulatory body, TGA, that followed. It’s hard not to cry with her while listening to it.
On to the comments and some less serious discussion…
As to the question of what factors could be affecting the death rates (apart from vaccination rates) suggestions included latitude, which is basically related to climate. A web search finds various papers, eg this one from Nature, where a detailed analysis of this relationship was performed. There are a lot of confounding factors that have to be controlled to confirm the relationship. Of course, factors like the economic status and the health system play a role. Latitude likely factors into the high rates of death in Eastern European countries that we observed, but we have to compare that with Scandinavian countries which are also very cold. Obviously economic status makes a difference between Eastern Europe and Scandinavia.
In Australia we are mid latitude, with a temperate climate, in the two main cities. But this will be confounded as we had a big wave in our Summer (just finished), after 2 years of lockdowns and no influenza that normally occurs in our Winter. Therefore the population had minimal natural immunity and a factor like that must have an effect.
Population density was proposed as a factor. As a country Australia compared to Israel is 3 people per sq km vs 400 per sq km. But Australia is mostly empty space and most people are in the main capital cities. Sydney has a population density of 400 people per sq km, so similar to Israel as a whole. Tel Aviv though has a population density of 7600 people per sq km. For comparison London and Tokyo both have a population density of approx. 6000. New York has 38,000! So clearly there are orders of magnitude difference in this parameter. It’s logical that the virus will spread more rapidly in densely populated areas.
We could possibly identify all the parameters that might have an influence and generate a regression model to predict death rates but it’s possible that there are so many factors and interaction effects that it is not practical. I may get round to having a look at this.
This leads on to the twitter comments which are less specific.
One comment said I was “pusillanimous”. I thought oh-oh I’m venturing into social media land and better have thick skin. This wasn’t a term I was familiar with. According to dictionary it means “contemptible timidity”. I didn’t like the sound of the contemptible part, but I take it to mean that I should have been harder hitting with drawing conclusions. “Lacking courage” is another definition. However, I’ll keep writing in a similar manner so that articles can be used as reference information, for example as it was in the Senate hearing I referred to above.
One twitter comment referred to a well-known demonstration that correlation does not imply causality, see the graph below, implying the graphs of COVID death rates vs vaccination rates show nothing. This is from a tongue in cheek article written in the New England Journal of Medicine relating chocolate consumption to number of Nobel Laureates awarded to a country.
The premise made was that dietary flavonoids could improve cognitive function and the test of this would be how many Nobel prizes were won by individual countries.
There is a very clear linear relationship in the graph above which is statistically significant. Interesting.
This paper spawned many others, further analysing this data, looking into the confounding factors and investigating the general subject of causality. Coffee and alcohol consumption were investigated. Coffee went the other way. Of course, this was an exercise to demonstrate a concept. The original paper used chocolate consumption over a 2-year period, compared to Nobel laureates accumulated over a century, a ridiculous proposition.
So I think the implication from the twitter comment was that the data, plotting vaccination rates and death rates, means nothing. However, let’s think about it further. Can we learn something about achieving Nobel prizes from the graph above?
We can look at the countries performing high on the graph, forgetting about what is good or bad, and then dig in further. Well Sweden is high. Hmmm. Nobel prize is awarded in Sweden. Possible conflict of interest? Vaccines were developed where? How does that country rank?
Other top-ranking countries for Nobel Laureates include Switzerland, UK and Germany: all with long histories of education and high economic status. No surprise that there’s high academic achievements by these countries. If we take away chocolate from the Swiss will they achieve lower? We suspect it won’t make a difference, because it is probably a surrogate for something else.
See down the bottom, China, the emerging economy, not in the game for as long. It looks like zero Nobel prizes. Low chocolate consumption. Would we say to China you need to eat more chocolate? And you will go higher up the scale. No, we would say there is something going on here. Leave them alone.
So back to the World Vaccine Effectiveness graph…
Should we say to African countries, take this intervention so you can achieve high results on the graph? Unfortunately, high on this scale is bad. No, you would do nothing and look into it further.
What is important is that performing these sorts of analyses are a step in the process of understanding relationships in data which are fundamentally very complex. We know not to draw immediate conclusions but rather to look for further insights. Can we unequivocally say vaccination is leading to high rates of death? No. There is relevance though according to the Bradford Hill criteria for causality (out of scope for this article). Irrespective, we know there is something very concerning in the data, that can’t be ignored.
We would say there are other factors going on. For example, we should investigate how do particular countries treat the disease? Were early treatments precluded in countries with high vaccination rate? Is there a low natural immunity in those countries caused by other factors? Is this something affecting primarily the elderly and should we be focused on protecting them (eg as per the Great Barrington Declaration)? Could it be affecting the poor in the rich countries?
Interestingly a few days after the newsletter on world analysis Dr John Campbell did an African update. The update points out the very low rate of COVID deaths in Africa. Commentators on their laptops in New York have claimed that Africans don’t keep good records and that deaths are much higher. This is insulting to Africans where every death is important and recorded. Doctors on the ground know where and how people are dying. Africa has bigger problems than COVID. For example, malaria and death during childbirth. There is a long list from the African doctors described in the video. The suggestion is that if the West wants to spend money don’t put it into vaccines. There are other fundamental things that are better support for African health. Dr John’s African correspondent, Wefwafwa, also gives an update reminding Mr Bill Gates that it is not “unfortunate” that Omicron has generated natural immunity.
We look at the graph of COVID deaths and vaccination rates for different countries in the context of other data from the countries. We are seeing emerging trends of excess (non-COVID) mortality in countries with high vaccination rates. Higher rates of infection in vaccinated vs unvaccinated. We need to look into these further.
An article that intentionally tries to misuse an observed correlation to draw a conclusion, came up recently on my news feed, from legacy news media, showing death rates from COVID being higher in US states that voted for Trump versus those that voted for Biden. Ok it was from the US ABC News (ie Disney)/CDC – say no more.
My version of the particular graph is below.
This article irritated me because clearly there is a partisan message being pushed, nothing remotely helpful to understanding health outcomes. On the graph in the actual article you can see Trump and Biden are displayed in big letters. So, the intended message is that people who voted for Trump are more likely to die. Like it serves them right. I note in Australia we don’t have the same level of partisanship as the US. We generally dislike the politicians from both the major parties, and while that might be thought to be a good thing, unfortunately a level of apathy has probably caused us big problems. Many people always vote the same way. We will have an election some time next month in May. The outcome will be interesting.
I went through the exercise of reproducing the US data myself, just to check. US death rates per state as of March 28 were found here. Note that I have used death rates from across the whole pandemic and the ABC/CDC article uses a period of 10 months covering vaccination. My aim is to demonstrate a point rather than reproduce the article’s graph exactly.
Those states that voted Republican (red) appear to have higher COVID death rate than those voting Democrat (blue). So, it does appear that if you voted for Trump you are more likely to die, though it’s not statistically significant result. Vaccination rate is shown along the horizontal axis and is lower in those states that voted for Trump. I note that the range of vaccination rates across all states in this graph only ranges from 50% to 80%, a relatively narrow range, and a standard trick to make some relationships appear more obvious. Would you expect a vaccination rate difference of 65% to 75%, where the crossover appears to be, in a population to make a difference? So, let’s explore what we think is really the underlying factor. I suspect that the distribution of elderly is similar across US states, although I haven’t checked, I know Florida is where a lot of people retire so probably there are more elderly there. New York has amongst the highest vaccination rates and it is up with the highest death rates. Population density, discussed above, could be in play.
My first impression is let’s look at some economic or poverty index. It’s clear that the poor have worse health outcomes. I took US poverty data from here. The Poverty Rate is the fraction of people in the state living in poverty. Mississippi is the highest at 19.6%. On the other hand, the lowest is New Hampshire's poverty rate at 7.3%. The following graph replaces vaccination rate with poverty rate.
The line is the other way round. High poverty, high death rate. So, as we suspected, there is a relationship between socio-economic status for a state and COVID death rates. I note that the correlation coefficient, though not high in either graph, is greater for poverty rate than vaccination rate. So it may be poverty rate leading people to voting for a particular party and then being associated with a particular leader. It’s quite possible that the population vaccination rate (in the range 50-80%) has little to do with death rates.
Presumably the most vulnerable, ie the elderly, are going to have a higher proportion of vaccinated than the overall population and that number may be unrelated to the overall population rate of vaccination for each state. When I look at the Mayo Clinic site vaccine tracker there is a vaccination breakdown by age groups and state. For 65+ the vaccination rates are similar across all states at 90+%. So, it appears the difference to the total population vaccination rates is driven by differences in vaccination of younger ages, eg 5-17. We know the COVID deaths are dominated in the 65+ age group and there are significantly less deaths in young age groups. So the graph could be implying that lower vaccination rates in children in Republican states is driving COVID deaths. Absurd. So maybe the CDC/ABC graph tells us nothing! Except that a narrative is in play.
The purpose of this exercise was to demonstrate the complexity of underlying relationships leading to the sad outcome of death from COVID.
It seems the legacy media and many bureaucrats have no appetite for vagueness and uncertainty. Saying that vaccination, or who you vote for, solves everything brushes over all the other variables that are clearly influencing the result. This does nothing to improve healthcare.
Let’s keep exploring…