And how does that compare to harm?
N.B. This article has been edited to provide more detail and explanation.
Risk is notoriously difficult to communicate effectively. It is especially hard when referring to an emotive subject like the risk of dying as the emotional response prevents rational interpretation of complex numbers. To simplify understanding of the benefits of interventions the number of people who need to be treated to prevent a covid death can be measured, the number needed to treat (or “NNT”). The same calculation can be applied to vaccination to give a number needed to vaccinate to prevent a covid death.
The data available to carry out such a calculation is flawed, but even if we assume it is perfect the answer we get is alarming.
For the Delta wave the UKHSA published deaths by covid vaccination status. From this the risk of dying if vaccinated or unvaccinated could be calculated as a proportion of the whole vaccinated or unvaccinated population. The vaccinated population can be accurately measured via records of those vaccinated, but the size of the unvaccinated population is by definition only an estimate, and the problems with the figures used have been discussed before. However, we are going to use the UKHSA estimates for the unvaccinated population. If we assume that the covid death rate in the unvaccinated had applied to the vaccinated then the number of covid deaths prevented by vaccination will be evident. Using the number of people vaccinated we can then calculate the number that needed to be vaccinated to prevent each covid death.
Doing this calculation for each age group gives the following results for the duration of the Delta and first Omicron waves. The Delta data is for a period of 16 weeks where Omicron is only for 12 weeks because data was not published thereafter. This in itself is rather irregular. It very much has the appearance that the data is not being published because it does not show what the prevailing narrative insists upon. This lack of transparency is worrying and does not increase public confidence in government statistics.
|Age||Covid deaths prevented based on differences in covid death rates per 100k DELTA (27th Aug – 16th Dec 2021)||Number needed to vaccinate per covid death prevented based on differences in covid death rates per 100k DELTA||Covid deaths prevented based on differences in covid death rates per 100k OMICRON (3rd Jan – 27th Mar 2022)||Number needed to vaccinate per covid death prevented based on differences in covid death rates per 100k OMICRON|
Problems estimating the size of the population mean that the data based on death rates could be flawed. To carry out a sense check further calculations were done.
Firstly, the number needed to vaccinate to prevent a death could be calculated by comparing the case fatality rate for the unvaccinated to the rate for the vaccinated, using cases two weeks before deaths. Because of potential differences in testing rates the number needed to vaccinated to prevent a death based on hospital fatality rates is also included. (See appendix for methods).
|Number who need to be vaccinated to prevent a covid-labelled death based on Case Fatality Rate Delta vs unvaccinated and unlinked||Number who need to be vaccinated to prevent a covid-labelled death based on Hospital fatality rate Delta vs unvaccinated and unlinked||Number who need to be vaccinated to prevent a covid-labelled death based on Case Fatality Rate Omicron vs unvaccinated and unlinked||Number who need to be vaccinated to prevent a covid-labelled death based on Hospital fatality rate Omicron vs unvaccinated and unlinked|
Using the population who had only had one dose as a control group to compare to those who had three or more doses also gives very high numbers needed to be vaccinated to prevent one death.
|Number who need to be vaccinated to prevent a covid-labelled death based on Case Fatality Rate Delta for two or more doses vs one dose||Number who need to be vaccinated to prevent a covid-labelled death based on Hospital fatality rate Delta for two or more doses vs one dose||Number who need to be vaccinated to prevent a covid-labelled death based on Case Fatality Rate Omicron for two or more doses vs one dose||Number who need to be vaccinated to prevent a covid-labelled death based on Hospital fatality rate Omicron for two or more doses vs one dose|
When applied to drugs, published data suggest that only half of doctors would prescribe a drug with a number needed to treat of 200. That would imply that half of doctors would not recommend it even for over 80 year olds during the Delta wave. For Omicron the number is in a different league.
If the intervention lasted longer than the Delta wave then the number needed to vaccinate would fall as benefits continued. However, the length of time that the vaccine could be said to have imparted a benefit can surely be no longer than the duration of a wave of covid. It is therefore hard to make that claim.
Any underestimate of the size of the unvaccinated population (as seems likely) would make the number needed to vaccinate even higher than these figures. Furthermore, the data does not distinguish between people with co-morbidities and those who are healthy. Overall 95% of covid deaths have been in people with pre-existing conditions so the number needed to vaccinate in the healthy population will be far higher still than the numbers given here.
The above tables and analysis only considers numbers needed to treat to prevent deaths from (or with) covid. This needs to be balanced by the harms caused by the intervention. Obviously, an intervention which increased morbidity and / or mortality overall (ie from all causes) would not be considered worthwhile, even if it did reduce severe covid outcomes.
Reliable all-cause data on mortality rates between vaccinated and unvaccinated groups is difficult to find, likely to be significantly confounded, and also subject to fundamental categorisation errors. More reliable (ie controlled) data is however available from the Pfizer trial.
Hence, as well as understanding the number needed to vaccinate in order to prevent a covid death, it is important to also consider the number needed to vaccinate in order to cause a non-fatal harmful event. Data to estimate this is still difficult to come by. However, Phil Harper has carried out this calculation using the Pfizer trial data. For every covid hospitalisation prevented from vaccination there were an additional 1.3 hospitalisations for serious adverse events related to the vaccine. That data was available to regulatory bodies in December 2020. More recently, analysis of trial data as a whole has been published showing the increased risk of serious adverse events was higher than the reduction in risk for covid hospitalisations for both Moderna and Pfizer.
Because Omicron is milder the ratio between benefit of vaccination and harm from vaccination will have swung even further towards harm. It is becoming harder and harder for medical professionals to claim they are ignorant of these facts. The data is in and the harms outweigh the benefits.
The numbers in the tables were calculated from UKHSA vaccine surveillance report data in the following way:
- The total covid deaths for vaccinated and unvaccinated in each age group over the whole wave were summed. The total covid deaths in the vaccinated were summed regardless of doses given.
- For Delta reports for 2021 weeks 35-38, 39-42, 43-46 and 47-50 were used. For Omicron reports for 2022 weeks 1-4, 5-8 and 9-12 were used.
- UKHSA estimates for the number who had been vaccinated were taken as the mean of the numbers at the start and end of the period.
- Case fatality rate was calculated based on the total deaths in the period divided by the total cases reported two weeks earlier. The unvaccinated and unlinked deaths and cases were combined to account for people who are not in the system when diagnosed but are registered before they die.
- Hospital fatality rate was calculated based on the total deaths in the period divided by the total hospital admissions reported one week earlier. The hospitalisation rate is based on the UKHSA definition of cases presenting to emergency care (within 28 days of a positive specimen) resulting in an overnight inpatient admission one week earlier.