From a New Zealand article: Once again the headlines have screamed that vaccination provides a risk reduction of hospitalisation of between 200% and 1200%. Is this really the case?
Firstly, no adequate randomised placebo controlled trials have been done with regard to vaccination and the Omicron variant.
Secondly, the most alarming figures have been done using modelling. There is just no way to prove or disprove this. It is akin to reading tea leaves.
Thirdly, no mention is made of treatment options. How many hospitalisations could be prevented with adequate treatment? Evidence from overseas indicates that large numbers of hospitalisations can be prevented with early treatment.
Fourthly, we have gnarly issue of relative and absolute risk reduction that has been repeatedly used to make the risk appear worse than it is. Here is Dr. Guy Hatchard’s analysis.
Are Unvaccinated People at Serious Risk From Omicron in NZ? Many people have been asking me about numerous recent articles in mainstream media reporting high risks for unvaccinated people and enhanced protection for the boosted. Accurate statistical analysis takes time to compile, fortunately I have been helped by people working hard behind the scenes. According to an assessment of Ministry of Health data from 9th March 2022, unvaccinated individuals are 200% more likely to be hospitalised than the boosted. At first this sounds impressive.
Epidemiologist Dr. Michael Baker thinks so. He says (11th March, 2022) that, without vaccination, New Zealand would have much higher hospitalisations and deaths: “It would be much higher if it wasn’t for that. It would be not manageable, absolutely unmanageable if it weren’t for vaccination.”
It is actually a relief to see that Michael Baker is referring to Ministry of Health data. A few days earlier, modeler Dr. Shaun Hendy’s team, at government funded Te Punaha Matatini, sought to frighten the whole population, just as they did when they famously predicted 80,000 NZ covid deaths. They suggest that the unvaccinated are 1200% more likely to be hospitalised than people with two mRNA doses and 2700% more likely to go to hospital with Omicron than the boosted (Stuff, 17th February, 2022).
The scientific flaws in their argument are so large and obvious, it was certainly irresponsible of Stuff to report this. The critical components of successful modelling are accurate assumptions and real world data comparisons. Neither were satisfied in this case. Te Punaha Matatini’s predictions are completely contradicted by the data referred to by Michael Baker.
There are also flaws in the arguments presented by Michael Baker et al., although they are not so immediately obvious. Baker reports relative risks, without commenting on absolute risks. This is a statistical sleight of hand used to make a risk factor appear more significant than it actually is. You see it all the time in news stories about food: “Why eating just one sausage a day raises your cancer risk by 20 per cent”
Twenty percent is the relative risk, and it sounds scarily high. The reported finding actually translates to an absolute risk increase from 6 people in 10,000 up to 7 people in 10,000. You might drop your sausage for 20%, but if you knew the increased absolute risk was only 1 part in 10,000, you might be tempted to hang on to it, and even add a slice of bacon.
What is the absolute risk difference for Covid-associated hospitalisation? The graph below (data from MoH – spreadsheet here) shows that there are 4 hospitalisations per 100,000 unvaccinated people, compared to 2 per 100,000 boosted people, and 3 for any vaccination on March 9th 2022.