Testing claim of no link of MMR to autism
Post #857
This post relates to a prior post regarding an autism safety signal for MMR vaccine in February 1995.
If MMR vaccine had no link to autism, then you would expect VAERS reports for autism to be in rough proportion to the relative share of all vaccines made up by MMR. To test the assumption that MMR vaccine is not linked to autism, a chi-squared Goodness of Fit test can be used, along with expected proportions:
[click to enlarge]
At left is a test for reports with any onset date, but more valid would be a test for reports with an onset date of autism within 120 days of receiving vaccine (right side). For the years of 1993, 1994, and 1995, the expected count of reports of autism within 120 days of a vaccine were all 5 or above (required for valid inference).
The p-value for 1993 is in cell Q6, and it is so small that there would be a decimal followed by 308 zeroes before ending in “231” (p-value = 0.000 … … … … … 000231). If it really were true that MMR vaccines were not linked to autism, then you would not see such lopsided reporting.
It was a year when 7.9% of all net vaccine doses delivered were comprised of MMR doses. For 1993 with MMR, 10 reports were expected, but 124 of them observed. For all other vaccines, 117 were expected with a grand total of just 3 reports observed. That kind of lopsidedness — as shown by the p-value — doesn’t happen by chance.
The p-value is so low that if you took every second on Earth, for 4.5 billion years, and in each individual second, you ran an equivalent experiment, then there would still not even be a 1% chance to see the 1993 autism results even one time — when there had been a new trial every single second, for all of Earth’s age.
Total # of seconds (or # of trials) = 142,009,200,000,000,000 seconds (trials)
1993 proportion of all vaccine doses comprised by MMR
The proportion of the total doses for 1993 comprised by MMR is:
The total of reports of autism within 120 days of vaccines taken in the 1993 year was 127 of them, so the expected proportion of all autism reports ascribed to MMR vaccines for that year would be (127 * 0.079 =) 10 of them — but 124 out of the 127 reports were for MMR vaccine, a share of all reports twelve times what was expected.
Another way to say this is that MMR vaccines made up 8% of all doses given, but made up 98% of all reports of autism following vaccines. Notably, this is before Andrew Wakefield popularized the possible link between MMR and autism — so it is authentic, rather than being the result of an “increased tendency to report.”