Astonishing Safety of Measles in 1962
Post #1248
RETRACTION: As explained in this follow-up, inference from the BMJ data below has proved to be invalid. In short, the journal had stopped reporting measles deaths.
A prior post on this topic is here.
The most objective measure of the safety of an illness is the infection fatality rate, which can be interpreted as “the percentage of people who die after getting infected.” Bureaucrats often like to sum up the history of medical knowledge in short quips, such as:
“Measles is a thing which kills 1-in-a-thousand”
By thinking of measles as this “thing” that has this certain “fatality” — because the bureaucrat “said so” — we can then supposedly move forward by making rational decisions about how to go about protecting ourselves. But while it can be helpful to have short quips to guide us, what if they are completely and utterly false? E.g.,
“The Earth is a thing that is flat”
When I went back in history in order to uncover the actual case rates and actual death rates from measles in late 1962 in England and Wales, my head almost came off. I was finding measles cases, but for the life of me, I wasn’t finding any measles deaths (from those cases) in the medical literature. Here is an example of weekly measles cases:
This screenshot shows the top half of the weekly infectious disease report for all of England and Wales, showing the weekly cases found, but the bottom half shows you the deaths found, and measles deaths were zero for these 19,990 measles cases showing at bottom left of the data table. I then expanded the search:
The third bar from the right is the one in the screenshot, with 19,990 measles cases in England and Wales for the week ending on 5 Jan 1963. But all 13 of these consecutive weeks did not come with any measles deaths reported. However, measles might take 2 weeks to kill you, so I went forward by two extra weeks, never finding any deaths.
It goes without saying that it is not possible that:
“Measles is a thing which kills 1-in-a-thousand”
… when, in 1962/63, measles had been a thing which did not kill even one person in 146,000 infections. When applied to those weeks I found, the statement that:
“Measles is a thing which kills 1-in-a-thousand”
… overstates the lethality of measles by 146-fold. That is a lot of error, and I will continue the search, expanding the weeks, until I find at least one single measles death in all of England & Wales. In the meantime, we can use what is called the Rule-of-Three in order to get the “real” lethality of measles for that season:
The Rule-of-Three states that, if you looked and looked, but failed to find any events (e.g., failure to find even one single measles death), then just go ahead and imagine that there were 3 of them that were found — and you will then be estimating the upper bound on the event rate, by putting in a “3” as a placeholder for where “0” was.
This makes the upper bound of the “real” lethality of measles out as something that kills approximately one person for every 49,000 people infected. That’s much different from the short quip that bureaucrats use — i.e., one per 1,000 — but unlike the bureaucrat’s number, it is based on publicly-available, published mortality data.
In other words, it is an estimate that “has receipts.”
Statements divorced from evidence
Here is a really big estimate on measles fatality, one that fails to integrate the history of mortality data:
If I attempted to run a computer simulation of the 1962 measles mortality data, but using the estimate above (kills 3 per 1,000), then the computer software might “break” when trying to compute the probability — because this estimate is just too far from the reality on the ground in England and Wales in 1962.
Reference
BMJ -- Vital Statistics or Epidemiology or Medico-Legal
Wk ending 27 oct 1962
https://pmc.ncbi.nlm.nih.gov/articles/PMC1926588/?page=2
Wk ending 3 nov 1962
https://pmc.ncbi.nlm.nih.gov/articles/PMC1926703/
Wk ending 10 nov 1962
https://pmc.ncbi.nlm.nih.gov/articles/PMC1926742/
Wk ending 17 nov 1962
https://pmc.ncbi.nlm.nih.gov/articles/PMC1927057/?page=2
Wk ending 24 nov 1962
https://pmc.ncbi.nlm.nih.gov/articles/PMC1926823/
Wk ending 1 dec 1962
https://pmc.ncbi.nlm.nih.gov/articles/PMC1926931/?page=2
Wk ending 8 dec 1962 <--last issue called Vital Statistics
https://pmc.ncbi.nlm.nih.gov/articles/PMC1926966/?page=2
Wk ending 15 dec 1962 <--first issue called Epidemiology
https://pmc.ncbi.nlm.nih.gov/articles/PMC2123084/?page=2
Wk ending 22 dec 1962
https://pmc.ncbi.nlm.nih.gov/articles/PMC2122591/
Wk ending 29 dec 1962
https://pmc.ncbi.nlm.nih.gov/articles/PMC2123363/?page=2
Wk ending 5 jan 1963
https://pmc.ncbi.nlm.nih.gov/articles/PMC2123674/
Wk ending 12 jan 1963
https://pmc.ncbi.nlm.nih.gov/articles/PMC2123790/?page=2
Wk ending 19 jan 1963
https://pmc.ncbi.nlm.nih.gov/articles/PMC2123799/
Wk ending 26 jan 1963
https://pmc.ncbi.nlm.nih.gov/articles/PMC2123964/?page=3
Wk ending 2 feb 1963
https://pmc.ncbi.nlm.nih.gov/articles/PMC2123426/?page=2